(2022) Études des ressources en eau: Région métropolitaine de Port-au-Prince, République d'Haïti

[page 1] Development Bank 4 Water Resource Investigations Port-au-Prince Metropolitan Region Republic of Haïti Water and Sanitation Division Authors: TECHNICAL NOTE N° James K. Adamson IDB-TN-2446 Javan Miner Pierre-Yves Rochat Editors: Sergio Perez Monforte Maria Rodriguez March 2022 [page 2] LA Inter-American / . Development Bank / 4 Water Resource Investigations Port-au-Prince Metropolitan Region Republic of Haïti Authors: James K. Adamson Javan Miner Pierre-Yves Rochat Editors: Sergio Perez Monforte Maria Rodriguez Inter-American Development Bank Water and Sanitation Division March 2022 [page 3] Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Adamson, James K. Water resource investigations : Port-au-Prince metropolitan region, Republic of Haïti / James K. Adamson, Javan Miner, Pierre-Yves Rochat; Sergio Pérez Monforte, Maria Rodriguez. p. cm. — (IDB Technical Note; 2446) Includes bibliographic references. 1. Water resources development-Haiti. 2. Watershed management-Haiti. 3. Water supply- Environmental aspects-Haiti. 1. Miner, Javan. Il. Rochat, Pierre-Yves. III. Pérez Monforte, Sergio, editor. IV. Rodriguez, Maria, editor. V. Inter-American Development Bank. Water and Sanitation Division. VI. Title. VII. Series. IDB-TN-2446 Keywords: Water resource, hydrology, source monitoring, source protection, groundwater flow model JEL Codes: L95, Q25 htto://www.iadb.orq Copyright © [2022] Inter-American Development Bank. This work is licensed under a Creative Commons 1GO 3.0 Attribution- NonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 1GO) license (http://creativecommons.org/licenses/by-nc- nd/3.0/igo/legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB’s name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-1GO license. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. © BY NC ND [page 4] _ __ __ WATER RESOURCE INVESTIGATIONS Port-au-Prince Metropolitan Region Republic of Haïti AUTHORS: James K. Adamson, PG Javan Miner, EIT Pierre-Yves Rochat EDITORS: Sergio Perez Monforte Maria Rodriguez March 2022 [page 5] | : : : Port-au-Prince Metropolitan Region R lic of Haïti epublic of Haïti With support from: Northwater International & Rezodlo, S.A, DINEPA, OREPA-Ouest and CTE-RMPP WIDB C:::… S ras \H) _ 7 Repubie of Austri Swiss Agency for Development and Cooperation SDC [page 6] 2 TEE IE 0W La) _ 72°20'W 72°10'W 72°0W 71°50'W Ecuba 0 50100 20 dm ge FE £ pre TE ? R " ÉR Er 7 F EN Haiti À RP RS eur HE FVUTÉS Dominican Republic DCS, : /2 : 63 At 2 ° # LE TS a 740W 720W 7OUW uste < \ Ë CbE porcs prince] 2/0 & : EME Vr NA (plainelae] one NN Re. el Ë # = SCÉ F — à TEA ro gaie Mmratstshiit die lle Selle 5 OT 72°40'W —_ 7230 W u 72°20'W 72°10W — 72°0W 71°50'W €A Study Area Hydrogeological Environment PVC bord St SEE @M Unconsolidated: recent alluvial deposits :#, Primary alluvial aquifers er. Canal | Reef Carbonate: uplifted reef deposits 2 EPGFZ Spring, Massif de la e Semi-Consolidated: marl, detrital limestone, — Faults * Selle Ssandstone, siltstone > Lake e Interior Sedimentary: limestone (karstified in National Road areas), hard chalk < h Secondary Road Igneous: volcano-sedimentary rocks, basalt nor water Water Resource Investigations - Study Area. Table of Contents PREFACE AND SUMMARIES Hydrogeological Investigation of Tunnel Diquini: Characterization of Hydrology and Guidance for Source Monitoring and Protection. Hydrogeological Investigation of Source Mariani: Characterization of Hydrology and Guidance for Source Monitoring and Protection. Plaine du Cul-de-Sac Groundwater Flow Model. [page 7] Preface and Summaries - The Plaine du Cul-de-Sac aquifer is an unconsolidated alluvial aquifer. There are This work is financed with support from 26 municipal wells that take 737,000 m*/ the AquaFund. The AquaFund is the IDB's day based on 2015-2018 data. Municipal thematic fund for water and sanitation, and PUmMping can potentially be increased by has been the main financing mechanism to 745,000 m°/day through new proposed wells support the Banks investments in the sector and rehabilitation of inactive wells. since its creation in 2008. The AquaFund has contributed to the achievement of the Based on this information, there is a significant Millennium Development Goals for water and gap between water supply and demand. This sanitation in Latin America and the Caribbean has prompted the urgent need to protect and will play a crucial role in supporting existing resources and secure additional the region's governments in achieving the Water supplies. new Sustainable Development Goails. It has done so by facilitating investments to In 2018 - 2019, Northwater International and increase the provision of water and sanitation, … Rezodlo SA were contracted to perform three water resources management, solid waste water resource investigations in the Port- management, and wastewater treatment, au-Prince region. The Government of Haïti while contributing to the sustainability and Supported the research both in the field and accessibility of these services for low- With historical data compilation. income populations. It also supports the Bank's client countries in adaressing the Theinvestigations included: emerging challenges of climate change, rapid degradation of freshwater ecosystems, and 1. Characterization of Tunnel Diquini increasing water insecurity. The AquaFund is 2. Characterization of Source Mariani financed through the IDB's own resources and 3, Groundwater Flow Model for the Plaine du resources from donor partners, namelÿ the Cul-de-Sac aquifer Government of Austria, the Spanish Agency for International Development Cooperation Tunnel Diquini is a “15 km long tunnel AECID, the PepsiCo Foundation, and the Swiss constructed in 1940 that collects groundwater Agency for Development and Cooperation from fractures and a fault in the Massif de SDC and the State Secretariat for Economic |a Selle carbonate aquifer system. lt is the Affairs SECO. single largest water source for the Port-au- Prince municipal water system, accounting Background and context. The metropolitan for approximately 26% of the supply. area of Port-au-Prince has an estimated population of 2.8 million and projected to Source Mariani is the second largest naturally increase to 3.5 million by 2030 (CIA 2020). flowing source that serves the Port-au-Prince With an estimated water demand of 365,000 municipal watersystem. Thespringis aprimary m?/day, protecting existing supplies and discharge of the Massif de la Selle carbonate securing additional water supply is an urgent aquifer system, the same aquifer complex as priority for the region. Since the 19805, there Tunnel Diquini. Due to its distal location from has been insufficient investment in advancing Port-au-Prince and low elevation, it requires knowledge of the primary resources that a pumping system. When in operation, the Port-au-Prince greatly relies on. spring accounts for approximately 17% of the Port-au-Prince water supply. The Port-au-Prince water supplies include: + Massif de la Selle aquifer system, a karst The Plaine du Cul-de-Sac aquifer is one of carbonate bedrock aquifer system that feeds Haitis most productive and largest aquifers fifteen springs and one well that supply the (-360 km? aerial extent). This multi-layer network. The average supply from these Unconsolidated alluvial aquifer has a thickness sources is approximately 120,000 m°/day. of more than 200 m, and supports municipal, [page 8] l private and agricultural wells. The Port- knowledge is important to guide future studies au-Prince municipal system includes © 26 and monitoring of the tunnel and to aid the wells that serve the lower-lying areas of the Centre Technique d'Exploitation de la Région metropolitan region. Historical peak aquifer- Métropolitaine de Port-Au-Prince (CTE-RMPP) wide abstraction was estimated to be nearlÿ in water use planning and management. 300,000 m°/day in the 1970s and 1980s as a The investigation was accomplished by using result of the sugar cane industry, thus suffering a combination of literature review, satellite declines in the water tables and increases in and topographic imagery analysis, and salinity. Current abstraction is significantly field reconnaissance. A brief field mission lower than historical peaks due to unreliable to the tunnel was conducted on April 15, power and lack of commercial agriculture. 2018, including: (i) physical and chemical À groundwater flow modeling exercise was sampling, (ii) stable isotope sampling, performed to establish a preliminary estimate (iii) chlorofluorocarbon (CFC) and sulfur ofrenewable quantities of groundwater,andto hexafluoride (SF6) sampling, (iv) flow rate improve the understanding of recharge origins measurement and (v) visual inspection of and the interaction between groundwater tunnel geology. and surface water. These parameters are important to guide water supply planning Note: Additional data collection and research and development for Port-au-Prince and, at since the Tunnel Diquini study was conducted the same time, to manage and protect these Warrants revisiting and updating some of resources. the interrogations and findings of the report. Le. . . . This summary presents data primarily based The objective of this research is to IMPTOVE on the original report, with the exception of the understanding of Port-au-Prince's : Fe : an increased range of recharge rate and a critical water resources and to guide informed planning and investments to secure decreased catchment area based on updated sustainable water supplies to satisfy the leSearch. growing demand from the region. These RER studies guide further research and reveal D) +1 important monitoring needs to strengthen 1% $e resource characterization and data-driven ï if decision-making and resource management. ‘ \4 These investigations are preliminary in nature 4» : à and limited due to the scarcity of data. 1 PAU È 1. Tunnel Diquini Summary Ce Photo 1. Limestone exposed in the main tunnel. Tunnel Diquini is the largest single source of water for the municipal water system ; er of Port-au-Prince. Based on records from >, \ Là 2014-2018, the tunnel supplies an average of cd 29,449 m*/day to the metropolitan Port-au- à Prince water system. The tunnel accounts N TR for 26% of total municipal production, and À 37% of all gravity-fed spring flow that the metropolitan area obtains. An inspection was G performed to characterize the hydrology of the tunnel waters and to better understand the origin of its flow and its relationship with groundwater and surface water systems. This Photo 2. The tunnel portal facing into the tunnel. [page 9] Flow Characteristics 1. Recharge to the Massif de la Selle carbonate aquifer system and tunnel appears to be largely affected by high intensity and high-volume rainfall events such as hurricanes and tropical storms. lt appears to be a 3-to-7-year cycle of recharge trends partially influenced by El Niño and La Niña events. 2. Tunnel discharge varies seasonally with recorded flows ranging from 11,085 to 73,265 m/d, with a geometric mean for all known recorded flows of 27,987 m*/d (1980-2018 dataset). 3. Decreases in tunnel flow result from extended periods of normal precipitation and consecutive years without high intensity rainfall periods such as tropical storms and hurricanes. a. The recharge dynamics and flow regression can influence flow trends over periods of several years. Tunnel flows do not appear to have been decreasing over the long-term. b. Limited historical data from 1959 suggests that dry season low-flow conditions were comparable or perhaps lower than current low-flow conditions and strongly influenced by major recharge and drought events. c. The response time of the aquifer to major recharge events such as hurricanes may be shortening, possibly due to land cover and climatic changes. 800 Direct recharge in 700 karst terrain Ê 800 Normal fault LA | g 5 500 s: Gentil na. Ë 5 È Baptiste, Diquini © 300 È Direct recharge in Mahotiere and va, = of mr] ë Source Karst terrain : Gorossol mm 1 from A Fo 200 & Mariani ë 0 1 2 3 4 5 6 7 8 9 10 11 North Horizontal Distance (km) South Figure 1. Conceptual cross section of Diquini Tunnel groundwater flow. [page 10] Groundwater / Surface Water Interaction 1. Based on water chemistry and tracers, the tunnel and Riviere Froide are connected to the same regional karst limestone aquifer and gain flow from it. The Riviere Froide may recharge the aquifer at various spatial and temporal extents, and this could result in a possible link between the tunnel and river system. 2. Based on geology and structure, Tunnel Diquini does not appear to have a hydraulic connection to the Riviere Momance, whichis located farther to the south in the mountains and flows westerly to the Plaine de Leogane. a. The EPG fault zone and a perpendicular fault appear to direct groundwater in the Momance basin either into the Riviere Momance or into the lower reaches of the Riviere Froide, below where recharge to the tunnel would likely occur. Spatial Distribution of Groundwater Recharge 1. The long-term mean annual recharge rate in the karst terrain is estimated greater than 40% of annual precipitation. Recharge rates can be higher in years with tropical storms and hurricanes, and less than 15% in years with normal or low precipitation. 2. Recharge rates are higher during high intensity precipitation events, and lower during periods of average and low precipitation. 3. Aquifer storage of the ‘spring shed' of the tunnel is estimated between 265 and 327 Mmé. 4. The aquifer is well mixed and has an average groundwater age of 26 to 32 years based on a single sampling event. 5. Recharge area of the tunnel flow appears to in the range of 12.4 km. a. Recharge area is dependent on the rate and duration of Riviere Froide leakage to the regional carbonate aquifer. b. The average recharge elevation is estimated at 650m, indicating that a portion of tunnel flow may originate from river leakage from the Riviere Froide to the regional carbonate aquifer. [page 11] Aquifer Vulnerability 1. Due to the high permedability and high infiltration rates typical in karst limestone environments, the tunnel waters are vulnerable to contamination. For example, fecal coliform, E. coli, and salmonella contamination was reported by Eptisa in March 2014. 2. Urbanization and land use changes in the hills south of the tunnel portal may have negative impacts to tunnel water quality and flow. The lack of centralized waste management and sanitation in karst environments increases the risk of aquifer contamination. Increase of impervious surfaces and loss of soil decreases recharge to the aquifer that contributes to tunnel flows. 500 ne D El MIE CNE INT 5 RE 4 Ë 22 i 255 800 : À - É E 13 È È 5 ‘ ME En D ee Z 6 Él È À Ë & À J À È É 500 G = É j ” 2500 à a 400 Ë À nn 1e h F 2000 = 300 « | | É 200 f 1500 < 100 (] Ste ENS DUB EE DS M 90 AIS Me eme Daelim © BE Sn : mu Station (UHM 2019) ——Tunnel Diquini ——Source Mariani Figure 2. Tunnel Diquini and Source Mariani discharge with ENSO climate events, 1980 - 2018. IL Source Mariani Summary de la Selle aquifer system and to aid CTE- RMPP in water use planning, development, Source Mariani is currently the most distal Monitoring and protection, source of water that supplies the CTE-RMPP ., . . water system. Itisthe largestnaturallyflowing This investigation was accomplished using spring and second largest single water source 4 combination of literature and data review, that supplies the Port-au-Prince municipal Satellite and topographic imagery analysis, water system. When the pumping station is and field reconnaissance. A brief field in operation, an average of "19,000 m°/day mission to the spring was conducted in April is available to supply “17% of total municipal 2019, including: (i) physical and chemical production, and 24% of all spring flow Sampling, (ii) stable isotope sampling, supplying metropolitan Port-au-Prince region (iii) chlorofluorocarbon (CFC) and sulfur (based on CTE-RMPP data 2014-2018). The hexafluoride (SF6) sampling, and (iv) visual spring discharges from limestones that drain observation of local geology. At a follow-up a portion of the Massif de La Selle carbonate visit to the spring was conducted in January aquifer system, west of the Riviere Froide and 2020 to verify more recent flow monitoring north of the Riviere Momance. The objective data received from CTE-RMPP, two nearby of this evaluation is to better understand the SPrings were ,supsequently document that spring flow characteristics and the origin of lepresent "25% of the overall flow from the the waters to guide future study of the Massif Mariani spring system. [page 12] È Fe — PNA. 20e 1 The long-term mean annual recharge rate in the karst terrain is 1. The groundwater recharge area that estimated greater than 30% of annual contributes to the spring flow appears precipitation. Recharge rates can be to be approximately 13.3 km? but may higher in years with tropical storms and be larger. hurricanes, and less than 15% in years With normal or low precipitation. a. Thisuncertaintyintherecharge area is largely due to the complexity of 1. Aquifer storage of the ‘spring shed'is groundwater flow in karst environments estimated between 155 and 259 Mm. and limited datasets regarding tracers and hydrochemistry. 2. The limestone karst aquifer that feeds the spring is well mixed and has an 2. The average recharge elevation is average groundwater age of between estimated at 580 m above mean sea 21 and 35 years based on a single level with a corresponding temperature sampling event. of 22.7 C. This suggests the possibility that some spring flow may originate — — name from distal zones in the regional sé Ee S < se t ke oi J carbonate aquifer such as within the CT pe * à | Fe Riviere Momance basin. 62e à Am £ Ce El : AMP ANT UNE ep > de, APR is © \@ 7,5 en Ar [page 13] : — Stream Topographie Sinks Road À : :PointThor HT SA oi ZZSping Catchment pen < 5 m — primary À he. 1: ZWatershed Boundary pm 5 - 10 m = secondary Catchment N | Em 10m De 7 1:50,000 + FH RE K , ° 1 2 4 es SK Frs ANT Sous Ambas ) [MARIANI] {: E Geology ? | É FH Stike/Dip } b am AÆ—— Normal Fault EX: £ j | É— Strike/Slip Fault } É D — — Unknown, Inferred V4 v fan FH Bedding fe [Cxauoea] Geology Qa Qam / Qac f, ein REré © [wtérshed) Mi 7 por ée Par Em D £ ES nl © Figure 3. Geologic Map of Interpreted Catchment Area of Source Mariani Flow Characteristics 1. Recharge to Massif de la Selle aquifer system appears to be largely affected by high intensity and high-volume rainfall events such as hurricanes and tropical storms. There appears to be a 3-to-7-year cycle of recharge trends partially influenced by El Niño and La Niña events. 2. Spring discharge displays mild seasonal variability, with monthly average flows typically ranging between 14,500 and 25,000 m°/d with an average of 19,500 mÿ/dl. a. Instantaneous (daily) flows display greater variability, ranging from 7,600 to 30,700 mÿ/d. b. Based on the spring catchment infrastructure as observed in 2019, spring flow is measured from a single water meter. Since this method does not account for overflow, some high spring flows could be underreported. 3. The spring flow is most vulnerable to extended periods of average or below average precipitation and to consecutive years without high intensity rainfall periods such as tropical storms and hurricanes. [page 14] a. This recharge characteristic and resulting flow regression that extends from 2014 to 2019, may foster perceptions that the spring flow has been decreasing over the long-term or that acute impacts have occurred. b. Historical flow data from 1925 and 1933 has similar flow rates as the present. 4, The cyclic and multi-annual recharge characteristics are important for water managers to understand in order to balance water use allocations from the different water sources of CTE-RMPP. 900 2,000 800 1,800 £ 300 = 700 1.600 & 2504 = + = + e © ] ne Z 200 ms © 500 7 S = + + E 400 1000 À & 150 £ 800 © £ # 300 un | 60 À 5 100 E] Q . = 2 200 À. dl 40 3 & 50 +1925-1933 m2008 - 2020 100 : 200 À 0 0 0 ÉÉÉRÉÉÉÉEÉE 123456789101 RRRSRARTRRRRRRRRSERS Month —— Source Mariani —— Tunnel Diquini —#—Petion-Ville Precipitation Figure 4. (i) Source Mariani and Tunnel Diquini flow compared with annual precipitation, (ii) average monthly flow of Source Mariani by month. Groundwater / Surface Water Interaction 1. Source Mariani flows from the regional Massif de la Selle carbonate aquifer system. a. Source Mariani is the lowest elevation terrestrial outlet known for the aquifer and appears to emanate from a topographic exposure of the main aquifer lithology rather than as a contact spring. This may act to sustain flows even when higher elevation springs exhibit reduced flows. b. Source Mariani essentially serves as a drain for the western portion of the Massif de la Selle aquifer. 2. Source Mariani does not appear to have a significant hydraulic connection to the Riviere Momance or Riviere Froide. This is supported by the isotope and tracer sampling and analysis of recharge catchment size. [page 15] Aquifer Vulnerability 1. Due to the high permeability and rapid infiltration rates typical in karst limestone environments, the spring vulnerable to contamination. For example, fecal coliform, E. coli, and salmonella contamination was reported by Eptisa in March 2014. 2. Urbanization and land use changes in the hills south of the spring may result in negative impacts to water quality and flow. The lack of centralized waste management and sanitation in karst environments increases the risk of aquifer contamination. Increase of impervious surfaces and loss of soil decreases recharge to the aquifer that contributes to spring flow. Ill. Plaine du Cul-de-Sac Summary Recharge (in) The Plaine du Cul-de-Sac (PCS) aquifer is one of the largest aquifers in Haïti and currently 1% provides at least 25% of the water supply for 5% 11% Port-au-Prince from 26 municipal wells. The aquifer is also an important water supply 12% for private, agricultural and industrial wells. À numerical groundwater flow model was developed to better understand the hydraulic 71% parameters and behavior of the PCS aquifer and support water supply development planning for the Port-au-Prince metropolitan © Direct recharge (Rech) region. The model effort was preceded with © Riviere Blanche infiltration (Rsw) data mining and research to support the © Riviere Grise infiltration (Rsw) model construction and calibration. © _inflow from bedrock aquifer (Rgh) @ Riviere Batard infiltration (Rsw) The primary goal of the modeling exercise was to better understand i) the sustainable Discharge (out) and renewable quantities of groundwater available from the aquifer, ii) the complex CR recharge dynamics, and iii) surface water/ 10%1 74 groundwater interactions between lakes 1 and river systems. MODFLOW1 2005 code was selected for modeling the PCS aquifer. 33% Viewlog software from Earthfx Inc. was applied to build the model, this software directly integrates with the borehole database 53% for building, developing and refining the model. Groundwater Vistas Advanced, version. 6 was used to run the model simulations and © Lao AnseŸieg Bumentre Doi 1 ( ul scenarios. © Trou Cainen Dos) © Ocean (Dsea) © Pumping (ABS) @ Canal Boucambrou (Dsw) [page 16] galbrated Baseline Groundwater Flow roundwater Flow Model Details The groundwater flow shows similar trends as has been illustrated in 1. Aquifer extent of 363 km? previous reports. The hydraulic gradient is steepest in the southern limits of the 2. Maximum thickness greater than 200 aquifer where the Riviere Grise and meters. Riviere Blanche enter the plain and recharge the aquifer. A groundwater 3. Multi-layer aquifer system, silty sand, divide bisects the aquifer in the east- and sandy gravel layers. central portion where groundwater flows either westward towards the 4. Current pumping simulation: 71,600 ocean or north and eastward into Trou m$/day from 141 wells, 26 of which are Caiman, Canal Boucambrou, and Lac municipal wells. Azuei. zzvw row row ro ro row rom eo rev raw : RSS à SET ET & FE LE PRG PUSERMENENNENNS RME JE NN NN $ Faro Se SE PR eg NN 17: POS Aquifer Boundary canal { È > RES je #2) pes [Es Re #Æ "27 7) | — Simulated potentiometric lines (Layer 3) — river 2 PSE sur À 7} MX © Wells used for calibration _ stream Re =" PAS ds RS A nn | M Primary Stream infiltration Zones [1 Lake _ RS vs + LE > = Primary Recharge Zones LS [his NE ee pe 5 ee | _ ——— |: -* RE? a Fe ee ei 6 LES AE LE 1 Sibert AD Lkergesr ss Trou j. + ET Ti 0€ è f te DASCAUS Î 4 @Psscrer e 27 Gaiman Ve Pr Fe 0e : # = ? Cotes, DES e DZ: Des A # 5 2 Cr Repos ° fé V5. \, Lac. nn ï J # à Azuei 5e 2° Î e, 8 “ Baie 0% Drouillard 2 S is La Serre ’ Es Port À ù Je je eee ©, “| au 4 A à 5 50: NS Prnes Er ns ss ; "+" © rS Dee, De z =: tu. SES 2 É / 19 Ee NÉ e° se A AN À Ne : l'IPLLSDE eo / JS ae 2 È Port-au-Prince | / LE ge OF YAË EU FA - 3Q Foi din se Pi es ê, CA ”] : HAN EN CE NS FPE UINE CE à À HER 4 13 VS LAIT 4 A hi set DRRe ns PAIN À É TRS ve MC REE Ÿ RAR GERS SE La ENT x RER LE + ME 457.0 Mia ssif de} aÿ Se Te NAS RQ AR LA S AY È CR — AMATEUR ROZ ZE 1 PAS (5: +0 51 AS [ Noïthwater International,2019 Figure 5. Simulated groundwater piezometry under steady-state conditions. 1 MODFLOW is the United States Geological Survey (USGS) three-dimensional finite-difference groundwater model. [page 17] iv fl b j 200 Coast Duvivier Sibert Croix des Bouquets / Airport Tabarre Dumay a LAN ; 10€ \ Re a N & : GS Po Du ÊS o # Es ÉSEMNNNeRRRRS SS S & ES CS RAR EE NE J [y] ÿ Ci] rt PR A A AA nc em: LEE FRA BERHESSEET 0 EE — Re 9 ES BÉE ES ii TR TT A CA ES SEE Se Era se PR Ce LRRRES DAS EREt TR EL MR A ANR A A PPS AAA PU M M eu RQ roldediandifaultedN ; M u A ne ondrconfined sans ane eee" ARCC ECC Droductuity 222" tn MIN IV DO dUtIVE Eee r rer EN tr ER RSR NN F Le DS D D D D D OS OS DS D D D D D D DE nu A AR RAR RP RAR AE configurations BRRE CT ER EE EL A AAA AAA ——. A — Recent sands and gravels | . ; | NI semiconsoldatedsands ana graves | A oxrasional sandstone andimestone 25,000 20,000 15,000 10,000 5,000 0 Figure 6. Conceptual hydrogeologic cross section of the Plaine du Cul-de-Sac aquifer. . Takeaways and Insights Based on the steady-state model and initial model run the following observations are noted regarding the water balance: - Renewable recharge inputs to the aquifer are on the order of 135,000 m3/day. If current pumping is 71,600 m°/day, this would imply a 0.53, groundwater development ratio. + 83% of the aquifer inputs are from infiltration of the Riviere Grise and the Riviere Blanche, consistent with historical findings. - Canal Boucambrou appears to be a drain from the PCS aquifer. This relationship needs to be examined through investigation and monitoring. - Lac Azuei does not appear to receive a significant proportion of its water budget from the PCS aquifer, and in fact the simulation suggests 21 L/s (1,838 m°/day). Lac Azuei appears influenced by stream infiltration of the Riviere Blanche. Groundwater that discharges to the eastern portion of Canal Boucambrou may flow into Lac Azuei. The relationship between the aquifer, Lac Azeui and Canal Boucambrou be further investigated through studies and monitoring. [page 18] l - Based on the steady state simulation and assumed pumping schemes, saltwater intrusion does not appear to be a major factor at present for the primary aquifer layer. Dry season stress periods may enhance the risk, and the shallow layer is most susceptible. The coastal area of the aquifer has few wells; further, there was limited data to calibrate the model along the coast. - Trou Caiman appears to receive water from the PCS aquifer, at a range that the model simulation suggests of 45 L/s (3,890 m°/day). Model Scenario Results Summary Model scenarios suggest that impacts should be anticipated from both climate change and increased pumping. + Rehabilitation of existing wells and addition of new municipal wells may drawdown the water table, thus affecting nearby wells. This may also create a stronger gradient between the Riviere Grise and the aquifer. + Pessimistic climate change scenarios appear to have a regional impact on the aquifer, since this is largely sensitive due to a decrease in river flows that in turn reduce the volume of the river water available to infiltrate into the aquifer. - The modeling exercise indicates that the Riviere Grise and the Riviere Blanche are critical components of the aquifer and its ability to sustain groundwater abstraction and flows to surface water bodies. The recharge from the river drives the hydraulic gradient, replenishes the aquifer when there is pumping or climate change stress, and it mitigates saltwater intrusion risk in coastal areas. Aquifer impacts from schemes related to river diversions and/or dams need to be understood and mitigated. [page 19] IV. Conclusions groundwater abstraction, water quality, environmental isotopes and meteorological Summary parameters. Establishing and strengthening hydrological and hydrogeological monitoring The three investigations were effective in Programs with systematic procedures for advancing preliminary understandings of data management and dissemination is an Tunnel Diquini, Source Mariani and the PCS important recommendation that spans not aquifer. just the three study areas, but the country as a whole. Theinvestigations sharedcommon challenges, … . as all were limited due to the scarcity of Tunnel Diquini and Source Mariani data and knowledge to perform detailed . . . . hydrogeological studies. Uncovered data was These studies provide a preliminary basis often poorly documented. Significant efforts to inform planning and decision-making were necessary to synthesize, verify and "egarding the use, sustainability, and utilize the few datasets that were available Protection of the sources, so they continue to to support these studies. The 40-year record be an important source of water supply in the of flow collected and maintained for Tunnel future. Diquini should be strongly commended. Considering the regional importance of these A key finding from these studies is the Water supplies, additional investments are importance of the Massif de la Selle carbonate Warranted and fall into three categories: aquifer system. lt is arguably Haïtis most . di important aquifer systern, as it is responsible 1. Strengthening of flow, precipitation, and for the provision of a significant proportion of Water quality monitoring programs are outlined water supply to Port-au-Prince due toits large for both sources and nearby rivers to address springs and the added benefit of gravity. The important data gaps and to strengthen the rivers originating in the Massif supply the bulk Understandingofthesprings andthe associated of recharge to the Plaine du Cul-de-Sac, and aquifer system. For example, monitoring is perhaps the Plaine de Leogane aquifer as well. required to better understand connectivity High intensity precipitation events and ENSO between Riviere Froide, the aquifer, and the cycles appear critical for recharging both the Tunnel Diquini. bedrock and alluvial aquifers. RMPP resource quantities are especially vulnerable during El 2: Water source protection and enhancement Nino periods and consecutive years without -recharge protection areas should be pulses of recharge from tropical storms and delineated and protected to preserve the hurricanes. The water quality of the aquifer quantity and quality of the waters. Public systemsis also a concern due to changingland education, land use planning, zoning, and use and inadequate waste management and Controlled development in these areas is sanitation. Potential impacts on the aquifer necessary especially as Haïti's population has from schemes related to river diversions and/ grown and waste management and sanitation or dams need to be understood and mitigated Practices are lacking, given that there is a strong connection between the aquifers and rivers. 3. Further study of both sources is necessary. The delineation of recharge areas and Scientific characterizations of the aquifer Understanding of interactions with river systems in Haïti are poorly developed, largely Systems requires more detailed geological due to the lack of monitoring and data Mapping andisotope/tracer studies, availability. This study and future studies will continue to be limited without time-series/ Monitoring is important to advance the temporal datasets on spring flows, river flows, understanding and characterization of the tunnel and the Massif de la Selle regional [page 20] l aquifer that supports it. Using the data and Saltwater intrusion risk in the coastal areas findings in this study, the potential exists should also be further investigated with for source protection and enhancement monitoring, programs in key zones of the tunnel watershed. Additional studies are necessary The importance of temporal monitoring to better understand the interaction between of water levels and water quality in wells the tunnel and the nearby Riviere Froide. is important to support groundwater If any hydraulic or significant watershed flow modeling and simulations. Temporal changes are proposed for the Riviere Froide, monitoring of flow and stage along multiple we recommend comprehensive studies to reaches of the Riviere Grise, Riviere Blanche evaluate and quantify the tunnel's impacts. and Canal Boucambrou is also important considering the three systems’ relevance Plaine du Cul-de-Sac in the aquifers dynamics. Well pumping estimates and monitoring also present a The steady-state groundwater flow model significant data gap that could be addressed presented serves as a good tool to support through future activities, in the same manner the next steps of groundwater development that the estimation of current aquifer-wide and management in a regional context. abstraction was based on limited data. The model is structured to support steady- state simulations of various groundwater abstraction, environmental and climate change scenarios. The resulting model suggest a renewable groundwater resources’ use on the order of 130,000 m’/day, thus further validating the importance of the Riviere Grise and Riviere Blanche streamflow infiltrations the recharge and groundwater flow dynamics of the aquifer system. Although a significant volume of data was compiled to support model development and calibration, the quality and reliability of data is variable. Further, a limited quantity of time- series or temporal data was available for river/ stream stages and water levels in wells. The development of transient and stress period models instead of should be considered, but must be supported with additional data mining, and a focused monitoring campaign of surface water flows and groundwater levels. Scientific characterization needs to be strengthened in the northeast, east and southeast portions of the aquifer to better understand lithology and the surface and groundwater interactions related to Lac Azuei, Canal Boucambrou and Trou Caiman. These will support model refinement and result in a greater level of confidence for these zones of the aquifer. [page 21] pu HYDROGEOLOGICAL INVESTIGATION OF TUNNEL DIQUINI Characterization of Hydrology and Guidance for Source Monitoring and Protection Department Ouest, Republic of Haïti Final Report October 2018 Note: Additional data collection and research since Updating som Of the analysis and findings Of this report. Prepared for: Inter-American Development Bank & DINEPA Prepared by: Northwater International and Rezodlo S.A. Rs [page 22] Keywords Tunnel Diquini, Plaine du Cul-de-Sac; Groundwater:; Haïti; Port au Prince; hydrogeology; water supply; Massif de la Selle Latitude, Longitude 18.517N, 72.393W Citation Northwater International and Rezodlo. 2018. Hydrogeological Characterization of Tunnel Diquini: Port-au-Prince, Haïti, Inter-American Development Bank, Technical Report, HA-T1239-P001 Original report in English, French translation available. Authors James K. Adamson, PG Javan Miner, EIT Pierre-Yves Rochat [page 23] Table of Contents EXECUTIVE SUMMARY 20 SECTION 1.0 - INTRODUCTION AND PHYSICAL SETTING 21 SECTION 1.1 - GEOLOGY 22 SECTION 2.0 - METHODS AND RESULTS 24 SECTION 2.1 - HYDROLOGY 25 SECTION 2.2 - WATER QUALITY AND HYDROCHEMISTRY 27 SECTION 2.3 - STABLE ISOTOPE AND TRACER 29 SECTION 2.4 - GROUNDWATER AGE 30 SECTION 2,5 - AQUIFER STORAGE 30 SECTION 2.6 - GROUNDWATER RECHARGE 30 SECTION 3.0 - DISCUSSION 32 SECTION 4.0 - RECOMMENDATIONS FOR CONTINUED ACTIVITIES 33 À - STRENGTHENING ONGOING MONITORING EFFORTS 33 B - WATER SOURCE PROTECTION, ENHANCEMENT AND RIVER MONITORING 34 C - FURTHER HYDROGEOLOGICAL CHARACTERIZATION 35 SECTION 5.0 - CONCLUSIONS 37 REFERENCES 38 [page 24] l EXECUTIVE SUMMARY from the Riviere Froide to the regional carbonate aquifer. Tunnel Diquini is the largest single source of water for the municipal water system Groundwater Budget of Port-au-Prince, with an average supply of 29,449 m#/day to its metropolitan water 1: The long-term average annual recharge system, based on records from 2014-2018. rate in the karst terrain is estimated at 26% The tunnel accounts for “26% of all the of annual precipitation. During high intensity municipal production of water, and for 37% rainfall periods, the recharge rates are of all the gravity-fed spring flow that supplies Substantially above 26%, while during normal the metropolitan area. An investigation was Of low precipitation periods the recharge performed to analyze the hydrology of the Could be lower than 10%. tunnels waters and better understand the à . . origin of the flow andits relationship withthe 2: Aquifer storage relative to the tunnel is groundwater and surface water systems. estimated between 265 and 327 million ms. Knowing this is important to guide future studies and monitoring the tunnel, and also to 3. The limestone karst aquifer that feeds aid the water use planning and management the tunnel is well mixed and has an average by Centre Technique d'Exploitation de la groundwater age of 26 to 32 years based on Région Métropolitaine de Port-Au-Prince @ single sampling event. (CTE-RMPP): Flow Characteristics The research was undertaken using a combination of literature review, satellite 1. Recharge to the regional aquifer and and topographic imagery analysis, and particularly to the tunnel appears to be largely field reconnaissance. À brief field mission Gffected by high intensity and high-volume inside the tunnel was conducted on April 15, rainfall events such as hurricanes and tropical 2018, including: (i) physical and chemical Storms. There appears to be a 3-to-7-year sampling, (ii) stable isotope sampling, cycle of recharge trends partially influenced (iü) chlorofluorocarbon (CFC) and sulfur by El Niño and La Niña events. hexafluoride (SF6) sampling, (iv) flow rate measurement, and (v) visual inspection of 2. Tunnel discharge is seasonally variable, with tunnel geology. recorded flows ranging from 11,085 to 73,265 m*/d, and a geometric mean for all known Based on the study, the key results and recorded flows of 27,987 m?/d,. conclusions are summarized below: Spatial Distribution of Groundwater 3. The tunnel flow is most vulnerable to Recharge extended periods of normal precipitation and consecutive years without high intensity 1. The groundwater recharge area that lainfall periods such as tropical storms and contributes to the tunnel flow appears to hurricanes. range between 22 and 55 km£. a This recharge characteristic, a. This recharge area depends on the Combined with the resulting flow regression rate and duration of Riviere Froide leakage to that can extend over periods of years, may the regional carbonate aquifer. foster perceptions that the tunnel flow has been decreasing over the long-term or that b. The average recharge elevation is Some events have had an acute impact on it. estimated at 650 m. above mean sea level, , , indicating the possibility that some of the b. Limited historical data from 1959 tunnel flow may originate from river leakage Suggests that dry season low-flow conditions [page 25] are comparable or perhaps lower than current the risk of direct contamination of the aquifer il low-flow conditions and strongly influenced and tunnel waters. Increase of impervious by major recharge or drought events. surfaces and loss of soil associated with urbanization increases runoff and decreases c. The response time of the aquifer to recharge to the aquifer that contributes to major recharge events such as hurricanes tunnel flows. may be shortening, possibly due to land cover and climatic changes. a. Land use planning, zoning, and managed development of the area south of the tunnel portal is necessary in order to Connection to Regional protect the tunnel water from future water Groundwater and Surface Water quality and flow impacts. 1 Both the tunnel and Riviere Froide are connected to the same regional karst Conclusions and Recommendations limestone aquifer, from where they both receive flow. The Riviere Froide may recharge This study provides a preliminary basis to the aquifer at various spatial and temporal inform planning and decision-making with extents, and this could result in a possible link regards to the sustainability and protection between the tunnel and river system. of Tunnel Diquini so that it continues to be an important water supply in the future. If further a. Further study and monitoring is work is planned in the tunnel watershed or required to better understand the complex more information is needed concerning the hydraulic links between the Riviere Froide, the tunnel, recommendations are provided at regional aquifer and the tunnel. the end of the report about water source protection, compiling historical data, and 2. Tunnel Diquini does not appear to have a monitoring of climate, flow and water quality. hydraulic connection to the Riviere Momance. This is supported by the nature of geological structure and faulting. a The EPG fault zone and a perpendicular fault appear to direct groundwater in the Momance basin either into the Riviere Momance or into the lower reaches of the Riviere Froide, below where recharge to SECTION :.0 - INTRODUCTION the tunnel would likely occur. and Physical Setting This study is part of a coordinated effort to Aquifer Vulnerability better understand the existing and potential water supplies to serve the metropolitan area 1. Due to the high permeability and rapid of Port-au-Prince. The intent of this study infiltration rates typical in karst limestone is to determine the hydrology of the tunnel environments, the tunnel waters have high waters and better understand the origin of vulnerability to contamination. its flow and its relationship with groundwater and surface water systems. Specifically, it 2. Urbanization and land use changes in the is important to better understand whether hills south of the tunnel portal are considered significant recharge of the tunnel occurs from the greatest risk to the tunnel water quality either the Momance or Froide rivers. and flow. The lack of centralized waste management and sanitation, combined with Tunnel Diquini was completed in 1940 by the karst hydrogeology, significantly increases the J.G. White Engineering Corporation. It is [page 26] l currently the largest single water source of possible that the hydrology of the area of the Port-au-Prince municipal water system. study had largely adjusted to deforestation by Tunnel flow accounts for approximately 24% the early years of the tunnel. of total municipal production, and 38% of all gravity-fed spring flow that supply the Section 1.1 - Geology metropolitan region. Although design and construction documents for the tunnel were The geology of the tunnel area is primarily not available, it has been assumed that its composed of carbonates that range from primary target was an east-west trending lower-Miocene to Paleocene age. Most of the normal fault, approximately 1.5 km south of tunnel appears to be bored through upper to the tunnel's portal. It is believed that this fault middle Eocene-age limestones that are hard drains a sizeable portion of the Massif de La and well bedded with low bedding attitude. In Selle carbonate aquifer system in this area. the area of the tunnel portal, beds of limestone were observed to be near horizontal. To the The tunnel is reportedly 1.5-km in length along south of the portal, the Eocene limestones its main shaft. It appears to have an alignment are dissected by fault-controlled valleys, and approximately southward, although several primarily consist of detrital limestones and minor changes in bearing were noted over the chalky limestones of lower to Upper Miocene first several hundred meters from its entrance. age. The entire length of the tunnel is within At least one secondary tunnel branches from the hanging wall fault block (stratigraphically the main tunnel toward the southeast. The offset by the normal fault that traverses main tunnel is approximately 2.4 m wide with east-west 1.5 km south of the portal). The rectangular cross-section. Although minor northern wall of this fault is downthrown, and roof collapse had occurred in several places, _ it likely impounds groundwater and fosters no constrictions to flow were observed. preferential groundwater flow paths along the Below the normal fault, additional flow enters fault trace through the higher permeability the tunnel from its sidewall, and roof seeps limestones. This is believed to be the primary in fractures, merging with the main channel target and main source of groundwater to the flowing toward the portal. The tunnel floor tunnel. Figure 1 displays the geology around is rough, with fractured limestone bedding the tunnel and associated watersheds based planes protruding into the water course. At on country-wide geologic mapping (CERCG, the portal, the final length is concrete lined 1989) and faults based on mapping by as the flow is channeled into a large pipe to Pubellier (2000) and Cox et al (2011). Figure 2 supply the municipal system. The entire tunnel displays a generalized geologic cross-section length is reportedly inspected annually by Mr. along the tunnel alignment southward to Mackenson Louis of CTE-RMPP. the Riviere Froide drainage at the Enriquillo- Plantain-Garden Fault. The tunnel watershed ranges from the portal elevation at 140 meters to over 1,800 meters in the upper reaches of the Riviere Froide watershed. Average annual rainfall ranges from 1,400 mm/year near the portal to 2,100 mm/year in the upper reaches of the watershed. Land cover in the watershed is variable, with steeper slopes tending to be covered with scrub and flatter areas used for subsistence agriculture. Woodring (1924) described the watershed above Source Diquini as primarily scrub vegetation, indicating the possibility that land cover has not changed considerably in this watershed over the last 100 years. Given this land use history, it is [page 27] ævw Raw Raw Rm0w r1gow r1g0w EE - = A DE A ny pe ru Re ee ie = ES ES TRE RAP ANET TE. 1 É [| P pare 7, ee € moe Lu Pa) Hydrogeologic Mapof |? ir Là PAIE" 7 | :}/$Sal AT ACER _ ‘à * s: CE k | fee ou Jing Pam 7 D LE Tan D À À Major Springs Dar PTE 7e QE 4 Dr OUT | Catchment x P] roximate Recharge PAT EN Cr dre LR ON 4 PRE | À me sared ty ravel Be M d'A PATY, PAYS #7 AAC RS f] 275000 [| f] and Froide Watershed NY 7 à PAR V F a à AN / F Ù popaprrorimate Recharge > HR AALL KI r "À oo 1 2 3 k H — Stream LR as L BOUT PERTE DE ic A H L:Zwstershed Boundary US RE Pr PA 4 5 nn I Doines L ete KW 321) a F £ ES " 7 RUE E| Surficial Geology LAN F —— Sr IT 1 [| =— Strike/Siip Faut 7 À | pp) ct LME ps [| — — Unknown, Interred EVE Le 1 sen £ { Qec, Quatemanage coame Lf 4#y,, #70) 6 a eme] © EME À ILES 2%" nr T4, 14) Mi, Lower Miocene-age chak, 2 #7 TT » td LA Lo { mari and limestone ARE © 13 Ur. broiet dép Emi, Eocene-age mari and Fès af atershet PP Up ce Æ gs = A 2; En Eveceneagereniy A I APR, AE AE #5 DE Wstéréhed) 78 | breccatdhardimestne ME 2 TP PTS. NA. Cia pt | Es, Midele to Upper Eocene- MT 7 A, CR Le fe OS dif s SPL P) (à 1} \ 1. ' age hard limestone AT ne 7 147 LA #11) MAC PS Be U mil Cb, Cretaceous-age volcanic 922152 ir APT NS Pots AL NN? RE ? ds VAE CODE - DM ré voicanosedimentay unis Em) | (TTC A A) ! De À, A Erercarreféi, ofthe Dumisseau formation DES à dj ses CEE - £ 3 À a vb ph iteitenanone 201) | Ë Figure 1 - Geologic map of tunnel and associated watersheds with interpreted recharge zones. 800 no Direct recharge in karst terrain 7 —— TE TT Ÿ SPRL) $ coo LR > ë CLRLRRLLRLLLRRLLLOONRLLLLLLLODN S /- pi Normal faut WRRRRRRRLLOON LL) $ ST € 500 ee /1517000011114001111000111000011111111 ÈS LE FE get by PR RSR LCR LÉO) ES £ 0770 ntec te 112000011041100110011111100) EE É RIRE = RER CORP L I 0 0000000 0000100011101110011001111111107 AAC 5 400 NERO 000000000000 junte $ re MÉTRO LOL LÉ TS ET RENE: SR 100000041000 0 0000041727 00000 ,19100077002015191 RE 8 200 Tunnel Diquini CR RME MORIN ARE ê q LS TRS NAN OO PIN MR É PAR RSR RTS PA LR LL MO LLX 200 CR RNCS NO IT NS POI S LR RERO LLLCNT ON ANA AIT SN POINTES RS M PNR RÉ ÉÉ TSI LIL LL É LU 120000000010 001110011155104111100001111 NS 14 RL 000000011000 00001000011000000111000 100, EL ICLÉL LÉ 0 0111560 0011110011110001111 RE N RTS 0 RER RRRRRORRRRROR ROLL N LR LRRORRRRRTRRRRRRRRLIEREEESSS 0 500 1000 1500 2000 2500 3000 3500 4000 4500 North Horizontal Distance (meters) South Figure 2 - Generalized cross section of tunnel geology and hydrology. [page 28] : Section 2.0 - Methods and Results Water was sampled several meters into the A brief field mission in the tunnel was tunnel, as sampling farther into the tunnel was conducted on April 15, 2018, including: (ip not feasible. Mr. Mackenson Louis believed physical and chemical sampling, (ii) stable that our sampling event was representative of isotope sampling, (ii) chlorofluorocarbon à lower flow condition for the tunnel. A 12V (CFC) and sulfur hexafluoride (SF6) sampling, Sampling pump with flexible tygon tubing was (iv) flow rate measurement, and (v) visual Used to collect low-flow samples where the inspection of the tunnel's portal geology. All tunnel flow was considered laminar. Samples activities at the tunnel were performed under for physical, chemical, and stable isotope the supervision of the caretaker of the tunnel, analysis were collected by filling prepared Mr. Mackenson Louis. sample bottles provided by the laboratories of analysis (First Environmental and Isotech). The tunnel flow was measured at the Chlorofluorocarbons (CFCs) and Sulfur tunnel's portal just prior to the point where Hexafluoride (SF6) were collected to age- the flow leaves the open-rock channel with date the groundwater discharging from the a Marsh-McBirney Flo-Mate 2000 portable tunnel. Samples for CFC-11, CFC-12 and electromagnetic velocity meter. The open- CFC-113 were collected using the glass bottle channel width was approximately 2.44 meters Method with copper tubing as described with an average water depth of 0.215 meters. PbY USGS and Reston Chlorofluorocarbon The calculated flow rate was 1,300 m?/h, or Laboratory. Samples for SF6 were collected 361 L/Ss, corresponding to a stage-height of Using 1-Liter plastic-coated safety amber 15.5 centimeters on the staff gauge affixed glass bottles with polyseal cone-lined caps, to the east side of the tunnel portal (this GlSo applying methodologies developed by staff gauge does not extend to the channel USGS and University of Utah Noble Gas Lab. bottom). All sampling bottles and excess air tubes for CFC's and SF6 were provided by the Dissolved The field team was escorted approximately and Noble Gas Lab at the University of Utah. 300 meters into the tunnel for geological Samples for excess air analysis were also inspection. The limestone appeared to be collected in #4-inch copper tubes with clamps; well-bedded, moderate to hard, and with a these samples support .correction of the near-horizontal bedding attitude. Numerous SF6 data. Upon completion of sampling, all seeps entered the tunnel from fractures and Samples were wrapped in insulating materials cavities daylighting the walls and ceiling along and transported to the US for shipment to the the 300 meters length inspected. Many of analysis labs. these seeps were less than 1 L/s, although several were estimated to flow at more than 2 FRS +. 5 L/s and one was estimated at 20 L/s. At 4 ue | approximately 150 meters from the tunnel er, Us portal, a secondary smaller tunnel enters œ SR. the main tunnel from the east. This branch # produces significantly cooler water than the main tunnel flow. This smaller tunnel was RS. barricaded with cobble that surrounded a à concrete pipe. Mr. Mackenson Louis reported that each November/December he walks the g full length of the tunnel to inspectit. His father (now deceased) was the original caretaker of the tunnel since it was constructed. Based on their observations, they believe that the tunnel's flow rate has been decreasing over the last several decades. Photo 1. View inside tunnel portal. [page 29] er DANCE Re 0 Photo 2. Secondary adit approximately 150 m from ESEREEENS 063 pr : 70 PES portal, noticeably colder flow. fe, LE, va LT ARCS ATEN ARE Deer) 1 £ ES …. … CP vs Ve 8 j'e “ j À : à CA 144 À à 4 NET 4 Photo 5. Example of seep in east tunnel wall. UE DO+ ! 4 OPEL Photo 3. View of limestone geology looking toward Section 2.1 - Hydrology tunnel portal. Tunnel discharge varies significantly based _ : on the intensity and duration of recharge Fa. 0 4 -#] events and the transit time through the D". mes. : à 5 aquifer. À discontinuous tunnel flow dataset CUITE se was compiled from various sources spanning TS : 7 between 1980 and 2018. The data was De TT: : Ligues : : : : ; ‘ += ; primarily provided by Engineer Pierre Colon x x 4 , +: Geffrard of CTE-RMPP. Flow measurements . ARE , = dE were provided as average monthly flow Eaghe SS LA 2 data from 1980 to 2010, single monthly flow 3 LAPS g y * AN ul measurements estimated using a spinner v Ya k un à velocity-meter from 2010 to 2014 and monthl 3" A ÿ y At D flow measurements using an electromagnetic pt ET velocity-meter from 2014 to 2018. The 1: “te % project team was informed that a CTE-RMPP : Ds. “à technician continues to measure flow rate Photo 4, Example of near horizontal bedding,. at the tunnel each month. One older flow [page 30] L a — —— — —————————" = : à Là ‘À Len %, VAR : Re : À à V4 OAMELE 2 ge, ? s, à TE. » Ÿ ( 7" s LL Us. Photo 6. Sampling for CFC's and SFé. Photo 7. Existing staff gauge. measurement that was located for the tunnel, shortening. Decadal climatic variations may from August 1959 (Waite, 1960) recorded 217 also be illustrated by the flow data, showing L/s. Based on the available data, tunnel flow a downward trend in flow rate from 1980 to rates are seasonallÿ and annually variable 1990, a slight upward trend from 1990 to 2000, and range from 128 L/s to 848 L/s with a a potentially strong upward trend from 2000 geometric mean of 324 L/s. to 2010, and a decreasing trend from 2010 to 2018. Trends from 2010 to 2018 may be Recharge to the regional aquifer and to the misleading due to the difference in monthly tunnel in particular appear to be largely average vs. single-event flow measurements. affected by high intensity and high-volume rainfall events such as hurricanes and tropical There was insufficient data available to storms, something that is typical of karst establish clear long-term trends in the aquifers. There appears to be a 3-to-7-ÿear tunnel's discharge since its construction. The cycle of recharge trends partially influenced common understanding is that the tunnel by El Niño and La Niña events. Conversely, flow rates have been decreasing through extended periods of declining flow rates occur time, however this trend is not apparent in the during drought and El Niño years. Figure 3 1980 to 2018 flow data. À decreasing trendis shows flow data from 1980 to 2018 along also not present when comparing the August with recorded hurricanes and droughts. A 1959 flow rate (217 L/s) to average August lag time of several months or longer is visible flow rates between 1980 and 2010 (324 L/s) between the rainfall event and an increase or 2011 and 2018 (404 L/s). While the trend in discharge; however, the response time of is not clear for Tunnel Diquini, Source Diquini, the aquifer to major recharge events may be shows a 44% decrease in flow over ninety [page 31] years. The average recorded flow of Source tunnel. Further study of these long-term and il Diquini between 1923 and 1938 was 62 L/s, cyclic flow dynamics of the tunnel and other and the 2014 monitoring period recorded an major Port-au-Prince springs would provide average of 34.4 L/s. However, Source Diquinis valuable insights for water management and decreased flows may be related to changes planning. in hydrology related to dewatering near the & À à É MERE SUR Shi «3 2 228 LE D E È SNA SIMS SIN ES RE [7 5€ È 2 | 3500 700 AE 5 î DÉERS MiE #4 54 ä so © & 500 = ” n 2500 À 5 400 Jan É A . Le Au 200 L f 1500 < 100 [ 0 1000 ARS SSSR an Ses à ass aessse esse ose Ie emPetionville Station (UHM 2019) === Tunnel Diquini Source Mariani Figure 3. Tunnel Diquini discharge with major meteorological events, 1980 - 2018. Section 2.2 - Water Quality and Hydrochemistry & . Tunnel Diquini is considered to have excellent s/ \\ 2 physical and chemical quality as a drinking Fe water source, with a fairly dilute groundwater (214 mg/L total dissolved solids). The groundwater is a CaHCO3 type water typical of a limestone aquifer (Figure 4). Nitrate was . the only potential parameter of concern from © the sampling event (8.64 mg/L as NO3). Past ä + È analysis from 2014, and monitoring between #/ Ne S7 WE 2006 and 2013 also reported elevated nitrate > (o concentrations, although nitrates were still below the USEPA (10 mg/I as NO3-N), WHO (50 mg/las NO3) and DINEPA (50 mg/las NO3) ca ci potable water guidelines (Table 1). Biological CATICNS ANICNS sampling and analysis were not performed as a part of this study. The total organic ” ” : — carbon (TOC) measurement of 0.4 mg/L Ada diagram of Tunnel Diquini water, [page 32] l is a typical concentration in groundwater, to surficial contamination within the recharge particularly in the tropical climate zone. There areas, especially near the tunnel portal. was also a measurable amount of barium Further study concerning the aquifer and in the groundwater at 0.116 mg/L. Barium tunnel hydrology would help locate the occurs in the open ocean at a concentration primary recharge areas and locations of 0.05 mg/L and is a group-two element as where the aquifer is most susceptible due calcium. Barium substitutes for calcium during to increased conduit flow and concentrated limestone formation, and, when the limestone recharge. Table 1 provides the water quality dissolves, barium ends in the groundwater. and hydrochemistry data available from Other trace metals were below detection previous studies and the sampling event from limits, indicating a low likelihood of current April 15, 2018. industrial or commercial contamination. Due to the karst-conduit and rapid recharge nature of the aquifer, the tunnel is susceptible USEPA Tunnel Diquini DINEPA WHO USEPA Parameter Units Secondary 18.517 Standard Guidelines MCL MCL 72393 Date Collected 13-Jan-14 4-Nov-14 15-Apr-18 Alkalinity, Total (Caco3) mg/l 500 _ _ _ 200 200 230 Bicarbonate (CaCO3) mg/L — - _ _ 0 0 <5 Chloride mg/l 250 250 _ 250 11.99 15.5 6.2 Conductivity umhos/cm — _ _ _ 390 387 382 Fluoride mg/l 2 1.5 4 _ 0.27 Ammonia (N) mg/L _ _ _ _ 0 0.013 <0.01 Nitrite (NO2) mg/L 3 3 1 _ 0.023 0.033 Nitrate (NO3) mg/l 50 50 10 _ 16.82 12.4 8.64 pH@25°C Units _ _ _ … 7.29 7.6 7.42 Sulfate mg/l 250 500 _ 250 4 5 <15 Silica ( SiO2) mg/L — _- _ _ 12.9 TOC mg/L _ _ _ = 0.4 Antimony mg/l _ 0.02 0.006 _ < 0.006 Arsenic mg/l — 0.01 0.01 _ <0.01 Barium mg/l — 0.7 2 _ 0.116 Beryllium mg/l _ _ 0.004 _ < 0.004 Cadmium mg/l _ 0.003 0.005 _ < 0.005 Calcium mg/l 100 _ _ _ 78.47 76.6 744 Chromium mg/L _ 0.05 0.1 _ < 0.005 Copper mg/l 1 2 1.3 _ < 0.005 Iron mg/l 0.2 _ - 0.3 <0.05 Lead mg/l 0.01 0.01 0.015 _ < 0.005 Magnesium mg/l 100 _ _ _ 2.91 4.86 4.1 Manganese mg/l _ 0.5 _ 0.05 < 0.005 Potassium mg/L _ _ _ _ 0.6 0.7 <0.5 Silver mg/L _ - _- 0.1 < 0.005 Sodium mg/l _ _ _ _ 3.18 2.07 3.1 Thallium mg/l _ _ 0.002 _ < 0.01 Zinc mg/L 3 _ _ 5 <0.01 Total Hardness (CaCO3) mg/L 300 _ - - 202 Mercury mg/l _ 0.0005 0.002 _ < 0.0005 TotaDissoed mg 600 1000 - 500 186.3 214 Note: 2014 sampling based on EPTISA database (2016) Table 1. Hydrochemical analyses of Tunnel Diquini waters. [page 33] Section 2.3 - Stable Isotope and Blanche and Froide, and their geographic and Tracer topographic similarities, an assumption can be made that the river water in the Riviere Froide Stable isotopes of oxygen (8:80) andhydrogen MAY have an isotopie composition displaying (SD) were sampled to aid in evaluating the same relationship between ô#O and recharge dynamics. Gonfiantini and Simonot elevation. Such a calculation indicates that (1988) observed a linear trajectory of © the Froide may have abaseflow è 80 of around versus elevation from samples collected south -3.34. This 5*O result is nearly identical to that of Port au Prince. They found that there is a Of Tunnel Diquini (-3.36 or -3.28). -0.9 per mil change for every 1000 meters of elevation gain for water points on the plain, However, based on topography, the average and estimated a slope of -1.4 per mil change elevation of the Riviere Froide watershed for every 1000 meters of elevation for the is roughly 150 m higher than the tunnel springs south of Port-au-Prince. This linear … Watershed. This apparent elevation difference trajectory can be applied to Tunnel Diquini, as is likely due to the evaporative fractionation of it was one of the originally sampled points in 8"O© during surface flow in the river channel. 1987 and has similar characteristics to springs When these pieces are put together, this in the area. indicates two possibilities for how the tunnel and Rivière Froide are related: i) both derive The stable isotope composition of the sample their discharge from roughly the same collected in April this 2018 was 520 = -3.28 per regional carbonate aquifer zone: or ii) stream mil and $D = -14.4 per mil, while Gonfiantini and losses in the Froide infiltrate the normal fault Simonot (1988) measured the groundwater and end up discharging into the tunnel. In any discharging from the tunnel in 1980s and Case, the hydrologic dynamics of the two are their results were 5120 = -3,36 per mil and D likely linked. Further study including isotopic = -14 per mil. These results are essentially the Sampling and flow rate measurements at same, since the analytical uncertainty in the various locations and hydrologic conditions on measurement is +0.1 520, and +1 D. Applying the Froide are necessary to further evaluate the 8:20 value to the -0.9 and -1.4 per mil/1000 these possibilities. À better understanding of meter regression slopes, the likely minimum this link is a key aspect of future studies to recharge elevation is 200 meters above sea better define the tunnel. level and the average recharge elevation is 650 meters above sea level. The aquifer water has had essentially the same isotopic composition over the past 31 years. That implies that the aquifer is well mixed before it emerges from LL the tunnel. à": 35 Ce ee. 46 0 = 0.9 per mil1000m The stable isotope values recorded for this . . s study and by Gonfiantini and Simonot (1988) so . plot above the Global Meteoric Water Line, 50 ” which typically indicates that the groundwater 46 . has been subject to limited evaporation fractionation. However, the 1988 study and sole the analysis for this study indicate greater ® complexity in drawing conclusions from the a nent Nes En value of O0. Gonfiantini and Simonot (1988) 0 400 800 1200 observed that water from the Riviere Grise Elevation (meters) and Riviere Blanche followed a similar trend to the carbonate springs, with an average à!#O of -3.66 for the Grise and -4.23 for the Blanche. Based on the mean elevations of the Grise, [page 34] l typically is). Taking these results together, we a an on on OO du un en de can assume a groundwater age for the tunnel à ad Di | : during an average flow regime of 29 years. era 20) | Further sampling of CFC's and SF6 during * nan | : ° higher flow and lower flow events could * Conan as) 5 further help illuminate the recharge dynamics ea | n » 8 of the aquifer. à Ammearcrer M . ë Lu e à Section 2.5 - Aquifer Storage : . With the available data, a planning level estimate of storage in the aquifer can be Figure 5 - Stable isotope data plotted with Global made. Recognizing that the isotope data Meteoric Water Line (GMWL). from 1988 and 2018 indicate that the aquifer is well mixed, meaning dispersion is high in the aquifer porous structure, then the annual output of the spring multiplied by the age of the groundwater equates to a qualitative storage estimate. Using the calculated annual discharge from 1980 to 2000 of 9.36 million m#/year, this method estimates between 243 and 300 million m3 of storage. This evaluation . is very approximate: refinements are likel Section 2.4 - Groundwater Age possible th further analysis of discharge, sampling data, and tracer tests. Chlorofluorocarbons (CFCs) and Sulfur Hexafluoride (SF6) were collected to age-date the groundwater. These estimates are based on CFCs and SF6 accumulating in air during Section 2.6 - Groundwater Recharge the 20th century, measuring their solubility in water and extrapolating back to the The groundwater recharge rates associated atmosphere (see Appendix B). Both methods With the tunnel discharge were estimated require assumption or measurement of other Using the chloride mass balance method and parameters. For CFCSs the primary adjustable à custom GIS-based direct recharge model parameter is the recharge temperature, developed by Miner and Adamson (2017). which affects solubility of the CFCs. For Forty-one chloride measurements for the SF6 the primary adjustable parameters tunnel waters and 13 rainfall chloride samples are excess air, recharge temperature, and were available for the mass balance (Table recharge elevation. For both analyses, the 2). The geometric mean rainfall chloride is recharge temperature was set at 25°C, and 2,5 mg/l and the tunnel geometric mean is 9.8 for SF6 the recharge elevation was set at 500 g/L, yielding an average annual recharge meters, and the excess air at 2 cc/l, which rate of 26%, or 434 mm of the roughly 1,700 is a common value for most groundwater. mm/year, which falls in the probable recharge The CFC calculated recharge age date was area. Combining this recharge rate with the 1986, indicating the water is 32 years old. The average annual tunnel discharge, indicates calculated recharge date for the SF6 was that the recharge area of the tunnel aquifer 1992, indicating the water is 26 years old. It could be approximately 22 kme. is not uncommon for these two age-dating methods to be in slight disagreement (the The GIS-based recharge model developed older the water the more discrepancy there by Miner and Adamson was calibrated to [page 35] historical average annual baseflow in the baseflow from the Riviere Froide, so the Riviere's Grise, Blanche and Momance that chloride levels of the tunnel reflect the flank the tunnel, and Riviere Froide to the east chloride and recharge dynamics of the larger and west. Average modeled recharge in the aquifer that supplies the Riviere Froide. In any Riviere Froide watershed was 206 mm/year, case, it is likely that the spatial extent of the less than half that has been indicated by the tunnel recharge area ranges between 22 chloride mass-balance. This would equate to km? and 55 km2. This large potential range a roughly 55 km? drainage area. One possible points to the need of more comprehensive reason for this discrepancy could be the and diagnostic studies to better delineate the scarcity of rainfall chloride measurements, tunnel recharge area and flow pathways, and especially at the elevations where primary also to characterize the nature of the Riviere recharge is occurring. À second potential Froide along the reaches south of the tunnel. reason could be that the tunnel is receiving Site Type Date Lat (dd) Long (dd) __ Elevation CI (mg/L) Tunnel Diquini Spring 4/15/2018 18.517 -72.393 140 6.2 Tunnel Diquini Spring 1/13/2014 18.517 -72.393 140 12 Tunnel Diquini Spring 11/4/2014 18.517 -72.393 140 15.5 2006 - Tunnel Diquinis Spring 2013 18.517 -72.393 140 9.7 Momance River River 8/1/2018 18.475 -72.407 305 6 ………. Froïde River River 8/1/2018 18.487 -72.412 280 6.2 Thomassin 36 Rainfall 10/6/2018 18.482 -72.317 1025 1.24 Thomassin 36 Rainfall 10/2/2018 18.482 -72.317 1025 0.71 Thomassin 36 Rainfall 9/28/2018 18.482 -72.317 1025 1.78 Anse-a-Galet Rainfall 8/26/2018 18.834 -72.868 20 4.04 Anse-a-Galet Rainfall 8/14/2018 18.834 -72.868 20 34 Anse-a-Galet Rainfall 8/10/2018 18.834 -72.868 20 8.86 Anse-a-Galet Rainfall 8/5/2018 18.834 -72.868 20 10.5 Bas de Delmas Rainfall 8/1/2018 18.563 -72.340 5 2 Petionville Rainfall 8/1/2018 18.511 -72.290 380 7.1 Petionville Rainfall 8/2/2018 18.511 -72.290 380 13.6 Petionville Rainfall 8/2/2018 18.511 -72.290 380 9.89 Laboule Rainfall 10/19/2016 18.495 -72.315 850 <0.7 Laboulet Rainfall 10/4/2016 18.495 -72.315 850 <0.7 Laboule Rainfall 10/24/2016 18.495 -72.315 850 <0.7 Clercine 12 Rainfall 10/30/2016 18.575 -72.277 42 <0.7 Cabaret #1 Rainfall 9/16/2015 18.736 -72.418 45 1.32 Cabaret #4 Rainfall 3/6/2016 18.736 -72.418 45 3.91 Cabaret #2 Rainfall 2/2/2016 18.736 -72.418 45 14 Lafito Rainfall 2/11/2016 18.697 -72.349 29 22 Anse-a-Galet Rainfall 10/17/2015 18.834 -72.868 20 0.8 Geometric Mean for Rainfall 2 2.7 Geometric Mean for Tunnel Diquini 10.5 1 Sample taken during Hurricane Matthew 2 Assumes chloride values for Laboule and Clercine 12 are approximately 0.5 mg/L Note CTE-RMPP records contain 38 chloride samples; the geometric mean of these is presented. Table 2. Rainfall and tunnel water chloride values. [page 36] l Section 3.0 - Discussion m/d, with a geometric mean for all known recorded flows of 27,987 m°/d,. Based on the study, the key results are outlined and discussed below: 3. The tunnel flow is most vulnerable to extended periods of normal precipitation Spatial Distribution of Groundwater Recharge and consécutive years without high intensity rainfall periods such as tropical storms and 1. The groundwater recharge area that hurricanes. contributes to the tunnel flow appears to range between 22 and 55 km£. a This recharge characteristic, combined with resulting flow regression that a. This recharge area depends on the Can extend over periods of years, may foster rate and duration of Riviere Froide leakage to Perceptions that the tunnel flow has been the regional carbonate aquïfer. decreasing over the long-term or that acute impacts have occurred. b. The average recharge elevation is . Le. estimated at 650 m above mean sea level, b. Limited historical data from 1959 indicating the possibility that some tunnel Suggests that dry season low-flow conditions flow may originate from river leakage from are comparable or _Perhaps lower than the Riviere Froide to the regional carbonate Current low-flow conditions and are strongly aquifer. influenced by major recharge or drought events. Groundwater Budget c. The response time of the aquifer to 1. The long-term average annual recharge Major recharge events such as hurricanes rate in the karst terrain is estimated at 26% May be shortening, possibly due to land cover of annual precipitation. During high intensity and climatic changes. rainfall periods, the recharge rates are . . substantially higher than 26%, while during Connection to Regional normal or low precipitation periods the Groundwater and Surface Water recharge could be lower than 10%. 1. Both the tunnel and Riviere Froide are 2. Aquifer storage relative to the tunnel is Connected to the same regional karst estimated between 265 and 327 million m5. limestone aquifer, and both receive flow from the aquifer. The Riviere Froide may recharge 3. The limestone karst aquifer that feeds the the aquifer at various spatial and temporal tunnel is well-mixed and has an average extents, and this could result in a possible link groundwater age of 26 to 32 years, based on between the tunnel and river system. a single sampling event. a. Further study and monitoring is Flow Characteristics required to better understand the complex hydraulic links between the Riviere Froide, the 1. Recharge to the regional aquifer and to regional aquifer, and the tunnel. the tunnel in particular appears to be largely affected by high intensity and high-volume 2: Tunnel Diquini does not appear to have a rainfall events such as hurricanes and tropical hydraulic connection to the Riviere Momance, storms. There appears to be a 3-to-7-year This is supported by the nature of the cycle of recharge trends partially influenced geological structure and faulting. by El Niño and La Niña events. a The EPG fault zone and a 2. Tunnel discharge is seasonally variable with Perpendicular fault appear to direct recorded flows ranging from 11,085 to 73,265 groundwater in the Momance basin either into [page 37] the Riviere Momance or into the lower reaches A - Strengthenin Ongoing of the Riviere Froide, below where recharge to Monitoring Efforts the tunnel would likely occur. . eve Historical monitoring data collected by Aquifer Vulnerability CTE-RMPP provided key insights into the tunnel dynamics. lt is our understanding that 1 Due to the high permeability and rapid current monitoring efforts, when performed, infiltration rates typical in karst limestone include (i) monthly flow measurement at environments, the tunnel waters have high the tunnel portal and (ï) collection of a vulnerability to contamination. water sample for physical and chemical analysis including conductivity, salinity, PH, 2. Urbanization and land use changes in the temperature, turbidity, hardness, alkalinity, hills south of the tunnel portal are considered calcium, magnesium, chloride, sulfate, nitrate, the greatest risk to the tunnel water quality Hitrite, and iron. Data gaps exist in terms of and flow. The lack of centralized waste What was provided to our team by DINEPA management and sanitation combined with and CTE-RMPP. There are multiple ways that the karst hydrogeology significantly increases monitoring efforts could be further reinforced the risk of direct contamination of the aquifer and improved. and tunnel waters. Increase of impervious surfaces and loss of soil associated with . We commend CTE for the data that urbanization increases runoff and decreases has been collected. Data that was particularly recharge to the aquifer that contributes to useful to this study included flow rate, tunnel flows. chloride, conductivity, turbidity, and nitrate measurements. We recommend that, at a a. Land use planning, zoning, and minimum, these parameters continue to be managed development of the area south of Hneasured monthly. the tunnel portal is necessary to protect the tunnel water from future water quality and + Data format consolidation - it appears flow impacts. that existing records are maintained in several different formats including paper and . . digital spreadsheets, and that the retrieval Section 4.0 - Recommendations for of data or analysis is a challenge. The most Continued Activities consistent and accessible records were hand-written notes. Simplicity, clarity, and Due to Tunnel Diquinis importance as the accessibility are key to ensuring CTE-RMPP largest single water supply to Port-au-Prince, has the data needed to properlÿ manage its additional work may be warranted to guide water resources. lt was our impression that water source protection and enhancement, the current digital data management scheme water use planning, and future water supply was unclear or overly complicated, which led development in the Massif de la Selle aquifer. {to difficulty in locating and compiling data Based on the findings of this study, this Whenit was requested. Developing a standard section provides recommendations in three data architecture and recording and archiving categories for (1) improvements to ongoing method is recommended along with training monitoring efforts, (ii) water source protection for CTE-RMPP or DINEPA employees involved and enhancement, and (ii) options for in water monitoring. additional study. Any future activities would be greatly aided by increased availability of + Monthly flow measurements can be temporal datasets for climate, discharge, augmented and eventually made simpler hydrochemistry, and stable isotopes. Table 3 by the incorporation of the staff gauge into and Table 4 summarize the recommendations monitoring and development of a stage- by category, and the narrative provides discharge curve. This would also make supplemental detail. daily flow measurements more feasible [page 38] l and allow for automatic flow monitoring of sampling which is recommended. This using pressure transducers. An increase monitoring should begin as soon as possible in monitoring frequency would allow for and be incorporated into a permanent better understanding the complex recharge monitoring program so that data is available dynamics when coupled with rainfall data. for water use planning or future studies. - Table 3 provides recommendations for measurement parameters and timing P Quarterl 'arameter Daily Weekly Monthly y Yearly Notes Ti 1 FL Ï | Î | Î | Select and document a standard location and methodology. Gus - | | l x | | | Electromagnetic velocity probe or industry standard jé | | | | | | equivalent recommended. | | T j | | Existing staff gauge is difficult to read and does not indicate Tunnel Flow | | | | Î | actual water depth. Recommend installation of new staff Height (stage) | Î Î | Î | gauge with easy to read centimeter scale. Over time, stage | l L L l | can be used to estimate flow on a daily basis. eee trheeerreeeccrreeteete==LPPRERRPRREEE CE pH | | X Î | | | Low cost conductivity/PH field probe, calibrated as required. Temperature | H x | | H | Î | | | | | Low level chloride analysis often required, suggest detection Chiens | | | h | | | limit of 1 mg/L or lower. Nitrate | | | x | | | CTE-RMPP lab analysis Turbidity | | x | | | CTE-RMPP lab analysis EnERSe D DSERE IEEE DER DDDE DRE BDD DER D ee pee 1e JERES E. coli | | | X | | | sufficient and more economical than a full analysis at the Hydrochemistr | | | | x Î | CTE-RMPP lab analysis of Ca, Mg, Na, CI, K, CO3, HCO3, l'A l Î | Î | SO4, NO2, Fe, TDS, Hardness, Alkalinity AE Compilation | Î | Î | contains a table of the measured results be published online and publication | | Î | | X | so that data is easily available. This is also a good interval of data | | | | | | to review issues with data collection and revise the program 1 in order to reduce data gap. Table 3. Tunnel monitoring program recommendations. B - Water Source Protection, and reduce sources of contamination near e PPT Enhancement and River Monitoring the tunnel. Water source protection and enhancement + Land use practices In the recharge planning - This study provides an improved drea can be mapped and reviewed to guide understanding of discharge dynamics and focused interventions, such as forestation, recharge areas for the tunnel which highlight … terracing of steep slopes, and development of interim insights useful for guiding next steps to Exclusion zones. protect and enhance it. . . . - Sinkhole delineation can be performed - Recharge protection areas can be to locate zones of concentrated recharge. delineated to protect critical areasofrecharge Fences could be built to keep wildlife and [page 39] livestock out and to eliminate potential of the aquifer, there is likely both diffuse and il contamination sources. concentrated recharge occurring. Potential study methods include: - The ridgetop area southeast of the tunnel portal is becoming increasingly - À monitoring program using stable urbanizedi efforts at sanitation planning and isotopes of dt#O and dD along with chloride infrastructure in this area would help protect collected over a multi-year period to record the tunnel from biological contamination. changes occurring due to both drought and hurricane-induced recharge events. River Froide monitoring - The Riviere Froide is likely a major component of the - Periodic streamflow measurement at hydrogeological system that supplies the multiple locations along the upper reaches of tunnel. Temporal and spatial data of river flow the Riviere Froide to locate zones of gain and and water quality are needed to better define loss. the relation between the Froide and Tunnel Diquini. While short-duration measurement - Dye tracer testing of karst dolines and campaigns may be incorporated into later the Froide River is perhaps the most definitive studies, the most useful data would come from method for delineating the recharge area, a permanent and well-defined monitoring although transit times may be prohibitive. program. + Sampling of various major seeps - Periodic streamflow measurement at and the fault face in the tunnel may provide multiple locations along the upper reaches of _ insight into the spatial extent and amount of the Riviere Froide to locate zones of gain and recharge. loss. Although measurements ideally should be conducted weekly or monthly, quarterly Refinement of tunnel recharge and discharge measurement is a good starting interval based dynamics - As previously mentioned, the on seasonal flow variations. tunnel system appears to be largely driven by large climatic events of 3-to7year cycles. + Measurement should include flow rate, field water quality including pH, - À concerted and coordinated effort conductivity, and temperature, and sample should be made by CTE-RMPP, DINEPA, and collection for low-level chloride analysis. It is BME to locate any documents related to also recommended collection of grab samples historical flow rates from the first decades for stable isotopes of d'fO and dD during high after the tunnel was constructed. This will help and low flow events at least bi-annually when in understanding how and if the tunnel has corresponding flow and chloride data is also been affected by the land use changes and available. climate changes that have likely decreased the flow in other springs in the area, such as Source Diquini. - Careful analysis of these recharge C - Further Hydrogeological and discharge trends may also help to predict Characterization the future effects of climate change on the tunnel and major carbonate spring discharge Refinement of the tunnel recharge area - to Port-au-Prince's water supply. This study indicates a large uncertainty in the spatial extent of the tunnel's recharge. Refinement of tunnel geology and The improvement of the knowledge of this Structure aspect is important to guide water source protection, land use planning, and future uses - À concerted and coordinated effort of the Riviere Froide. Due to the karst nature should be made by CTE-RMPP, DINEPA, and [page 40] l BME to locate any documents related to the this study. If such data is located, an analysis tunnel design and detailed local geologic could be performed comparing rainfall mapping. intensity over periods of recorded flow rates to better understand recharge thresholds + Geologic mapping along the tunnel and the conditions conducive to diffuse or adit to better understand the ways in which concentrated recharge. lithologic and structural changes affect the occurrence of groundwater flow into the -Implementation of arainfall monitoring tunnel. Such mapping may also be useful if program that also samples rainfall for 810, future efforts to secure a reliable source of ÿD and chloride. Potential localities for such a water for Port-au-Prince include the possibility program include Degand, south of the tunnel, of similar tunnels. and Fermate, in the Upper-eastern reaches of the Froide watershed. Such rainfall intensity . CPR . : Increased rainfall monitoring and measurements coupled with meteoric je : : sampling hydrochemistry will allow for better recharge estimates, and will be a valuable resource to - Daily rainfall data suitable to determine all future hydrogeological studies in the Massif rainfall intensity was not made available to de la Selle. De Financial Expertise à : Institutional Category Activity Costs Required Equipment Required Involvement | | | | Flow meter, staff gauge, | | itoril Î | ; pH/Conductivity probe | CTE-RMPP A | ne PUS ae (see Table | Low(annual) | Technician | field probe, CTE-RMPP | technician and data | | | | lab, low-level chloride | manager | Water Source Protection and | pl g | sanitation | Minimal | .-. dination. | Enhancement Planning | High | parmis rs | | community and local B ! OO + Cmpromentaon) | Ê | | leader support 1 Hydrologist À D | | | | A CTE-RMPP | Riviere Froide Monitoring Program | Low (annual) | ue | pes Ll | Hhieenen data | | | operation | | g i i Î : : Î CTE-RMPP pe | Hydrologist/ | DYe racing materials and | &pnician, DINEPA | Refinement of Tunnel Recharge Area | High (one-time) | Hydrogeologist | ati ne | approval of dye | Moderate or Low | | | Refinement of Tunnel Recharge and | if sufficient ! Hydrologist/ ! Minimal | CTE-RMPP data Discharge Dynamics | monitoring has ! Hydrogeologist | ue | manager Refinement of Tunnel Geology and | Moderate (one- | ue | Minimal | CTE-RMPP, BME Structure | time) | g log | | records review Moderate (one- | Hydrologist | Two to three telemetric | CTE-RMPP, Rainfall Monitoring and Samplin time setup) | Setup, | Wealherstations Wii | LARNDR technician g ping | | Technician | simple sample collection | and support }__Low (annual) ! operation | mechanism l Ppoi Table 4 - Phased recommendations and resource needs. [page 41] Section 5.0 - Conclusions in the country that take advantage of the carbonate bedrock geology that benefits This study applied discrete datasets to derive from high recharge rates, and the topography an understanding of the tunnel hydrology that supports gravity-fed water supplies. and hydrogeology. We believe this study is sufficient in characterizing the tunnel on an Limitations of Investigation interim basis from which to inform planning and . decision-making with regards to guiding the Aspects of the assessment were especially sustainability and protection of Tunnel Diquini. limited by the unavailability of data and Disciplined monitoring and the associated leSources regarding the Massif de la Selle temporal datasets are important to advance aquifer. The lack of consistent monitoring the understanding and characterization of and records of discharge, streamflow, and the tunnel and the Massif de la Selle regional Precipitation made it especially challenging to aquifer that supports it. Using the data and quantify recharge rates and size of the tunnel's findings in this study, the potential exists aquifer. A focused and basic level of analysis for source protection and enhancement and synthesis was applied throughout the programs inkey zones ofthetunnelwatershed. report with the primary objective to provide Additional studies could also be focused to initial insights into the tunnel dynamics and better understand the interaction between "ecommendations for further study. the tunnel and the nearby Riviere Froide. If any hydraulic or significant watershed À limited amount of historical data was changes are proposed for the Riviere Froide, available to support this analysis, andthis study we recommend comprehensive studies to included a single sampling event which is just evaluate and quantify tunnel impacts. a snapshot of a dynamic system. Conclusions in this report are preliminary and presented to The Massif de la Selle carbonate aquifer Gid interim planning and decision-making and is arguably Haitis most important aquifer to guide any future study and characterization. system, as it is responsible for providing a significant proportion of water supply to Port- This assessment was performed using au-Prince fromitslarge springs withthe benefit … Professional care and skill ordinarily exercised, of gravity and its rivers supply the bulk of Under similar circumstances, by experienced recharge to the Plaine du Cul-de-Sac aquifer. geologists and hydrogeologists practicing Further characterizing and understanding in this or similar locations with very limited the aquifer as a whole would enable future Sources of data and resources, Changes in informed planning and operations to protect Analysis and interpretations can and will oceur and enhance the important resources. with the acquisition and analysis of new data, such as monitoring reports, water quality There are many lessons learned from this data, and tracer and isotope data. Analysis study and implementation of the tunnel and interpretations presented in this report that can help to guide future water supply Mustbe considered fluid and subject to review exploration and water supply development and revision as additional data is compiled. elsewhere in Haïti and in other areas of the Analysis and interpretations described in this Massif de la Selle. It is our opinion that Tunnel report may be invalidated wholly or partially Diquini is a favorable case study to warrant by the results of continued data collection and the evaluation of other tunneling opportunities ©bservations, [page 42] REFERENCES Adarnson, JK, Jean-Baptiste, G., and Miner, W.J, 2016, Summary of groundwater resources in Haïti, in Wessel, GR, and Greenberg, JK, eds. Geoscience for the Public Good and Global Development: Toward a Sustainable Future: Geological Society of America Special Paper 520, p. 1-22, doi10.1130/2016.2520(14). BME [Bureau des Mines et de l'Energie], 1993, Notice Explicative de la Carte Géologique d'Haïti: Port-au- Prince, Bureau des Mines et de l'Energie. BRGM [Bureau de Recherches Géologiques et Minières], 1988, La Synthèse géologique notamment dans ses parties stratigraphiques et tectoniques: Bureau des Mines et de l'Energie, Port-au-Prince, Haïti. BRGM [Bureau de Recherches Géologiques et Minières], 1989, Étude des ressources en eau de la région de Port-au-Prince. Butterlin J., 1960, Géologie générale et régionale de la République d'Haïti [General Regional Geology of the Republic of Haiti]: Institut des Hautes Etudes de l'Amérique Latine, 194 p. CERCSG [Centre d'Etudes et de Réalisations Cartographiques Géographiques], 1989, Carte Géologique de la République D'Haïti [Geologic map of the Republic of Haïti]: Bureau des Mines et de l'Energie, Port-au- Prince, scale 1:250,000, 1 sheet. Cox, et al, 2011, Géologie de Port-au-Prince. 1:250,000 scale. Déll, P. and Fiedler, K, 2008, Global-scale modeling of groundwater recharge: Hydrology and Earth Systems Sciences, no. 12, p. 863 - 885. Gonfiantini and Simonot, 1988, Isotopic Investigation of Groundwater in the Cul-de-Sac Plain, Haïti. International Atomic Energy Agency, IAEA-SM-299/132, Pg 22. Hiimans, R.J, S.E. Cameron, JL. Parra, P.G. Jones and A. Jarvis, 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965-1978. LGL, 2011, Actualisation du Schéma Directeur d'Alimentation en Eau Potable de la Région Métropolitaine de Port-au-Prince: PHASE 1 : Collecte des données et Analyse Diagnostic, Rapport no 1.4 Étude des ressources en eau. Pg 151, N/D: SLI 608471, LGL 211374. Miner, W.J, and Adamson, J. (2017). Modeling the Spatial Distribution of Groundwater Recharge in Haïti using a GIS Approach, Geological Society of America 2017 Annual Meeting, Seattle, Washington, doi: 10.1130/abs/2017AM-297120. Moliere, E, and Boisson, D. 1993, Coupes Géologiques d'Haïti, in, Notice Explicative de la Carte Géologique : Port-au-Prince, Bureau des Mines et de l'Energie. Oxfam, 2014, Carte Geologique de Canaan, Jerusalem, Corail et Onanville (CROIX DES BOUQUETS, Haïti), Evaluation technique des menaces naturelles et vulnerabilite de la commune de Croix des Bouquets, Port- au-Prince. 1:10,000 scale. Pubellier, M. 2000, Plate boundary readjustment in oblique convergence: Example of the Neogene of Hispaniola, Greater Antilles. Tectonics, Vol. 19, No.4, p 630-648. Scanlon, BR, Healy,R. and Cook, P.G. 2002, Choosing appropriate techniques for quantifying groundwater recharge: Hydrogeolology Journal, no. 10, p. 18-39. Suez, 2013, Travaux prioritaires de renforcement de la production d'eau. Pg 178. Taylor, G.C. and Lemoine, R.C. 1949, Ground water in the Cul-de-Sac Plain, Haïti: US. Geological Survey Open-File Report, 59 p. UNDP [United Nations Development Program], 1990, Carte Hydrogéologique République d'Haïti [Hydrogeologic Map of the Republic of Haiti]: United Nations Development Program, New York, scale 1:250,000, 1 sheet. United Nations, 1991, République d'Haïti : Programme des Nations Unies pour le Développement : Développement et Gestion des Ressources en Eau. [Government of Haïti and Organization of the United Nations, Department of Technical Cooperation for Development]: Report HAI/86/004, vol. 6. US. National Aeronautics and Space Administration. Hispaniola region, Landsat 8: spectral bands 1 through 7. Product L1T. Vacher, HL, and Ayers, JF. 1980, Hydrology of Small Oceanic Islands - Utility of an estimate of recharge inferred from the chloride concentration of the fresh-water lenses: Journal of Hydrology, vol. 45, p. 21-37. Waite, H. 1960. Reconnaissance Investigations of Public Water Supplies of Port au Prince and in 12 Villages in the Department du Nord, Haïti. United States Geological Survey. Woodring, W.P. Brown JS, and Burbank, WS, 1924, Geology of the Republic of Haïti. Department of Public Works, Port-au-Prince, Haïti. World Health Organization, 2011, Guidelines for drinking-water quality, 4th edition: WHO, Geneva, Switzerland. [page 43] - es HYDROGEOLOGICAL INVESTIGATION OF SOURCE MARIANI Characterization of Hydrology and Guidance for Source Monitoring and Protection Department Ouest, Republic of Haïti Final Report October 2018 Revised March 2020 Note: Additional data collection and research Prepared for: Inter-American Development Bank & DINEPA Prepared by: Northwater International and Rezodlo S.A. pes | [page 44] Keywords Source Mariani, Plaine du Cul-de-Sac; Groundwater: Haïti; Port au Prince; hydrogeology; water supply; Massif de la Selle Latitude, Longitude 18.535N, 72.427W Citation Northwater International and Rezodlo. 2019. Hydrogeological Characterization of Source Mariani: Port-au-Prince, Haïti, Inter-American Development Bank, Technical Report, HA-T1239-P001 Original report in English. Authors James K. Adamson, PG Javan Miner, PE Pierre-Yves Rochat [page 45] Table of Contents EXECUTIVE SUMMARY 42 SECTION 1.0 - INTRODUCTION AND PHYSICAL SETTING 44 SECTION 1.1 - CLIMATE AND LAND COVER 44 SECTION 1.2 - GEOLOGY 45 SECTION 2.0 - METHODS AND RESULTS 46 SECTION 2.1 - HYDROLOGY 50 SECTION 2.2 - WATER QUALITY AND HYDROCHEMISTRY 53 SECTION 2.3 - STABLE ISOTOPE 56 SECTION 2.4 - GROUNDWATER AGE 57 SECTION 2,5 - AQUIFER STORAGE 58 SECTION 2.6 - GROUNDWATER RECHARGE 58 SECTION 3.0 - DISCUSSION 60 SECTION 4.0 - RECOMMENDATIONS FOR CONTINUED ACTIVITIES 61 À - STRENGTHENING ONGOING MONITORING EFFORTS 61 B - WATER SOURCE PROTECTION, ENHANCEMENT AND RIVER MONITORING _ 63 C - FURTHER HYDROGEOLOGICAL CHARACTERIZATION 63 SECTION 5.0 - CONCLUSIONS 65 REFERENCES 66 [page 46] l EXECUTIVE SUMMARY fluctuations in precipitation volume and intensity have on the recharge rates. Source Mariani is currently the most distal Le, source of water that supplies the CTE-RMPP à. The average recharge elevation is water system. lt is the largest naturally flowing SStimated at 580 m above mean sea level with spring and the second largest single water G Corresponding temperature of 22.7 C. This source that supplies the Port-au-Prince Suggests the possibility that some spring flow municipal water system. When the pumping may originate from distal zones in the regional station is in operation, an average of “19,144 carbonate aquifer such as within the Riviere m3/day spring flow can supply 17% of total Momance basin. municipal production, and 24% of all spring flow supplying metropolitan Port-au-Prince region (CTE-RMPP data 2014-2018, The Groundwater Budget spring discharges from limestones that drain a portion of the Massif de La Selle carbonate 1: Thelong-termaverage annualrechargerate aquifer system, west of the Riviere Froide and is estimated at 38% of annual precipitation. north of the Riviere Momance. The objective During high intensity rainfall periods, the of this evaluation is to better understand the recharge rates may approach 50%, while spring flow characteristics and the origin of during normal or low precipitation periods the the waters to guide future study of the Massif recharge could be approximately 15%. de la Selle aquifer system and to aid CTE- . . oo, RMPP in water use planning, development, 2: Aquifer storage relative to the spring is monitoring, and protection. estimated between 155 and 259 million ms. This investigation was accomplished by a 3. The limestone karst aquifer that feeds combination of literature and data review, the spring is well mixed and has an average satellite and topographic imagery analysis, 9groundwater age of between 21 and 35 years and field reconnaissance. A brief field mission based on a single sampling event. to the spring was conducted in April 2019 which included: (i) physical and chemical ee sampling, (ii) stable isotope sampling, Flow Characteristics (ii) chlorofluorocarbon (CFC) and sulfur . à hexafluoride (SF6) sampling, and (iv) visual 1: Recharge to the regional aquifer appears observation of local geology. A follow-up visit to be largely affected by high intensity and to the spring was conducted in January 2020 high-volume rainfall events such as hurricanes to verify more recent flow monitoring data and tropical storms. There appears to be a received from CTE-RMPP. 3-to-7-year cycle of recharge trends partially influenced by El Niño and La Niña events. Based on the study, the key results and conclusions are summarized below: 2. Spring discharge displays mild seasonal variability with monthly average flows typically ranging between 14,500 and 25,000 m°/d with Spatial Distribution of Groundwater an average of 19,500 m/d. Recharge a. Instantaneous (daily) flows display 1. The groundwater recharge area that ÿreater variability, ranging from 7600 to contributes to the spring flow appears to be 30,700 m°/d. approximately 15 km? but may be as large as 34 km. b. Based on the spring catchment infrastructure as observed in 2019, total spring a. This uncertainty in the recharge flow is measured from a single water meter. area is due to the large effect that annual However, this method does not account for [page 47] overflow. As a result, some high spring flows d. Source Mariani essentially serves as il could be underreported. a drain for the western portion of the Massif de la Selle aquifer. 3. The spring flow is most vulnerable to extended periods of average or below 2. Source Mariani does not appear to have a average precipitation and consecutive years significant hydraulic connection to the Riviere without high intensity rainfall periods such as Momance or Riviere Froide. This is supported tropical storms and hurricanes. by the isotope and tracer sampling and analysis of recharge catchment size. a. This recharge characteristic combined with the recent flow regression that extends from 2014 to 2019 may foster Aquifer Vulnerability perceptions that the spring flow has been decreasing over the long-term or that acute 1. Due to the high permeability and rapid impacts have occurred. infiltration rates typical in karst limestone environments, the spring waters have b. Limited historical data from between high vulnerability to contamination. This 1925 and 1933 suggests that average is confirmed by the elevated nitrate levels spring discharge remains relatively stable or consistently measured in spring discharge. perhaps has even increased due to increased precipitation intensity. 2. Urbanization and land use changes in the hills south of the spring are considered the 4. The cyclic and multi-annual recharge greatest risk to groundwater quality and flow. characteristics of the regional carbonate The lack of centralized waste management aquifer are important for water managers and sanitation, combined with the karst and planners to understand and utilize in hydrogeology, significantly increases the risk balancing the different water sources of CTE- of direct contamination of the aquifer waters. RMPP. a. Land use planning, zoning, and managed development of the area south of Connection to Regional the spring in an area larger than the existing Groundwater and Surface Water spring protection perimeter is necessary in order to protect the spring water from future 1. Source Mariani flows from the regional water quality and flow impacts. Massif de la Selle carbonate aquifer. a. The regional aquifer also supplies Conclusions and Recommendations many of CTE-RMPP major springs, Tunnel Diquini, and provides base flow to river This study provides a preliminary basis from systems. which to inform planning and decision- making with regards to the sustainability and b. Monitoring is required to better protection of Source Mariani, so it continues to understand the complex hydraulic be an important water supply into the future. relationships between these major outlets of Recommendations are provided at the end of the aquifer. the report regarding water source protection, compilation of historical data, and monitoring c. Source Marianiis the lowest elevation the climate, flow, and water quality. Significant terrestrial outlet known for the aquifer and increases in study efficiency would be gained appears to emanate from a topographic by combining the recommendations of exposure of the main aquifer lithology rather this study with those of the Tunnel Diquini than as a contact spring. This may act to characterization (Northwater International sustain flows even when higher elevation and Rezodlo 2018). springs exhibit reduced flows. [page 48] l SECTION 1.0 - Introduction and with block perforations to allow inflow from Physical Setting the colluvial deposits that transmit the groundwater to the surface. This study is part of a coordinated effort to better understand the existing and potential . . water supplies that serve the metropolitan Section 1.1 - Climate andLand Cover area of Port-au-Prince. Its intent is to . characterize the hydrology of Source Mariani Average annual rainfall ranges from 1,300 waters and better understand the origin and MM/year near the spring to 1,700 mm/year characteristics of its flow. in the upper reaches of the Riviere Froide watershed. Based on data from the Petion- Source Mariani is currently the most distal Ville meteorological station (UHM, 2018), two source of water that supplies the CTE-RMPP distinct rainy seasons occur in the catchment, water system. lt is the largest naturally flowing the first peaking in May and the second in spring and second largest single water September/October (Figure 1). Precipitation source for the Port-au-Prince municipal Varies from year to year, with periods ofintense water system. When the pumping station rainfall and hurricanes spaced between is in operation, Source Mariani accounts for Periods of relative drought. These cycles approximately 12% of the total municipal GPPear to occur on 3-to-7-year rotations production, and 15% of all the spring flow linked to El Niño and La Niña events. A slight supplying metropolitan Port-au-Prince region increase in annual precipitation is apparent (based on 2014 data). The spring discharges in data from 1980 to 2016 for the Petion- from limestones that drain a portion of the Ville station (Figure 1). Recently, unusually Massif de La Selle carbonate aquifer system, high rainfall and intense hurricane seasons west of the Riviere Froide and north of the between 2007 and 2010 were followed by Riviere Momance. decreased precipitation from 2011 to 2016. The Source Mariani catchment ranges Land cover in the catchment is variable, with from the outlet at 24 m to over 650 minthe Steeper slopes tending to be covered with karst plateau north of the Riviere Momance. SCrub, and flatter areas used for subsistence The spring catchment infrastructure was Agriculture and sporadic forest, Woodring reportediy constructed im 1992, although (1924) described the watershed area as recent improvements have been made to Primarily scrub vegetation, indicating the the overflow, pumping and power stations, possibility that land cover has not changed and supply line up to the reservoir. À spring considerably in the southerly hills over the last protection area approximately 45-hectaresin 100 years. Given this, perhaps the hydrology size has been fenced and reforested around of the area had adjusted to deforested the spring and corridor up to the reservoir. Conditions when spring flow measurements The catchment is a broad concrete structure Were first collected in the 1920s and 1930s. Average Monthly Rainfall in mm (data: UHM. 2018) 2,000 Years 1960-2016 Petion-Ville _ 1,800 250 È 1,600 200 Ê & 1400 150 3 & 1,200 100 Ë 1,000 800 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC TN NRA Figure 1. Average Monthly Precipitation at Petion-ville UHM Station and Annual Precipitation from 1980 to 2016. [page 49] Section 1.2 - Geology rates of the spring, which serves as a ‘drain’ for a large portion of the aquifer. The geology of the spring catchment is : . composed of carbonates that range from lo the southofthe spring outlet, approximately lower Miocene to middle Eocene age. The 78 km? (65%) of the 15.2 km° catchment spring outlet is primarily surrounded to the is composed of hard, bedded limestones west, south and east by weathered Eocene of upper to middle Éocene age. Regional age marl and chalky limestone with relatively folding has created a triangular wedge which low permeability. More recent colluvial Widens westward and is composed of detrital deposits cover the bedrock formations limestones of lower Miocene age. These north of the spring. Several outcrops of hard limestones transgress into the marls and limestone were observed along the south Chalks ofthe upper Eocene formations (Figure and southeastern side of the spring and are 2). The southern portion of the catchment interpreted to be middle to upper Eocene APpears to be altered by q high degree of age, indicating that the spring result from a Karst weathering at elevations between 400 topographie intersection with the piezometrie and 600m. These karst features likely promote surface associated with regional limestones high infiltration and recharge rates through that are hydraulically connected through the upper to middle Eocene limestones into faults and fractures to the recharge areas the regional carbonate aquifer. within the Massif de la Selle. Source Mariani . . | is the lowest elevation terrestrial outlet known Figure 2 displays the geology of the interpreted for the Massif de la Selle aquifer. The geologic SPring catchment and associated watersheds and topographie intersection may be a key based on the adaptation of various sources explanation for the consistentlÿ high flow 9f data (CERCG 1989, Eptisa 2015, Pubellier 2000 and Cox et al 2011). Eee 7 7 7, OURS ARE" = RÉ ee A A ; RARE SENTE, Ro RS SR ee GUERRE Tr) es : Photos 1 and 2. Detrital limestone, chalk and marl of lower Miocene or upper Eocene age, outcrops to west and south of spring. PRES. EN Se: AR SN ANR Ro SEA Er QE He RO DS UN T PO Man is D RS RTE LE ETS Pa %e Nan. LUS Photos 3 and 4. Hard micritic, well bedded limestone of middle to upper Eocene age, outcrops on southeast side of spring and in majority of catchment. [page 50] Hydrogeologic Map of | an : Topographic Sinks Road | À SENS | Source Mariani nee ==pho mu ” _ Catchment x 7 mn + : [| 1:50,000 + PRET ET PE anee n i V Geology a À 0 & SAND DA Vi | (2 D) 5 (eo or FT au RE FH Strike/Dip k ANT PA LAORE 5 2 Normal Faut } 71 LT CAT sopbsefer LME D Ne 9 2 ne. | | J Ed ) L D 10} (PP PT D 2 Æ T FA à ere } pe nd 7 DL ANS 4 a CTI IL SIL 5 » DS ND] bp EL CAD AP 2 1 0) Em LP LILI SI LD) ie ! AT AT SAN GARE, RP HS 194 De LS LA DS DT A Watérshes JA) DES A Ni | Be D NA IS LLS AA A | LÉ al © FAT] 2 150 FLN Le NE } ALT ee | D LAPS ARR De | CT Va à GA 4] ar VE Din NO) ee | Figure 2. Geologic Map of Interpreted Spring Catchment and Associated Watersheds. Section 2.0 - Methods and Results discharge, the overflow from the spring catchmentis not monitored, so high flows may Two brief field visits to the spring were not be accurately recorded. The January 16, conducted as a part of this study. The first 2020, flow measurement was taken in a canal was conducted on April 2, 2019 and included: downgradient of the pumping station that i) physical and chemical sampling, ii) stable … "eceives the flow when the pumping station is isotope sampling, ii) chlorofluorocarbon nOtin operation. (CFC) and sulfur hexafluoride (SF6) sampling . . and, iv) visual observation of local geology. The field team was allowed to enter the spring The second visit was conducted on January Catchment to visually inspect the construction 16, 2020 and included: i) spring flow rate and nature of the water seepage during the measurement, ii) review of CTE-RMPP flow April 2, 2019, visit. À layer of silt and fine sand meter readings, and ii) visits to two nearby With some gravel was noted to cover most springs. All activities at the spring were of the catchment floor. According to CTE performed under the supervision of Jean Staff, the catchment floor is cleaned twice Jimmy Cyndigue or Ing. Pierre Colon Geffrard Per year. À washout portal was observed, of CTE-RMPP. although internal concrete dividers may limit its functiondlity. The overflow portal flows Most flow data for this report is derived from from the catchment into a rock and concrete- a totalizing flow meter between the spring lined open channel and is diverted to and the pumping station. The reading on surface drainage below the pumping station. this meter is documented monthly by CTE- Groundwater flow into the catchment is RMPP technicians. The flow meter measures AChieved via offsetting gaps in the bottom four all the flow diverted to the pumping station. layers of concrete block along the south and Unfortunately, during periods of high southwest corners of the catchment. Colluvial [page 51] gravel and cobble deposits were visible : Samples for SF6 were collected using 1 Liter il through the gaps, along with extensive roots amber plastic-coated safety glass bottles that likely originated from trees surrounding With polyseal cone-lined caps, also employing the catchment. These roots are reportedly cut methodologies developed by the USGS and during the bi-annual maintenance. CTE staff University of Utah Noble Gas Lab. All sampling noted that, after cleaning, an increase in flow bottles and excess air tubes for CFCs and SF6 occurs. However, this may reflect temporary were provided by the Dissolved and Noble adjustments to the hydraulic gradient due to Gas Lab at the University of Utah. Samples lowering of the catchment floor after sediment for excess air analysis were also collected removal. in %4-inch copper tubes with clamps: these samples support correction of the SF6 data. Water sampling was undertaken in the spring Upon completion of sampling, all samples catchment, adjacent to the gaps in the blocks were wrapped in insulating materials and where groundwater seepage occurs. À 12V transported to the US for shipment to the sampling pump with flexible tygon tubing was respective laboratories. used to collect low-flow samples. Samples for physical, chemical and stable isotope analysis Two additional springs were visited on 16 were collected by filling laboratory prepared January 2020 to aid in the characterization of sample bottles. Chlorofluorocarbons (CFCs) the local and regional hydrology. The springs and sulfur hexafluoride (SF6) were collected arelocally known as Tet Sous and TiSous Amba as a means to age-date the groundwater and had a combined flow of approximately 58 discharging from the spring. Samples L/s which seep from the semi-consolidated for CFC-11, CFC-12 and CFC-113 were Pliocene formations that overlie the limestone. collected using the glass bottle method with It is possible that portions of these flows are copper tubing as described by USGS and return flows from Source Mariani. the Reston chlorofluorocarbon laboratory. : de L. it A ET INTER Photo 5. Panoramic view of spring catchment. RUE ; Es _ 7 = fi + ” | Photo 6. Panoramic view of inside spring catchment, outlet to distribution at bottom right. [page 52] 4 | Photo 7, West outlet to pumping, station. Photo 8. East outlet to pumping station. RS er ou Qu < ÉEEee L S Pr | M an GE DO SMS Photo 9. Inside view, overflow portal to surface drainage. Photo 10. Outside view, overflow portal to surface drainage. MS GE M à 1 Photo 11. Historical washout portal. Photo 12. Flow in downstream canal when pump station not in operation. [page 53] he DR te Le 5 RS æ D. , , OR | 4 RE TD CES. 2% = rs. RE D DS CR ! SR k \ y - V6 - LL L TVR Photo 13. Sampling in April 2019. Photo 14. Totalizing flow meter measuring spring flow to pumping station. RSS RES NET OU 4 nr NC x Le a 2e MN UE À a (os FN MAN RNE L Sr RE se LL /ES + US — Photo 15. Roots entering the spring catchment. Photo 16. Clean-up in progress to remove fallen debris g pring p In prog from slope above catchment. | | FRÈRES jai 2 SU à MR £ KG |: f = ÊR— = ESS LE À VE SE Photo 17. Pumping station. Photo 18. Overflow and drainage canals leading from ping g g catchment. [page 54] : TRE Tr #1 FA AE # & À : Û 1H RAT. eue se SU , ie che | ÉMVRE, | ES" 7 # ; ; Ne CR A 5 he SE: LE Be t: LA Pr: ; Cr x oe 1 ES S 1933 (Direction Generale des Travaux Publics, 1918 - 1938), ranging from 170 L/s to 280 Spring flow varies based on the intensity and L/S with an average of 217 L/s. The recent duration of recharge events and the transit … discharge data ranges from 88 L/s to 355 L/s time through the aquifer. À discontinuous flow With an average of 225 L/s. This indicates that dataset was compiled from CTE-RMPP and SPring discharge trends may not have varied Eptisa (2015) spanningintermittently between Substantially in the past ninety years. While October 2008 and July 2019. The data @ long-term trend of decreasing discharge is was primarily provided by Ing. Pierre Colon not supported by the data, a short-term trend Geffrard of CTE-RMPP. Flow measurements Of decreasing discharge is apparent between were provided as average monthly flow data 2009 and 2015. Average spring discharge was by CTE and as discrete measurements by 288 L/s in the 2008 - 2009 data and only 217 Eptisa. Mr. Jean Jimmy Cyndigue of CTE- L/sin the 2014 - 2019 data. This recent trend RMPP continues to record monthly flow on is believed to be the result of adjustments to the totalizing flow meter between the spring the 2007 s 2008 period, when unusually high and pumping station and at a meter between andintense precipitation occurred (Figure 3). the pumping station and the reservoir. Some . . inconsistency and confusion are apparent Based on the available data and a previous regarding some of the historical data, as Study of nearby Tunnel Diquini (Northwater several measurements indicate the spring International and Rezodio. 2018), recharge to overflow, but not the total flow, andthe current the regional aquifer and to the spring appears flow meter does not measure overflow events. to be largely affected by high intensity and high-volume rainfall events such as hurricanes Seven older discrete discharge measurements and tropical storms. There appears to be a were documented for the spring from 1925to 3-to-7-year cycle of recharge trends partially [page 55] influencedbyEINiño andLaNiñaevents(figure flow rates have been decreasing through il 3). Conversely, extended periods of declining time, however, this trend is only apparent in flow rates occur during drought, El Niño or the short term due to the above mentioned El normal years. Figure 3 shows discharge data Niño and La Niña cycles. Comparison of flow from 2008 to 2019 for both Source Mariani between the 1920s and 1930s andrecent data and Tunnel Diquini, along with annual average suggest remarkably stable average discharge precipitation as measured in Petion-ville. À and hint at the possibility of increasing flow. similar trend is apparent between Source Figure 4 shows average monthly discharge for Mariani and Tunnel Diquini that parallels the the 2008 to 2019 data compared to monthly annual precipitation curve, lending evidence averages for the 1925 to 1933 data. Further to the hypothesis that aquifer discharge study of these long-term and cyclic flow rates are highly linked to years of increased dynamics of Source Mariani, Tunnel Diquini precipitation intensity and volume. The high and other major Port-au-Prince springs variability in flow rates in 2014 could result would provide valuable insights for water from the difference between average monthlÿ management and planning. It is also worth flows and instantaneous flows. The variability noting that a combined additional flow of 58 may also be affected by the partial capture of L/s was measured from Source Tet Sous and flow prior to catchment and pumping station Source Ti Sous Amba downgradient of Source rehabilitation. Mariani. The similarity in field water quality between these springs and Source Mariani Insufficient data was available to establish may indicate that they are connected to the clear long-term trends in spring flow. The regional aquifer, or that their flow is actually common understanding is that the spring recirculated waters form Source Mariani. 900 2,000 Lm] 800 1,800 E 700 | 1.600 £ 600 J 1,400 £ a 2 © 500 l 1200 € ÉA | 1,000 £ = 400 À 2 £ | 800 2 ä 500 +1 GE 600 à 2 CIF ; = 400 È Li] EI 100 200 à 0 0 = OO DO + M © DE D OO © m0 M + M © CO A © 2 L2LSLSLSLSS mm mn nm nm mm SSL SDL2DSCOCDLCOCLCLCLCSCOCCOoLCce NU —— Source Marian —— Tunnel Diquini —#—Petion-Ville Precipitation Figure 3 - Source Mariani and Tunnel Diquini flow with annual precipitation at Petion-ville Station, 2007 - 2016. [page 56] 300 HE 250 = = + D > 4 LS E & 200 = "… & 150 & E 100 A 50 +1925-1933 m 2008 - 2020 0 1 2 3 4 5 6 7 8 9 10 11 12 Month Figure 4- Source Mariani average monthly discharge, historical comparison. : : : Average Percent RMPP Spring Latitude Longitude Elevation Discharge Spring Flow (dd) (dd) (m) (L/s) (%) CARREFOUR-FEUILLES 18.52211 -72.33881 103.2 76.7 6% CHAUDEAU 18.51719 -72.38315 125.6 474 3% COROSSOL 18.52349 -72.40638 123.2 55.3 4% DESPLUMES 18.50099 -72.28644 518.8 12.3 1% DIQUINI SOURCE 18.52120 -72.39130 84.5 37.7 3% DIQUINI TUNNEL 18.51680 -72.39285 136.0 432.4 31% DOCO 18.50966 -72.25284 360.0 13.4 1% FRERE 18.51684 -72.25465 228.8 78.0 6% LECLERC 18.52361 -72.36055 91.1 28.2 2% MADAME BAPTISTE 18.52604 -72.40326 77.6 68.5 5% MAHOTIERE 18.52656 -72.40609 86.5 110.7 8% MARIANI 18.53525 -72.42699 244 225 16% METIVIER 18.50717 -72.24198 406.0 22.0 2% MILLET 18.48277 -72.28716 906.8 16.9 1% PLAISANCE 18.51746 -72.29767 271.5 54.3 4% TÈTE DE LAU 18.50228 -72.28631 481.3 37.1 3% TURGEAU 18.52473 -72.31933 195.4 73.3 5% Note: for comparative purposes, only average monthly discharges provided by CTE-RMPP were used in this table; the average discharge and percentage of total for Source Mariani and Tunnel Diquini are slightly different than reported elsewhere. Most discharge data is from 2010, 2011 and 2014. Table 1 - Springs of the Massif de la Selle used by CTE-RMPP. [page 57] Section 2.2 - Water Quality and l Hydrochemistry EXPLANATION + Jan, 2014 : Source Mariani is considered to have good RU EU . Ÿ Se physical and chemical quality as a drinking ® Mar, 2015 s / e. water supply. Its groundwater is fairly dilute bo \ / L (240 mg/l total dissolved solids), and it is ° 2 \/ a CaHCO3 type, which is characteristic of /\ a limestone aquife. Nitrate was the only À potential parameter of concern from the 4 VAS April 2019 sampling event (8.4 mg/l as NO:). & NUR /. en Nes: Monitoring by CTE-RMPP from 2008 to 2016 Ÿ \ S/ Ÿ similarly reported slightly elevated nitrate \ / concentrations with an average of 12.5 mg/l ; (as NO3) which is below USEPA (10 mg/l as Tu Ta NO3-N), WHO (50 mg/l as NO3) and DINEPA CATIONS ANIONS (50 mg/l as NO:) potable water guidelines (Table 2). Biological sampling and analysis Figure 10 - Piper diagram of Source Mariani Waters. was not performed as a part of this study. Trends in monthly groundwater conductivity The total organic carbon (TOC) measurement (Figure 6) and chloride (Figure 7) illustrate a of 34 mg/l is slightly higher than the typical several month lag-time between the onset of for groundwater, possibly indicating surficial the rainy season (April and September) and contamination. There was also a measurable the corresponding decreasein dissolvedsolids. concentration of barium at 0.174 mg/l. Barium The relatively small variation in conductivity occurs in the open ocean at a concentration indicates that the aquifer is generally well of 0.05 mg/l, and it is a group two element, mixed. Seasonal variations in chloride are the same as calcium. Barium substitutes for intriguing and may point to the variations in calcium during limestone formation, and recharge rate throughout the year, with higher when the limestone dissolves barium, it ends recharge rates occurring during months with in the groundwater. Other trace metals highrainfall asitistypicalof karst aquifers. Both were below detection limits, indicating a low conductivity and chloride show an increasing likelihood of current industrial or commercial tend from the 2008-2009 data through the contamination. end of 2016, the same period during which spring discharge was generally decreasing,. Due to the nature of the aquifer, the spring is Plotting conductivity and chloride relative susceptible to surficial contamination within to measured spring discharge confirms this the recharge areas, especially near the trend (Figure 8). The increased conductivity outlet, as evidenced by the nitrate and TOC and chloride, and decreased flow, result from data. Further study concerning the aquifer lower recharge and slower groundwater flow and spring hydrology would help locate quring decreased annual precipitation and the primary recharge areas and locations rainfallintensity. This trend may be associated where the aquifer is most vulnerable to with the multi-year flow and recharge cycles contamination due to increased conduit flow mnentioned in Section 2.1. Current conditions and concentrated recharge. Table 2 provides gre likely more indicative of typical flow and the water quality and hydrochermnistry data Water quality, whereas the 2008-2009 data available from previous studies and the April js more indicative of increased recharge and 2, 2019, sampling event. flow conditions. Water quality monitoring data provided by Conversely, both turbidity and nitrate CTE-RMPP allows for analysis of several increase with the flow rate. This result is not parameters on a monthly and annual basis. unexpected, since surficial contamination [page 58] l occurs most during periods of high rainfall frequently as urbanization near the spring and runoff. Turbidity and nitrate do not has increased. Turbidity may be in part due appear to be correlated (Figure 8), possibly to aquifer hydrology and karst muds, while indicating different sources. Some of the nitrate appears more anthropogenic in origin. highest turbidity measurements happened Mobilization of the sediment build-up in the shortly after the intense 2007-2008 hurricane floor of the catchment is also likely a source of season, while peaks in nitrate occur more high turbidity during larger flow periods. WHO USEPA Source Mariani Parameter Unis DINEPA Guideine PEP Secondar 18.535 s yMCL -72.427 Date Collected 23-Mar-15 29-Jun-15 6-Aug-16 2-Apr-19 Are: Co mg 500 _ _ _ 200 200 190 Bicarbonate (CaCO3) mg _ _ _ _ 244 244 232 Chloride mgf 250 250 _ 250 17 18 18.5 9.7 Conductivity umhos/cm _ _ _ _ 420 406 390 400 Fluoride mgA 2 1.5 4 = 04 Ammonia (N) mgl = = w _ 0.12 0.03 < 0.045 Nitrite (NO2) mgl 3 3 1 - 0.1 Nitrate (NO3) mgl 50 50 _ _ 10.6 111 9.7 84 Nitrate (NO3-N) mgl - - 10 - 1.9 pH@25°C Units _ _ _ _ 75 77 76 78 Sulfate mg/ 250 500 _ 250 8 11 4 5.9 Silica {SiO2) mgA = _ _ = 202 TOC mg/l _ _ _ _ 34 Antimony mgf _ 0.02 0.006 - < 0.000387 Arsenic mg _ 0.01 0.01 _ < 0.0076 Barium mgl _ 0.7 2 _ 0.174 Berylium mgl _ _ 0.004 _ < 0.00016 Cadmium mg = 0.003 0.005 = < 0.00036 Calcium mg/ 100 _ _ _ 75.3 139.34 721 732 Chromium mgf _- 0.05 0.1 - 0.016 0.01 <0.0014 Copper mg/ 1 2 13 _ 1.84 0.88 0.00518 Iron mg 0.2 _- _ 0.3 0.03 0.08 0.12 0.052 Lead mgf 0.01 0.01 0.015 _ <0.0031 Magnesium mg/l 100 _ _ _ 4.86 22.34 5.39 Manganese mg _ 0.5 _ 0.05 0.001 0.141 0.00404 Potassium mg/ _ _ _- _ 0.9 1.5 14 0.796 Siver mg/l _ _ _ 0.1 <0.0019 Sodium mg =. = eu " 12.13 6.82 6.11 5.91 Thallium mgl - _ 0.002 = <0.00011 Zinc mg/ 3 _ _ 5 <0.0044 Total cos mg 300 _ _ _ 188 348 180 180 Mercury mg/ _ 0.0005 0.002 _ <0.00015 Total Dissolved Solids _ mgf 600 1000 _- 500 197.3 187.7 240 Note: 2015 and 2016 data provided by CTE-RMPP Table 2. Hydrochemical Analyses of Source Mariani Waters. [page 59] 420 500 410 LI LI _ 2 450 Ë 400 = 0 = : , mn a Ë 300 ï = ZE 400 um rs, Ë . | ee 5 380 ." = Ë 350 js” æ = D = E 370 Ê * h 300 360 Li] 350 250 1 2 3 4 5 6 7 8 9 10 1 12 Ë À À € à & © A & à Month 5 8 8 OR 8 8 8 8 8 8 À À Figure 6. (left) Average monthly spring conductivity, (right) spring water conductivity time-series (data: CTE-RMPP). 20 30 18 " : 25 : 16 si é e ge — = 5 20 Ë 12 È ; = " Ê =: z dus u 3 10 E 15 = È 8 = : CR = = = Ë 6 Ë 10 Lei É 5 2 0 5 a + 5 æ 2 = à 2 © + L 2 3 4 5 ne 8 9 10 11 12 8 A A ë & & & ë & ë & & Figure 7. (left) Average monthly spring chloride, (right) spring water chloride time-series (data: CTE-RMPP). 60 30 500 a 450 50 25 RE] 400 _ a bee 350 à È # Ë 30 ÿ Ë 5 JS fe à 30 À £ 30 $ ef | 200 20 = Ë Re E Ÿ co 50 10 Be. an = ” ur + ÿ = 8 & £ 8 8 8 £ 5 ÿ Ê ñ ñ a ñ É 8 a ä S « + & = E es Es Dishcarge (Us) Turbidity (TU) +Turbidity mNitrate A Chloride _e Conductivity Figure 8. (left) Spring water turbidity vs. nitrate, (right) spring discharge vs conductivity, chloride, nitrate and turbidity (data: CTE-RMPP). [page 60] l Section 2.3 - Stable Isotope This possibility is coherent with an analysis of recharge temperature based on noble Stable isotopes of oxygen (5:20) andhydrogen gas sampling. Annual average temperatures (SD) were sampled to aid in evaluating in the catchment at 580 m elevation are in recharge dynamics. Gonfiantini and Simonot the range of 23° C to 25° C, which is slightly (1988) observed a linear trajectory of 510 above the average recharge temperature versus elevation from samples collected south Calculated for the CFC and SF6 analysis of of Port-au-Prince (Figure 9). They found that 22.7° C. Lower temperatures in the range of there is a -0.9 per mil change for every 1,000 20° Cto 23° C are found south of the probable meters of elevation gain for water points on "echarge catchment, in the Riviere Momance the plain, and estimated a slope of -1.4 per mil Watershed. This suggests that the Source change for every 1,000 meters of elevation for Mariani aquifer is connected to the greater the springs south of Port-au-Prince. Thislinear Massif de la Selle aquifer that supplies base trajectorÿ can be applied to Source Mariani, flow to the Riviere Momance andRiviere Froide. as it was one of the originally sampled points However, this does not suggest that Source in 1987 and has similar characteristics to the Mariani receives recharge from the Riviere springs in the area (Figure 10). Momance flows, which would likely occur in a warmer recharge temperature due to The stable isotope composition of the sample Warming of the surface flow. Given that both collected in April 2019 was 8!2O = -3.19 per Source Mariani and Riviere Momance derive mil and 8D = -14 per mil, while Gonfiantini and Q Substantial portion of base flow from the Simonot (1988) measured the groundwater Massif de la Selle aquifer, up gradient changes discharging from the spring in the 1980s and to the hydrology or recharge potential may their results were 80 = -3.21 per mil and 5D Alter flow rates in both waters. = -14 per mil. These results are essentially the same, since the analytical uncertainty in the measurement is +0.1 680, and +1 D. 3.0 : The aquifer water has essentially the same 4. . an 86 m5 perf isotopic composition over the past 32 years, -3.5 . a which implies that the aquifer is well mixed before it emerges from the spring. -4.0 . ° 50 ” ; Applying the #0 value to the -0.9 and -1.4 per 45 ° mil/1000-meter regression slopes, the likely minimum recharge elevation is 100 meters sole above sea level, and the average recharge : elevation is 580 m above sea level. This Le: Gonfientini and Simanot (1887) - Hat estimated average recharge elevation is near 0 400 800 1200 the high end of the elevations found within k the 15.2 km? recharge catchment, suggesting Elevation (meters) the possibility that some recharge would pe Figure 9 - Oxygen-18 versus elevation in groundwater derived from farther distances and at higher near Plaine du Cul-de-Sac, elevations in the Massif de la Selle aquifer. [page 61] 61#0 H20 -4,50 4.30 4,10 -3.90 -3.70 -3.50 -3.30 -3.10 -2.90 -2.70 -2.50 © Source Mariani 12 (Northwater, 2019) © Tunnel Diquini (Northwater, 2018) À -13 © Southwestern Karst Springs (Gonfiantini, 1988) + Riviere Momance 1 -15 (Northwater, 2019) + GW near Blanche (Gonfiantini, 1988) © 14-17 À + GW near Grise = (Gonfiantini, 1988) 4 8 à Average of CTE T Wells ] 40 A Average of CTE F Wells A Average of CTE D Wells [21 —— Linear (GMWL) 1 23 l 25 Figure 10 - Stable Isotope Data Plotted with Global Meteoric Water Line (GMWL). Section 2.4 - Groundwater Age elevation was set at 580 m and the excess air at 1.5 cc/Il based on the laboratory results. Chlorofluorocarbons (CFCs) and sulfur Only CFC-113 and the SF6 results were within hexafluoride (SF6) were collected as a means USeable limits. The CFC calculated recharge to age-datethe groundwater. Theseestimates age date was 1984, indicating the water is 35 are based on CFCs and SF6 accumulating in Years old. The calculated recharge date for air during the 20th century, measuring their the SF6 was 1998, indicating the water is 21 solubility in water and extrapolating back to Years old. It is not uncommon for these two the atmosphere (Appendix B). Both methods age-dating methods to reach slightly different require assumption or measurement of other results (the older the water, the greater the parameters. For CFCs the primary adjustable … discrepancy). Averaging these results, we can parameter is the recharge temperature, ASsume a groundwater age for the spring of which affects solubility of the CFCSs. For SF6 28 years. This result is similar to that measured the primary adjustable parameters are excess for Tunnel Diquini, which also flows from the air, recharge temperature, and recharge Same aquifer. Further sampling of CFC's and elevation. For both analyses, the recharge SF6 during higher flow and lower flow events temperature was set at 22.7 C based on Could further help illuminate the recharge the noble gas analysis. For SF6 the recharge dynamics of the aquifer. [page 62] l Section 2.5 - Aquifer Storage The GIS-based recharge model developed by Miner and Adamson was calibrated to With the available data, a planning level historical average annual base flow in the estimate of storage in the aquifer can be Rivieres Grise, Blanche and Momance that made. Recognizing that the isotope data flow from the Massif de la Selle aquifer to the from 1988 and 2019 indicate the aquifer is South and west. Average modeled recharge well mixed, meaning dispersion is high in the in the 15 km spring catchment was 15%of aquifer porous structure, then the annual @nnual precipitation, or 240 mm/year. This output of the spring multiplied by the age of is less than indicated by the chloride mass- the groundwater equates to a qualitative balance. The discrepancy could be due to storage estimate. Using the calculated the lack of rainfall chloride measurements, average annual discharge from 2008 to 2016 especially in the high elevation areas, since of 7.4 million m/year, this method estimates P'ecipitation varies greatly across the between 155 and 259 million m° of storage. Mecharge area. The GIS model does not This estimate is approximate: refinements are Account well for the anomalous events of expected with further analysis of discharge lainfall intensity and duration, such as during and sampling data and tracer tests. hurricanes and tropical storms. In any case, it is likely that the spatial extent of the spring recharge area ranges between 15 and 34 Section 2.6 - Groundwater Recharge km°. This range highlights the importance of additional monitoring and diagnostics to The groundwater recharge rates associated lefine the understanding of the spring, with the spring discharge were estimated using the chloride mass balance method and The chloride time series data allows for an a custom GIS-based direct recharge model analysis of potential recharge rate variations developed by Miner and Adamnson (2017). from year to year, and a comparison to Twenty-nine chloride measurements for the actual precipitation volumes. Over a two- spring waters (Table 3) and 12 rainfall chloride Year period from 2007 to 2008 there were samples from the Massif de la Selle (Table 4) Seven months with precipitation higher than were available for the mass balance. The 250 mm, while over a six-year period from average chloride values for rainfall and spring 2009 to 2018 there were only a total of five water are 5.5 mg/l and 14.4 mg/l respectively Months with precipitation over 250 mm. The for all available years. However, the average latter period corresponds to a decreased flow chloride of spring water during the high flow both at Source Mariani and at Tunnel Diquini. period from 2008 through 2009 was only Considering the average rainfall chloride value 11.2 mg/l, while the average for the normal Of 5.5 mg/l and the average spring chloride flow years between 2011 and 2016 was 16.7 of 11.2 mg/l (2008-2009), it is possible that mg/l. Based on these ranges of spring water annualrechargerates approach 50% of annual chloride, recharge rates may vary between precipitation in years of high rainfall volume 33% and 49% of annual precipitation, with an and intensity. The 38% average is largely average recharge rate of 38%. This equates influenced by the anomalous high years and to 600 mm of recharge from the 1,580 mm/ could be lower than 20% of precipitation in a year of rainfall in the recharge area. The typical year, as indicated by the GIS-based combination of the low and average recharge "echarge model. rate with the average annual spring discharge shows that the recharge area of the spring aquifer is likely between 12.3 and 15.2 kme£. [page 63] Chloride Chloride Date Date (mg/l) (mg/l) Jul-08 11.0 Mar-12 16.5 Aug-08 9.7 Jun-12 16.5 Sep-08 13.0 Jul-12 17.5 Oct-08 11.0 Jan-13 16.5 Nov-08 11.0 Feb-13 13.0 Dec-08 8.5 Jun-13 16.0 Jan-09 10.5 11/12/2013 18.5 Feb-09 9.5 Jan-14 16.0 Mar-09 11.0 10/6/2014 24.5 Apr-09 11.5 10/12/2014 16.5 Sep-09 13.5 Nov-14 12.5 Nov-09 14.0 3/23/2015 17.0 Jul-11 15.0 6/29/2015 18.0 Aug-11 12.0 8/61/2016 18.5 4/11/2019 19.4 Average (2008 - 2019) 14.4 Average (2008-2009) 11.2 Average (2011-2019) 16.7 Table 3. Groundwater Chloride Measurements for Source Mariani. Date Location Zone rs ad 8 a Fa © 10/4/2016 Laboulet? MassifdelaSele <1 18495 -72315 850 10/19/2016 Laboule? Massif de la Selle <1 18.495 -72.315 850 10/24/2016 Laboule? Massif de la Selle <1 18.495 -72.315 850 8/1/2018 Petionville Massif de la Selle 7.1 18.511 -72.290 380 8/2/2018 Petionville Massif de la Selle 13.6 18.511 -72.290 380 8/21/2018 Petionville Massif de la Selle 9.89 18.511 -72.290 380 9/28/2018 Thomassin 36 Massif de la Selle 1.78 18.482 -72.317 1025 10/2/2018 Thomassin 36 Massif de la Selle 0.71 18.482 -72.317 1025 10/6/2018 Thomassin 36 Massif de la Selle 1.24 18.482 -72.317 1025 01/04/19 Petionville Massif de la Selle 7.64 18.511 -72.290 380 02/04/19 Petionville Massif de la Selle 18.1 18.511 -72.290 380 03/22/19 Petionville Massif de la Selle 4.65 18.511 -72.290 380 Average 5.5 1 Sample taken during Hurricane Matthew 2 Assumes chloride values for Laboule and Clercine 12 are approximately 0.5 mg/l Table 4. Rainfall Chloride in Massif de la Selle. [page 64] l Section 3.0 - Discussion 2. Spring discharge displays mild seasonal variability with monthly average flows typically Based on the study, the key results are outlined "anging between 14,500 and 25,000 m°/d with and discussed below: an average of 19,500 m*/d. Spatial Distribution of Groundwater Recharge a. Instantaneous (daily) flows display greater variability, ranging from 7,600 to 1. The groundwater recharge area that 30,700 m°/d. contributes to the spring flow appears to be approximately 15 km2 but may be as large as b. Based on the spring catchment 34 km. infrastructure as observed in 2019, total spring flow is measured from a single water a. This uncertainty in the recharge Meter. Since this method does not account area is due to the large effect that annual for overflow, some high spring flows could be fluctuations in precipitation volume and Underreported,. intensity have on the recharge rates. 3. The spring flow is most vulnerable to b. The average recharge elevation is extended periods of average or below estimated at 580 m above meansealevelwith Average precipitation and consecutive years a corresponding temperature of 22.7 C. This without high intensity rainfall periods such as suggests the possibility that some spring flow tropical storms and hurricanes. may originate from distal zones in the regional . . . carbonate aquifer such as within the Riviere a. This recharge characteristic combined Momance basin. with the recent flow regression that extends from 2014 to 2019 may foster perceptions that the spring flow has been decreasing over the Groundwater Budget long-term or that acute impacts have taken place. 1.Thelong-term average annualrechargerate is estimated at 38% of annual precipitation. b. Limited historical data from between During high intensity rainfall periods, the 1925 and 1933 suggests that average recharge rates may approach 50%, while SPring discharge remains relatively stable or during normal or low precipitation periods the Perhaps has even increased due to increased recharge could be approximately 15%. precipitation intensity. 2. Aquifer storage relative to the spring is 4 The cyclic and multiannual recharge estimated between 155 and 259 million mi. characteristics of the regional carbonate aquifer are important for water managers 3. The limestone karst aquifer that feeds and planners to understand and utilize in the spring is well mixed and has an average balancing the different water sources of CTE- groundwater age of between 21 and 35 years RMPP. based on a single sampling event. Connection to Regional Flow Characteristics Groundwater and Surface Water 1. Recharge to the regional aquifer appears 1. Source Mariani flows from the regional to be largely affected by high-intensity and Massif de la Selle carbonate aquifer. high-volume rainfall events such as hurricanes . . . and tropical storms. There appears to be a a. The regional aquifer also supplies 3-to-7-year cycle of recharge trends partially Many of CTE-RMPP's major springs and influenced by El Niño and La Niña events. Tunnel Diquini and provides base flow to river systems. D © [page 65] b. Monitoring is required to better Section 4.0 - Recommendations for understand the complex hydraulic Continued Activities relationships between these major outlets of the aquifer. Due to Source Mariani's importance as the second largest single water supply to Port- c. Source Marianiis the lowest elevation au-Prince, additional efforts are warranted terrestrial outlet known for the aquifer. It to guide water source protection and appears to emanate from a topographic enhancement, water use planning and future exposure of the main aquifer lithology rather water supply development in the Massif de than as a contact spring. This may act to la Selle aquifer. Based on the findings of sustain flows even when higher elevation this study, recommendations fall into three springs exhibit reduced flows. categories that include: (i) improvements to ongoing monitoring efforts, (ii) water d. Source Mariani essentially serves as source protection and enhancement, and a drain for the western portion of the Massif (iii) options for additional study. Future de la Selle aquifer. activities would be greatly aided by increased availability of temporal datasets for climate, 2. Source Mariani does not appear to have a flow, hydrochemistry, and stable isotopes. significant hydraulic connection to the Riviere The recommendations by category, and the Momance or Riviere Froide. This is supported narrative provides supplemental detail. The by the isotope and tracer sampling, and recommendations provided closely parallel analysis of recharge catchment size. Tunnel Diquinis. À significant improvement in efficiency would be gained by combining these efforts. Aquifer Vulnerability A-StrengtheningOngoiïngMonitoring 1. Due to the high permeability and rapid Efforts infiltration rates typical in karst limestone environments, the spring waters have Historical monitoring data collected by high vulnerability to contamination. This CTE-RMPP provided insights into the spring is confirmed by the elevated nitrate levels dynamics. It is our understanding that current consistently measured in spring discharge. monitoring efforts, when performed, include: (i) recording the portion of monthly discharge that 2. Urbanization and land use changes in the is diverted to the pumping station, ii) recording hills south of the spring are considered the monthly pumping volumes up to the reservoir, greatest risk to groundwater quality and flow. and) periodic collection of water samples for The lack of centralized waste management physical and chemical analysis, often including and sanitation combined with the karst conductivity, salinity, PH, temperature, turbidity, hydrogeology significantly increases the risk hardness, alkalinity, calcium, magnesium, of direct contamination of the aquifer waters. chloride, sulfate, nitrate, nitrite, and iron. Many data gaps exist in the material provided to a. Land use planning, zoning, and our team by DINEPA and CTE-RMPP. There managed development of the area south of are certainly multiple ways that could further the spring in an area larger than the existing reinforce and improve monitoring efforts. spring protection perimeter is necessary in order to protect the spring water from future - We commend CTE-RMPP for the data that water quality and flow impacts. has been collected. Data that was particularly useful to this study included: flow rate, chloride, conductivity, turbidity, and nitrate measurements. We recommend that, at a minimum, these parameters continue to be measured monthly. [page 66] l measure overflow rates is recommended. - Since a comprehensive flow meter measuring Overflow measurements, when they occur, spring discharge already exists, an increase in could be added to the pumping rate data in monitoring frequency would be relatively easy order to estimate total spring flow rates. to implement. Daily recording of discharge at the meter between the spring and pumping - Discharge data collection and storage should station would allow for better understanding clearlÿy delineate the flow measurement type of recharge dynamics. as i) spring discharge to pumping station, ii) spring discharge overflow, iii) total spring o It appears that an electronic data logger discharge, andiv)pumped volume to reservoir. has been installed on the totalizing flow meter, but staff did not use this feature and did not - Table 5 provides recommendations for know how to operate it. Therefore, training measurement parameters and timing of and implementation of a scheduled read-off sampling. Monitoring should begin as soon as from data logger is recommended, to reduce possible andbe incorporatedinto a permanent the need for manual dailÿ recording on the monitoring program so that data is available meter face and reduce clerical errors. for water use planning or future studies. + Spring catchment overflow is not currently measured. Construction of a method to Parameter Daily Weekly Monthly Quarterl Yearly Notes ——— — ———— — ——— ——ñ# discharge to | x | | | | | Either read manually from totalizing flow meter or pumping | | | | | | downloaded periodically from data logger installed on meter. SRE LT PERS, RS SSSR RSS ERRSE RRERS S Spring | | | | | | Requires simple construction of weir in overflow canal and discharge | ? SR | | | | | installation of staff gauge to measure water height behind Total spring | x | | | | | Simple addition of discharge to pumping station and overflow discharge | | | | | | discharge. Pumped | x | | | | | Read manually from totalizing flow meter installed between volume | | | | | | pumping station and reservoir. a pH | | X } | | | Low-cost conductivity/PH field probe, calibrated as required. Temperature | | X | | | | | | | | | | Low level chloride analysis often required, suggest detection SO À ee Eee Turbidity | | ox | | | CTE-RMPP lab analysis. | | | Simple tests such asthe comparimentbagtestmaybe E. coli | | | X | | } sufficient and more economical than a full analysis at the Sn Hydrochemistr | | | | x | | CTE-RMPP lab analysis of Ca, Mg, Na, CI, K, CO3, HCO3, | | | | | | Recommend that a short memorandum that primarily Compilation | | | | | } contains a table of the measured results is published online and publication | | | | | X | sothat datais easily available. This is also a good interval of data | | | | | | to review issues with data collection and review the program i ( i i l }_to reduce data gaps. Table 5 - Spring Monitoring Program Recommendations. [page 67] B - Water Source Protection, development. Due to the karst nature of Enhancement and River Monitoring the aquifer, there is likely both diffuse and concentrated recharge. Potential study Water source protection and enhancement Methods include: planning - This study provides an improved . understanding of discharge dynamics and : À Monitoring program utiizing stable recharge areas for the spring that highlights isotopes 5*O and 8D along with chloride interim insights useful to guiding future steps Collected over a multi-year period to record to protect and enhance the spring. changes happening due to both drought and hurricane-induced recharge events. - Recharge protection areas can be delineated to protect critical areas of recharge and ‘ Dye tracer testing of karst dolines and the reduce contamination sourcesnearthe spring. _Momance and Froide River are perhaps the most definitive methods for delineating the - Land-use practices in the recharge area can lecharge area. Transit times, however, may be mapped and reviewed in order to guide be prohibitive. focused interventions such as forestation, terracing of steep slopes, and development ‘ Geologie mapping of the recharge exclusion zones. catchment to better understand the ways in which lithologic and structural changes + Sinkhole delineation can be performed affect the recharge, occurrence, and flow to locate areas of concentrated recharge. 9f groundwater to the spring. Such mapping Fences to keep wildlife andlivestock out could Would expand the coverage and detail of eliminate potential contamination sources. geological and hydrogeological mapping performed by Eptisa (2015). - The area immediately south of the spring protection zone is becoming increasingly : Hydrochemical, isotopic, and flow urbanized. Efforts at sanitation planning and Characterization of Source Ti Sous and Source infrastructure in that area would help protect Tet Sous below Source Mariani to determine the spring from biological contamination. their connection to the regional aquifer and relationship to Source Mariani - Regular schedules for spring catchment . . . inspection and cleaning can be implemented, … Refinement of spring recharge and discharge perhaps quarterly. At a minimum this includes dynamics - As previously mentioned, the the sediment and root removal from inside the legional aquifer system appears to be largely catchment and sediment removal from on top driven by large climatic events on 3-to-7-year ofit. cycles. - Installation of sediment barriers within the À coordinated effort should be made by CTE- catchment may help reduce the mobilization RMPP and DINEPA to locate historical flow rates of sediment into the supply line. for the spring. This will aid in the understanding of how andif the spring has been affected by changes in land use changes and climate. C - Further Hydrogeological Characterization - Careful analysis of these recharge and discharge trends may also help predict the Refinement of the spring recharge area future effects of climate change on the major and geology - This study indicates some carbonate spring that discharges to Port-au- uncertainty in the spatial extent of the spring Princes water supply. recharge. Refining the knowledge of this area is important to guide water source protection, land use planning, and future water supply [page 68] l Increased rainfall monitoring and and chloride is advisable. Potential localities for sampling such a program include Morne Boyer, Morne Chandelle, and Berot. This would provide - Daily data suitable for determining rainfall P'ecipitation data at various elevations within intensity within the spring catchment was the catchment and would be a complement not made available for this study. If such data to the rainfall monitoring in Degand and is located or collected, an analysis could Fermate that has been recommended to be performed comparing rainfall intensity Support Tunnel Diquini characterization over the periods of recorded flow rates (Northwater International and Rezodlo, 2018). to better understand recharge thresholds Such rainfall intensity measurements, coupled and the conditions conducive to diffuse or With meteoric hydrochemistry, will allow for concentrated recharge. better recharge estimates and be a valuable resource to all future hydrogeological studies - The implementation of a rainfall monitoring in the Massif de la Selle. program that also samples rainfall for 5t#O, 5D Financial Expertise Institutional Category Activity Costs Required Equipment Required Involvement | l pH/Conductivity field | î : Î probe, CTE-RMPP lab, CTE-RMPP A Spring Monitoring NS (see Table ! Low (annual) Technician low-level chloride probe, : technician and data Î Î installation of weir and | manager | | staff gauge | | | Moderate Hydrogeologist, | DINEPA, CTE- B Î Water Source Protection and | (planning) sentaron Minimal Î RAP ARNDR Î planner, inimal coordination, | Enhancement Planning | High landuse | community and local | | (implementation) planner leader support —_—— ———@" —"——_—— ———— _———— — ———————_————— | | Dye tracing materials and | CTE-RMPP | Refinement of Spring Recharge Area | High (one-time) Hroge get D pb ; technician, DINEPA | and Geology | 1 Geologist reconnaissance | spprovel of de QUO | equipment VOS | | Moderate or Low | (a | Refinement of Spring Recharge and | if sufficient Hydrologist / Minimal : CTE-RMPP data Î Discharge Dynamics ! monitoring has Hydrogeologist manager | | Moderate (one- Hydrologist Two ü jee ielemeute | CTE-RMPP, i Rainfall Monitoring and Sampling ime setup) TI setup, Weather stations witi MARNDR technician i Î ‘echnician simple sample collection d rt | | Low (annual) operation mechanism | and Suppo! Table 6. Phased recommendations and resource needs. [page 69] Section 5.0 - Conclusions Limitations of Investigation Thisstudy appliedlimitedanddiscrete datasets Aspects of the assessment were especially to derive an understanding of the spring limited by the Unavailability of data and hydrology andhydrogeology. The analysis and resources regarding the Massif de la Selle insights of this characterization were greatlÿ aquifer. The lack of consistent monitoring aided by previous efforts to characterize and records of discharge and precipitation the nearby Tunnel Diquini. We believe this made it especially challenging to quantify study is sufficient in characterizing the spring recharge rates and extent of the recharge on an interim basis from which to inform basin. À focused and basic level of analysis planning and decision-making with regards and synthesis was applied throughout the to guiding the sustainability and protection report with the primary objective to provide of Source Mariani. Disciplined monitoring initial insights into the spring dynamics and and the associated temporal datasets are recommendations for further study. important to advance the understanding and characterization of the spring and the Massif À limited amount of historical data was de la Selle regional aquifer that supports it. available to support this analysis, andthis study Using the data and findings in this study, the _ included a single sampling event that is just a potential exists for source protection and snapshot of a dynamic system. Conclusions enhancement programs in key zones of the in this report are preliminary and presented to spring's watershed. Additional studies could aid interim planning and decision-making, and also focus on better understanding and to guide any future study and characterization. characterizing the recharge catchment to the spring. This assessment was performed using the professional care and skill ordinarily exercised, The Massif de la Selle carbonate aquifer is under similar circumstances, by experienced arguably Haiïitismostimportant aquifer system, geologists and hydrogeologists practicing as it is responsible for providing a significant in this or similar locations with very limited proportion of water supply to Port-au-Prince sources of data and resources. Changes in fromits large springs withthe benefitof gravity, analysis andinterpretations can and will occur and its rivers supply the bulk of recharge to with the acquisition and analysis of new data, the Plaine du Cul-de-Sac aquifer. Further such as monitoring reports, water quality characterizing and understanding the aquifer data, and tracer and isotope data. Analysis as a whole would foster informed planning and interpretations presented in this report and operations into the future to protect and must be considered fluid and subject to review enhance these important resources. The and revision as additional data is compiled. results of this study and of Tunnel Diquinis Analysis and interpretations described in this provide important insights that can be usedto report may be invalidated wholly or partially guide such further studies. by the results of continued data collection and observations. [page 70] : REFERENCES Adarnson, JK, Jean-Baptiste, G., and Miner, W.J, 2016, Summary of groundwater resources in Haïti, in Wessel, GR, and Greenberg, JK, eds. Geoscience for the Public Good and Global Development: Toward a Sustainable Future: Geological Society of America Special Paper 520, p. 1-22, doi:10.1130/2016.2520(14). BME [Bureau des Mines et de l'Energie], 1993, Notice Explicative de la Carte Géologique d'Haïti: Port-au- Prince, Bureau des Mines et de l'Energie. BRGM [Bureau de Recherches Géologiques et Minières], 1988, La Synthèse géologique notamment dans ses parties stratigraphiques et tectoniques: Bureau des Mines et de l'Energie, Port-au-Prince, Haiti. BRGM [Bureau de Recherches Géologiques et Minières], 1989, Étude des ressources en eau de la région de Port-au-Prince. Butterlin J., 1960, Géologie générale et régionale de la République d'Haïti [General Regional Geology of the Republic of Haiti]: Institut des Hautes Etudes de l'Amérique Latine, 194 p. CERCSG [Centre d'Etudes et de Réalisations Cartographiques Géographiques], 1989, Carte Géologique de la République D'Haïti [Geologic map of the Republic of Haïti]: Bureau des Mines et de l'Energie, Port-au- Prince, scale 1:250,000, 1 sheet. CGIAR. (2007). Global Geospatial Potential EvapoTranspiration and Aridity Index, Methodology and Dataset Description. Available at: https://cgiarcsi.community/data/global-aridity-and-pet-database/. ARC/INFO Grid format, 30 arcs seconds. Cox, et al, 2011, Géologie de Port-au-Prince. 1:250,000 scale. Direction Generale des Travaux Publics [Republic of Haiti Public Works]. (1918 - 1938). Bulletin Hydrographique, Les Eaux de Surface de la Republique d'Haïti. Port-au-Prince, 16 bulletin reports in the series. Dôll, P. and Fiedler, K, 2008, Global-scale modeling of groundwater recharge: Hydrology and Earth Systems Sciences, no. 12, p. 863 - 885. Eptisa. 2015. Réalisation d'études hydrogéologiques sur la région métropoliaine de Port-au-Prince (RMPP), Carte Géologique. Gonfiantini and Simonot, 1989, Isotopic Investigation of Groundwater in the Cul-de-Sac Plain, Haïti. International Atomic Energy Agency, IAEA-SM-299/132, Pg 22. Hiimans, R.J, S.E. Cameron, JL. Parra, P.G. Jones and A. Jarvis, 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965-1978. Miner, W.J, and Adamson, J. (2017). Modeling the Spatial Distribution of Groundwater Recharge in Haïti using a GIS Approach, Geological Society of America 2017 Annual Meeting, Seattle, Washington, doi: 10.1130/abs/2017AM-297120. Moliere, E, and Boisson, D. 1993, Coupes Géologiques d'Haïti, in, Notice Explicative de la Carte Géologique : Port-au-Prince, Bureau des Mines et de l'Energie. Northwater International and Rezodlo. 2018. Hydrogeological Characterization of Tunnel Diquini: Port-au- Prince, Haïti, Inter-American Development Bank, Technical Report, HA-T1239-P001 Oxfam, 2014, Carte Geologique de Canaan, Jerusalem, Corail et Onanville (CROIX DES BOUQUETS, Haïti), Evaluation technique des menaces naturelles et vulnerabilite de la commune de Croix des Bouquets, Port- au-Prince. 1:10,000 scale. Polidori et al. 2014. Eleboration du Référentiel Hydrographique d'Haïti à Partir d'un MNT Aster. Pubellier, M. 2000, Plate boundary readjustment in oblique convergence: Example of the Neogene of Hispaniola, Greater Antilles. Tectonics, Vol. 19, No.4, p 630-648. Scanlon, BR, Healy,R. and Cook, P.G. 2002, Choosing appropriate techniques for quantifying groundwater recharge: Hydrogeolology Journal, no. 10, p. 18-39. Suez, 2013, Travaux prioritaires de renforcement de la production d'eau. Pg 178. Taylor, G.C. and Lemoine, R.C. 1949, Ground water in the Cul-de-Sac Plain, Haïti: US. Geological Survey Open-File Report, 59 p. UNDP T[United Nations Development Program], 1990, Carte Hydrogéologique République d'Haïti [Hydrogeologic Map of the Republic of Haiti]: United Nations Development Program, New York, scale 1:250,000, 1 sheet. United Nations, 1991, République d'Haïti: Programme des Nations Unies pour le Développement: Développement et Gestion des Ressources en Eau. [Government of Haïti and Organization of the United Nations, Department of Technical Cooperation for Development]: Report HAI/86/004, vol. 6. US. National Aeronautics and Space Administration. Hispaniola region, Landsat 8: spectral bands 1 through 7. Product L1T. Woodring, W.P. Brown JS, and Burbank, WS, 1924, Geology of the Republic of Haïti. Department of Public Works, Port-au-Prince, Haïti. World Health Organization, 2011, Guidelines for drinking-water quality, 4th edition: WHO, Geneva, Switzerland. Yeh, Hsin-Fu et al. (2014). Identifying Seasonal Groundwater Recharge Using Environmental Stable Isotopes. Water. 6, 2849-2861. Doi: 10.3390/w6102849. [page 71] _ __ __ © Plaine du Cul-de-Sac Groundwater Flow Model Department Ouest, Republic of Haïti Final Report Prepared for: Inter-American Development Bank & DINEPA Prepared by: Northwater International and Rezodlo S.A. [page 72] Table of Contents 1.0 - INTRODUCTION 69 2.0 - CONSTRUCTION OF THE MODEL 69 2.1 - MODEL AREA SETTING 69 2.2 - DATA 69 2.3 - METHODOLOGY 71 2.4 - EXAMPLE CROSS SECTIONS WITH HYDROSTRATIGRAPHY 76 3.0 - DEVELOPMENT OF THE NUMERICAL MODEL 76 3.1 - GENERAL APPROACH 76 3.2 - SELECTION OF MODEL CODE 76 3.3 - ASSUMPTIONS 77 3.4 - MODEL EXTENT 77 3.5 - MODEL DISCRETIZATION 77 3.6 - SELECTION OF LAYERS 79 3.7 - BOUNDARY CONDITIONS 79 3.8 - MODEL PARAMETERIZATION 86 4.0 - MODEL CALIBRATION 86 4.1 - GENERAL APPROACH 86 4.2 - CALIBRATION TARGETS 87 4,3- ADJUSTED PARAMETERS 89 4.4- CALIBRATION RESULTS 90 5.0 - MODEL RESULTS 91 5.1 - GROUNDWATER FLOW 91 5.2 - GROUNDWATER BUDGET 92 5.3 - GROUNDWATER STORAGE 93 6.0 - MODEL LIMITATIONS AND SENSITIVITY 93 7.0 - CONCLUSIONS AND CONSIDERATIONS 94 REFERENCES 95 [page 73] 1.0 - NTRODUCTION andits code has been extensively tested in various environments and conditions. ltis widely accepted, À numerical groundwater flow model was and the theory behind it is well documented, developed to better understand the hydraulic €9Sy to replicate, and can be applied to realistic characteristics of the Plaine du Cul-de-Sac (PCS) Conditions and adapted for future developments aquifer and support water supply development of the model. The ViewLog software from Earthfx planning for the Port-au-Prince metropolitan area. Inc. was applied to build the model. This software The model effort was preceded by data mining directly integrates with the borehole database andresearchthatresultedinaborehole database for building, developing and refining the model. that was important for the development of the Groundwater Vistas Advanced, version 6 was model. The database is presented in a separate Used to run the model simulations. report (Northwater International 2018). The modeling was also supported with recent data . and characterization of the aquifer (Northwater 2.0 - Construction of the Model International and Rezodlo 2017), . 2.1 - Model AreaSetting The primary goal of the modeling exercise was to . . . . support the Haïtian government in understanding: Îllustrates the geologic and physiographic setting i) the sustainable and renewable quantities of Of the study area with primary rivers, strearns, groundwater available from the aquifer: i) the Canals, lakes, and the regional drainage basins complex recharge dynamics: and ii) surface that are relevant to the Plaine du Cul-de-Sac water / groundwater interactions between lakes (PCS). The PCS groundwater model boundary and river systems. With a regional groundwater that is the primary focus of this report is outlined model, the water supply development planning inred. The main aim of this reportis to document can be informed and potential impacts of the groundwater flow modeling effort. Further development and climate change scenarios can details of the study area and aquifer can be found be considered. in the supporting literature. Upon completion of the hydrogeological 2.2-Data database, the modeling process included The data used to build the model was obtained the following: from geological and subsurface information . , compiled from previous reports and borehole logs 1. Construction of the geologic and conceptudl om various sources that were integrated into a models. database (Northwater International 2018). As detailed in the database report, the reliability and 2. Development of a steady-state groundwater integrity of datasets is variable, and professional flow model. judgement was important in terms of how to . : . insert data into the model. Table 1 outlines primary 3. Parameterization and calibration of the Sources of data used in the modeling process. groundwater flow model. The current conditions of the annual average , precipitationrates across the Plaine were modeled 4. Presentation of results. using the WorldClim Version 1.4 dataset produced . . . by Hijmans (2005). Precipitation was primarily 5. Running of model scenarios as guided by the jcorporatedinto the model through the recharge Inter-American Development Bank (IDB). boundary conditions. Figure 2 displays recorded : annual average precipitation at the Petion-Ville MODFLOW 2005 code was selected for modeling Hydrometeorological Unit of Haïti (UHM) station the PCS aquifer. MODFLOW is the United States from 1960 to 2016: the WorldClim Version 14 Geological Survey (USGS) three-dimensiondl dataset is based on climate normals from 1960 finite-difference groundwater model. MODFLOW _ & 1900, Figure 3 shows the model area and the is considered an international standard for data locations that supported its development simulating and predicting groundwater conditions, Gnd calibration, D © [page 74] * de, PF ee CT DR CR fn Sas = fi ME. es Le ; ET |. È PTS 2 TRE 277 A 1 Ve D = $ CL ? LL D ss SEULE Ÿ 8 PP NN LL) 5 4 TP LS COLA 70 L AA an EBres Aquifer Boundary | © — TK RE 02 CDDDIÉS .; >LIRÀ g dd £ £ YA 47 ES £ COR Pre ee à (Lu 5 asins È | Fe, \ TS Loi Ÿ . [Le Lake A Re St D ve CS Ë { es FN \ : à af rs z ; ss. se): y ETYS î a Ve. ne —— river & f ES X % — — stream ë Surficial Geology È Mc Pr Mm D Es D Ba z filos <5p | "De s N EN Qam MEL Elo : + ‘ = à NV Mio: Mis © Mc £ a —— or ——— 4 1:300,000 4 Northwater International 2019 è Figure 1. Geologic setting of the PCS Aquifer and the Basin. Annual Average Precipitation, Petion-Ville Station (UHM, 2018) 2.000 1.800 a È 1.600 E 1.400 E 5 F 1.200 8 à 1.000 800 e $ co a © = + co a © e = ce GS] © “3 © D D æ co co a a = e = = a EN a a a a a EN a a = e 2 a e — + — — et — — — et — [Sl [Si [a [al [a Figure 2 - Annual average precipitation at Petion-Ville Station, 1960-2016 (UHM, 2018). [page 75] [Data Sous [| Summy | 2013-2014: Well inspection, testing and rehabilitation of 17 municipal wells. Drilling of 17 monitoring piezometers, with lithology and water levels. SIND ! 2001 — 2017: Well lithology, well construction, pump test data, and water levels for the G wells. CTE-RMPP: Water quality monitoring data for select wells from 2006 — 2016. Northwater mn 72 private well drilling records that include lithology, well construction, International / Haiïti d stati ter levels. S Ils h test data and field wat Foratech / an M atic water levels. Some wells have pump test data and field water Geotechsol / Drilltech | 12 it Drilling records for 73 wells, include lithology, well construction, static water level, and airlift yield. Blue Ridge Missions Drilling records for 150+ additional wells are available but do not have coordinates available. These were not entered into the database. Haïitian American Well completion, lithology, water level, pump test records for 57 high Sugar Company capacity irrigation wells. Historical records provided by Foratech (HASCO) Environnement. EPTISA 2016 SIGES | Water quality data and water level measurements for several of the CTE- DATABASE RMPP production wells and monitoring piezometers. Living Water Locations of over 250 hand pump wells, but no data regarding well International construction, lithology, or water levels. Private Drilling a: : ar: . Select drilling records from private drilling companies that shared data. Table 1 - Summary of key data sources in the hydrogeological database. 2.3- Methodology The stratigraphic framework contains the units to be modeled, and describes their The geologic modeling approach was based lithology and stratigraphic relationship to on identifying and correlating stratigraphie Other units in the framework. The units in the contacts in subsurface data (borehole logs framework are derived from those contained and analyses of previous data) to produce unit in the surficial geology map, which also surfaces across the study area. Isopach, or lists their stratigraphic position and textural sediment thickness contours, were generated Characteristics. Using datasets illustrated in by subtracting adjacent surfaces. Figure 3, Figure 4, and the reliable borehole records as the basis for the framework, General Approach ensures that the model conforms to the most accurate sources of geological information. The first step was to gain an understanding of The position of each unit in the framework the various surficial geological units, as defined is Unique, that is, the order of the units in the on the geology map, as well as their relative three-dimensional model is inviolable. Each geometries. À stratigraphic framework was Unit also has defined characteristics (e.g, silty developed using the geological mapping, clay with sandbeds). This enables appropriate available literature, borehole records, and an aquifer property values to be assigned to each analysis of the geomorphology of the plain. unit. [page 76] EC». rs o re fete … Primary Recharge Zones — river . Le LL emieme È EE se red andior plezometry rimary Stream Infitration — stream > bd Ffitec SN # heihs 2 :—: PCS Aquifer Boundary M ou Steam fiat m ue D Fr 7 S ne TS Primary Routes 1 canal CN > / DST RE ENST | — nc LE FEES LR ee le PT QUE? + RTL © Ce - yes ea TA TI Lam D NE nee = 7 7 NE av _. € ! ° ? K LA ET PS ) can CE TE Ses D ; e © ID © © og pe, FER 00 F £ À Br ANT Ro Den | re È SR EICr | Repos SE © \, Lac EDR Fe SE Azuei * Æ e 4 È À É «| Fe Le Lacs PS he | LS E ; oué TER Ne RSS Vissere 2 2 Ë Baie de € g PTE Ve 4 è Port j ” 2e + % fl _ = fre Er NS & au n we? .... Le - | & * de es” jee 5 0 gun F 2e 1 À } À = Y aa FD, KES ©) e_* e n EN ° Te e,,)? Ne x % kgs: LATE AE Se FT Delmas j \ IN | Fr) Port-au-Prince ? ie 4 (fax ù\ » 2 à . & #4 AFIA . : ÿ) = i } w# SUR > te à TP Genthier è È à AN RER LIN 2 3 DR. Ë 2 DJ PER RAA FS ù RS “Le & PA KL EX À « NS Qe A W ‘ eu LE D PT ORNE SU PR ER A P'REURNRE | RS TS 4 ME ! “ n—+ A : AL Mia s Q if 71 %È See SP. sé as #sS >, jN h Le TA ° Figure 3 - Plaine du Cul-de-Sac Model Area. Table 2 contains the simplified stratigraphic or small-yield wells are installed in these framework developed for the study area shallower sandier aquifer units. Interestingly, based on a review of 140 borehole lithology many of the wells drilled by NGOs terminated logs in the database. It was clear through upon intersecting the upper section of the the analysis of lithology logs that the alluvial more productive aquifer layers (layer 3). This aquifer has complex stratigraphy with dozens provided a valuable indicator for delineating of layers of various strata with variable the depth to more productive aquifer layers extent, thicknesses, and properties. Bulk throughout the plain. Beneath the sandier characterization was necessary in order to aquifer unit (layer 2), there is a prevalence of develop a regional groundwater flow model sand and gravel beds with clay interbeds that for the aquifer. store and yield significantly more groundwater than their Upper counterparts. It is from Nearly all the borehole records had a layer this hydrostratigraphic unit (layer 3) that a of finer-grained soils of various thicknesses, maijority of the higher yielding wells produce whereas the upper layer acts as an aquitard groundwater. Since very few wells intersect and its texture is variable depending on its the entire thickness of alluvium, a fourth layer depositional setting. The soils are thicker and was established beneath layer 3 to represent finer-grained further from the Riviere Grise a few of the deeper wells and the unknown and Riviere Blancheinlets to the plain. Beneath strata, to differentiate the layer and provide the soil, many borehole records indicate a silty additional modeling and calibration flexibility. sand aquifer unit (layer 2) that stores and This layer was set to have the same hydraulic yields smaller quantities of water. Many of the properties as layer 3, and the two interact as shallow boreholes intended for hand pumps one unit for all intents and purposes. [page 77] Model Layer (from top to Stratigraphic Framework Unit Surficial Geology Unit Bottom) see development and fine-grained strata 2 | Sitysand "| Quaternary Alluvium | ané on an ca mme | Miocene, Oligocene, and limestone Note: Stratigraphic units listed from youngest at top to oldest at bottom. Table 2 - Generalized Framework. Once the stratigraphic framework located in as shown on the surficial geology was developed and combined with an map. In some borehole records, the sediment understanding of the geology and geometry of of the uppermost unit did not agree with what the various units, it was possible to start coding was expected based on the surficial geology units and stratigraphic contacts in subsurface map. In these cases, the upper few meters of data. Local experience and knowledge sediment in such wells were coded as the unit guided the interpretation of borehole logs, and they were located in. a few guidelines, as described below, helped constrain it, Discontinuities, either due to non-deposition or erosion, in borehole records were The data inputs Used to code sediment accounted for by picking the elevation of intervals in wells and boreholes consisted of the missing surface at the same elevation the texture and position of sediment intervals, of the stratigraphically adjacent underlying the geographic location of a well or borehole, (ie. older) surface. This ensures that the and the map unit in which a well or borehole complete stratigraphic sequence is captured was located in, as shown on the surficial for each borehole and that zero-thicknesses geology map. are calculated for stratigraphic units at appropriate locations and depths. Coded sediment intervals had to be compatible with the stratigraphic framework: Since very fewboreholesreachedthebedrock, in other words, the texture and position of a an aquifer-wide depth to bedrock analysis sediment interval from a well or borehole had was performed to support the modeling and to be compatible with the unit that it was storage estimates. This was done using the coded as. delineated limits of the alluvium, the locations of boreholes that intersected bedrock, and To ensure that the model conformed to the our understanding of the structural geology of surficial geology map, the uppermost unit of the plain. each well and borehole record was coded as the same unit in which the record was [page 78] Ÿ CSA | 17: PCS Aquifer Boundary — River [_] Lake Alluvial Thickness (meters) | à us, PSP 3 Primary Routes __ Stream __ Historical Route of Riviere High : 200 CV EE TELX LE. Cri 7 Blanche RTS | LE PNR Low: 0 £ Pr EE _ =. 7 IX PRES De Ds = es : - < ss FPS = _ 4 FE . calmes} Ci ere 0 si [ibert] Ré or TS À E à amet, CEE Se. \ “üd Fe à E ce Pr none} " d Baie de ru ù [sare] 4 ë Port | \s eu * @ Prince à ne ee À z D D. y dE. À U £ Port-au-Prince N 1 :."% fly Ke AI RNA 2 #4 É JE, ; Ze a ne DER. TRE». 8 ATPEES $ A NMANTE de a \; a) F)% ; «! d ; 's 4 AIS À Dre PRCRE AS TON D'EN T4 SD : : * 1:150,000$% NS CARE" RS LAN A: Û L'EAU LEE + où 1 2 3 4 D) Jmas sir a: LME NN ES Fe ie É À £ J CAS ROBE) at ywater International}20{9| Figure 4 - Estimated Alluvial Thickness of the PCS Aquifer. Generation of Cross Sections in the eastern and northeastern portion of the aquifer which limits the hydrogeological The ViewLog software from Earthfx Inc. Understanding of connectivity between the was used to generate cross-sections, pick Gduifer and Lac Azuei, Trou Caiman and Canal stratigraphic contacts, and interpolate Boucambrou. stratigraphic elevation data points to produce . ., surfaces for each unit in the stratigraphic Geological picking was performed along a framework. 17x11 grid as shown in Figure 6 to generate the model layers. Layer picking using local Two cross sections across the whole model Knowledge and borehole logs extended to area were first generated in areas where Mapped bedrock exposures on either side there was a better coverage of boreholes of the alluvial deposits. À boundary line was to compare the description of sediment drawn around the overburden deposits at the intervals in higher quality records. These ©Verburden/bedrock interface as displayed two cross sections are labeled A-A' and B-B° On the surficial geology map. This boundary and their locations are shown in Figure 5. Was used to constrain the interpolation of unit Most borehole records were available in the Surfaces to within the area when overburden northwest between Canal Boucambrou and deposits were mapped - ie, to prevent Riviere Batarde, and in the south around the deposits from being interpolated in areas Riviere Grise. This information was then used Where the bedrock is mapped at the surface. as a guide in interpreting well records in areas This was achieved by adding elevation control where no, or few, boreholes were located. Points along the boundary. At each point, the The most notable gap in lithology data was €levation of each unit in the framework was [page 79] assigned the elevation of the bedrock surface, sections with well or borehole records were that is, all elevation surfaces at each control used as a guideline to infer the location of point were merged. Therefore, zero-thickness stratigraphic contactsinthese barrensections. values are calculated for each unit at these control points. Once all cross-sections had been completed, elevation surfaces for each unit were Cross-sections were completed by identifying generated by kriging stratigraphic contact and correlating stratigraphic contacts on well elevation data points independently for records and boreholes as described above. À each surface. Isopachs were generated by number of cross-sections had only phantom subtracting the elevation surface of a unit from boreholes and did not have any boreholes its immediate underlying (ie. stratigraphically with logs. In such cases, adjacent completed older) neighbor. raw ro rev rieou rev rvgow rive row rreow raw Fe “SI L | 5 ARTE £ C2 XD Boreholes with lithology mx Primary Recharge Zones —— river NV SE |: us ref e) and/or piezometry mm Primary Stream Infilration — - stream D, hmg( $ 3 bei) !Z: PCS Aquifer Boundary Zones 1lLake À Vs 27.4 > j Primary Routes canal (@ NS 4F ts PTS ET TRES ne = er" VAE CI om © RG) QU Ve > - ne, DL Re, TH VF. LS S ÿ PT Jen NET, UC PERS TT" || B na ee AE Q ° D De Fos É à Ses © NAN g 4e | % . qe e © # F2 pa @ : “ ë D £ Baie des Q DÈMS" use î È Port eo © °ee Q au 4 w? © .e, C3 RE = ñ 4 =. ro! e Foix des Princes ps S RE 07 . LD'ECR . # PA B’ - CR EN NAN ere 2 ! É s e CET Ne © FAP ÿ C7 Delmas Ar, NS 8. EN 722 \ PS Port-au-Prince V4 ae . e © 4 ; / Dan PAIN Neo ° DÉPRE PP £ 1 xs LAC es — + 22," £ ” 8, 1 PN FG D. DA DEN Ë A D'ÉPEX ° MANENSE NS PAC ER TES ù A AUD { {7 er AN ON 77/6 7 UE Le MA, À £ 4 mao Cabtes ETRN N ADS KI N D) 2 « ges +, Bof rÉE j à ss SEX EN Na à ar Ù k à k \ $ Ë us — vs im 1m AR op, laps ae ‘Ye me ASS Dr ù D LR | Figure 5 - Geologic cross section locations using reliable borehole logs. [page 80] A B C D E F G H 1 J K L M N O P aQ b 9 [7 © © $ NT Lo 3 LS Rene > AR | | LA 90% | “| BOUT ere 2 © 8. 4 = Batardèy_ ES A — |" —" — : : \ Kviere | | à 54 6 ©,e e RC 17 RS M TE RE nn Mn PRE] M! L, È 4 à | EE $ = 8 | L | | O £ De mel logs }!° Re" 9 gl » | LL | Ke 87 à | | PEL b Le | 6. | | | F 16 n9\ | ° | o|° Q © b be o © on 8 | SEX | D nn | ° t ous 2 9 ! | | —# N No = 2 $ Li F1 10Y Streèms: | Il | | | | ae L l } Constant Hdad Boundary 2|Boreholes with Lithology —— No Flow Boundary [1 Model Domain - Active Area \ 112 Genorai Head Boundary CG Model Boundary : : : : 780000 785000 790000 795000 800000 | 805000 810000 UTM Easting Figure 6. Grid of sections applied to build 3D model. 2.4 - Example Cross Sections with described in the geologic model presented Hydrostratigraphy in previous sections. Pumping data were compiled from CTE-RMPP for recent years Figure 7 and 8 show the four bulk t° estimate the actual pumping rate for the hydrostratigraphic units that were defined Municipal production wells. Extraction rates along the two primary cross sections (Figure for other non-municipal wells were estimated 20) The figures show the locations of based on professional judgement and local boreholes with lithology that was within 3 km knowledge of the aquifer and are further of the cross-section line. discussed in Section 3.7. 3.0 - Development of the Numerical 3.2 - Selection of Model Code Model Based on the local hydrogeological setting 3.1 - General Approach and study objectives, the United States Geological Survey (USGS) finite-difference A numerical model representing the prevailing Model MODFLOW was selected to simulate hydrogeologic conditions within the study ÿroundwater flows in the study area. area was developed. The model is based on MODFLOW is capable of simulating three- previous studies and the latest available data dimensional groundwater flows in saturated to simulate the hydrogeologic processes that Porous media. It is a widely used and well are mostrepresentative ofthelocalconditions. tested code that can effectively simulate The model includes the overburden as both steadyÿ-state and transient groundwater flows of various degrees of complexity. The [page 81] open source, non-proprietary program has Flow of water under the ground is generally il a number of different graphical interfaces laminar unless large-aperture fractures or void available for pre and post processing. One of spaces are present. The total area covered the main advantages of the MODFLOW codeis by this study is small enough to consider a that it maintains mass balance in each model constant density of water. Water level and cell, and therefore allows reliable advective transmissivity data of the immediate vicinity of particle tracking. the study area suggest that the groundwater flow patterns are mainly controlled by large Earthfx ViewLog 4, ESRI ArcGIS, and scale heterogeneities in transmissivity and not Groundwater Vistas Advanced version by the horizontal anisotropy in the aquifers. 6 were used as pre and post processing Therefore, orientation of the model grid is tools. Viewlog's advanced capability of insignificant for the study. integrated borehole data management and interpretation was used to refine the local hydrostratigraphy. ViewLog can efficiently 3.4 - Model Extent create MODFLOW input files and was used in accurately assigning stream bed elevations The model extent is approximately 363 square and water levels to the stream segments using kilometers and was selected to correspond to the digital elevation model. Custom Visual the extent of overburden in the Plaine du Cul- Basic (VB) utilities were developed to assist de-Sac. Where possible, the domain has been in model preparation, including assigning extended to physical boundaries. For example, the streams to the appropriate model layer, the western boundary extends to the ocean, in order to ensure model layer continuity, while eastern boundary extends to Lac Azuei. assign hydraulic conductivity to each model The plain is rimmed to the north and south by cell based on layer pinch-out, and assign semi-consolidated or consolidated bedrock appropriate wetting factors. MS Access 2010 that forms a natural boundary. was used to store project data in a relational database and as an analysis and querying tool. 3.5 - Model Discretization In a numerical model, the conceptual model's 3.3 - Assumptions domain is replaced by a discretized model consisting of an array of cells. The size of the The basic assumptions of the MODFLOW code cells is critical in the design of the grid. The are as follows: discretization of the grid in the horizontal dimension is a function of the expected 1. Flow is laminar and Darcy's law is valid hydraulic gradient as well as the scale of data available for the model. 2. Density of fluid is constant The horizontal extent of the model domain 3. Medium of flow is saturated has been discretized with rectangular finite difference grids. Total coverage of the model 4. Principal direction of horizontal hydraulie grids is 629.18 km? (32.6 km X 19.3 km). A conductivity or transmissivity is parallel to the Uniform grid of 100 m X 100 m has been model axes. specified in the entire area (Figure 9). [page 82] NW Section A-A' SE d] L à [ } d ÿ e 4 ü : ’ L | [l Vertical Exaggeration: 20x EÆMUpper Silty Clay Aquitard Eusity Sand Aquitard EUpper Sandy Gravel Aquifer ELower Sandy Gravel Aquifer Ÿo 5000 10000 15000 2000 Section Distance (m Figure 7 - Geologic section A-A' showing generalized interpreted hydrostratigraphy. E Section B-B' W a 1 D, | Ê us L a f ü d Vertical Exaggeration: 40x BA Upper Silty Clay Aquitard EiSity Sand Aquitard EZ3Upper Sandy Gravel Aquifer [Lower Sandy Gravel Aquifer un 0 10000 20000 300€ Section Distance (m) Figure 8 - Geologic section B-B' showing generalized interpreted hydrostratigraphy. [page 83] UTM Easting Figure 9 - Groundwater flow model domain and grid. 3.6 - Selection of Layers + Layer 1: Upper Silty Clay Aquitard + Layer 2: Silty Sand Aquifer Model layers were selected based on bulk Layer 3: Upper Sandy Gravel Aquifer hydrostratigraphic units interpreted from ‘Layer 4: Lower Sandy Gravel Aquifer borehole logs, expert knowledge of the aquifer, and surficial geological maps. The ee hydrostratigraphic units applied in the 3.7 - Boundary Conditions modeling are based on the geologic model discussed in Section 2.3. À number of specified heads have been assigned along the natural model boundaries. À low permeability bedrock layer has been À specified head boundary reflects a situation chosen as the model bottom. In areas where Where the water table or potentiometric individual stratigraphic units were absent Surface is pre-specified in time. The model from the stratigraphic sequence (e.g. calculates the flux across this boundary discontinuities, non-conformities, pinch-outs), ASSUming a pre-specified value of head at that a minimum thickness of 0.5 m was used to location. A specified head boundary must be ensure layer continuity across the model placed sufficiently far from stress points so as domain, and the hydraulic conductivity was not to be a source of unreasonable flux. changed to match that of the underlying unit à found immediately below. These corrections Figure 10 shows the model boundary were applied across the entire model domain conditions. The following sub-sections discuss using Viewlog's equation processing utility. the selection of different boundary conditions. Following are the model layers from top to bottom: [page 84] o o © 8 PR 2 ou ° 0° g° 9, TT > Bouc os Ne Se è Riviere. 3 8 à Cr . To ui ue Se Grise, ° . o à É à © D PA 2 © $ dE CS vone ” œ £ T0 F2 ° 8 el d à CA é É ° E ° o © LG 6) 074 © œ À o ON, 5 ® © d à æ 2 : . KE S oo 8 8 Y Streams Ë gl Constant Head Boundary © Boreholes with Lithology®} [7 Model Domain - Active Akega — No Flow Boundary ——— General Head Boundary CC Model Boundary N \ 780000 785000 790000 795000 ‘800000 | 805000 810000 UTM Eastina Figure 10 - Groundwater flow model boundary conditions. 3.7.1 - Lateral Boundaries de la Selle karst limestone aquifer system. The general head boundary was assigned as As discussed in Section 3.4, the model extent Shownin Figure 10 with a 0.04 gradient. was selected to follow the delineated extent of . alluvium and definitive contacts with bedrock. 3.7.2 - Surface Boundaries Ideally, the specified head boundaries are applied to large natural features with known The model bottom works as a no flow/zero heads, such as ocean or large lakes. Specified flux boundary while recharge has been head boundaries have been selected along GSsigned as a constant flux boundary at each the ocean (0 masl) in the west, Lac Azuei (20 UPpermost active cell except surface water masl) in the east, and Trou Caiman (23.4 masl) features. in the north. The northern boundary coincides with the 3.7.3- Recharge overburden valley outline where the bedrock outcrops, and has been assigned no flow Groundwater recharge represents the boundaries. amount of water entering the top of the model and is one of the input parameters required During preliminary model runs, it was for the numerical simulations. Richards (2007) determined that a general head boundary discusses the difficulties associated with the needed to be defined in the southern aquifer "eliable estimation of infiltration, and argues limit to account for subsurface groundwater that, because of the non-linear recharge flow entering the aquifer from adjacent Massif response with time, ‘recharge cannot be [page 85] described by a simple direct relationship The Riviere Grise basin was one of the il to precipitation, since not all precipitation calibration areas for the Miner and Adamson produces recharge”. Rather, recharge is (2017) modeling, as it was one of the few a component of the water budget that is basins in Haiti with historical streamflow data typically derived from an array of measured to support calibration. Based on the analysis, and derived parameters. the recharge ranges from less than 5 to over 230 mm/year across the aquifer and provides Direct aerial recharge from precipitation was an input of approximately 15,000 m3/year derived from country-scale modeling (Miner into it (Figure 11). Increased recharge also and Adamson, 2017). The estimates were occurs along the edges of the alluvial valley at developed with custom spatial datasets that bedrock contacts where surficial runoff from include geologic permedability, slope, land the mountains enters the plain. In general, cover/vegetation (NDVI), drainage density, recharge decreases in the lower elevations and evapotranspiration. The Climate Normals due to reduced precipitation, increased (1961-1990) were applied to derive mean evapotranspiration, and less permeable soils. gridded estimates of the long-term average annual effective infiltration. GA es VW] := PCS Aquifer Boundary — River [_] Lake Effective infiltration (mm/year) [À NN Ve+yy 217 fes D] ; ne __. SN f ary Er Primary Routes - Stream _ _ Historical Route of Riviere æ High : 238 ERP SET RS canal ÉERERS f sn re pe 71 a To 2: e "1 CES ÉD RÉ Z 4 NE PERS 4 ire. cé bn PR PE LR FA UC 18 Lereboürs "RCE 7) e *. {isa Azuei | © EN SRE ES a: Drome é À È Baie de “gens Mes | à i en k a Port ie PF ES sé à SE x. * Ê & au ü ; BE ,. Prince, À mn] É _ ns » R ES Xe r f $ FE A NES 0 de Qi - \ : Ë TS 7 © pets 4 HE SN QI LE ï. Pa. LS À CL Qurr /Port:ä-Prince EST; LI =" e\ a F. + & SN PAIE - ANS Ne > PS) 2 20 1) AIN SR LD w DE - = —— : ; km. M APE ire ÿ De So pa À fÉ Re, NS Ps Se à S ! ae SALE Figure 11 - Groundwater recharge by infiltration. [page 86] l 3.7.4 - Rivers and Streams hydraulic conductivity of the bed materials etc.) of drains and streams are not available Streams have been defined with MODFLOW for every cell. However, several cross sections ‘river’ and "drain" packages. The flux at ariver and field observations were made available cell can be either discharging (water exiting from previous reconnaissance activities by the model domain and entering the surface the team. The 1.5m LIDAR elevation dataset water regime) or recharging (water entering Was valuable for interpolating cross sections the model domain). Depending on the along the length of the streams. surrounding groundwater conditions, which are calculated by the model, the river cells . will have a positive or negative flux. The flux 3.7.5 - Pumping Wells at a drain cell can only be discharging (water . exiting the model domain into the surface There NE PURES ne currently i i incorporate into the model, pumping water regime). When the water level in the approximately 75,000 ms/day (Figure 12 and model is below the elevation of the drain cell, Th 5 P ÿ 9, \ ÿ (Figure : there is no interaction between the drain cell able 3). Pumping wells were assigned to and the model domain, and consequently GPPropriate hydrostratigraphic layers based the flux is zero. Canal Boucambrou has been on their actual or inferred screen elevations. modelled with the drain package, while the Grise, Batarde, and Blanche rivers have been - There are 26 municipal CTE-RMPP production modeled with the river package. wells. Pumping rates were established based on recent data obtained upon request. The daily Actual data defining the properties (such as pumpng rates range from 4,230 m°/day (F2) to width,bedthickness, waterlevel, bedelevation, 0 m°/day (G wells, and new Canaan well). TN — on 61e) eamprou AU ON OCR CL Lee CAS . à, RRNEE, ee St el Q = à Riviere. ve es 468 à L Dé ° de a uns _. Ces CS 2 hu ee, EU : 5 lé Re 07 we © 8 FA Ds S RC CE La 6 Ë % Le ° eo" a © ORATS . o x QE Û A DR cu cs wa © . ŒUUE D 8 Streams : â F Constant Head Boundary _* PUMPINg Wells À [__] Model Domain - Active Akega —— No Flow Boundary —— General Head Boundary GŸ Model Boundary NX \ 780000 785000 790000 795000 800000 | 805000 810000 UTM Eastinq Figure 12 - Pumping Wells Incorporated into the Model. [page 87] - There was no pumping data available for il agricultural wells in the plain. For the purpose of initial model runs, a pumping rate of 250 m?/day was assigned to 37 agricultural wells spread across the plain for which borehole records were available. - There was no pumping data available for commercial and private wells. Pumping rates were assigned based on a World Bank survey of several large capacity truck filling stations in the plain and the reported pump sizes that were installed on private wells from well records. There are 50 commercial / private wells included. + Community wells refer to wells that are typically equipped with hand pumps or small submersible pumps; there are hundreds of such wells across the plain. We assigned a 15 m*/day pumping rate to 27 of these wells to account for this demand in initial model runs. [page 88] Elev Total Bottom Screen Screen Pump Well ID Type UTM X UTM Y (mas!) Depth Elevation Top Bot Rate (m) (masl) (masl) (masl) (m°/d) AG1 Agriculture 788641 2062246 20.3 98.80 -78.5 -30.4 -75.2 -250 CAG2 7 T Agriculture T 782556 T 2058919 T° 8.0 T 89.00 T -810 7-75 T -403 T -250 ] AG3 Agriculture 791870 2061715 25.8 97.50 -71.7 19.6 -68.8 -250 AG5 Agriculture 793847 2062038 25.4 91.40 -66.0 3.5 -63.2 -250 AG7 Agriculture 793655 2062118 24.9 156.00 -131.1 -101.1 -131.1 -250 AG9 Agriculture 781499 2055464 64 96.20 -89.8 -8.3 -41.3 -250 AG11 Agriculture 784898 2063668 9.8 97.50 -87.7 6.8 -83.7 -250 AG13 Agriculture 784401 2062680 11.1 173.70 -162.6 -93.0 -128.4 -250 AG15 Agriculture 781817 2060105 7.2 50.30 -43.1 -13.6 -57.2 -250 AG17 Agriculture 798091 2058111 34.3 110.60 -76.3 1.5 -73.8 -250 AG19 Agriculture 784784 2056937 20.0 TT.A40 -57.4 2.2 -53.6 -250 AG21 Agriculture 785752 2062410 13.6 103.50 -89.9 -5.3 -60.2 -250 AG23 Agriculture 786404 2062791 13.6 149.39 -135.8 1.0 -88.1 -250 AG25 Agriculture 789670 2062918 20.6 108.60 -88.0 -18.1 -TTA -250 CAG26 TT Agriculture T 793447 T 2061852 T 26.0 T 152.00 T -1260 1 -960 T -1260 T -250 ] AG27 Agriculture 793725 2062082 25.0 83.80 -58.8 -28.8 -58.8 -250 AG29 Agriculture 796274 2057910 43.3 102.10 -58.8 15.5 -54.6 -250 AG31 Agriculture 797541 2059296 29.0 121.90 -92.9 -63.3 -88.6 -250 AG33 Agriculture 791988 2062862 21.8 125.00 -103.2 -73.2 -103.2 -250 AG35 Agriculture 800265 2057732 42.0 102.40 -60.4 18.2 -57.4 -250 CAG36 TT Agriculture T 799607 T° 2057785 © T 40.4 T 105.20 T -648 1 98 T -608 T -250 ] AG37 Agriculture 798953 2057926 37.6 111.28 -73.7 0.0 -70.0 -250 [DT T Municipal T° 782636 T 2059170 T 9.0 T1 57.60 T -48.60 1 -2032 T -428 T -250 ] D2 Municipal 782221 2059502 7.0 58.00 -51.00 -19.60 -43.9 -1193 [D4 7 T Municipal T° 783376 T 2058750 T 110 T 100.00 T -89.00 T7 -1152 T -605 T -2800 ] D5 Municipal 783057 2058499 9.0 100.00 -91.00 -81.31 -88.4 -2401 F2 Municipal 787396 2056780 37.0 83.00 -46.00 -37.58 -46.0 -4230 F4 Municipal 788018 2056389 41.0 45.00 -4.00 16.93 -4.0 -3296 F6 Municipal 788848 2055817 50.0 62.00 -12.00 12.58 -7.8 -3371 G1 Municipal 794499 2051304 118.5 121.30 -2.8 72.1 -0.4 0 LG2 7 Municipal | 794943 T 72051429 7 1286 | 97.80 [308 1 828 1 326 T 0 ] G3 Municipal 795487 2051581 127.7 110.20 17.5 84.2 23.3 0 [G4 7 T Municipal 1 795926 T 2051680 T° 127.1 1 10490 T 222 1 845 1 396 T 0 7] G5 Municipal 796442 2051437 125.3 125.70 -0.4 55.7 0.6 0 LG6 7 Municipal ! 796546 [2051718 T 1216 | 130.00 [ -8.4 1 490 1-61 T 0 ] G7 Municipal 793248 2051196 113.2 118.00 -4.8 45.5 -3.8 0 T2 Municipal 790783 2054031 74.0 100.03 -26.03 -19.76 -25.7 -2324 T4 Municipal 791989 2052481 99.0 110.00 -11.00 60.17 7.9 -242 T6 Municipal 792781 2051939 11.1 96.22 -85.09 -33.67 -85.1 -2916 CT8 1 Municipal l 792321 [2053224 T 910 T1 10000 T -9.00 1 5973 T1 -60 | -1266 | Wo1 Private 783602 2057135.4 11.4 70.0 -58.6 -16.6 -56.6 -2422 CW02 7 T Private T° 784539 T 20571823 T 167 1 700 T -53.3 1 -113 T -513 T -4542 ] Wo3 Private 785694 2057400.9 22.7 70.0 47.3 -5.3 -45.3 -1696 Cwo4 7 T Private | 788462 T 20561234 T 459 T 700 T -241 T 179 OT -221 7 -407 ] Wo5 Private 790764 2057183.1 52.6 70.0 -17.4 37.6 -12.4 -1817 CWo06 TT Private T° 785446 T 20575011 T 20.8 T 60.0 T -39.2 1 5.8 T -342 T -814 ] Wo7 Private 792002 2053613.1 84.4 60.0 24.4 69.4 29.4 -163 CW08 7 T Private T° 789406 T° 2056108.6 T° 51.9 T7 60.0 T -8.1 7 36.9 T -3.1 T -182 ] [page 89] Elev Total Bottom Screen Screen Pump Well ID Type UTM X UTM Y (masi) Depth Elevation Top Bot Rate {m) (masl) (masl) (masl) (m‘/d) Wo09 Private 786004 20576000 22.5 60.0 -37.5 7.5 -32.5 -569 CW10 7 T Private T 784831 T° 20569823 T 20.2 T° 600 | -398 | 5.2 T -348 T -569 7] w11 Private 785370 2057621.8 19.7 60.0 -40.3 4.7 -35.3 -2023 CW12T Private T 786417 T° 2057669 T° 26.0 T° 60.0 T -340 T 110 T -290 T -569 7] W13 Private 790143 2057062.8 50.9 60.0 -9.1 35.9 -4.1 -569 CW14 7 T Private T 787369 T° 2057477 T 320 T 600 T1 -28.0 | 170 T -23.0 T -1453 ] W15 Private 790782 2057731 48.0 60.0 -12.0 33.0 -7.0 -2023 CW16 7 T Private T 788401 T° 2057556 1 37.0 T1 60.0 1 -23.0 | 220 OT -18.0 T -56 7] W17 Private 788604 2055455.3 50.3 60.0 -9.7 35.3 4.7 -569 CW18 7 T Private T 786206 T° 20575568 T° 24.2 T° 600 T -35.8 T 0.2 T -30.8 T -1453 ] W19 Private 785098 2057495.8 20.2 60.0 -39.8 5.2 -34.8 -569 CW20 7 T Private T 789132 T° 2057509 © T 42.0 T° 60.0 1 -18.0 | 270 T -130 T -136 7] W21 Private 789862 2057112 48.0 60.0 -12.0 33.0 -7.0 -136 CW2277T Private T 785115 T° 2058240 T 18.8 1 60.0 T -412 | 3.8 T -36.2 T -250 7] W23 Private 788608 2058080 37.5 60.0 -22.5 22.5 -17.5 -388 CC1 7 T Municipal T 786227 T° 20625289 T 14.2 T° 820 T -678 | 12 T -648 T 0 ] W35 Communiti 790974 2057252.8 52.2 67.1 -14.9 -2.7 -14.9 -15 CW36 7 T Community | 794224 T° 2058853.8 T 39.7 T° 549 1 -15.2 | -30 T -15.2 T -15 ] W37 Communiti 807614 2053859.6 58.4 54.9 3.6 15.8 3.6 -15 CW38 7 T Community | 796596 T° 2055789 T7 64.3 T7 54.9 T 9.4 [216 T 94 T -15 ] W39 Communit 796994 2057346.2 44.6 54.9 -10.3 1.9 -10.3 -15 CW40 7 T Community T° 787814 T° 2056650.7 T 39.6 T° 48.8 1 -9.1 7 [3.1 T 91 T -15 W41 Communiti 794779 2057090 56.1 48.8 TA 19.6 TA -15 CW427T Community T° 789292 T° 2063651.8 T 18.7 T° 29.0 T -10.2 | -41 T -10.2 T -15 ] W43 Communiti 789292 2063651.8 18.7 36.6 -17.8 -5.6 -17.8 -15 CW44 7 T Community T7 788124 T° 2064077 T 23.7 T 103.7 | -80.0 | 237 [237 T -15 ] W45 Communiti 788124 2064077 23.7 85.4 -61.7 23.7 23.7 -15 CW46 7 T Community T 796828 T° °2054463.4 T 04.6 T° 610 1 33.6 | 458 T 336 TT -15 ] W47 Communiti 787167 2064505.5 15.4 33.5 -18.1 -12.0 -18.1 -15 Cwa48 T Community | 789697 T° 20578979 T° 376 T° 549 T -172 [ -50 T -172 T -15 ] W49 Communiti 791299 2056703.9 56.3 54.9 14 13.6 1.4 -15 CW50 7 T Community | 794495 T° 2054980.9 T° 83.5 T7 48.8 T 347 | 469 T 347 T -15 ] W51 Communiti 792718 2053734.9 87.4 54.9 32.6 44.8 32.6 -15 CW52 7 T Community | 791833 T° 2056379.8 T° 61.0 T° 48.8 [122 | 244 [122 T -15 ] W53 Communiti 795174 2052111.3 119.2 61.0 58.2 70.4 58.2 -15 Cw54 7 T Community | 795174 T° 20521113 T° 119.2 T1 610 1 58.2 [704 TT 582 T -15 ] W55 Communiti 789266 2058445.1 39.8 54.9 -15.1 -2.9 -15.1 -15 CW56 T Community | 798675 T° 2051169 T° 112.3 T° 01.5 1 208 | 30.1 [269 T -15 ] W57 Communiti 794785 2056757.8 58.9 54.9 4.1 16.3 4.1 -15 CW58 7 T Community | 787196 T° 20626229 T 15.7 1 44.2 7 T -28.5 | -224 TT -285 T -15 ] W59 Communiti 801921 2053103.5 86.4 61.0 25.4 37.6 25.4 -15 CW60 T Community | 802135 T° 20528854 1 022 T° 549 1 373 | 495 TT 373 TT -15 ] W61 Communiti 784875 2062477 12.4 45.7 -33.4 -21.2 -33.4 -15 CW62 77 T Commercial [788300 T° °2059427.3 T 30.9 T° 610 T -30.1 | -17.9 T -30.1 T -500 7] W63 Private 787823 2056097 41.4 48.8 -7.4 4.8 -7.4 -50 CW64 7 T Private T 806280 T° 20514007 T 160.2 T° 115.9 1 443 | 626 T 443 T -50 ] W65 Private 805965 2051284.9 130.4 54.9 75.6 87.8 75.6 -50 CW66 T° Private T 706286 T° 20553411 1724 T° 67.1 [53 [17.5 [5.3 T -50 7] W67 Private 795432 2055881.7 68.2 64.0 4.2 14.8 4.2 -50 CW68 7 T Private T 708107 T° 20547064 T 628 T 915 T -286 | -164 T -286 T -50 ] W69 Private 790446 2064112.6 20.5 42.7 -22.2 -10.0 -22.2 -50 W71 Private 820631 2046429.8 139.9 97.6 42.3 54.5 42.3 -50 W73 Private 816627 2052013.4 23.5 68.6 -45.1 -39.0 -45.1 -50 W75 Private 796699 2055901.4 62.3 61.0 1.3 13.5 1.3 -50 W77 Private 789412 2055788.9 53.6 54.9 -1.3 10.9 -1.3 -50 W79 Private 789328 2054347.6 65.8 54.9 10.9 23.1 10.9 -50 CW80 7 T Private T 821151 T 20460025 T 854 T7 79.3 1 6.2 | 854 T 854 T -50 ] W81 Private 788257 2055328.2 49.7 67.1 -17.4 0.9 -11.3 -50 CW82 7 T Private T 796907 T° 20561263 T° 58.6 T° 54.9 1 3.8 T 160 T 3.8 T -50 7] W83 Private 787404 2055869.1 42.9 67.1 -24.2 -12.0 -24.2 -50 CW84 7 T Private T 821364 T° 20468853 T 78.2 T7 854 T -7.1 [51 [ -7.1 [ -50 7] W85 Private 789462 2059445.1 32.7 61.0 -28.3 -16.1 -28.3 -50 CW86 T Commercial [805732 T° 2052610.5 T 02.0 T7 73.2 7 1 189 [311 [189 TT -50 ] W87 Commercial 790121 2057793.6 46.4 61.0 -146 -2.4 -146 -50 CW89 T Agriculture T 799591 T° 2053399 T° 82.7 T 109.8 T -270 | -26 T -209 T -500 7] W90 Commercial __ 820704 2048425.3 26.6 64.0 -37.5 -25.3 -37.5 -500 Table 3 - Pumping wells incorporated into the model. [page 90] l 3.8 - Model Parameterization Calibration is the process of adjusting the model parameters within reasonable limits Initial hydraulic conductivity (K) values for the to obtain a good match between the model model were compiled from 35 pump tests results and the estimates derived from actual throughout the aquifer (Northwater 2018) Observations, and derived from specific capacity estimates and published literature (Freeze & Cherry To evaluate the model calibration, the resulting 1979) where necessary. heads generated from the simulations were compared to the water table measurements Hydraulic conductivity was adjusted as Gt different observation wells. There were 131 required during model calibration to improve Observation points throughout the model area results, as described in the following section. in hydrogeological database (Northwater AS previously mentioned, a general head 2018). The locations of these observation boundary was added to the southern aquifer Wells are shown in Figure 14. The targets limits during the calibration process as Were assigned to appropriate model layers hydraulic conductivity modifications were not based on the reported or inferred screen a practical solution given the ranges of values bottom elevations, À number of targets either from wells. did not have any screen or well construction information available. In these situations, a first attempt was made to assign layers based on 4.0 Model Calibration the bottom of the well. If the bottom elevation of the well was also not available, the target 4.1 - General Approach was assigned to the primary production layer of the aquifer. Ë s PL . Sert A7 QUE RA Trou Cadet © AR A" BOUCLE « LS o® CRE k sen, OP dre n ès, R 7 Ë DRE ou or —_ Se F Ë 2 D pi, È ss Q UE o" … ÿ 8 3 oc” œ se. a Ces 6 £ £ 7e sa 200 OT ; SE —_ o' \g gré 9006 o" È È © 5 07 2 co QE © ñ oc" er sf Dur 0° #|_ 2 © Le Ê CS ri] Fan 8 0° CR Le o) É à german os [Bu © _ FE 2 Rois Rigoles o de Coe (malbson À À o" ne OM an germnrenus nr o É * [el GS Jone Masseau > 8 ” TA Pare F1 Streams o Ÿ à é Constant Head Boundary © ©bServation Wells À — No Flow Boundary [7] Model Domain - Active Aka ——— General Head Boundary A Model Boundary NX \ 780000 785000 790000 795000 800000 | ‘805000 810000 UTM Easting Figure 13 - Observation wells used for calibration targets. [page 91] The Root Mean Square (RMS) error is an overall simulated and measured values to verify the il measure of the differences between values reasonableness of the resulting simulations. predicted using a model and the observed values. The values of the residual were plotted L : à el . at the location of each observation point and 4.2 - Calibration Targets compared with the contoured potentiometric surface. The cumulative probability of the Figure 13 illustrate the groundwater elevation residuals was also plotted to monitor the calibration targets that were assigned for the relative degree to which the simulation model calibration process. The elevations are matches the field data. While calibrating based on discrete measurements of static the model, groundwater flow directions, water levels in wells from various well testing water budget, and hydraulic head gradients or monitoring campaigns. were compared qualitatively between Well Screen Screen Target Well Name SM Ou ESS Depth Top Bottom Head {m) (m) (masl) {m) (masl) (masl) (masl) Bon Repos 3 788641 2062246 20.3 98.8 -30.4 75.2 18.2 [Menelas 177777777777 782556 | 2058919 [7 8 T7 89 T7 -7.5 [403 TT 62] Pasher 1 791870 2061715 258 975 19.6 68.8 25.2 LPasher 2777777777 T 791283 | 2062957 [216 1 063 T7 165 1 -712 TT 196 7 Dessources 4 193847 2062038 25.4 914 3.5 “63.2 245 CMoleard 2777 T 781738 | 2060895 T7 4.8 7 914 7-5 7 T -818 T 4277 Dessources 2 793655 2062118 249 156 23.7 Duvivier 781499 2055464 64 96.2 8.3 413 3 Sibert A3 784898 2063668 9.8 97.5 6.8 -83.7 77 Sibert A1 784401 2062680 111 173.7 -93 “128.4 8.7 Vaudreuil NRR 2 781817 2060105 7.2 50.3 “13.6 -57.2 48 Vaudreuil Hasco 4 798091 2058111 343 110.6 15 73.8 29.7 L'Etoile 2777777 T 784740 | 2056950 [19.5 1 104.8 T7 -14 TT -816 [143 77 Etoile 3 784784 2056937 20 TTA 22 -53.6 14.8 LSibert A4 T 783933 | 2063011 T7 8.47 7 201 [04 7 -1036 TT 59 77 Sibert B1 785752 2062410 13.6 103.5 53 60.2 8.1 LSibert A2 T 784680 | 2063226 [7 9.271 61 7 T7 7477 T7 443 TT 8377 Sibert B3 786404 2062791 13.6 149.39 1 -88.1 10.9 Bon Repos 2 789670 2062918 20.6 108.6 “18.1 14 FA [Dessources 177777777777 703447 T 2061852 [7 26 T7 182 TT TT 26 7] Dessources 3 793725 2062082 25 83.8 249 [La Moriniere 477777777777 T 794988 | 2057476 [53.7 1 111.2 T7 148 1 -436 T 537 7] La Serre No.2 196274 2057910 43.3 102.1 15.5 -54.6 43.3 [La Serre 237777777777 T 782648 | 2053828 [53.9 1 1021 [26.1 T7 -44 7 T 539 77 La Serre 2 4 797541 2059296 29 121.9 -63.3 88.6 29 Pasher 5 791988 — 2062862 218 125 21 Vaudreuil Hasco 1 800265 2057732 42 102.4 18.2 “514 42 [Vaudreuil Hasco 2777777771 709607 T 2057785 | 404 T 105.2 T7 9.8 1 -608 T7 335 7] Vaudreuil Hasco 3 798953 2057926 376 111.28 0 -70 376 DT TT 782636 | 2059170 [7 9 7 T 576 TT -20.32 T7 -428 TT 55 7] D2 782221 2059502 7 58 “19.6 43.9 57 (Da 7 783376 | 2058750 [7 4177 100 T -1182 T7 -60.5 TT 92 7] T8 792321 2053224 a 100 59.73 6 59.2 12 790783 2054031 74 100.03 “19.76 25.7 558 LE 791290 2053485 84 109 44.35 23.9 58.9 (Ds 7 T 783057 | 2058499 [7 9 7 7 100 T -8131 1 -884 T 89 77 F3 787619 2057552 36 111.8 -70.22 758 33.6 T6 792781 2051939 111 96.22 =33.67 -85.1 0 C2 7 787396 | 2056780 | 7 37 T7 83 7 T° -87.58 | 46 | 342] F7 787206 2057700 31 60 18.6 22.7 29.2 CT4 7 7 701989 | 2052481 T7 99 T7 110 T 60.17 T7 79 7 T 91477 F1 788385 2057349 39 110 5.17 68 38.6 F4 788018 2056389 ai 45 16.93 4 37.9 LG1 7 7 794499 | 2051304 T7 118.5 [121.8 T7 72.1 T7 -04 T 916 7 62 794943 2051429 128.6 97.8 828 326 04 3 T 795487 | 2051581 [1277 [110.2 T7 842 1 233 TT 949 77 Ga 795926 2051680 127.1 104.9 84.5 39.6 95.2 [page 92] Well Screen Screen Target Well Name pus pi (mes) Depth Top Bottom Head (m) (masl) (masl) (masl) CGs TT 796442 T 2051437 T° 1253 T 125.7 T° 55.7 TT 06 T 81] G6 796546 2051718 121.6 130 49 -6.1 71.8 Sibert 784890 2059729 14.9 11.13 CHatte Lathan 77 7 786595 T° 2064221 7103 77 74 7] Carrefour Shada 786418 2058752 25.3 18.5 CDespinos 7777777777 T 704282 T 2061187 [294 1 TT 158] Cesseles 791805 2059835 37.6 29.33 CSanto 2087777 7 790207 T 2057636 [481 [1 TT 3229 | Bellanton 794931 2057235 55.4 30.44 [Croix des Bouquets 7777 T 792694 T° 2056172 T° 644 7 TT 7 T 3231] Tapage 790796 2053856 75.6 64.2 [Trois Rigoles 777 T 702593 T 2053360 [83.2 7 1 TT 7818 ] Soisson 791265 2053295 83.5 80.57 CGreffin 7 7 792041 T° 2052291 | 96.477707 02 7] Galette Greffin 792692 2051667 106.8 103.15 Cri Moulin 7 7 708232 T 2051008 1 1188 [TT 115.49 ] Trou Caiman 799868 2064483 26.1 18.66 CChambrun 7 T 707182 T 2062255 [255 [TT TT 2111 | Lassere 8 798175 2059472 30 28 [Lassere 7 T 7e6901 T 2059080 1 324 7 TT 31] La Tremblay 3 797926 2054929 58.5 40.23 (La Tremblay 127777777777 798723 T 2054267 1 822 [1 TT 4926] Delmar 799469 2052075 94.6 81.72 Lone 7 T 799826 T 2050020 T 971 7 TT 7367 | Drouillard 801644 2057430 43.7 29.2 [Beauge T1 802835 | 2054304 [731 7 TT 6137 ] Ti Mache Campo 801821 2052641 92.6 71.35 CMasseau 777777777777 7 801292 T 2050881 1 109.9 77 TT 63.06] Placement 804834 2050183 160.7 104.45 CHatte Cadet 777 T 806068 T 2063306 1725 1 TT 154] Jouaneau 806723 2058757 33.9 28.23 CCotin 7 T 807006 T 2056311 1 424 1 TT 3616] Balan 809252 2054950 44.2 20.16 [Madame Beauge 771 807284 T 2053544 [54.17 7 TT 3724 7] Bonnet 805517 2053386 78.3 53.15 CCarrefour Beauge 7777777777 T 806598 T° 2052606 1 64.477707 TT 625 7] Pacharles 805510 2050240 153.5 108 946 7 781700 T 2058300 8.5 1 TT 25] 167 783200 2056400 10.2 8.5 a00 7 785700 1 2064600 [61 LT TT 7] 460 785600 2064500 6 5 708 7 786900 7 2064000 7122 7 TT 8 7] 489 785800 2061300 15.4 15 Ca76 7 786500 7 2059800 [204 7 1 TT 185 ] 466 789900 2059300 34.8 28 170 7 786600 T 2058900 257 [TT 215 | 223 786100 2056300 30.5 27 081 7 786700 T 2055300 1 48.7 7 TT 30 7] 464 792500 2064600 20.8 16.5 250 7 792700 T 2056200 [63.771 32 7] 468 790700 2053700 78.3 49 ag 7 792400 T 2058500 [789 [TT 51] 463 796500 2063100 24.9 25 329 7 798300 T 2061200 1 803 1 TT 265] 328 798262 2060400 29.2 28.5 39 7 795700 T 2058500 [394 7 TT 30] 422 796700 2057900 41.6 31 256 7 795400 7 2055370 [7520 7 TT 35 7] 355 797800 2055300 56.3 36 361 7 799000 T 2058700 [822 [TT 56] 307 798600 2053500 91.9 55 375 7 795200 T 2052700 [4487 [TT 71] 353 800400 2059400 36.3 30 382 7 801300 T 2057600 1 447 7 TT 32 7] 376 803700 2057300 424 36.5 366 T 802100 T 2056400 [85.2 7 TT 39 7] 362 803200 2054700 66.7 55.5 Cg64 7 7 808200 7 2054200 7753770 59 7] 320 801300 2054100 70.7 56 363 T 801800 T 2053600 [794 7 TT 585 ] 1021 801800 2052600 93.8 65 319 7 800400 7 2052400 [83.5 7 TT TT 66 ] 378 809800 2064000 24.2 23 877 7 805500 T 2062700 1 254 [TT 7 235] 374 805600 2052400 98.9 58 341 7 809400 T 2051700 999 7 TT 70] Table 4 - Model Calibration Targets. [page 93] 4.3- Adjusted Parameters established and applied to calibrate the model. These zones were selected based on the The following model parameters were Specific capacity and hydraulic conductivity adjusted from their initial values to calibrate data from over 45 locations in the aquifer that the model: were available in the database, combined with a geomorphic analysis of the plain. During the 1. Hydraulic conductivity of model layers: calibration process, hydraulic conductivities 2. River bed conductance: had to be adjusted lower in many zones to 3. Stagesinrivers, streams and canals. achieve model calibration. This is attributed to the fact that specific capacity and pump A uüniform value of 0.1 was assumed for the test data were mostly available for higher vertical anisotropy ratio (Kv/Kh) for all the Capacity production wells. Table 4 presents hydraulic conductivity zones. Figure 14 shows the calibration parameters, calibrated value the hydraulic conductivity zones that were and their range of variation. 8 RC SES : a : - Ë 2 Riviere 3 \ & à n Éo Grise >) ( à £pûs ( | 0 2 va \ Cù ‘ FE —\, Ai < 2" >) JE 2 \ \ \ \ AO | É gl € 8 Streams ——— General Head nd GŸ Model Boundary N \ 780000 785000 790000 TUE ñ 800000 | 805000 810000 asting Figure 14 - Hydraulic conductivity zones developed for calibration. LE [page 94] Zone Bulk Hydraulic Conductivity (m/d) | Layer) | Layer2 | Layer3 | Layer4 | Dot | | 0 | #æ® | w | [21 | 15 | 20 | 20 | [81 | 10 | 15 | 15 | [4] | 8 | 10 | 10 | [51 1 50 | 50 | 50 | [6e | | 2 | 5 | 5 [7 || 0.864 | 0.864 | 0.864 | [8 | | 0864 | 00864 | 00864 | [9 |" | 0.0864 | 0.0864 | 0.0864 [10 | 000864 | | | 1417 7 10 1 30 1 30 | Table 5 - Hydraulic conductivity values by zone. 4.4- Calibration Results Figure 15 presents a scatter plot showing the goodness of fit between the observed and simulated heads. The 45° line represents the perfect match between observed and simulated heads while a random distribution of the points around the line indicates that the simulated heads are not over or under predicted across the study area. The root mean squared error (RMSE) value is 12.04 and absolute residual mean is 8.43 m. Considering a large range of observation points and the high gradient of the aquifer, the calibration obtained is very good. As a usual calibration Obsened vs. Computed Target Values practice, root mean squared residual (12.04 . Le m) should be scaled by dividing it by the range : : 3 à Les of calibration points (115.5) to estimate a #0 : RSC scaled root mean squared value that can be TA evaluated for the goodness of calibration. In . 1: | our case, the scaled root mean squared value ë !. L h is 0.104 m. 3 j : 2 #0 n - re The correlation coefficient between observed ee ‘ and model heads is 0.92. While a better m0 Li | calibration is represented by a correlation Léle coefficient close to 1, Spitz and Moreno (1996) FE suggest that the correlation coefficient should lou A pa uA A un lie between 0.7 and 1.0 for a calibrated flow Observed Vale model. Figure 15 - Scatter Plot of Observed vs Simulated Heads. 0 [page 95] 5.0 - Model Results confidence in terms of the groundwater flow model and groundwater balance: 5.1 - Groundwater Flow | - The eastern zone of the aquifer, between the Figure 16 shows the calibrated potentiometrie Simulated N-$ oriented groundwater divide to surface map of the primary aquifer zone. The the boundary of Lac Azuei. groundwater flow shows similar trends to . what has beenillustrated in previous literature. _: The northeastern zone of the aquifer, near The hydraulic gradient is steepest in the the boundary with Trou Caiman, Canal southern limits of the aquifer where the Riviere Boucambrou and the bedrock contact in the Grise and Riviere Blanche enter the plain and northeast. recharge the aquifer. À groundwater divide . bisects the aquifer in the east-central portion The south zone of the aquifer, where, due to where groundwater flows either westward bedrock underlying alluvium, it was difficult to towards the ocean or north andeastwardinto interpret if observation wells were influenced Trou Caiman, Canal Boucambrou, and Lac by hydraulic headbs in the bedrock units Azuei. Due to the limited potentiometric andlithology data, several areas have a limited level of zzow row reteow row row ro regow reoow rev resov : PS RER SRE EE RE AT TT SE SN $ CEE at F7 AS AE y ERA, | | !T: PCS Aquifer Boundary = canal É Ë : ILES je F7) ee RS # a FF | — Simulated potentiometric lines (Layer 3) — river 2 Pl CRT "A? » 7}, 79 | © Wells used for calibration —-- stream ge LT RÉ RS A RÉ | M Primary Stream infiltration Zones [1 Lake > Res EE an % Le # = Primary Recharge Zones RES ANR RE st | - 7 A] © - CNE S / + me 45 ST LES LT ra rer D BE A + f L ne Dnene, Trou re PTT RSS ÿ | ‘ e side © d ons Le (re cainen, NES S : # = But € ge, se EE ue 4 4 a Aarebours La * Ê RC Enr : DE Se Repos e e Ÿ SS fe D; Î 21 é À = 2 © ne © 0, . 8 IST A à z RES Pres. LMoriniers @ Sie de ‘ a Baie der Drouillard % PS È en Se (5: / ë Port À À ° Je > VER } & au ñ É & NS Prince a s— RE Er Es x =" e eo s 5 ‘ z De SZ ENS | Ë ÿ f \E ROUES A9 HE, EN EN À FES Delmas * \ Ê > j A$) Port-au-Prince 224. fie RAS e + [IE Le 4 RUA - ; RTS Ê PR © En, ë HT PE ° TA VEN a j 42 retomité ! a NA Q SE 4 DEN 72 47 PE &e À A 4 AK, # SOPE re 2 %S ef 4 ; os LEA 6 DES Ar 7 AT re ere POP CE, (REC T ' >) A { dé 8 z Rs Er RS CS TRANS NS | Nea LA ' L ÿ f \ $ ë Ÿ 0 1 2 3 4 ju /Massit de "12 s PAU F2 NS à D FN à, LL pas ë Figure 16 - Simulated heads of the calibrated model. [page 96] l 5.2 - Groundwater Budget Based on the steady-state model and initial model run, the following observations are Based on the calibrated model output, noted regarding the water balance: the steady-state groundwater budget is . . presented in Table 6. The groundwater budget : Renewable recharge inputs to the aquifer for the model area can be expressed by the are on the order of 135,000 m°/day. If current following equation which outlines the inputs PUmping is 71,600 m?/day, this would imply a and outputs, 0.53, or 53% groundwater development ratio - 83% of the aquifer inputs are from infiltration = of the Riviere Grise and Riviere Blanche, which MODEL RON NTI is consistent with historical findings IN (md) OUT (mÿ/d) Riviere Blanche (Rsw) 16,253 - - Influx to the PCS alluvial aquifer from Massif de la Selle limestone aquifer is a moderately General Head (Rgh)__ 6,226 . important input (5% of total recharge simulated in the model). This input supported Nere : = calibration along the southern boundary of Trou Caiman (Dsw) = 3,890 - Canal Boucambrou appears to be a drain Î TTC from the PCS aquifer. This relationship needs Table 6. Groundwater budget simulation results. to be further researched and monitored. - Based on the steady state simulation and _ assumed pumping schemes, saltwater Rech*Rsw +Rsea=Dsw+Dsea*ABS intrusion does not appear to be a major factor at present for the primary aquifer layer. Dry Where: season stress periods may enhance the risk, and the shallow layer is most susceptible. Rech . ee . The coastal area of the aquifer has few wells; Groundwater recharge (directinfiltration). further, there was limited data to calibrate the model along the coast. Rsw Groundwater recharge from stream infiltration +04, Caiman appears to receive water from (river leakage). the PCS aquifer, at a range of 45 L/s (3,890 m*/day) according to what is suggested by Rsea model simulation. Seawater entering aquifer (saltwater intrusion). - Lac Azuei does not appear to receive a Dsw : significant proportion of its water budget Discharge to surface water bodies. from the PCS aquifer. The model simulation suggests a range of 21 L/s (1838 m°/day). It Dsea appearsinfluenced by streaminfiltration of the Discharge to the sea. Riviere Blanche. Groundwater that discharges to the eastern portion of Canal Boucambrou ABS . . may flow into Lac Azuei. This hydrological Discharge via well abstraction. relationship between the aquifer, Lac Azuei, and Canal Boucambrou needs to be further investigated with studies and monitoring. [page 97] 5.3 - Groundwater Storage l Total aquifer storage is estimated to be in the range of 6.32E+9 m3 (6.32 km), differentiated into three different model layers (Table 7). The estimates were based on layer volumes calculated from the model and a limited dataset of PCS aquifer storage properties from pump tests. Specific yield values from published literature (Morris and Johnson, 1967) were applied. Sitty Sand Aquifer Upper Foi nes Lower Fat es (L2)' (L3)'2 (L4)'2 average of 6.59E-4 Table 7 - Groundwater storage estimates. 6.0 - Model Limitations and Sensitivity - All the elevations measurements are derived . . . from the topographic surface of the 1.5m MTN Uncertainty Is a factor in any groundwater flow LiDAR digital elevation model and then averaged model, especially for regional models in areas #& Gbtain values for the grid cells in the model. with limited spatial andtemporal datasets. Errors This factor adds a certain level of uncertainty associated with model inputs can be associated nd variability in simulated head conditions. The with factors such as errors In measurement, lack of surveyed conditions of the study areato a scale, origin, data, and calculation. Because the specific datum also introduces a source of error, development of a conceptual and numerical as the observed heads and river dimensions model often relies on synthesizing and analyzing are largely based on coarse granular surface data from diverse sources and datasets, there jévations from the digital elevation model. are many opportunities for the modeling results to be affected by sources of error. - The model was built in a regional context; the , , , many and complex heterogeneities identified : Although the calibration process achieved fm borehole log analysis are not captured due the targets and resulted in a good RMSE and to the goal of simulating a regional system. This correlation between simulated and observed may result in differences between simulated heads, the residual error is not equally distributed and observed conditions as more local level over the whole model area. Simulated heads simulations are performed. The numerical had higher errors In some areas, largely due model was developed in a manner that supports to uncertain boundary conditions and the future refinement of the geological model when possibility that vertical and horizontal model more localized simulations are desiredl. boundaries and modeled conditions may not correspond to the aquifers natural physical | Error associated with the groundwater boundaries. . AS previously mentioned, the balance is always a factor in simulating flow eastern portion of the aquifer had limited data conditions, and it is important that the model for building and calibrating the model. considers uncertainty. Data used for recharge [page 98] l and streamflow infiltration were based on further steps of groundwater development observations and calculations from past and management in a regional context. The studies. The model was largely calibrated model is structured to support steady-state to simulate discrete measurements of simulations of various groundwater abstraction, groundwater elevations and to accommodate environmental, and climate change scenarios. characterized recharge dynamics of the river Its results suggest that renewable groundwater systems. Continuous streamflow monitoring of resources are on the order of 130,000 m°/day:; the rivers and wells, and monitoring of chloride these results further validate the importance concentrations in groundwater, surface water, of the Riviere Grise streamflow infiltration in the and precipitation will allow additional calibration recharge and groundwater flow dynamics of the targets and future refinements of the model. aquifer system. - À brief sensitivity analysis was performed to Although a volume of borehole and well data understand the uncertainty in the calibrated was available to develop and calibrate the model by the estimation of parameters, model, the quality and reliability of data is boundary conditions, and stressors. The spatially variable. Further, a limited quantity of purpose of such an analysis is to understand the time-series or temporal data was available for model response when parameters are varied. river/stream stages, and water levels in wells. Hydraulic conductivity, recharge, and riverbed The development of transient and stress period parameters (stage/conductance) were the models should be considered under the directive key parameters evaluated by multiplying each of afocused objective, and a specific data mining, parameter by various multipliers. The RMSE research, and monitoring campaign can be done changes for the variations indicate that the to support such transient model development model is most sensitive to: (i) river bed and flow and validation. stage of the Riviere Grise and Riviere Blanche, and (ii) hydraulic conductivity of the defined We recommend that scientific characterization aquifer units. The greatest sensitivity of the be performed in the northeast, east, and model appears from changes in the stage and/ southeast portions of the PCS aquifer to better or riverbed conductance of the Riviere Grise. understand lithology and the surface and groundwater interactions related to Lac Azuei, - The modeling effort is also sensitive to the Canal Boucambrou, and Trou Caiman. These will geological model, as it is a large factor in support model refinement and result in a greater defining the surface water and groundwater level of confidence for these zones of the aquifer. connections. Especially to better understand potential impacts on surface water bodies and saltwater intrusion - The model does not account for interflow vulnerability of the aquifer to the aquifer from the aquifer bottom or northern limits. This is considered a conservative Saltwater intrusion risk in the coastal areas should assumption in that if interflow does occur from also be further investigated with monitoring. these areas, recharge to the aquifer could be larger than modeled. The importance of temporal monitoring of water levels and water quality in wells should - The modeling is steady-state and does not be considered a priority to support future and accommodate either simulations of changes in advanced groundwater flow modeling and storage or transient conditions during extreme simulations. The temporal monitoring of flow climate events or stress periods. and stage in the Riviere Grise, Riviere Blanche, and Canal Boucambrou is also recommended considering their importance in the dynamics of 7.0 - Conclusions and Considerations the aquifer. Well pumping/abstraction estimates and monitoring is also a large data gap that could The steady-state groundwater flow model be addressed with future activities, as current presented serves as a good tool to support aquifer-wide abstraction was estimated based on limited data. [page 99] References Anderson, MP. and W.W. Woessner, 1992. Applied Groundwater Modeling, Simulation of Flow and Advective Transport, Academic Press, 381 pp. Freeze, R.A. and Cherry, J.A, 1979, Groundwater, Prentice-Hall Inc. 29 pp. Hijmans, R.J, SE. Cameron, JL. Parra, P.G. Jones and A. Jarvis, 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965-1978. Miner, W.J, and Adamson, J, 2017. Modeling the Spatial Distribution of Groundwater Recharge in Haïti using a GIS Approach, Geological Society of America 2017 Annual Meeting, Seattle, Washington, doi: 10.1130/abs/2017AM-297120. Morris, D.A. and Al. Johnson, 1967. Summary of hydrologic and physical properties of rock and soil materials as analyzed by the Hydrologic Laboratory of the US. Geological Survey, U.S. Geological Survey Water-Supply Paper 1839-D, 42p. Northwater International, 2018. Plaine du Cul-de-Sac Interim Hydrogeological Database, Version 1.0: Port-au-Prince, Haiti, Inter-American Development Bank. Northwater International and Rezodlo, 2017. An Evaluation of the Plaine du Cul-de-Sac aquifer and its potential to serve Canaan: Port-au-Prince, Haiti, United States Agency for International Development and American Red Cross, Technical Report, cooperative agreement no. AID- 521-A-15-00010, 40 p. Richards, P.A, 2007. The Importance of Accurate Hydrogeological Conceptualization - Are we Correct?, CGS/IAH Joint Conference Proceedings, Ottawa, ON, Oct. 2007, pp. 123-130. Spitz, K, and J. Moreno, 1996. A Practical Guide to Groundwater and Solute Transport Modeling, John Wiley & Sons, Inc., New York, NY. [page 100] REPORT APPENDICES [page 101] APPENDIX A - HYDROGEOLOGICAL INVESTIGATION OF TUNNEL DIQUINI: LABORATORY REPORTS Analysis Reports. ——_ — — Environmental à Laboratories, Inc. IL ELAP / NELAC Accreditation # 100292 ———— 1600 Shore Road + Naperville, Illinois 60563 » Phone (630) 778-1200 + Fax (630) 778-1233 Analytical Report Client: NORTHWATER CONSULTING Date Collected: 04/15/18 Project ID: Diquini and Cap Haitien Time Collected: 15:00 Sample ID: Tunel Diquini Date Received: 04/20/18 Sample No: 18-2208-002 Date Reported: 05/11/18 mo Analyte Result RL. Units Flags à Alkalinity, Total (CaCO3) Method: 2320B 1997 Analysis Date: 04/27/18 10:00 Alkalinity, Total (CaCO3) 230 5 mg/L Co Alkalinity, Bicarbonate (CaCO3) Method: 2320B 1997 Analysis Date: 04/27/18 10:00 Alkalinity, Bicarbonate (CaCO3) __< 5 5 mgL Ammonia (as N) Method: 350.1R2.0 Analysis Date: 05/03/18 Ammonia (as N) < 0.10 0.10 mg/L Chloride by IC Method: 300.0 Analysis Date: 05/03/18 Chloride 6.20 3.00 mgL NS Conductivity Method: 2510B 1997 Analysis Date: 04/27/18 9:00 Conductivity @ 25°C 382 5 umhos/cm Fluoride Method: 4500F,C 1997 Analysis Date: 04/30/18 11:00 Fluoride 0.27 0.10 mg/L Total Hardness, as CaCO3 Method: 2340B 1997 Analysis Date: 05/01/18 Total Hardness, as CaCO3 202 : 3 __. mgL Nitrite (as N) Method: 4500N02,B 2000 Analysis Date: 04/26/18 9:00 Nitrite (as N) : < 0.01 0.01 mgL H Nitrate (as N) Method: 353.2R2.0 Analysis Date: 04/30/18 Nitrate (as N) 1.95 0.10 mgL [page 102] a First — _— Environmental ÿ ; Laboratories, Inc. IL ELAP / NELAC Accreditation # 100292 = 1600 Shore Road + Naperville, Illinois 60563 + Phone (630) 778-1200 + Fax (630) 778-1233 Analytical Report Client: NORTHWATER CONSULTING Date Collected: 04/15/18 Project ID: Diquini and Cap Haitien Time Collected: 15:00 Sample ID: Tunel Diquini Date Received: 04/20/18 Sample No: 18-2208-002 Date Reported: 05/11/18 Analyte Result RL. Units Flags Sulfate Method: 375.2R2.0 Analysis Date: 04/30/18 Sulfate < 15 15 mgL TOC Method: 5310C 2000 Analysis Date: 05/03/18 TOC 0.4 0.2 mg/L Total Mercury Method: 7470A Analysis Date: 04/27/18 Mercury < 0.0005 0.0005 mgL Total Metals Method: 6010C Preparation Method 3010A Analysis Date: 05/01/18 Preparation Date: 04/30/18 Antimony < 0.006 0.006 mgL Arsenic < 0.010 0.010 mg/L Barium 0.116 0.005 mgL Beryllium < 0.004 0.004 mgL Cadmium < 0.005 0.005 mgL Calcium 74.4 0.5 mgL Chromium < 0.005 0.005 mg/L Copper < 0.005 0.005 mgL Iron < 0.05 0.05 mgL [page 103] Lead < 0.005 0.005 mgL Magnesium 4.1 0.5 mg/L Manganese < 0.005 0.005 mg/L Potassium < 0.5 0.5 mgL Silver < 0.005 0.005 mgL Sodium 3.1 0.5 mgL Thallium < 0.010 0.010 mgL Zinc < 0.010 0.01 mg/L Total Dissolved Solids Method: 2540C 1997 Analysis Date: 04/30/18 9:30 Total Dissolved Solids 214 10 mgL cFc11 | crc12 | crc Recharge ÎRechargel Elev. ea air [eq air CFCA1 ones LEE en fe) ce fr ep ER TT Lee [page 104] AIISOTECH ISOTECH LABORATORIES INC ANALYSIS REPORT Lab #: 663361 Job #: 38243 IS-90371 Co. Job#: Sample Name: Tunnel Diquini Co. Lab#: Company: Northwater Consulting APlWell: Container: 125ml bottle Field/Site Name: 18009 Location: Tunnel Diquini Formation/Depth: Sampling Point: Date Sampled: 4/15/2018 15:00 Date Received: 5/09/2018 Date Reported: 5/16/2018 ëD of water - -14.4 % relative to VSMOW à'#0 of water 3.28 % relative to VSMOW Tritium content of water--------- na ô'3C of DIC 7 na 14C content of DIC 7 na 8'5N of nitrate ee na 880 of nitrate um na 5#S of sulfate ee na 580 of sulfate = na Vacuum Distilled? * = No Remarks: nd = not detected. na = not analyzed. “Indicates if vacuum distillation was utilized for hydrogen and oxygen isotopic analysis of water [page 105] APPENDIX B - HYDROGEOLOGICAL INVESTIGATION OF TUNNEL DIQUINI: CFC l AND SF6 METHODOLOGIES Chlorofluorocarbons (CFCs) Chlorofluorocarbon (CFC) compounds have been synthesized on an industrial scale since 1931. They have primarily been used as refrigerants and aerosol can propellants, but also as foam blowing agents, solvents, and in insulation. Production reached its peak during the 1970s and 1980s before it was recognized that CFCSs contribute to the destruction of the Earth's ozone. Production was subsequently banned in the 1990s as part of a global agreement. Three principal CFC compounds were used during the 20th century: trichlorofluoromethane, dichlorodifluoromethane, and trichlorotrifluoroethane, whose trade names are CFC-11, CFC-12, and CFC-113, respectively. The CFCs production and release to the atmosphere have been measured and reconstructed back to 1940 (McCarthy et al, 1977; Gamilen et al, 1986; Wisegarver and Gammon, 1988; Fisher and Midgley, 1993; Fraser et al. 1996). Atmospheric CFC Concentrations since 1940 _ 600 …— & 500 & 400 ° CFC- 5 300 | ps LISE 8 Di CFC- L 100 É ' < 0 di - 1930 1950 1970 1990 2010 Year The basis for age-dating with dissolved CFC measurements in groundwater is based on comparing the measured values to those of the atmospheric concentrations at the time of recharge. This is accomplished by recognizing that the dissolved concentration Ci is Ci = Kai where KH is the Henry's constant and pi is the partial pressure of the CFC in air. The concentration is related back to atmospheric concentration through pi Pi = Xi(P — Pro) where xi is the dry air mole fraction of the CFC, P is the atmospheric pressure and PH20 is the water vapor pressure. Henry constants have been carefully measured for the three CFCSs of interest and solubility determined as a function of temperature and salinity. À number of comparative age-dating [page 106] l studies have shown the reliability of the CFC approach (Busenberg and Plummer, 1992; Busenberg and Plummer, 1993; Ekwurzel et al. 1994; Cook and Solomon, 1997), Sulfur Hexafluoride SFs SFe is used as electrical insulator in high-voltage switches and transformers. It is also used as a blanket gas in the production of magnesium metal. Production of SFs began in 1953, and ever since SFs is building up concentration the atmosphere. SFs has a lower solubility in water compared to the CFCs at 30 ppm. Solubility of SFs is temperature and salinity dependent. Solubility is also dependent on elevation and on any excess air in the water. Excess air origin is originated by from rapid recharge that traps air in the vadose zone and carries that air into the saturated zone where it solubilizes. If trapped in pockets of air formed in the aquifer space, SFs will readily partition into that trapped air because its low solubility. The basis of using SFe as an age dating tool relies on Henry's constant of SFs with respect to water. The Henry's constant for SFs is 0.00024 mols/kg-bar. The measured concentration in groundwater can be compared to the atmospheric concentration through the use of its Henry's constant, resulting in an age date. There are natural sources of SF associated hot springs and fumaroles. Sometimes these sources can interfere with age dating of groundwater. 1 SF À in the atmosphere } Les 4 & 80 ] LL 1 L + [e] 4 5 60 , £ © 1 o 4.0 - | 1 8 W 20 + n 4 : Li 0.0 1950 1970 1990 2010 Year [page 107] APPENDIX C - HYDROGEOLOGICAL INVESTIGATION OF TUNNEL DIQUINE l COMPILED DATASETS Compiled Discharge Data for Tunnel Diquini, Source Diquini and Riviere Froide RC RE ES Name Stage (cm) Flow (L/s) (m3/h) (m3/d) Data Source BR SRE DR ERREUR A PL I PA EE D Jam] men | | À ve Vue Los | 4/15/2018 15.5 cm 12.7 5,687 359 1,292 31,004 2018 EE Ce a RE RE SRE Er Cr BL PS RS PS RS RS CE Ce Ra RS RS ER RER a RSR RSR RE SR RES RUS EE Ce RES RSR SUR REUREUN | Cr D Er er RO RS ES RE SRE Ce BB SE SE RER Er Ce Ce Er D RS RE RE DR RE RUE EC BL RS PS LS RES RSR RSR RSR RES a CE Ce a nn Ce Ce Ban Cu PS LS D A 2 D BE [page 108] 2) ne [mem] [em l em | À | | sm | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source A A A D BL A BR PE PS ES RS ER BR D PS SN A A A LL UM Joue Que | ue Lee | mue | emaan 4 7,031 445 1,602 38,448 EPTISA (2016) A SL A PS D SR RSR A D I A D D D A A A I D A A A A PS A A A LL A LL A A A D Up) Joue ques | L'un | | emaame 3 14.8 6,905 437 1,573 37,757 EPTISA (2016) RS ER ER ER ES ER ERETS 3 22.4 10,475 663 2,387 57,283 EPTISA (2016) D A A A A A A A A A A A A A A D A D A [page 109] Te [mem] mm [em l mm | À | EE | mm | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source A A SP A A D A A A SP A A A A A PS A A A D A A A A A D A A A A A A BP D A D PS A A A A D D DL er A D D BE A A A SA A D A D D A A A D PS BL A A PS DS RS RE A D PS D Dr A D D PS DL BR D A SR BA A D D LR [page 110] HE] om [men mm [mm l en | PE | EE | sum | Name Flow (L/s) (m3/h) (m3/d) Data Source A D SR A A BR PS OS RES RS SR SR SRE A Re Se SR SRE BR RS NE BA Ro PS SR Re RS SN NE RO CE A D A D A RES RS RS SR NE ER A A D CE BA RE PS RO Re RS RS RO CE BR RS SR NRC CE BR RS SR NERO A D A A BL Ro SR ER A PL AA A ER A De RES RS SRE RE A BR RS SE SR SERRES A RSR RS SN RS ER RES RS RS SRE RO D [page 111] ne [meme [mem l em | Ph | | sum | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source A D D SR A RS ES SR US ER A PP SE De BR RS A A GE A PS A D SE BR PP A BR A PS A PS A A BR A A D A D A A D A A A A A PL A De Ba PS RER BR D PS ES Sn A A RS A PS RS BR D PS ES RS ER DL A GE BE D A D SG A A I D D [page 112] RS RTE Name Flow (L/s) (m3/h) (m3/d) Data Source AA D CL A D A A A SE RS EN BR RS NN NERO RS RS SENSUEL Da ee A A LA PE A D A Re PS CL A D A BB PS RS ER A A A RS RS SR ER AR D PP I AA PS SR BG RS ER RS A Da er Ro SE SR RE RR A A A A A A LE AA A D RS RS RS SR NE RO D [page 113] ee [meme femme] 5 | ds | eme | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source SP ES RS ER A BR PE SE A BAS D PS ES SR A A A M SE A A RO SR RSR A D SL A A A A PO A A A BR PE PS A A A PS A PS A A A A A BP D A D A A A BR D AR A ER A A A PS A A A A A A A A A D A A A D A A EL A A D [page 114] Le far eme [emm le] 5 | 6 | eme | Name Flow (L/s) (m3/h) (m3/d) Data Source A A A PP A Re el SR ER A A BASE ES RS SR SRE A D D A D Re RS SN RE RU A A A PS D D RES RS SN RE RU D BR RS SN ERR BR RS RS SRE ER A A D RS RS SERRE BR A A RES RS SN SERRE AA A D AR A A A A DE Re SO DS SR RS RS SE SR SERRES A D EN A A A SL RO A A SO A BA PS ER A A D [page 115] ee [meme [mm l en | 6 | ds | oem | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source A A A D A A A A A A A A A A D BL A A A A A A BR A A D PS A D PS A A D D D A A A PS SO A A A D A A PP A A A A A A A D A PB A A A A A ES A A A A A A PS A BR D PS ES RS ER BR D PS ES RS ER D [page 116] BR CRT ES Name Flow (L/s) (m3/h) (m3/d) Data Source A A BA Be DS D RE A Be A D A a RES RS SN RER RS RS RS SRE RO A A A Ro RS US ES RSR Ro RS SR RO RU ES RE Be PR US ES RS A D A PS SE A A A A A A A D A EP A A A D A SE RS SR SE SEE A A A A A AA A D [page 117] om fume eme [mm] em] 5 | ds | eme | Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source A D D D A A M A D D A D A A SR BAS Re PS SR D A A D A BA Po SR A A D RS M RSR A A A A A Po D A A A BR PS A A A PS BE A A A BR A A A A A A A A A I D DE BAS Po PS ES RE ER A D A A D D LE D D A A D [page 118] Lee fan eme [ml | 5 | 6 | eme | Name Flow (L/s) (m3/h) (m3/d) Data Source RS RS SR NERO BU ES RE A D RSS RS RE SR BA RE PS RS A A AS A PS A A AA DS ES A A A RS RS RS RER A BP CS SR RO A A AA Re A CS A CS A A A A BC ER RS PS RER AA Re ER Ce A D D RS SR RER A D A A SP D Le RS RS SR SR NRC A A D D A D D A [page 119] Water Point Flow Flow Name Stage (cm) Flow (cfs) | Flow (gpm) | Flow (L/s) (m3/h) (m3/d) Data Source Tunnel Diquini 111987 |__| | 7,246 1,651 39,623 CTE-RMPP Tunnel Diquini 10/1981 |__| | 7,892 | 50 1,798 43,157 CTE-RMPP Tunnel Diquini 9/1981 |__| "6. 6,913 1,575 37,800 CTE-RMPP Tunnel Diquini 8/1981 |__| "6. 6,913 1,575 37,800 CTE-RMPP Tunnel Diquini 71981 |__| "6. 6,913 1,575 37,800 CTE-RMPP Tunnel Diquini 671981 |__| | 6,913 1,575 37,800 CTE-RMPP Tunnel Diquini 5/1981 | | "9 | 7,892 | 0 1,798 43,157 CTE-RMPP Tunnel Diquini 47981 [| 7. 6431 1,465 35,165 CTE-RMPP Tunnel Diquini 3/1981 |__| 7 | 6,431 1,465 35,165 CTE-RMPP Tunnel Diquini 271981 |__| 7 | 6,431 1,465 35,165 CTE-RMPP Tunnel Diquini 11981 |__| "7 | 6,431 1,465 35,165 CTE-RMPP Tunnel Diquini 121980 | | "7 | 5,961 1,358 32,599 CTE-RMPP Tunnel Diquini 111980 TT | 6,913 1,575 37,800 CTE-RMPP Tunnel Diquini 10/1980 |__| | 5,961 1,358 32,599 CTE-RMPP . |“. Fe 7 MT ulletins PT |. T° TT Fe Bulletins M | *#. " Fe Bulletins a Hydrographic .. |. TT TU PTT |“. . TP Bulletins De |". 7 si PT Bulletins PT | #. " Fe Bulletins PU | %. nu 7 Te Bulletins a Hydrographic _ |. . . nu . Source Diquini | 4/6/1926 761 48 173 4,162 Hydrographie ource Diquini , Bulletins os | # . . 6 Bulletins . | #. . ulletins PT je) +) #i ulletins .. |? FT |“ |. . ulletins [page 120] Water Point Flow Flow Name Stage (cm) Flow (L/s) (m3/h) (m3/d) Data Source … | #. …. . ulletins 2. | #. n 3 Bulletins .…. RIRES ulletins . _|"|#|#/)"*] Ce [page 121] Compiled Water Quality Data for Tunnel Diquini and Source Diquini ivi idil ; Nitrate Water Point Name Temperature C Conductivity | Turbidity | Chloride (mg/L as Data Source (uS/cm) (NTU) (mg/L) NO3) Tunnel Diquini 6/25/2015 7.5 24.7 641 EPTISA (2016) Tunnel Diquini 1/16/2015 7.2 23.6 367 EPTISA (2016) Tunnel Diquini 12/12/2014 7.1 24 381 EPTISA (2016) Tunnel Diquini 11/5/2014 EPTISA (2016) Tunnel Diquini 9/25/2014 7.3 25 413 EPTISA (2016) Tunnel Diquini 8/29/2014 24 390 EPTISA (2016) Tunnel Diquini 7/22/2014 79 25 415 EPTISA (2016) Tunnel Diquini 6/30/2014 25 394 EPTISA (2016) Tunnel Diquini 5/20/2014 25 402 EPTISA (2016) Tunnel Diquini 4/11/2014 8.7 24 413 EPTISA (2016) Tunnel Diquini 2/27/2014 TA 25 407 EPTISA (2016) Tunnel Diquini 1/27/2014 8.2 24 408 EPTISA (2016) Tunnel Diquini 11/14/2013 79 24 413 EPTISA (2016) [page 122] ELEC EEE NO3) A D D il Ben En SC PE El Bas RE RO El A D D il Creer ea ef em | Den a SC D D El Cesar fe fe fe fe amer | A D D il A A A D D A SP A D D A PL A D A A A D PP A a Bon RE RO RS El A D D A A A D A EE il a A PP A PE Ce me en lee | A RE D ae Te er ee er] [page 123] ELLES E NO3) Ce ee | AP D RE Ce ee fe fn | Ce eme ee fe fr | AP PR RE Ben RU PR D RE El A A SP El A PE D El A D A PR El A D D Ben PE D El PE D D RE RE n D PR PP El A D D il D PS PP El A A PR A Ce ee A D D D RC PE RS El A El Ce ee Ban RC El A D Ce ee D il Bas on EN SR RE RE Ce ee A D D D SSSR: (2016) RSS RER E (2016) [page 124] LL essE 5 (2016) SERBE SENE.E (2016) a, | (2016) BE ERR:E (2016) tt | | (2016) tt (2016) A ES A El A A D RS BR RS D RE BREL A PB D A D BR A A A ES D A DS BE SL DS BE ES BE A ES RE D D ES A D A SR A SE A D SE A D A SR EL A A SE A El [page 125] APPENDIX D - HYDROGEOLOGICAL INVESTIGATION OF SOURCE MARIANI: l : LABORATORY REPORTS Analysis Reports www.encolabs.com Analvte Results Haa MDL POL Units Method Notes Barium - Total 0.174 0.00110 0.0100 mg/L EPA 200.7 Calcium - Total 732 0.0390 0.100 mg/L EPA 200.7 Calcium Hardness 180 0.018 0.0 mg/L SM 2340B-2011 Chloride 9.7 19 5.0 mg/L SM 450001 E-2011 Copper - Total 0.00518 J 0.00160 0.0100 mg/L EPA 200.7 Huoride 0.40 0.0097 0.20 mg/L EPA 300.0 Q-01 Iron - Total 0.0520 0.0220 0.0500 mg/L EPA 200.7 Magnesium - Total 5.39 0.0290 0.100 mg/L EPA 200.7 Manganese - Total 0.00404 1 0.00150 0.0100 mg/L EPA 200.7 Nitrate as N 19 0.041 0.10 mg/L EPA 353.2 Nitrate/Nitrite as N 19 0.041 0.10 mg/L EPA 353.2 Nitrite as N 0.031 1 0.017 0.10 mg/L EPA 353.2 Q-02 pH 7.8 LO LO pH SM 4500H+B-2011 Q-01 Potassium - Total 0.796 0.150 0.500 mg/L EPA 200.7 Silica (SiO2) - Total 20.2 0.0270 0.214 mg/L EPA 200.7 Sodium - Total 5.91 0.400 0.500 mg/L EPA 200.7 Specific Conductance (EC) at 25 Deg € 400 10 10 umhos/cm SM 2508-2011 Sulfate as SO4 59 2.9 5.0 mg/L EPA 300.0 Q-01 Temperature for pH (deg. C) 20 pH SM 4500H+B-2011 Q-01 Total Alkalinity as CaCO3 190 14 15 mg/L EPA 3102 Q-02 Total Dissolved Solids 240 50 5û mg/L SM 2540C-2011 Q-02 Total Organic Carbon 34 0.34 10 mg/L SM 5310B-2011 www.encolabs.com Description: Source Marlani Lab Sample ID:CC06218-01 Received: 04/19/19 11:00 Matrix: Surface Water Sampled:04/02/19 13:00 ‘Work Order: CC06218 Project: Source Mariani Sampled By: Javan Miner/Maxwell Pierril À - ENCO Cary certified analte [NC 591] Analyte [CAS Number] Results Flag Units DE MDL PQL Batch Method Analyzed By Notes Antimony [7440-36-0] 0.00037 U mg/L 1 0.00037 0.00100 9D22019 EPA 200.8 04/26/19 11:32 CMK Arsenic [7440-38-2]° 0.00760 U mg/L 1 0.00760 0.0100 9D25010 EPA 200.7 04/27/19 11:16 JDH Barium [7440-39-3]% 0.174 mg/L 1 000110 0.0100 9D25010 EPA 200.7 04/27/19 11:16 JDH Beryllium [7440-41-71* 0.000160 U mg/L 1 0.000160 0.00100 9D25010 EPA 200.7 04/27/19 11:16 JDH Cadmium [7440-43-9]* 0.000360 U mg/l 1 0.000360 0.00100 9D25010 EPA 200.7 04/27/19 11:16 JDH Calcium [7440-70-2]° 73.2 mg/L 1 0.0390 0.100 9D25010 EPA 200.7 04/27/19 11:16 JDH Chromium [7440-47-3]* 0.00140 U mg/L 1 0.00140 0.0100 9025010 EPA 200.7 04/27/19 11:16 JDH Copper [7440-50-81] 0.00518 J mg/L 1 0.00160 0.0100 9D25010 EPA 200.7 04/27/19 11:16 JDH Iron [7439-89-6]* 0.0520 mg/L 1 00220 0.0500 9025010 EPA 200.7 04/27/19 11:16 JDH Lead [7439-92-1]° 0.00310 U mg/L 1 0.00310 0.0100 9025010 EPA 200.7 04/27/19 11:16 JDH Magnesium [7439-95-4]* 5.39 mg/L 1 0.0290 0.100 9D25010 EPA 200.7 04/27/19 11:16 JDH Manganese [7439-96-5]* 0.00404 J mg/L 1 0.00150 0.0100 9D25010 EPA 200.7 04/27/19 11:16 JDH Mercury [7439-97-6]° 0.000150 U mg/L 1 0.000150 0.000200 9D25029 EPA 245.1 04/26/19 12:09 RLF Potassium [7440-09-7]* 0.796 mg/l 1 0.150 0.500 9D25010 EPA 200.7 04/27/19 11:16 JDH [page 126] Silica (Si02) [763-18-69]* 202 mg/L 1 00270 O214 OD25010 EPA2007 O04/27/1911:16 JDH Sliver [7440-22-47 000190 mg/l 1 O0:00190 00100 9D25010 FPA2007 0427/1911:16 JDH Sodium [7440-23-5]* 5.91 mg/L 1 0400 0500 ©D25010 EPA2007 G04/27/1911:16 JDH Thallum [7440-28-07 0.000110 U mg/l 1 0000110 000100 9D22019 FPA2008 04/26/1911:32 CMK Zinc (7440-66-6]* 00040 mg/L 1 000440 00100 ©D25010 EPA2007 G04/27/1911:16 JDH Classical Chemistry Parameters À ENCO Cary certihed anaïyte [NC 591] Analyte [CAS Number] Results Flag Units DE MDL PQL Batch Method Analyzed BY Notes Ammonia as N [7664-41-7]" 0045 u mg/L 10045 010 OE070 © EPA3SO1 05/07/191355 MKS Q-01 Calcium Hardness 180 mg/l 1 O0I8 00 925010 SM23408-2011 04/27/911:16 JDH Chloride [16887-00-6]* 97 mg/L 119 50 9D20019 SMASO0CIE-2011 04/29/1913:59 MKS Huoride [16984-48-8] 0.40 mg/l 100097 020 9E06020 EPA3000 O05/07/1904:28 MKS Q-01 Nitrate as N [14797-55-8] 19 mg/L 1 0041 OO10 [CAL © EPA3532 04/30/1913:18 MKS Nitrate/Nitrite as N* 19 mg/l 1 0041 O0 OD30028 EPA3532 O04/30/1013:18 MKS Nitrite as N [14707-65-0]* 0051 1 mg/L 107 010 GOD19013 © EPA3S32 (04/20/1910:21 MKS Q-02 pH 78 pH 110 10 9D26010 5MASOOH4B-2011 04/26/1012:57 ASC Q-01 Specific Conductance (EC) at 25 Deg 400 umhosm 1 10 10 ©D23033 SM2510R2011 04/23/1915:24 OC € Sulfate as $04 [14808-79-8]* 59 mg/L 129 SO OE06020 EPA3000 05/07/190428 © MKS Q-01 Temperature for pH (deg. C) 20 pH 1 9D26010 5M4500H+B-2011 04/26/1912:57 ASC Q-01 “Total Alkalinity as CaC03 [471-34-1]° 190 mg/L 1 14 15 OD20018 EPA3IO2 04/29/191431 MKS Q-02 Total Dissolved Solids* 240 mg/l 1 50 50 OD20004 SM2540C-2011 04/2/1911:28 JOC Q-02 Classical Chemistry Parameters = ENCO Orlando certlled anale [NC 424] Analyte [CAS Number] Results Faq Units DE MDL POL Batch Method Analyzed By Notes Total Organic Carbon 34 mg/L 1 O3 10 G©D23024 SM53108-2011 0424/1015:53 SiR ENCO www-encolabs.com FLAGS/NOTES AND DEFINITIONS B The analyte was detected in the associated method blank. D The sample was analyzed at dilution. 3 The reported value is between the laboratory method detection limit (MDL) and the laboratory method reporting limit (MRL), adjusted for actual sample preparation data and moisture content, where applicable. U The analyte was analyzed for but not detected to the level shown, adjusted for actual sample preparation data and moisture content, where applicable. E The concentration indicated for this analyte is an estimated value above the calibration range of the instrument. This value is considered an estimate. MRL Method Reporting Limit. The MRL is roughly equivalent to the practical quantitation limit (PQL) and is based on the low point of the calibration curve, when applicable, sample preparation factor, dilution factor, and, in the case of soil samples, moisture content. PQL PQL: Practical Quantitation Limit. The PQL presented is the laboratory MRL. N The analysis indicates the presence of an analyte for which there is presumptive evidence (85% or greater confidence) to make a ‘tentative identification". P Greater than 25% concentration difference was observed between the primary and secondary GC column. The lower concentration is reported. [page 127] [CALC] Calculated analyte - MDL/MRL reported to the highest reporting limit of the component analyses. J-06 The associated laboratory control sample exhibited low bias; the reported result should be considered to be à minimum estimate. Q-01 Analysis performed outside of method - specified holding time. Q-02 Sample received outside of method - specified holding time. QM-07 The spike recovery was outside acceptance limits for the MS and/or MSD. The batch was accepted based on acceptable LCS recovery. QM-08 Post-digestion spike did not meet method requirements due to confirmed matrix effects (dilution test). A ÆAIISOTECH ISOTECH LABORATORIES INC ANALYSIS REPORT Lab #: 715541 Job#: 41344 1S-90371 Co. Job#: Sample Name: Source Mariani Co. Lab#: Company: Northwater Consulting APl/Well: Container: 125ml bottle Field/Site Name: Source Mariani Characterization Location: Mariani, Haiti Formation/Depth: Sampling Point Date Sampled: 4/02/2019 13:00 Date Received: 4/16/2019 Date Reported: 4/29/2019 ôD of water 14.0 %. relative to VSMOW 580 of water ee -8.19 % relative to VSMOW Tritium content of water na 5"C of DIC ce na 14C content of DIC me pa 5!5N of nitrate ue pa 5180 of nitrate a na 5%S of sulfate a na 5'80 of sulfate me pa Vacuum Distilled? * --—-- No Remarks: nd = not detected. na = not analyzed. “Indicates if vacuum distillation was utilized for hydrogen and oxygen isotopic analysis of water [page 128] : APPENDIX E - HYDROGEOLOGICAL INVESTIGATION OF SOURCE MARIANI: CFC AND SF6 METHODOLOGIES Chlorofluorocarbons (CFCs) Chlorofluorocarbon (CFC) compounds have been synthesized on an industrial scale since 1931. They have primarily been used as refrigerants and aerosol can propellants, but also as foam blowing agents, solvents, and in insulation. Production reached its peak during the 1970s and 1980s before it wasrecognized that CFCs contribute to destruction oftheEarth's ozone. Production was subsequently bannedinthe 19905 as part of a global agreement. Three principal CFC compounds were used during the 20th century: trichlorofluoromethane, dichlorodifluoromethane, and trichlorotrifluoroethane, whose trade names are CFC-11, CFC-12, and CFC-113, respectively The CFCs production and release to the atmosphere have been measured and reconstructed back to 1940 (McCarthy et al, 1977; Gamlen et al, 1986; Wisegarver and Gammon, 1988; Fisher and Midgley, 1993; Fraser et al, 1996). Atmospheric CFC Concentrations since 1940 _ 600 Ë 500 R 400 o Ë 100 PA 1930 1950 1970 1990 2010 Year Atmospheric concentration of three principal CFCs produced since 1940 based on annual measurements from approximately 1980 to present and reconstructed based on release rates prior to 1980. Source: University of Utah Noble Gas Lab. The basis for age-dating with dissolved CFC measurements in groundwater is based on comparing the measured values to those of the atmospheric concentrations at the time of recharge. This is accomplished by recognizing that the dissolved concentration Ci is Ci = Kxpi [page 129] where KH is the Henry's constant and pi is the partial pressure of the CFC in air. The concentration is Ï related back to atmospheric concentration through pi Pi = Xi(P — Pro) where xi is the dry air mole fraction of the CFC, P is the atmospheric pressure and PH20 is the water vapor pressure. Henry constants have been carefully measured for the three CFCSs of interest and solubility determined as a function of temperature and salinity. À number of comparative age-dating studies have shown the reliability of the CFC approach (Busenberg and Plummer, 1992; Busenberg and Plummer, 1993; Ekwurzel et al. 1994; Cook and Solomon, 1997), Sulfur Hexafluoride SFs SFsis used as electrical insulator in high voltage switches and transformers. It is also used as a blanket gas in the production of magnesium metal. Production of SFs began in 1953, and ever since SFs has been building up concentration the atmosphere. SFs has lower solubility in water compared to the CFCs at 30 ppm. Its solubility is dependent on temperature, salinity, elevation, and any excess air in the water. Excess air origin is originated by rapid recharge that traps air in the vadose zone and carries that air into the saturated zone where it solubilizes. If trapped in pockets of air form in the aquifer space, SFs will readily partition into that trapped air due to its low solubility. The basis of using SF as an age dating tool relies on Henry's constant of SFe with respect water. The Henry's constant for SFsis 0.00024 mols/kg-bar. The measured concentration in groundwater can be compared to the atmospheric concentration through the use of its Henry's constant, resulting in an age date. There are natural sources of SFs associated with hot springs and fumaroles. Sometimes these sources can interfere with age dating of groundwater. 10.0 SF, in the atmosphere ë 8.0 — 5 ë 6.0 = 8 40 9 u° 2.0 o 0.0 = 1950 1970 1990 2010 Year [page 130] L Li] APPENDIX F - HYDROGEOLOGICAL INVESTIGATION OF SOURCE MARIANI: COMPILED DATASETS. Compiled Discharge Data for Source Mariani cu LES Type Data Source IR BR PS PP PS A PP PS A PP PS BR PP PS A PS SP M A A A A A A A QE PS A A PP A A A PS A PS A A M AR A M A ES A M A AE A 2 M De AR PP PS A PP PS A PP PS AS PP PS BR PS SP A ES A M A ES A M A M A A 2 M A M AA PP PS A PP PS BR PP PS A PS PS M A A M BR AE M D 2 [page 131] [I L _- Type Data Source Re Le A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A D D [page 132] L [I _- Type Data Source Re A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A D A [page 133] Compiled water quality data for source Mariani. : 8 ls ls 18 Le | ë ri da s 2 e 3 s 3 3 E 5 5 H = n 5 $ |à ë A A A A A A A A A A A A SE RE A A A A A SA A SE A ES ES EG D A A A A A A A A A A A A A A A ES A A A PE A A A A A A A A SE A A A A A A A A A A ES ES A A A A A A A A EE A ES ES A A A A A A A A RE SE A A A A A EE RE PE A PE A A SA A A A A ES M A A A A A A A A A SE M A A A A A A A SE M A A A A A A A A A A A A A A A A A A A A A A A A A RE D A A A A EE A A A A A A A A A A SA A A A A A A AE A PS A A SA A A A A ES ES M A A A A A A A A A A A A M A A A A A A A A A ES EE A A A A A A RE A A A M A A A A A A A A A A A A M A A A A A A SA A A A A A A M A A A A A A A A A A A EE A A A A A A A A A A A A A A A AE A A A A A A A EE A A A A A EE A A A A A A A A A A A A A M A A A A A A A A A M A A A A A A A A A A A A A ES EE A RE A A A A EE A A A A M A A A A A A A A A A A A A ES A A A A A A A A A A A A M A A A A A A A A A A A A A M A A A A A A A A A EE A A A A ES A A A A A A A A A A A A A A A A A A [page 134] + 2 |$8 g |g |8 |» < = le a = Ë 5 JS 5 lé SE ls V5 |s LE [£a 8 |8 2 | lé |# | ë |< ÊË Ë H ë | Ë 8 |6 ë É È A A AE A A A A A A A A A A A A A A A EE A A A RS A A A A A A A A A A A EE A A A A A A A A A A A A A ES A A A A RS A A A A A A A A A A A ES EE M A A A A A A A A A A A A A EE M A A A A A A A A A A A A A ES M A A A A A A A A A A A EE RE A SE A SE EE A A A A A M D D A A A ES RS A A A A A RE A A A A A A A A A A A A A A A A A A A A A A EE RE AE AE A SE SE A SE A A A A A A A A A A A A A EE RE A A A A A A A A A A A A EE CRE TT TR TONI VTT] A A A A A A A A A A A A A A ES EE RS A A A A A A A A A A A A A EE A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A A A EE A A A A A A A ES A A EE A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A ES M AA A AE A SE A SE A ES EE A A A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A A EE RE A A A A ES A A A A PS PS RS RE A A A A A A A A A A A A A A EE A PE AE A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A ES RE AE A EE EE SE RS A A A A M D A A A A A RS A A A A A RE A A A EE M A A AE A SE SE A A RE A A A A A A A A A A A A A A A EE A A A A A A A A A A A A A EE PE A A A A A A A A A A A A ES A A A A A A A A A A A A ES ES [page 135] 2 & & & da = 5 EP PE A RE A D EE RE A PE A A A A PE DE A A A A A A A A AE A A A A A ES I A A A RE A PE A A AE A A A A A A A ES PE A A RE A D A A A AE A A A A A A ES A ES RE EP PE A D EE A AE A A A A A A A PE ES EE A A RE A A EE EE A RE RE A A A A A A A A PE PE PE A D A A A PE A A A A A PE ES D A A RE PE A PE A A A A A A A PE EE A A A A A RS A A RE A A A A A A A A EE RE AE RE A A EE EE AE RE RE A A A A A A EE AE A EE A RE EE A RE A A A A A EE EE A RE A A RE A A RE A A A A A EE I PE PE PE A D A A A AE A A A A A PS EE I A A A A A A A AE A A A A A A A A ES A A A RE A A A A AE A A A A A A ES I A A A A A A AE A A A A A A A ES I A A A A A A A AE A A A A A A ES I A A A A A A A AE A A A A A A PE ES EE A RS A A EE EE A AE RE A A RE RE A RE RE A A A A A A A A EE PE PE A D EE A AE A A A A A A PS D EE PE A A RE PS PS A A A A A A ER A A A A A A EE A AE A A A A A EE EE A A RE A A EE EE A RE RE A A AE EE PE AE RE A A RE EE RE A RE A A A A PE PE A A A A EE A AE A A A A A A ES EE I A A A RE A A A A A AE A A A A A A ES EE A RE A A RE A RE A A A A A A AE EE A A EE A EE A AE A A A A A A A EE I EP D A A D A A A A PE D ER A PR A D EE SE EE EP A A RS SE A A A A EE PE D PE A AR A A A A EE A A A A RE ER A A A D A A RE PS A PE A A A A A A A A PE EE A A RE A A EE EE A RE RE A A A A A A AE PE PE PE A D A A RE A PE A A A A A ES ES EE EE RE A RE A A RE A RE A A A A A A A RE PE PE A A A A A A A A A A A PE ES EE [page 136] Hi 8 Hi ; £ 5 8 5 £ 88 | à È £ Ë Ë $ Ê 8 F 4 8 + 5 JE [5 ls |E | ë |6 |$ |$ Ë A A A A A A A A A A A A A A SE A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A EE A A A A A A A A A A A A AE ES EE ES A A A A A A A A A A A A A A A EE A A A A A A A A A A A A ES SE EE M A A A A A A A A A A A A A A EE RE A AE A SE SE A A A A A A PE A A A A A A A A A A A A A A A A A A A A A A EE RE A AE A SE SE A A A PE A A A A A A A A A A A EE EE A A A A A A A A A A A A A A ES A A A A A A A A A A A EE A A A A A A A A A A A A A A A EE PE A A A A A A A A A A A A A EE A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A EE A A A A A A A A A A A A A EE EE A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A EE RE AE AE A SE SE A A A A A A A A A A A A A A A EE EE A A A A A A A A A A A A A A A A A EE A A A A A A ES EE EP A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A EE I A A A A A A A A A A A A A EE A A A A A A A A A A A A A A A A EE CRETE TR TRIER NT | M PA AE AE SE SE A A A A RE A PS A A A RE A A A A A A A ES RE A SE A SE SE A A A A A A A A A A A A A A A AE EE A A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A EE A A A A A A A A A A A A A A ES ES [page 137] _ = 8 € & 8 s 2 _ 5 _ 5 à 5 A A A A A A A A A A A A A EE A PE A A A EE A EE A A A A A A RE A A AE A A A A EE A EE A A A A A RE RE AE AE A A A A EE A EE A A A A A A RE SE DE RE A A A A A A RE AE AE A A A EE A EE A A A A A A EE EE A A AE A A A A A A SE EE A A A A A EE AE AE A A A A RE A EE A A A A EE RE A AE PE A A A EE A EE A A A A EE RE EE AE AE A A A EE A EE A A A A EE A AE A A A A A A EE A A AE A A A A EE A EE A A A A A A RE A PE A A A A EE A EE A A A EE RE EE RE A A A A A A RE AE AE A A A EE A EE A A A A A RE RE AE AE A A A EE A EE A A A A EE A RE RE AE AE A A A A RE A EE A A A A A A RE SE AE RE A A A A A A M A A A A A A A A A RE AE AE A A A EE A EE A A A A A A RE A D RE A A A A A A A A EE AE AE A A A A EE A EE A A A A A RE M AE AE A A A A RE A EE A A A A A RE A A A A A A A A A A AE AE A A A EE A EE A A A A A RE RE AE AE A A A A RE A EE A A A A A A RE RE AE AE A A A A RE A EE A A A A A A RE A GE A A A A A RE AE AE A A A EE A EE A A A A A RE [page 138] : APPENDIX G - PLAINE DU CUL-DE-SAC: GROUNDWATER FLOW MODEL The regional steady-state groundwater flow model developed for the Plaine du Cul-de-Sac aquifer allowed for simulations of groundwater flow and for an understanding of the aquifer's groundwater budget. Using this base model, model scenarios were provided by the IDB, and included three groundwater management alternatives with eight climate change scenarios. It is important to note that decreases and increases of groundwater flow presented and discussed are not relative to the complete water balance of each of the individual hydrological systems, but only the change in proportion to the PCS aquifer. For example, a 5% decrease in flow to Canal Boucambrou does not indicate that the canal flow will be 5% lower, but only that the groundwater contribution is 5% lower. Quantifying the water budget of the surface water systems is important to better understand the potential impact of the reduced groundwater inputs. Groundwater Flow Model Scenarios Based on the IDB-led analysis of 36 climate change models and associated projections, four unique climate change scenarios were selected to be incorporated into the groundwater model runs. Each climate change scenario included a change in annudal precipitation, temperature, and streamflow. To complement the climate change scenarios, three different groundwater management alternatives were simulated: 1. Base: Current situation of groundwater abstraction/pumping as in the base model. This results in pumping approximately 70,000 m3/day from the aquifer 2. Alternative 1: DINEPA CTE-Wells P1, and G1 - G7 are commissioned to pump approximately 32,000 m3/day. Currently these wells exist but are not in service. This results in pumping over 100,000 m3/ day from the aquifer 3. Alternative 2: DINEPA CTE drills 12 new production wells (G8 - G19), with an additional production of 45,000 m3/day. This results in pumping of over 140,000 m3/day from the aquifer Groundwater flow model scenarios Optimistic Climate Change Scenario ° 1° C increase in temperature E2 Alt1: Climate change with increase of groundwater pumping by 1.5X + __1.2% increase in streamfow Central Climate Change Scenario E4 Base: Climate change with baseline model °_1,5° C increase in temperature E4 Alt1: Climate change with increase of groundwater pumping by 1.5X ° _4.7% decrease in streamflow E4 AIt2: Climate change with increase of groundwater pumping by 2X Central-Pessimistic Climate Change Scenario E6 Alt1: Climate change with increase of groundwater pumping by 1.5X + _9.6% decrease in precipitation + 2.1° Cincrease in temperature E6 Alt2: Climate change with increase of groundwater pumping by 2X °_10% decrease in streamflow Pessimistic Climate Change Scenario E8 Base: Climate change with baseline model 29° C increase in temperature E8 Alti: Climate change with increase of groundwater pumping by 1.5X °__ 24% decrease in streamflow E8 AIt2: Climate change with increase of groundwater pumping by 2X [page 139] Driving global climate change models considered for climate change projections. l Historical climate models Control scenario CTL [_csiro_mk3 6 0 [XX XX x | [_ giss e2 7] XX XX XX [__miroc_esm 1] X 1 XX XX Groundwater Management Alternatives Groundwater flow model output is presented to visualize the potentiometric surface of the two groundwater management alternatives compared to the base model. {Error! No se encuentra el origen de la referencia. following table includes the groundwater budget results from simulating the management alternatives. Results of groundwater management alternatives compared to base model. MODEL RUN BASELINE Base-Alt1 Base-Alt2 Recharge (Rech) 15.200 _ 15.200 _ 0% 15.200 0% Riviere Blanche (Rsw) 16.253 _- 16.653 _ 2% 17.717 9% Riviere Grise (Rsw)| 95.742 - 113.734 _ 19% 139.369 46% General Head (Rgh) 6.226 — 6.287 _ 1% 6.326 2% Canal Boucambrou (Dsw) _ 13.281 _ 12.812 é 4% 12.238 dé 8% Riviere Batard (Rsw) 1.154 _ 1494 - 30% 1.736 50% Lac Azuei/Etang Sumautre (Dsw) L 1.838 = 1.819 1% 1.779 -3% Trou Caiman (Dsw) _- 3.890 _- 3.804 2% 3.610 7% Ocean (Dsea) _ 43.808 _- 41.018 6% 39.171 11% Pumping (ABS) - 71.582 _ 103.582 45% 144.832 102% [page 140] : Groundwater management alternative simulations. Groundwater Management Alternative 1 8 new wells (G1 - G7, P1) + 32,000 m3/day pumping Trou JS KR e e e é = 0 Fr d Lac , 1 | 7 0 à Azuei Prince { ITA RE = Ÿ LCR LEZ Fm 80 Eee Ces “a” |commissioned in alternative 1 aù Groundwater Management Alternative 2 Alt 1, plus 12 new wells (G8 - G19) + 140,000 m3/day pumping Trou © 1 LU DE Lac @. D : nus Baie de @) *%e Lo CIE See EN N CES VAT 70e e [AA pin. ES N planation e CTE-RMPP Pumping Wells e Proposed CTE-RMPP Wells (G8 to G19) © Other Pumping Wells in Model SD Approxiate Aquiter Baseline Model Potentiometric Surface (m- RS Groundwater Management Alternatives: asl Potentiometric Surface (m-asl [page 141] ALTERNATIVE 1 l Alternative 1 includes the commissioning of eight existing wells that are not currently active to pump approximately 32,000 m3/day. The wells include P1, and G1-G7, all of which are in the southern portion of the aquifer near to the Riviere Grise. This alternative results in a 50% increase of groundwater pumping from the base model. À summary of the model results follows: - The additional pumping creates an oblong cone of depression of an approximate 1.5 km radius in the G-well area. The cone of depression extends more northerly than southerly due to the steep groundwater gradient. Based on the simulation, the potentiometric surface drawdown in the cone of depression averages 1.5m and reaches up to 8m - Diffuse impacts to the potentiometric surface are simulated, and direct drawdown from pumping dissipates approximately 4 km down from the well field. The overall aquifer system experiences a slight decrease in water levels as it reaches a new equilibrium with the new pumping condition - The groundwater budget indicates that that the increase in pumping is offset by an increase of river infiltration from Riviere Grise into the aquifer due to the increased hydraulic gradient between the river and the potentiometric surface of the aquifer. Based on the model simulation, Riviere Grise infiltration to the aquifer could increase up to 19% under scenario 1. Small decreases of flow into the ocean (6%) and surface water bodies (1 to 4%) were simulated ALTERNATIVE 2 Alternative 2 includes the subsequent addition of 12 new production wells (G8 - G19), adding 45,000 m3/day of additional pumping. When combined with alternative 1, this results in pumping approximately 140,000 m3/day from the aquifer. The alternative results in a 100% increase in groundwater pumping from the base model, and a summary of the model results are as follows: - The combined pumping of the G8 - G19 wells combined with G1 - G7 and P1 create a larger cone of depression with extends down-gradient (north) approximately 4.5 km and up-gradient (south) by half that distance due to the groundwater gradient. Based on the simulation, the potentiometric surface drawdown averages 3 m and is over 15 m (perhaps more) along the aquifer boundary associated with the outcrop area to the east of the well field - Diffuse impacts to the aquifer under this alternative are more significant than alternative 1. The potentiometric surface of the aquifer lowers by an average of 15 to 2.0 m, ranging from 15 min the area of the G wells to minimal in the eastern portion of the aquifer - The water budgetillustrates that the increase in pumping is offset by the increase of river infiltration from upper and middle Riviere Grise into the aquifer. Thisis causedbytheincreasedhydraulic gradient between the river and the potentiometric surface due to drawdown, both related to the cone of depression and the diffuse regional lowering of the potentiometric surface. Based on the model simulation, the Riviere Grise infiltration to the aquifer could increase by over 40% under Alternative 2. This simulation also indicates an 8% decrease of groundwater flow to Canal Boucambrou, a 7% decrease in groundwater flow to Trou Caiman, and 3% decrease in groundwater flow to Lac Azuei. - The vulnerability to localized seawater intrusion slightly increases in the coastal areas due to the regional lowering of water tables, although the simulation, as run for this study, does not suggest a regional occurrence of seawater intrusion. Adding pumping wells near the coast in the PCS aquifer should always be proceeded with due diligence and caution. [page 142] : Climate Change Scenario Results OPTIMISTIC CLIMATE CHANGE SCENARIO The optimistic climate change scenario included a 7.8% increase in precipitation, 1°C increase in temperature, and a 1.2% increase in the Riviere Grise streamflow. The results of the climate change simulation and groundwater management alternatives are presented below. This climate change scenario under current groundwater management conditions results in an increasedrecharge, with correspondingincreases of groundwater flow to the canal, lakes, andocean. Riviere Batarde infiltrates less into the aquifer due to the higher water tables. The optimistic climate change conditions mitigate impacts to the groundwater budget when simulating the pumping alternatives. Under Alternative 2, the increased pumping is predominantly offset by increased river infiltration. Groundwater flow to Trou Caiman and Canal Boucambrou in the two management alternatives decreases by 3% and 5%, respectively. Groundwater flow to Lac Azuei does not appear to be reduced from baseline conditions in these scenarios. Optimistic climate change scenario results. MODEL RUN BASELINE E2-Base E2-AIti E2-AI2 Recharge (Rech) 15.200 - 16.385 = 8%] 16.385 8% 16.385 8%) Riviere Blanche (Rsw)| 16.253 - 16.412 - 1% 16.871 4%] 17.588 8%) Riviere Grise (Rsw)| 95.742 _ 96.969 _ 1%) 124.822 30%)| 162.202 69%) General Head (Rgh)| 6.226 - 6.096 - 2%) 6.157 41%] 6.196 0%| Canal Boucambrou (Dsw)| _ 13.281 - 13.732 r 3% 13.275 r 0,0%] 12.683 É 5%) Riviere Batard (Rsw)| 1.154 _ 965 - 416%) 1.305 13%) 1.542 34%! Lac Azuei/Etang Sumautre (Dswi)| - 1.838 _ 1.900 3% 1.878 2%| 1.832 0%) Trou Caiman (Dswi)| _ 3.890 _ 4.037 4%) 3.952 2%) 3.762 3%) Ocean (Dsea)| _ 43.808 - 45.190 3%) 42.410 3%) 40.613 7%! Pumping (ABS)| _ 71.582 _ 71.582 0%| 103.582 45%) 144.832 102% CENTRAL CLIMATE CHANGE SCENARIO The central climate change scenario included a 0.2% increase in precipitation, 1.5°C increase in temperature, and a 4.7% decrease in the Riviere Grise streamflow. The results of the climate change simulation and groundwater management alternatives are presented below. This climate change scenario under current groundwater management conditions results in a slightly diminished recharge from river infiltration, with corresponding decreases of groundwater flow to the canal, lakes, and ocean. The central climate change conditions (mostly 47% decrease in the Riviere Grise streamflow) magnify impacts to the groundwater budget when simulating the pumping alternatives. Under both alternatives, the increased pumping is predominately offset by increased river infiltration. Groundwater flow to Trou Caiman decreases by 10% and 15% in the two pumping alternatives, respectively and flow to Lac Azuei decreases by 9% and 11%, respectively. This scenario starts to expose the importance and sensitivity of the model to the Riviere Grise concerning its role in driving the recharge and groundwater flow of the aquifer. [page 143] Central climate change scenario results. l MODEL RUN BASELINE E4-Base E4-AIt1 E4-AIt2 IN (m3/d) OUT (m3/d) IN (m3/d) OUT (m3/d) % Change IN (m3/d) OUT (m3/d) _% Change IN (m3/d) OUT (m3/d) _% Change Recharge (Rech)| 15.200 _ 15.200 _ 0%| 15.200 0%] 15.199 0% Riviere Blanche (Rsw)| 16.253 _ 15.600 - 4% 16.152 41% 16.951 4%) Riviere Grise (Rsw)| 95.742 - 91.116 - -5% 119.298 25%| 153.080 60%) General Head (Rgh)| 6.226 - 6.526 - 5% 6.587 6%) 6.620 6%| Canal Boucambrou (Dsw)| = 13.281 _ 12.367 \É 7%) 11.954 r 10,0% 11.450 é 14% Riviere Batard (Rsw)| 1.154 _ 1.541 = 4%) 1.871 62% 2.080 80%| Lac Azuei/Etang Sumautre (Dsws)| = 1.838 = 1.705 7%) 1.669 -9%| 1.634 11% Trou Caiman (Dsw)| _ 3.890 = 3.572 8%) 3.487 10% 3.314 15% Ocean (Dsea)| _ 43.808 _ 40.909 7%) 38.163 13% 36.538 17% Pumping (ABS)| - 71.582 _ 71.582 0%)| 103.582 45%) 141.082 97%| The central-pessimistic climate change scenario included a 9.6% decrease in precipitation, 2.1°C increase in temperature, and a 10% decrease in the Riviere Grise streamflow. The results of the climate change simulation and groundwater management alternatives are presented below. This climate change scenario under current groundwater management conditions results in a significant decrease in recharge from both river infiltration and aerial recharge, with corresponding decreases of groundwater flow to the canal, lakes, and ocean. Without any increases in groundwater pumping, groundwater flow to Trou Caiman and Lac Azuei decreases by 18%. The impacts of the central-pessimistic climate change scenario to surface water bodies are on the same order of magnitude as the 2X groundwater pumping alternative under a more central climate change condition. The increased pumping under both groundwater management alternatives results in significant increases in river infiltration, partially offsetting the abstraction. Climate change conditions are responsible for a larger proportion of the impacts to the groundwater budget than increased pumping._ In the pumping alternatives, groundwater flow to Trou Caiman decreases by 20% and 25%, respectively, and groundwater flow to Lac Azuei declines by 20% and 22%, respectively. The importance of the Riviere Grise is even more apparent based on these simulations, since they show that decreased river flows can significantly affect the groundwater budget and water tables of the aquifer. The groundwater modeling does not simulate the Riviere Grise becoming completely dry, which becomes more of a seasonal possibility in these climate change conditions. Seawater intrusion vulnerability starts to increase in the coastal areas due to the regional lowering of water tables, although the simulations did not indicate its occurrence. Regionally, a positive flux is maintained from the aquifer to the ocean, and no flux from the ocean to the aquifer. Central - pessimistic climate change scenario results. MODEL RUN BASELINE E6-Base E6-AIt1 E6-AIt2 IN (m3/d) OUT (m3/d) IN (m3/d) OUT (m3/d) _ % Change IN (m3/d) OUT (m3/d) _% Change IN (m3/d) OUT (m3/d) _% Change Recharge (Rech)| 15.200 — 13.739 _ -10%| 13.738 -10%| 13.737 -10%| Riviere Blanche (Rsw)| 16.253 _ 14.518 _ -11%] 14.822 -9%| 15.876 -2%| Riviere Grise (Rsw)| 95.742 _ 88.359 - 8%| 116.454 22%| 150.304 57% General Head (Rgh)| 6.226 _ 6.719 - 8%| 6.797 9%| 6.893 11%] Canal Boucambrou (Dsw)| _ 13.281 _ 11.347 É 15%) 10.998 r 17%] 104817 21%] Riviere Batard (Rsw)| 1154 _ 2.008 _ 74%) 2.312 100%] 2.519 118%] Lac Azuel/Etang Sumautre (Dsw)| — 1.838 _ 1.505 -18%| 1475 -20%| 1.429 -22%] Trou Caiman (Dsw)| - 3.890 _ 3.195 18%) 3.114 -20%| 2.924 -25%] Ocean (sea 4808 - ssn ax sacs 20% s27 24% Pumping (ABS)| = 71.582 _ 71.582 0%) 103.582 45% 141.082 97%] [page 144] The pessimistic climate change scenario included a 25% decrease in precipitation, 2.9°C increase in temperature, and a 24% decrease in the Riviere Grise streamflow. The results of the climate chance simulation and groundwater management alternatives are presented below. The model became unstable with this scenario, as it was departing from its original calibration and could not converge with the 24% decrease in streamflow. We applied the maximum possible flow reduction while keeping the model stable, which was in the range of 18%. This climate change scenario under current groundwater management conditions results in a decrease in recharge from both river infiltration and aerial recharge, with corresponding decreases of groundwater flow to the canal, lakes, and ocean. Without any increases in groundwater pumping, groundwater flow to Trou Caiman and Lac Azuei decreases by 27% and 32%, respectively. The increased pumping under both groundwater management alternatives resultsin significantincreases in river infiltration, partially offsetting the abstraction. Climate change conditions are responsible for a larger proportion of the impacts to the groundwater budget than increased pumping, In the pumping alternatives, groundwater flow to Trou Caiman decreases by 29% and 38%, respectively, and groundwater flow to Lac Azuei declines by 34% and 37%, respectively. The importance of the Riviere Grise is apparent based on these simulations, since they show that decreased river flows can significantly affect the groundwater budget. The groundwater modeling did not simulate the Riviere Grise becoming completely dry, which becomes more of a possibility in these climate change conditions. Seawater intrusion vulnerability is higher in the coastal areas due to the regional lowering of water tables, and there is a 17% to 31% change in flux between the aquifer and ocean. Regionally, a positive flux is maintained from the aquifer to the ocean, and no flux from the ocean to the aquifer. Pessimistic climate change scenario results. MODEL RUN BASELINE E8-Base E8-AIt1 E8-AI2 IN (m3/d) OUT (m3/d) | IN(m3/d) OUT(m3/d) % Change IN (m3/d) OUT (m3/d) % Change IN (m3/d) OUT (m3) % Change Recharge (Rech)| 15.200 _ 11.391 _ -25%| 11.390 -25%| 11.262 -26%| Riviere Blanche (Rsw) 16.253 — 12.236 — -25%| 12.740 -22%| 7452 -56%| Riviere Grise (Rsw) 95.742 - 90.298 _ 6%)| 117.761 23%| 149.678 56%) General Head (Rgh)| 6.226 _ 6.746 _ 8%. 6.974 12%. 7.369 18%) Canal Boucambrou (Dsw) - 13.281 - 105 7 20% | 10179 7 23% 92037 -31%| Riviere Batard (Rsw)| 1154 _ 2.160 _ 87%| 2.491 116%) 2.902 152%) Lac Azuei/Etang Sumautre (Dsw)| - 1.838 - 1.248 -32%| 1216 -34%| 1.150 -37%| Trou Caiman (Dsw)| _ 3.890 _- 2.851 -27%| 2.743 -29%| 2.396 -38%| Ocean (Dsea)| - 43.808 - 36.453 17%! 33.583 -23%! 30.168 -31%| Pumping (ABS) s 71.582 — 71.582 0%) 103.582 45%| 141.082 97%] [page 145] Considerations Regarding Model to potentially more significant depending on Scenarios the climate change scenario. Quantifying the water budget of the surface water systems is The model scenarios suggest that impacts to important to better understand the potential the aquifer should be anticipated from both the impact of the reduced groundwater inputs. climate change and groundwater management Lu k , alternatives. The potential impacts range from 4 The pessimistic climate change scenarios minimal to more significant, especially when the appear to have a greater regional impact on pessimistic climate change scenario with more the aquifer than the groundwater management significant groundwater pumping is considered. alternatives. The groundwater budget and The results presented can be considered a "egiondl water tables are most sensitive to planning tool at the regional level to help inform changes in the Riviere Grise flow. roundwater development and management ractices that balance potential impasts with S. The Riviere Grise isa critical component ofthe economic and public health benefits. The aquifer, and so its ability to sustain groundwater scenarios also help guide what studies, modeling abstraction and flows to surface water bodies. and other activities should be considered and The recharge from the river drives the hydraulic prioritized in the future to facilitate improved gradient, replenishes the aquifer when there IS integrated management of the groundwater pumping or climate change stress, and mitigates resources. saltwater intrusion risk in coastal areas. e 6. Based on the evaluation of scenarios, the Severdl observations regarding the feasibility of the proposed well field of G8 - model scenario results are outlined G19 may warrant further evaluation in terms below: of well interference and the potential impacts to the Riviere Grise and the aquifer. An 1 Drawdown / cone of depression area in abstraction rate and number of wells should the zone of the G-wells from groundwater that considers the results and potential impacts management alternatives may affect other presented should be planned. Exploration and nearby wells and create a stronger gradient groundwater development could be considered between the Riviere Grise and the aquifer that in less developed areas of the aquïfer, perhaps may result in increased flow from the river the area recharged by the Riviere Blanche into the aquifer. Impacts are magnified under infiltration. pessimistic climate change scenarios, especially regarding decreases in the flow of the Riviere Should hydrologic projections of the Riviere Grise Grise result in sustained periods of flow below 1,500 L/s, transient or stress period modeling should 2. Scenarios result in a diffuse effect on the be considered to evaluate the implications of water balance of the aquifer due to the regional this condition, If the Riviere Grise. adjustment(lowering) of watertables.Thiseffect ranges from small to potentially more significant 7, does not flow for significant periods of time, depending on the climate change scenario and this would affect the dynamics of the aquifer in groundwater management alternatives a significant way. The pumping conditions in the two groundwater management alternatives 3. The regional lowering of water tables resulting are largely offset in the groundwater budget from climate change and groundwater by an increase in river infiltration. The Riviere management _ alternatives reduces the Grise diversions upstream of the primary groundwater flow to surface water systems recharge areas may also reduce the stage and (Ocean, Trou Caiman, Lac Azuei, Canal water available to infiltrate into the aquifer and Boucambrou). These effects range from small mitigate the increased pumping, [page 146] _ | | Inter-American Development Bank