Quantifying the Available Groundwater Resource in the Speightstown Catchment Barbados Anuradha Maharaj 1*, Karl Payne 2, Adrian Cashman 3 1 The Centre for Resource Management and Environmental Studies (CERMES), Cave Hill Campus, St. Michael, Barbados. Email: maharajanuradha27@gmail.com 1 ; adrian.cashman@cavehill.uwi.edu 3 2 Department of Civil and Environmental Engineering, University of South Florida (USF), Tampa, Florida. Email:karl.payne@mail.usf.edu 2 Abstract The island of Barbados is characterized as one of the world s most water scarce countries (Emmanuel and Spence 2009, UNEP 2010). The link between available water and a thriving economy cannot be disputed (Global Water Partnership 2014; Sanctuary and Tropp 2007; Bank and Tech 2014), with Barbados tourism-based dependence adding to the already strained water resources (Charara et al. 2011). It is therefore critical from a water management standpoint, to quantify aquifer volumes and their possible responses to the changing climate in order to ascertain water availability. The Speightstown area, located on Barbados West Coast, is a wellknown, high-end tourist hub and an area of on-going coastal development. The land elevation increases from sea level to a height of approximately 250 m inland; a gradient allowing for surface runoff and channeling of water; 15 30 % (Jones and Banner 2003b; Jones and Banner 2003a) of which infiltrates and recharges the Ashton Hall and Whim groundwater units below the Speightstown catchment. A three- dimensional numerical groundwater model of this unconfined aquifer system was developed using the MODFLOW software. A previously developed island-scale model was downscaled to provide the boundary conditions for the Speightstown model from which the available water can be quantified and allocated. Keywords: MODFLOW; water availability; water resources management 1. Introduction Barbados is a small island situated east of the Windward Islands of the Lesser Antilles. The island is almost entirely dependent on groundwater supplies. With available per capita natural water resources estimated at 350 m 3 per person per year, Barbados is classified as a water scarce country (Reid 1994). Its unique position is representative of the distinctive geology that underpins its development. This authochtonous Pleistocene Coral Rock, (Harper and Donovan 2002) is composed of mostly carbonate rock that has over time resulted in a very robust groundwater system. The groundwater flow dynamics in the aquifers of Barbados are largely 1
driven by regional and tropical climatic conditions (I. C. Jones and Banner 2003a; I. C. Jones and Banner 2003b); with most recharge to the aquifer system occurring during the rainy season (I. C. Jones and Banner 2003a). The karstic geology of Barbados is as such that with intense rainfall; there is little evaporation and therefore most of the precipitation infiltrates by discrete transport mechanisms into the groundwater system. According to Senn (1946), the controlling factors for groundwater occurrence in coralline area of Barbados are rainfall, permeable rock, impermeable bed, structure and sea-level. The water in the aquifer is ultimately derived from meteoric waters, principally rain (Senn 1946), and is therefore highly sensitive to change. It therefore must be guarded in light of climatic changes and the resulting influence. The annual average rainfall over the island of Barbados varies from about 1000 mm/year at the coastal areas to over 2000 mm/year in the central areas (I. C. Jones and Banner 2003b) due to orographic effects created by its terraces. The aquifer systems on the island are unconfined and are the primary source of potable water for the domestic demand as well as other water intense sectors such as tourism. Maintaining a sustainable water supply is therefore critical to the economic stability of the country. As such, quantification of the groundwater resources as well as data regarding projections of changes to the groundwater system by both natural and anthropogenic influences is vital to managing and safeguarding the water resource. Added to this is the enhanced effect of sea level rise and its impact on freshwater aquifers due to climatic changes combined with the socioeconomic implications of droughts. (Wellington and Moore 2001) The aquifers of Barbados hydraulically connected to the sea. A total of 86.4% of the island s drinking water is acquired from three coastal catchments, namely the St. Michael catchment (52.8% of total), the St. Philip catchment (20.2%) and the West Coast catchment (13.4%). (Wellington and Moore 2001) The wells of the West Coast catchment of which the Speightstown catchment is a part are about 992.80 m away from the sea, with water levels about 0.3m above sea level. About 51,000 persons along the West Coast and in the northern parishes of Barbados, are serviced by this catchment; highlighting the extent at which water shortages would be felt should this result. Therefore it is crucial that Barbados begin investigations into water supply augmentation, since any loss of output from the West Coast catchment will have serious implications for the island. (Wellington and Moore 2001) 2. Study Area Apart from the previously mentioned contribution to island-wide water supply, the Speightstown catchment, which is located within this west coast catchment area (Figure 1), is also integral to tourism and experiences high levels of coastal development. The town centre is located on the coast, which is at sea level; increasing in elevation inland; meeting the boundary of the Scotland District in the uppermost parts of the catchment. 2
The Speightstown groundwater model is provides data for and is integrated into a wider Caribbean-based project: the Water_aCCSIS Project 1, which focusses on assessing water availability (WA) at the catchment level for various watersheds in the region. The provision of groundwater data is a critical input into the WA models being developed for those countries dependent on groundwater as their primary water source. 2.1 Data Availability Most of the hydrogeological datasets required for model building and calibration was acquired from the BWA. Any additional data needs were supplemented with information extracted from consultancy reports and other secondary data sources. It must be noted that data is very limited and one outcome of the research is the need for more data collection with regards to more boreholes and frequent readings taken for hydraulic head data, which are needed for calibration purposes. 1 The shortened version stands for Sustainable Water Management under Climate Change in Small Island States of the Caribbean. http://www.water-accsis.org/ 3
Figure 1: Location of Study catchment area: Speightstown catchment, Barbados 3. Methodology 3.1 Regional Groundwater Model Setup There are several software packages available for simulating three-dimensional groundwater flow. MODFLOW-2000 (Modular Groundwater Flow) is a numerical modeling package that has been used in a numerous research and consulting projects for various hydrologic regimes and climatic conditions (Harbaugh et al., 2000). More specifically, there are several studies that have 4
used MODFLOW in studies investigating the impact of climate change on groundwater resources. The code numerically approximates the partial differential equation (PDE) describing groundwater flow which is given by: K x x h K x y y h K y z z h W z S s h t (1) where, x, y and z are the three spatial coordinates, t is the time dimension, h is the hydraulic head K x, K y, K z are the hydraulic conductivities in the x, y and z directions respectively, S s is specific storage and W accounts for sources and sinks. For complex geometries and boundary conditions as well as for three dimensional flow, an analytical solution to equation 1 precluded. MODFLOW-2000 uses the finite-difference technique to calculate the spatial and temporal variations of hydraulic head in the saturated zone MODFLOW discretizes equation x and solves the resulting algebraic equations iteratively. In developing the groundwater flow model for Barbados the following are further assumptions and simplifications made: 1. The occurrence of groundwater flow is three-dimensional. 2. Darcy s law is valid and therefore turbulent flow, associated with flow through karst conduits is ignored. 3. The limestone domain is a single continuum. 4. The limestone aquifer is unconfined with variable thickness. 5. The aquitard has a sufficiently low permeability to ignore fluxes across the interface between the base of the limestone and underlying aquitard. 6. The entire model domain is completely saturated and there are no variable density considerations. 7. The vadose zone has a relatively small thickness and the volume of water in the unsaturated zone is negligible relative to the aquifer. 8. The Scotland District is comprised of impervious geologic formations and the permeability of these lithologies is negligible. 3.1.2 Hydrostratigraphy The Pleistocene limestone aquifer of Barbados is a composite limestone island aquifer composed of Pleistocene coral reef limestone underlain by a deep-sea sedimentary aquitard (Jones and Banner, 2003). The island s carbonate is underlain by low permeability Tertiary units, which are exposed at the surface in the Scotland District. A more in depth discussion on the island s 5
geology can be found elsewhere. Nevertheless, the conceptual model in this study is comprised of one geologic layer; the Pleistocene limestone aquifer. The aquitard was not explicitly accounted for in the model due to its low permeability. 3.2.2 Boundary Conditions A complete description of a numerical groundwater model requires the assignment of boundary conditions. A boundary condition (B.C) is a mathematical statement assigning a value of the dependent variable (hydraulic head) or derivative of dependent variable (groundwater flux) at the boundaries of the model domain. Since hydraulic heads are generally easier to measure relative to groundwater flux, it is expedient to use heads for boundary condition specification. In general, the three types of boundary conditions used in hydrogeologic models are: specified head (Dirichlet B.Cs), specified flux (Neumann B.Cs), and head-dependent flow conditions (Cauchy B.Cs). Two physical boundaries were used for the island-scale model boundary conditions. These physical boundaries are: the coastal boundary and the interface between the base of the limestone and the underlying low permeability stratum. The coastline was assigned a Dirichlet boundary condition to account for a specified head of zero. A no flow boundary was set at the bottom of the limestone unit in order to simulate the low permeability aquitard. An assumption was made that there is no flow component normal to the contact between the two geologic units. 4. Results and Discussion Firstly, a steady state, calibrated, 3D numerical groundwater model of Barbados, the regional model (RM), was built using the MODFLOW model (Figure 3). The built-in Telescopic Mesh Refinement (TMR) tool in Groundwater Vistas was then used to carry out local grid refinement (LGR); whereby the grid is downscaled only to the area of interest, the steps are shown in figure 2 (Leake and Claar 1999; Mehl, Hill, and Leake 2006), in this case, to the Speightstown catchment area. The RM was used to set the boundary conditions for the localized TMR extracted model; with the blue cells being constant heads and the black cells are assigned no flow boundary cells. 6
Regional Model (RM) :All-island Barbados Model RM includes surface and subsurface elevations, BC's and aquifer layer interface location Telescopic Mesh Refinement (TMR) Use of Groundwater Vistas (GWV) for the downscaling to catchment level Speightstown Model (SM) extracted with boundary conditions (BCs) attached from RM Pre-defined BCs from the RM needed as prerequisite before BCs for SM are assigned in TMR Process Figure 2: Procedure for downscaling Regional Model (RM) to Speightstown Model (SM) Two separate models were run from the SM: a steady state and a transient model. The downscaled model was then calibrated against a set of 3 observed hydraulic heads acquired from the Barbados Water Authority (BWA) for the Ashton Hall borehole, Ashton Hall Pumping Station and Whim Pumping Station. These calibration targets are shown in Figure 4 and their values shown in Table 1. According to Jones 2000; I. C. Jones and Banner 2003a and I. C. Jones and Banner 2003b, only 9% to 20% of the average annual rainfall recharges the Barbados aquifer system. Jones 2000 states that recharge is restricted to the wettest 1-3 months of the year and there is less recharge with increases in elevation. Recharge occurs by rapid infiltration and is associated with a monthly rainfall threshold of between 190mm-200 mm (I. C. Jones and Banner 2003a). A uniform recharge was therefore assumed over the entire Speightstown catchment which was calculated from gauge data from the centrally located Orange Hill Gauging Station. The average annual mean rainfall for the years 1960-2004 was calculated as 1575.77 mm per annum. The lowest average recharge was assumed, that is, 9% of 1575.75 mm = 141.8 mm. This was them divided by 365 to attain daily recharge rate of 0.38 mm/day. The porosity, hydraulic conductivity, specific yield extracted during the downscaling process was remained the same for the Speightstown model due to limited available data at the catchment scale. 7
Figure 3: Regional Model with velocity vectors and location of the Speightstown Model 8
Table 1: Calibration Target location and value Figure 4: Spatial location of the calibration targets LOCATION Name X_Coord Y_Coord Target Layer Weight 1 Whim Pumping Station STW1 21660 82835 40 1 1 2 Warleigh Plantation STW3 21660 82830 43 1 1 3 The Whim in gully above Speightstown STW9 22008 83009 56 1 1 Several metrics were run to do initial calibration on the targets provided by the BWA and the model output. The results of these are shown in Table 2 below. These results show that the comparison between the model and observed data are statistically sound with a small difference in the actual hydraulic head value and that of the model generated value. Table 2: Calibration Target values Calibration Value Metrics 1 MAE 7.78 2 RMSE 8.76 3 R 2 0.95 9
5. Conclusion The model developed in this study is to the best of our knowledge, the first to describe threedimensional groundwater flow on the island of Barbados. Development of an island-scale model enabled a more detailed model of the Speighstown catchment model to be developed. We envision that the model will be utilized in the future to support transport models of saline intrusion, which will be critical in assessing the impact of salinization on freshwater resources. Moreover, future work will focus on coupling MODFLOW with WEAP to form a robust framework for assessing various hydrologic stresses on water availability. Acknowledgments This study forms part of the Sustainable Water Management under Climate Change in Small Island States of the Caribbean (Water_aCCSIS Project) which is funded by the International Development Research Centre (IDRC), Canada. 10
References Jones, Ian C., and Jay L. Banner. "Estimating Recharge Thresholds in Tropical Karst Island Aquifers: Barbados, Puerto Rico and Guam." Journal of Hydrology 278, no. 1 4 (2003): 131-43. Jones, I. C., & Banner, J. L. (2003). Hydrogeologic and climatic influences on spatial and interannual variation of recharge to a tropical karst island aquifer. Water Resources Research, 39(9), n/a n/a. http://doi.org/10.1029/2002wr001543 Jones, I. (2000). Estimating recharge in a tropical karst aquifer. Water Resources Research. Retrieved from http://onlinelibrary.wiley.com/doi/10.1029/1999wr900358/full Wellington, C., & Moore, R. (2001). Barbados First National Communications to the UNFCCC. Ministry of Physical Development Environment (p. 125). Retrieved from http://unfccc.int/resource/docs/natc/barnc1.pdf Leake, S., & Claar, D. (1999). Procedures and computer programs for telescopic mesh refinement using MODFLOW. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.400.2243&rep=rep1&type=pdf Mehl, S., Hill, M., & Leake, S. (2006). Comparison of local grid refinement methods for MODFLOW. Groundwater. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.1745-6584.2006.00192.x/full Rumbaugh, J. (2011). Tutorial Manual for Groundwater Vistas. Retrieved from https://scholar.google.com/scholar?q=groundwater+vistas+manual&btng=&hl=en&as_sdt= 0%2C5#5 Rumbaugh, O., & Rumbaugh, O. (2001). Groundwater Vistas User s Manual. Retrieved from https://scholar.google.com/scholar?q=groundwater+vistas+manual&btng=&hl=en&as_sdt= 0%2C5#2 Senn, A. 1946. Report of the British Union Oil Company Limited on Geological Investigations of the Groundwater Resources of Barbados, B.W.I. 11