Environmental Study of Agricultural Lands in the Island of Vieques, Puerto Rico

(Ingrid Padilla and Eric Harmsen, UPRM)


Vieques is the largest offshore island belonging to Puerto Rico. It is located about 7 mi east of Puerto Rico and has a population of over 9,000 people (2000 U.S. Census Bureau).The eastern and western parts of Vieques have been used by the U.S. Department of Defense for military training, including bombing practices, for over 30 years. Consequently, the residentsof Vieques live in the central portion of the island, where tourism, agriculture, and fishing dominate the economic market (ATSDR, 2001). Agricultural activities are practiced throughout the entire central portion of the island, but major agricultural efforts are concentrated on several areas considered to have high potential for agricultural development (Figure 1).

Figure 1. Location of study area, ViequesIsland, PR. (TEMPORARY FIGURE)

The island is approximately 23 km long and 5 km wide at its widest point (Figure 1). The island’s topography is characterized by hills and valleys, with the maximum elevation reaching nearly 300 m above mean sea level at Monte Pirata. Most of the island is underlain by highly weather, plutonic rocks and marine volcaniclastic rocks of Cretaceous age. These older rocks are overlain by limestone of tertiary age at several locations. Alluvial deposits occur in the Esperanza and Resolución valleys on the southern and northern sides of the island, respectively. The Esperanza and Resolución alluvium materials, consisting primarily of sands, silt and clay, are significant groundwater aquifers with maximum saturated thicknesses in the range of 25 to 30 m (Renken et al., 2002).

Historically, rainwater and groundwater have been used to supply water for domestic, agricultural, and industrial purposes in Vieques. Water supplies from these sources, however, have diminished since 1977, when an underwater drinking water pipeline from the mainland was built. Water losses from the pipeline (Torres-Gonzales,1984), drought periods, and other stresses on the system have prompted occasional groundwater and rainwater use for domestic and agricultural use.

The military practices in the Island of Vieques have caused potential contamination of soil, water, and biota by chemicals contained in explosive devices (e.g., metals, nitroaromatic compounds, perchlorate) and other military-related contaminants (e.g., solvents). If existent in agricultural lands, this contamination may enter the food chain and pose a great risk to the health of the residents and the environment. A real need, thus, exists to evaluate the potential extent of contamination in groundwater basins and watersheds containing agricultural lands in the island of Vieques.

To meet this critical need, theUniversity of Puerto Rico at Mayaguez has assembled a multidisciplinary team of scientist and engineers, which will investigate the extent of contamination in agricultural lands in the central portion of Vieques. The investigation will involve determining the spatial extent of contamination in biota, soil, surface water, and ground water (saturated and unsaturated), as well as the identification of potential contamination sources in the watersheds containing agricultural lands.


This part of the proposal addresses the investigation concerning potential contamination ofgroundwater and surface water in the watersheds containing agricultural lands in the Island of Vieques, Puerto Rico. The objective of this investigationis toevaluate the possible extent of groundwater and surface water contamination in areas that could impact agricultural practices in the island of Vieques, PR. To achieve this objective, field and computer studies will be conducted. Field studies will be done to evaluate the current extent of groundwater and surface water contamination in the study area. Computer simulation studies will provide a means of assessing migration pathways, the potential rate of chemical transport, and the potential for present and future contamination of surface and subsurface water resources in agricultural lands.

Specifically, this investigation will acquire and analyze all hydrologic and chemical information available, develop a conceptual hydrologic model of the area(s) using Geographic Information Systems (GIS) tools, characterize physical, hydrological, and hydraulic properties of surface and subsurface environments, identify contaminant sources and data gaps, collect necessary hydrologic and contaminant data, delineate the extent of contamination, build contaminant transport models, and evaluate remedial alternatives.

Existing hydrogeologic and water quality data will be compiled and analyzed for the purpose of constructing a site conceptual model. A conceptual model represents our best understanding, both qualitatively and quantitatively, of the hydrologic system and the state ofcontamination. Only after a sound conceptual model has been developed can the numerical model be developed. The study will emphasize the groundwater and surface water basins associated with the agricultural lands. However, areas outside of the agricultural lands will be incorporated into the study as needed to evaluate the potential for future contamination and assure a high level of accuracy in the numerical model. This is necessary because water systems are dynamic and can easily transport contamination across imposed study boundaries. In other words, transport of dissolved contaminants from contaminated areas outside of agricultural lands may cause future contamination of the study area.The site hydrologic conceptual model will include the following factors:

  • Surface conditions: topography and soils
  • Climate: temperature, humidity, radiation, wind speed, evaporation/evapotranspiration
  • Geology: unconsolidated material, bedrock and lithology
  • Hydrologic properties: aquifer thicknesses, confining layers, hydraulic conductivity, storativity
  • Groundwater: hydrogeology, groundwater levels (historical trends, tidal effects); groundwater flow directions, groundwater use, distribution of wells and pumping rates, aquifer recharge and discharge areas and their associated vertical hydraulic gradients
  • Surface-water/ground- water interactions
  • Surface-water: overland and stream flow

The conceptual model on the state of contamination addresses issues concerning the contaminant type, source, amount, location, extent, and mobility. More specifically the study proposes to:

  • Better identify the contaminants of concern, which may include metals, nitroaromatic and nitroamine organic compounds, perchlorate, volatile and semi- volatile organic compounds organics, pesticides and herbicides, radionuclides, and non-aqueous phase liquids
  • Identify sources of potential contamination
  • Evaluateand delineate the extent of groundwater and surface water contamination
  • Determine the fate and transport propertiesof contaminant in surface and subsurface environments

The data collected and evaluated for the conceptual models will be displayed, when appropriate, in map form using GIS.

After completion of the preliminary site conceptual model, data gaps, in terms of groundwater and surface water sampling, will be identified and a plan to collect and analyze complete sets of groundwater levels, streamflow, and water quality samples will be developed. Locations where wells are needed will be determined and wells will be installed. Monitoring stations for streamflow sampling activities will be selected and marked. Sets of groundwater levels, streamflow, and water quality samples will be taken at the selected sites during the same sampling periods so that they correspond to the same hydrologic conditions. Groundwater samples will be collected during the dry and wet season, but surface water samples will only be collected during the wet period, given that streamflow occurs intermittently from ephemeral streams. A rainwater collector will also be installed and used to determine water quality of rainwater near major agricultural lands. Sufficient samples and measurements will be takenfor each sampling period so that the results are statistically acceptable. Water quality sampling will follow a strict Quality Assurance/Quality Control Plan based on EPA established protocols. Water samples will be analyzed for contaminants of concern and for major ions. Analysis of the water samples will be conducted mostly at the Analytical Chemistry Laboratoryat the University of Puerto Rico, Mayagüez Campus.

The new groundwater elevation and contaminant distribution survey will be incorporated into the site preliminary conceptual model and the models will be redefined prior to a complete assessment on thepossible extent of groundwater and surface water contamination in areas that could impact agricultural practices. Complete assessment will be done by analyzing the redefined conceptual model using GIS tools and numerical models.


The following models will be utilized in the project:

  • One-dimensional vadose zone water flow and solute transport model (HYDRUS-1D)
  • This model will be used at various locations to estimate the rate of chemical migration from the surface soil to the water table.
  • Three-dimensional groundwater flow model (MODFLOW)
  • This model will encompass the entire groundwater basin associated with the study area.
  • Three-dimensional solute transport model (MT3DMS)
  • This model will be coupled with the groundwater flow model. In areas where the groundwater contamination is limited in extent, smaller-scale transport models may be developed in order to achieve a finer grid resolution.
  • Land surface hydrologic model (e.g., HSPF)
  • This model will encompass all watersheds associated with the study area and/or the area covered by the groundwater flow model. The model is capable of running continuous long-term scenarios, considering surface runoff, solute and sediment transport.

Vadose Zone Model

Pore water flow and solute transport in the vadose or unsaturated zone will be simulated using the HYDRUS-1D model. The model will be capable of estimating the rate at which selected chemicals enter the groundwater with rainfall percolation. Estimated pore water concentrations and loading rates can depend on most or all of the following transport processes: advection, dispersion, diffusion, adorption and degradation. The model requires as input, soil hydraulic and transport properties with depth, daily rainfall and potential evapotranspiration, contaminant properties, and the initial contaminant concentration distribution in the soil.

As a part of the project, zones of elevated soil contaminant concentrations will be delineated using ArcView to produce maps of soil chemical concentration distribution. Use of this information, along with soil and climate information, will be used to develop a map showing the distribution of chemical loading to the groundwater within the study area.

3. Groundwater Model

A fully three-dimensional, transient groundwater model will be constructed using the USGS Modular Three-Dimensional Ground-Water Flow Model (MODFLOW) developed by McDonald and Harbaugh (1984). Because of its ability to simulate a wide variety of systems and its rigorous USGS peer review, MODFLOW has become the worldwide standard groundwater flow model. MODFLOW is used to simulate systems for water supply, contaminant remediation and regional groundwater basin processes. MODFLOW has been the recognized standard model used by courts, regulatory agencies, universities, consultants and industry.

The GIS-based user interface GMS (Groundwater Modeling System) will be used to manipulate input and output databases. GMS was developed under the direction of the U.S. Army Corps of Engineers and involved support from the Department of Defense, the Department of Energy, and the Environmental Protection Agency. Tools are provided for site characterization, model conceptualization, finite-difference grid generation, geostatistics, telescopic model refinement, and output post-processing.

Generally the steps needed to develop a groundwater model within GMS include: 1) import map images, 2) delineate extent groundwater basin, 3) construct finite-difference grid, 4) develop aquifer recharge zones, 5) input layer data (e.g., aquifer thicknesses, hydraulic conductivities, etc.) and 6) assign initial and boundary conditions (e.g., flux or constant head boundaries, pumping wells, etc.).

Calibration, validation and sensitivity analysis are necessary steps in the modeling process. The calibration is performed in order to arrive at estimates for model parameters, which will reproduce field conditions. The purpose of the validation is to establish greater confidence in the model. During model validation the calibrated model is used to simulate a period for which conditions differ from the calibrated conditions. The sensitivity analysis is performed in order to evaluate the effect of parameter uncertainty on the calibrated model.

The groundwater elevation data set used for calibration (steady-state) of the model will be obtained from a groundwater survey of the area during the first year of the study. The groundwater data will be collected during the dry season (e.g. February 2004). Calibration will be achieved by adjusting aquifer properties within reasonable limits in order to match observed average groundwater levels and discharge rates. Hydrologic analyses of stream base flow (if any) and springs will provide discharge rates for comparison with the groundwater flow model. Model calibration procedures used in this study will comply with guidance provided in the Standard Guide for Comparing Ground-Water Flow Model Simulations to Site-Specific Information (D5490-93).

If sufficient historical data is available, a transient calibration will also be conducted as well. Approximately six months of continuous monitoring well data was reported by Torres-González (1989) for two wells in the EsperanzaAlluvialValley, which may be usable for the transient calibration.

Automatic calibration will be performed using the optimization software PEST (Parameter ESTimation, Doherty, 1994). For optimization of parameter values, PEST employs a robust nonlinear parameter estimation procedure. It runs the model as many times as necessary in order to find that parameter set that minimizes the weighted least squares discrepancies between model-generated heads and corresponding measured heads. PEST provides several statistical results (root mean squared error, upper and lower confidence limits on parameters, covariance matrix, correlation matrix, Eigenvectors and Eigenvalues) which can be useful in managing the calibration process.

The groundwater flow model will be validated by comparing simulated results against an independent groundwater elevation data set. The groundwater elevations will be measured during the second year of the project in November 2004 when conditions are normally wet in Vieques. Methods used for comparing the field and model-generated data will comply with guidance provided in the Standard Guide for Comparing Ground-Water Flow Model Simulations to Site-Specific Information (D5490-93).

A sensitivity analysis will be performed to identify which model inputs have the most impact on the degree of calibration and on the conclusions of the modeling analysis. This approach is recommended in ASTM Standard D 5611-94, Standard Guide for Conducting a Sensitivity Analysis for a Ground-Water Flow Model Application. In the analysis, up to five parameters (e.g., horizontal hydraulic conductivity, VCONT multiplier, aquifer recharge, constant head boundary elevations, evapotranspiration extinction depth, streambed conductance) will be considered. Parameter values will be analyzed at high and low extreme levels. As part of the sensitivity analysis, particle tracking will be performed using the computer program MODPATH. MODPATH computes three-dimensional flow paths using output from MODFLOW. The sensitivity of the particle tracking simulations will reveal how sensitive potential contaminant pathways are to variations in model parameters.

Groundwater Solute Transport

Evaluation of soil and groundwater samples will determine which the contaminants of concern are and which may need to be modeled. For the purposes of this proposal we assume that no more than three contaminants will be considered for solute transport modeling. Transport of chemicals within the groundwater system will be simulated using the computer model MT3DMS. MT3DMS is capable of simulating advection, dispersion, diffusion, adsorption and decay of chain-reaction coupled species in three-dimensional aquifer system systems.

As part of the groundwater model development, the finite-difference grid will be designed so as to avoid numerical problems during the transport simulation phase. Configuring the numerical transport model involves specification of aquifer and contaminant transport properties, assignment of source loading rates, and distribution of initial groundwater concentrations.

PEST will be used to calibrate the solute transport model. Variations in at least three transport parameters will be considered (e.g., dispersivity, partition coefficient and decay constant). Temporal concentration data needed for the calibration will be obtained from the four field samplings. Unfortunately, given the relatively short duration of the project it will not be possible to validation the solute transport model. A sensitivity analysis will be performed in which at least three transport parameters (e.g., dispersivity, partition coefficient and decay constant) will be adjusted to evaluate their effect on the predicted concentration at specific location of interest (e.g., irrigation wells).

Surface Hydrologic Model

Surface water flow and solute/sediment transport modeling will be accomplished using the computer model Hydrologic Simulation Program Fortran (HSFP). HSPF is an U.S. EPA program for simulation of watershed hydrology and water quality for both conventional and toxic organic pollutants. The model uses information such as the time history of rainfall, temperature and solar radiation; land surface characteristics such as land use patterns; and land management practices to simulate the processes that occur in a watershed. The result of this simulation is a time history of the quantity and quality of runoff from an urban or agricultural watershed. Flow rate, sediment load, and nutrient and pesticide concentrations are predicted. Input needed to run HSPF will be developed using the GIS-based pre/post processor software Watershed Management System (WMS). WMS provides tools for all phases of watershed modeling including automated watershed and sub-basin delineation, geometric parameter computation, hydrologic parameter computation (CN, time of concentration, rainfall depth, etc.) and result visualization.

The major tasks associated with configuring HSPF are: