GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 91-10
Presentation Time: 10:45 AM


REEVE, Andrew S.1, GÓMEZ ARÉVALO, Efrén David1 and GORDON, Ryan P.2, (1)School of Earth and Climate Sciences, The University of Maine, 5790 Bryand Global Sciences Center, Orono, ME 04469, (2)Maine Geological Survey, 93 State House Station, Augusta, ME 04333

About 160,000 inhabitants live in the 1600 sq. km. Carraipia River Basin in northeastern Colombia and northwestern Venezuela. Historically, water has been supplied to inhabitants of this arid coastal region by shallow dug wells that frequently is of poor quality due to the regions high potential ET (1700 mm/yr) relative to precipitation (1000 mm/yr). Colombian agencies seek more quantitative information on groundwater resources to alleviate severe potable water shortages that adversely affected the area.

Groundwater flow models were constructed to identify the most important hydrogeologic data needed in this region and create plausible baseline models for groundwater pumping scenarios. MODFLOW 2005 with the SWI2 package are used to simulate groundwater flow and saltwater intrusion. Idealized models of the basin's hydrostratigraphy are based on geologic maps, hydraulic well tests, field observations, and stratigraphy data from previously drilling water wells. Simulated hydraulic heads are compared to dug well data, and water-table position are compared to the ground-surface elevation to evaluate model plausibility. Modeling results indicate that the SWI2 package has limited impact on the modeling areas where water wells are located. Using a baseline model, poorly constrained hydraulic conductivities and hydraulic stress parameters (conductances and rates) are systematically varied. The results of these models are assessed to identify plausible models for future pumping scenarios. Results suggest that the most plausible models are produced by increasing the evapotranspiration rate, lowered recharge rates, and increased river conductance relative to the baseline values of 1.e-11 m/s, 2.5e-9 m/s and 1.4e-9*(river length) sq. m/sec, respectively. Many models that closely matched the measured water-table elevations were considered poor because the water-table was positioned above the ground surface in the highlands, where hydraulic head data were not collected. Increasing ET rates 10 to 100-fold produced the most plausible models, likely because ET removes larger amounts of water when the water table is near the ground surface and thus constraining the water-table position. Changing hydraulic conductivity in the poorly constrained deeper layers of the model had little impact of the water-table position.