Finite Difference Model For Evaluating The Recharge Of The Guaraní Aquifer System On The Uruguayanbrazilian Border

Abstract

Part of the precipitation that falls over a region infiltrates through soil layers and eventually
migrates deeper reaching the groundwater reservoir known as aquifer. The water that makes its way to
the aquifer becomes recharge. The magnitude of recharge constitutes an upper bound to the amount of
water that can be extracted from a groundwater reservoir without causing a major impact on it. The
recharge can not be measured directly therefore it has to be determined by an indirect method, and as
such, is one of the most difficult hydrological variables to calculate.
The Guaraní Aquifer System (GAS) is an international transboundary groundwater reservoir. Its
area is about 1200000 km2 distributed in Brazil, Argentina, Paraguay and Uruguay. This aquifer is
contained within the pores and cracks of sandstones (rocks of mostly sandy nature). The Uruguayan-
Brazilian border nearby the cities of Rivera (Uruguay) and Sant´Ana do Livramento (Brazil) is of
special interest as this area receives a significant amount of recharge, both direct from precipitation
and indirect from fractured basalts overlying the GAS, that compensates water abstractions. This
study focuses in this particular area of the GAS, covering about 650 km2 surrounding both cities.
The well known numerical code MODFLOW that solves the parabolic groundwater flow equation
by the finite difference method was used. A steady state condition was simulated representing the
current hydraulic behavior of the groundwater system. The finite different grid consists of 135 rows
and 156 columns, 250 m x 250 m in both, x and y directions. Vertically, the model contains three
layers coincident with the aquifer units defined in the conceptual model, which are basalts, shallow
sandstone aquifer and deep sandstone aquifer. The calibration was performed by the trial-and-error
method, matching simulated hydraulic heads with the observed data. Boundary conditions, recharge
rates, stream/aquifer interphase conductances and hydraulic conductivities were adjusted during the
calibration process.
The conceptual model was correctly validated. Model results approximately match existing data,
although they highlighted data scarcity and the dubious reliability on many available data. In the deep
aquifer the model reproduces adequately the cone of depression detected near the two cities, surveyed
in the field, caused by intensive groundwater pumping. Regarding flow directions, model results
would indicate that most of the streams in the area drain the groundwater reservoir, in agreement with
field evidence. Downward flows were simulated between model layers. In terms of the simulated
recharge rates for the current calibration of the model, the rate over basalts resulted in a meager 1.3
mm/year, and the rate over the outcropping areas of sandstones resulted in 140.2 mm/year, equivalent
to 0.08 % and 8.55 %, respectively, of the mean annual precipitation of 1639 mm estimated at Rivera.