This work is part of a project aiming to determine the origins of the
nitrate contamination in the urban area of Urnia (SP, Brazil) and the
objective is to make a conceptual and numerical model of the groundwater
flow system which will improve the understanding of the groundwater
circulation in the area.
The conceptual model is based on data from earlier studies in the area
complemented with data collected in the field. The model area is delimited
by streams and water dividers attribut-able to topographic highs. The upper
boundary of the model domain is the water table and the bedrock is an
impermeable lower boundary. Two geological formations are considered in the
model: a sandstone aquifer and a layer of fractured basaltic rock
underneath. It is assumed that the hydraulic conductivity of the sandstone
aquifer is isotropic in the plane and varies gradually from its lowest
value in the north of the area to its highest value in the south. The
hydraulic conductivity of the fractured basalt is estimated to be two
orders of magnitude lower than for the sandstone. Groundwater is assumed to
flow from topographic highs towards the streams. No evidence of a deep
groundwater flow underneath streams has been found and it is thus assumed
that there is no groundwater movement to or from the model domain.
The software Visual MODFLOW is used for the numerical model. A grid is made
with 106 columns, 97 rows and five layers, where the four top layers
represents the sandstone aquifer and the bottom layer represents the
fractured basalt. The streams of the conceptual model are simulated as
drain boundaries. To get the model to converge and in order to get
reasonable flow directions, constant head boundaries are located in layers
2 ? 5 in cells below drain boundaries. Static head values and stream flows
have been used as quantitative calibration targets whereas groundwater flow
directions and the distribution of hydraulic conductivities have been used
as qualitative calibration targets. A combination of automatic and manual
trial-and-error calibration have been performed, where hydraulic
conductivities and recharge have been automatically optimized, while drain
conductance, constant head elevation and the number and location of zones
with different hydraulic conductivities have been modified manually. An
effort has been made to calibrate the model when pumping of public wells in
Urnia is included, but no calibrated model has been obtained for the
pumping scenario. After calibration, the sensitivity of the model for
selected parameters has been analysed.
This study shows that groundwater circulation in Urnia is more complex
than was previously assumed and more data needs to be collected in order to
obtain a thorough understanding of the flow system. The horizontal
hydraulic conductivity of the Adamantina aquifer increases from north to
south, at least within the urban area of Urnia, but more data is needed to
evaluate the extent of this variation. Calibration data also need to be
improved. Boundary conditions are crucial for the reliability of the model
and are at present imperfectly under-stood. Vertical groundwater movement
need to be better understood in order to obtain a ...