Modelización del flujo subterráneo de dos cuencas de tamaño medio empleando dos estructuras (modelos) distintas

Abstract

The surface and groundwater hydrology of two medium size Belgian catchments(the Gete and Jeker, with an area of about 621 km2 and 465 km2, respectively) were modeled using the distributed physically based code MIKE SHE and the lumped conceptual code NAM. For the calibration of the stream flow, it was used data from the catchment outlets, whereas for the calibration of groundwater flow, it was used observed levels from piezometers distributed throughout each one of the study catchments. The validation process was carried out through a common split-sample (temporal) analysis for either of the hydrologic codes (MIKE SHE or NAM) as well as through a multi-site (i.e. spatial) analysis, for the distributed MIKE SHE code, using control stations (i.e. discharge and piezometric) that were not consider during model calibration. A sensitivity analyses was run varying parameters like the river-aquifer leakage coefficient, the drainage level, the reciprocal time constant and the hydraulic conductivity for the most influential layers in each catchment. The results of this investigation suggest that the NAM model is capable of better predicting the observed discharges; further, the simulation of low (i.e. base) flows suggest an acceptable simulation of groundwater flow rates. Nevertheless, the MIKE SHE predictions are acceptable, with the significant advantage of producing 3-D distributed prediction of both, river discharge as well as piezometric levels. The results encourage the use of either model of the study catchments, depending on the purpose of the modeling.