Impact of the model structure and calibration strategy on baseflow modeling in the German low mountain range

Abstract

Baseflow is a vital component of the water balance. The fractured hard rock aquifers of the German low mountain range are in danger of increased water stress due to climate change because they react rapidly to deficits in precipitation and groundwater tables decline sharply. Therefore, simulation software must be able to model baseflow accurately. Three soil moisture simulation and two monthly factor-based baseflow models are evaluated using two calibration strategies. Models were calibrated to total flow (S1) or stepwise to baseflow and then total flow (S2). Results were not significantly different for total flow. Regarding baseflow, S2 proved significantly better with median values (S1 calibration, validation | S2 calibration, validation) of SSE (20.3, 20.3 | 13.5, 13.8), LnNSE (0.15, 0.17 | 0.47, 0.34), and PBIAS (27.8, 21.6 | 2.5, −0.8). Parallel linear reservoir proved best at modeling baseflow with a median SSE (S2: 6.1, 5.9), LnNSE (S2: 0.64, 0.71), and PBIAS (S2: 3.8, 3.8). The new modified monthly factor approach is a simple and robust alternative with SSE (13.0, 13.3), LnNSE (0.61, 0.61), and PBIAS (9.8, −8.6). The results are useful regarding selection of baseflow model structure and calibration strategy in low mountain ranges with fractured hard rock aquifers.