Search for a command to run...
Due to the inherent complexity of hydrological systems, single observation data types rarely contain sufficient information for model calibration, particularly for distributed models with large parameter dimensions. Previous studies have demonstrated the ability of multi-objective calibration to constrain equifinality and promote good model practices. However, most of these efforts have focused mainly on predictions of discharge and ET, while lacking evaluations of other key components of ecohydrological systems (e.g., subsurface storages and fluxes, and ET partitioning), potentially leading to biased hydrological inferences. Although incorporating multiple objectives provides additional constraints on the modelling, degradations in model performance often occur due to trade-offs among observations. Evaluations of these trade-offs are necessary for robust modelling inferences. Therefore, we applied multiple calibration schemes combing discharge, isotope, spatial and temporal patterns of remote sensed ET in an ET-dominated catchment (the Mid-Spree, 2800km2) of the river Spree, NE Germany, to constrain a spatially distributed tracer-aided model (STARR) over a 20-year period. Since the Spree is a major water source supplying Berlin’s drinking water, agricultural irrigation and industrial needs, ensuring trustworthy hydrological modelling, realistic process representation and careful consideration of calibration strategies is of vital importance.Our findings show that compared to discharge-only based calibrations, additional incorporation of either isotope, temporal or spatial patterns of ET produced distinct process insights. These multi-variable calibrations revealed contrasting trade-offs, with slightly degraded discharge performance but clear improvements in the additional calibrated variables (i.e., isotope or ET patterns). Temporal patterns of ET contained similar information to discharge, and provided limited additional insights into catchment functioning. In contrast, incorporating isotopes and spatial patterns of ET in addition to discharge reduced simulated discharge volumes generated in the Mid-Spree region (>70% of discharge at the outlet of the catchment originated from the upper Spree), accompanied with slower lateral flow rates in the subsurface layer, reflecting a slow water celerity of the catchment. Isotope-aided calibrations also inferred large subsurface water storage to reproduce the damped isotope variations observed in the field, and higher ET peaks (compared to other calibration schemes) during summer. Conversely, calibrations constrained by the spatial patterns of ET indicated lower subsurface water storage, compared to calibrations constrained by other variables. This study demonstrated the implications and trade-offs of using multiple observational targets in model calibration in large scale, heavily anthropogenically influenced catchments, helping to identify more reliable parameterizations and to improve process-based understanding in distributed hydrological modelling.