Advancing global-scale hydrological modeling and the understanding of the global freshwater system by model development and calibration/data assimilation

点击翻译按钮获取中文摘要

During the first 28 months of the first phase of the research unit GlobalCDA, P2 finalized the new WaterGAP 2.2d version used in the research unit, which was adapted to enable EnCDA (Ensemble Kalman Filter calibration/data assimilation) and POC (Pareto optimal calibration). The development of a global gradient-based groundwater model was finalized in a stand-alone steady-state version; extensive sensitivity analyses and comparisons to other models and independent data identified scale-related difficulties for validating simulation results against groundwater well data. Multi-variable calibration of WaterGAP using EnCDA, POC as well as GLUE (Generalized Likelihood Uncertainty Estimation) for the Mississippi basin clarified conceptual and practical differences of the three ensemble-based calibration approaches. Calibration was successful in identifying behavioral and optimal parameter sets that lead to improved model performance as compared to the simple standard calibration procedure for WaterGAP. The main objectives of P2 for the second phase of GlobalCDA are 1) to refine, together with P1, P3 and P10, the calibration/data assimilation (C/DA) approach such that it can be applied globally to compute best estimates of WaterGAP output variables and their uncertainties, also by exploring together with P3 how to combine POC and GLUE, 2) to develop and improve components of WaterGAP (integration of gradient-based groundwater model and representation of man-made reservoirs) and to provide alternative model variants for C/DA, 3) to validate model outputs and 4) to assess water storages and flows in critical regions and globally as estimated by WaterGAP after C/DA, taking into account uncertainties.