Beyond Traditional Metrics: Unravelling Hydrological Systems with a Response Time Evaluation Approach

International audience ; Hydrological, land surface, or climate models are commonly assessed using classical metrics (e.g., Nash-Sutcliffe Efficiency, Root Mean Square Error, etc.) to evaluate their performance in simulating total water storage (S) and streamflow (Q) either separately or jointly. However, these integrative metrics may provide limited insights into the descriptive capabilities of system dynamics, as they heavily rely on the quality of boundary conditions. In contrast to traditional hydrological modelling and approaches that predominantly operate in the time domain, this study introduces a novel approach based on the concept of hydrological ”response time”, physical information linked to system intrinsic properties, which is computed as the ratio S/Q in the frequency domain.This approach explores how different components of the hydrological system respond to variations in input signals at different frequencies, offering reduced sensitivity to boundary conditions. An additional advantage of this approach is its ability to define dominant exchange processes between soil, rivers, and groundwater across various frequency ranges. This methodology contributes to a more robust and physics-based assessment of hydrological system behaviour, fostering a deeper understanding of water dynamics. To demonstrate the effectiveness of this approach, we evaluate Phase 6 of the Coupled Model Intercomparison Project (CMIP6) and the System Global Land Data Assimilation System (GLDAS)-driven Noah models against the Global Runoff Data Centre and the Gravity Recovery and Climate Experiment (GRACE) satellite time series of observed Q and S, respectively, across more than 50 large basins worldwide. The results will be presented and discussed.