Assimilation of GRACE/-FO data into global hydrological models for water use scenarios

Abstract

Global Hydrological Models (GHMs) are an invaluable tool for simulating the dynamics of our freshwater cycle and estimating its contribution to sea level rise. However, uncertainties of input data (e.g., meteorological forcing, water demand estimates) and empirical parameters, as well as errors in the model structure (e.g., simplified representation through conceptual equations) lead to hydrological predictions that can vary strongly between different GHMs. Validation with independent satellite and in-situ observations shows that no single model can outperform the others in all regions and in all periods. Thus, assimilation of hydrological state variables into GHMs is a great opportunity to reduce model prediction uncertainty and improve performance statistics. Over the last two decades, the assimilation of terrestrial water storage (TWS) anomalies observed by the satellite gravity missions GRACE and GRACE-FO into land surface and water balance models has shown great potential to improve predictive capacity.In this presentation, we focus on exploiting the potential of TWS data assimilation into GHMs to reveal the human water use impact on shifts of TWS patterns under changing climate. We compare two decades of TWS simulated by the open-source and Python re-programmed WaterGAP Global Hydrology Model (WGHM) version while (a) disregarding and (b) considering surface water and groundwater extraction to isolate the human impact on the terrestrial water cycle. The identified patterns are compared to GRACE/-FO-derived TWS long-term trends to identify their correlation for regions with intense water use activities. This enables to map regions under acute and chronic water stress, where water overconsumption shows significant impact on shifting TWS patterns. In addition, we perform a global TWS data assimilation experiment into the World-Wide Water Resources Assessment (W3RA) model using the open-source Python-based Global Land Data Assimilation system (PyGLDA) developed by our research group. A comparison of the W3RA predictions before and after data assimilation reveals the spatial and temporal signatures that are introduced from GRACE/-FO observations. Comparisons to the isolated human impact on the terrestrial water cycle from WGHM simulations enables the generation of a global map displaying the potential of GRACE/-FO data for improved representation of human activities on the water cycle.