An efficient data assimilation (DA) approach for integrating GRACE-C, NGGM, and MAGIC data into global high resolution hydrological models

The state-of-the-art Data Assimilation (DA) can be applied to take advantage of time-variable satellite gravity observations to constrain the vertical sum of water storage simulations of large-scale hydrological models. These DA implementations are, however, often performed regionally, and if globally, at low spatial resolution. This choice is made to manage the considerably high computational demands of hydrological DA at a global scale and to avoid numerical problems, e.g., instabilities related to the inversion of covariance matrices. To fully exploit the potential of future satellite gravity observations, including GRACE-C, NGGM, and MAGIC missions, we will test the performance of the AAU’s Python-based open-source DA system, known as PyGLDA, to integrate the global estimates of terrestrial water storage (TWS) from these missions with a global hydrological model at 0.1° spatial resolution. The main novelties of our experiments are: (i) testing the effect of spatial resolution and error structure of the future gravity missions on the global fine-resolution DA-derived water storage estimates; (ii) evaluating the effects of basin-scale and grid-scale strategies on the experiment 1; and (iii) evaluating the effect of domain localization and neighbouring-weighted global aggregation on the efficiency of global DA implementations. As observations, monthly GRACE-C-like, NGGM-like, and MAGIC-like TWS estimates with their full error covariance matrices will be used. The data sets are the outputs of closed-loop simulations of the ESA SING project, which are computed by considering both instrumental and dealiasing errors. The relevance of the experiments for operational hydrological applications will be evaluated by computing indicators such as wetness/dryness indices and storage deficits to explore the performance of the DA experiments in downscaling satellite gravity products.