Improving global evaporation estimation using GRACE and GRACE-FO satellite data assimilation

Abstract

The accurate monitoring and prediction of land water cycle components are crucial for applications in climate, hydrology, and agriculture. However, the remote sensing of ecohydrological variables, though essential, still faces challenges, especially in estimating non-directly observable factors like evaporation. Utilizing GRACE and GRACE-FO satellite data has the potential to improve global evaporation estimates and therefore to enhance our ability to understand and manage these components. Such advancements in global evaporation estimation can furthermore contribute to addressing future water management challenges, including mitigating the impacts of drought and potential groundwater reductions. To date, several remote sensing assets have been underused within the context of global evaporation estimation, such as the Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) missions, active since 2002 and 2018, respectively. These missions can play a key role in representing surface and subsurface processes related to water redistribution, providing estimates of Terrestrial Water Storage (TWS), and enriching our ability to navigate global water resource complexities.The goal of this research is to improve the estimates of evaporation from the Global Land Evaporation Amsterdam Model (GLEAM) by using GRACE and GRACE-FO observations. GLEAM is a set of algorithms dedicated to estimating terrestrial evaporation based on satellite observations of meteorological drivers of terrestrial evaporation, vegetation characteristics, and soil moisture (Miralles et al. 2011). In this regard, we use GRACE observations in a data assimilation approach, based on Newtonian Nudging with model and observation errors defined by triple collocation, to improve the evaporation estimates of GLEAM. The study period comprises 20 years, between January 2003 and December 2022. Preliminary results indicate that the data assimilation output is closer to reality, for instance for estimating evaporation changes in Brazil and South America.