The Famine Early Warning Systems Network (FEWS NET) Land Data Assimilation System v2 (FLDAS-2): Integrating modern datasets to improve hydrological modeling

The Famine Early Warning Systems Network (FEWS NET) Land Data Assimilation System (FLDAS) has been in operation since 2017 and is the predecessor to the FLDAS-2 model. Since 2017, there have been improvements in remotely sensed datasets, land surface models, and data assimilation methods. The creation of FLDAS-2 allows us to leverage these improvements to better model hydrological conditions, especially in food insecure areas of the world where in-situ data are scarce and droughts are frequent. FLDAS-2 model output are created at NASA Goddard for public use and are used by various partners to try and proactively provide humanitarian aid to vulnerable areas of the world that are affected by droughts and floods. This study shows how integration of new remotely sensed datasets (Integrated Multi-satellitE Retrievals for GPM (IMERG)), new land surface models (Noah-MP 4.0.1), and assimilation of remotely sensed data (leaf area index (LAI)) have all improved hydrological modeling in FLDAS-2. Soil moisture monthly anomaly correlations, which are calculated using data from the Soil Moisture Active/Passive (SMAP) dataset, are improved in FLDAS-2 by as much as 0.75 in certain subdomains (CONUS, Central Asia, and Southern Africa). A similar analysis with the operational Simplified Surface Energy Balance (SSEBop) model shows less consistent results with changes in evapotranspiration monthly anomaly correlations as high as 0.40 and low as -0.45 in the East Africa subdomain. Early testing of assimilating leaf area index into the NoahMP 4.0.1 dynamical vegetative model shows wide-spread improvements in surface soil moisture and evapotranspiration in the East Africa subdomain. These changes in FLDAS-2 are allowing us to provide the users of these data with more accurate representations of hydrological conditions on a global scale.