An Appraisal of the Local-Scale Spatio-Temporal Variations of Drought Based on the Integrated Grace/Grace-fo Observations and Fine-Resolution Fldas Model

Abstract

The gravity recovery and climate experiment (GRACE) observations have so far been utilized to detect and trace the variations of hydrological extremes worldwide. However, applying the coarse resolution GRACE estimates for local-scale analysis remains a big challenge. In this study, a new version of the fine resolution (1 km) Famine early warning systems network Land Data Assimilation System (FLDAS) model data was integrated into a machine learning model along with the GRACE data to evaluate the subbasin-scale variations of water storage, and drought. With a correlation of 0.99 and a root mean square error (RMSE) of 3.93mm of its results, the downscaling model turned out to be very successful in modelling the finer resolution variations of TWSA. The water storage deficit (WSD) and Water Storage Deficit Index (WSDI) were used to determine the episodes and severity of drought events. Accordingly, two severe droughts (January 2008 to March 2009 and September 2019 to December 2020) were discerned in the Kizilirmak Basin (KB) located in Central Turkiye. The characterization of droughts was evaluated based on WSDI, scPDSI, and model-based drought indices of the soil moisture storage percentile (SMSP) and groundwater storage percentile (GWSP). The results indicated discrepancies in the drought classes based on different indices. However, the WSDI turned out to be more correlated with GWSP, suggesting its high ability to monitor groundwater droughts as well.