Gap-Filling Between GRACE and GRACE-FO Missions Over Africa

The Gravity Recovery and Climate Experiment Follow-On (GRACE-FO) was launched in May 2018 and is intended to continue monitoring mass variations provided by the earlier GRACE mission (2002-2017) for a better understanding of mass exchange and transport between the Earth's continents, oceans, and ice sheets. A recurrent dynamic artificial neural network (ANN) model was developed to fill the expected one-year data gap between the two GRACE missions and to forecast drought events using 10 major African watersheds as test sites. The ANN model derives relationships between GRACE-derived terrestrial water storage (TWS_GRACE) record and the controlling and/or related climatic and hydrologic variables (e.g., rainfall, temperature, evapotranspiration, and NDVI). The ANN model was trained (training period: April 2002—September 2013), tested (prediction period: October 2013—September 2015), and used to forecast TWS_GRACE for the period from October 2015 through March 2016. The model performance was evaluated using the Root Mean Square Error (RMSE) and Nash-Sutcliff Efficiency (NSE). The performance of the model was found to be "very good" (NSE > 0.75; RMSE < 0.5) for 60% of the investigated watersheds, "good" (NSE > 0.65; RMSE < 0.6) for 10%, and "satisfactory" (NSE > 0.50; RMSE < 0.7) for the remaining 30% of the watersheds. For the period October 2015 through March 2016, the model successfully predicted no drought events over the Niger basin, the termination of the latest (March-October 2015) drought event over the Zambezi basin, and the onset of a drought event (January-March 2016) over the Lake Chad basin. Ongoing research activities are focused on: (1) forecasting TWS_GRACE data as well as drought and flooding events, for one year in advance, over the entire African watersheds, and (2) downscaling TWS_GRACE record to produce high-resolution estimates of hydrologic variables.