Skills in sub-seasonal to seasonal terrestrial water storage forecasting: insights from the FEWS NET land data assimilation system

Abstract. Accurate prediction of terrestrial water storage (TWS), the sum of soil moisture, groundwater, snow/ice, and surface water, is critical for informing disaster responses. Here we evaluated subseasonal to seasonal (S2S) TWS forecasts produced by the Famine Early Warning Systems Network (FEWS NET) land data assimilation system (FLDAS) over Africa using observations from the Gravity Recovery and Climate Experiment (GRACE) and its Follow-On (GRACE/FO) mission. FLDAS consists of two advanced land surface models, Noah-MP and the NASA Catchment Land Surface Model (CLSM), both of which simulate key TWS components including groundwater. Results show that CLSM generally outperformed Noah-MP, with relative operating characteristics scores exceeding 0.6 (the threshold for predictive skill) for tercile forecasts over >50 % of the study domain across the 1–6 months lead times, and stronger correlations with GRACE/FO data. The superior performance of CLSM is largely attributed to its reanalysis-based initial conditions, which more accurately captured interannual variability observed in GRACE/FO observations (correlation of 0.72 vs 0.56 for Noah-MP for domain averaged TWS). CLSM also simulates strong TWS temporal variability and thus temporal persistence, enabling skillful initial conditions to propagate across forecast lead times. Accurate representation of interannual variability is essential for S2S forecasts because TWS is a long memory process, and interannual variability also directly affects climatology used to determine anomalies. Although persistence provides a source of predictability, this study shows that inaccurate persistence, such as that associated with anthropogenic trends and misrepresented precipitation variability, can degrade forecast skill. TWS forecasts from both models are also highly sensitive to precipitation interannual variability, achieving higher forecast skill when driven by precipitation forecasts with lower interannual variability. These findings underscore strong impacts of model physics and the critical role of independent observations such as GRACE/FO for evaluating and improving TWS forecasts.