This dataset is linked to the manuscript titled "GRACE-FCast: a global long-lead forecast of total water storage for 2010-2024," which focuses on seasonal to annual forecasting terrestrial water storage anomalies across the global land surface. The dataset provides global forecasts of terrestrial water storage anomalies using machine learning, based on near real-time observations of precipitation, land and sea surface temperatures, sea surface salinity, soil moisture, evaporation, and sixteen climate indices. Forecasts are initiated monthly from December 2009 to April 2024, with lead times of up to twelve months. The data is available both with and without seasonal and trend components, at a one-degree grid scale, covering global land areas including Antarctica and Greenland. The dataset also includes multiple versions of the global mean TWSC time series for each lead time. We expect this dataset will contribute to drought and flood early warning services, sea level forecasting, constraints on hydrological and land surface model forecasts, and Earth orientation parameter forecasting, and address the latency issue in accessing standard GRACE TWSC products.

The zip folders (CSR, GSFC and JPL) contains: Forecasted TWSC by training machine learning on CSR, GSFC, and JPL mascons.