Global Land Water Storage Forecast Release 1 (GLWFC1.0) ----------------------------------------------------------------------------------- University of Bonn Authors: Fupeng Li, Juergen Kusche Institute of Geodesy and Geoinformation Nussallee 17 D-53115 Bonn Date: 26 May 2025 1. Data Inputs ----------------------------------------------------------------------------------- This dataset provides forecasts of Terrestrial Water Storage Anomaly (TWSA) generated by machine learning models trained against the GRACE/-FO mascon solutions from the Jet Propulsion Laboratory (JPL RL06.3_v04). The forecasts utilize forward-moving and lagged predictors identified from observational/reanalysis products of seven hydrometeorological variables: ERA5 Precipitation (P) ERA5 Temperature (T) ERA5 Evaporation (E) ERA5 Runoff (R) ERA5 Soil Moisture (SOM) MOHC Sea Surface Temperature (HadSST) MOHC Sea Surface Salinity (HadSSS) (Note: MOHC - Met Office Hadley Centre) 2. Methodology ----------------------------------------------------------------------------------- The TWSA forecasts are produced by combining: Forecasted interannual and sub-seasonal components, predicted using a Long Short-Term Memory (LSTM) model trained against the corresponding components separated from GRACE TWSA. Seasonal and linear trend components, extrapolated from prior GRACE seasonal and linear terms, derived over the period from April 2002 to the initialization date. 3. Product Coverage and Resolution ----------------------------------------------------------------------------------- Forecast Range: TWSA forecasts are provided up to 12 months ahead, initialized monthly from January 2024 onward. Temporal Resolution: Monthly Spatial Resolution: 1-degree grid covering global land areas, including Greenland and Antarctica. (Note: Caution is advised when interpreting results over glacier-covered regions, as model performance is reduced in these areas). 4. Others ----------------------------------------------------------------------------------- The dataset forecasts the full TWSA signal as captured by the JPL mascon solutions. Users of land TWSA or mascon TWSA maps can apply our forecasts directly, without the need for additional corrections such as glacial isostatic adjustment (GIA) trend removal or geocenter motion augmentation. 5. References ----------------------------------------------------------------------------------- Li, F., Springer, A., Kusche, J., Gutknecht, B., Ewerdwalbesloh, Y. (2025). Reanalysis and Forecasting of Total Water Storage and Hydrological States by Combining Machine Learning With CLM Model Simulations and GRACE Data Assimilation. Water Resources Research, e2024WR037926. Li, F., Kusche, J., Sneeuw, N., Siebert, S., Gerdener, H., Wang, Z., ... & Tian, kunjun. (2024). Forecasting next year's global land water storage using GRACE data. Geophysical Research Letters, 51(17), e2024GL109101. 6. Contact ----------------------------------------------------------------------------------- For questions or further information, please contact: Fupeng Li Institute of Geodesy and Geoinformation Nussallee 17 D-53115 Bonn Germany fpli@uni-bonn.de