Hierarchical Graph Networks for Forecasting Terrestrial Water Storage Anomalies

The availability of fresh water is vital to the ecosystem and communities. In a changing climate, the increased risk of droughts makes it more important to have an accurate view of changes in terrestrial water storage (TWS). Predicting changes in TWS is inherently difficult since it integrates the changes of all water compartments, with underlying processes that operate on vastly different temporal and spatial scales. In our work, we explore a novel design of a hierarchical graph using domain knowledge of hydrological basins to encode these processes in a latent feature sequence using an encoder-processor-decoder style graph neural network. The subsequent recurrent neural network then forecasts changes in TWS from the latent feature sequence and historical TWS evolution for up to six months ahead. The gridded product of the seasonal forecast of global TWS evolution shows improvement over a seasonal long-term mean.