Multi-source Hydrologic Ensemble Augmentation for CHPS-HEFS Forecast System

The project will develop, implement, and assess the techniques and data infrastructure to enhance the skills of the CHPS-HEFS ensemble hydrologic forecasts. The techniques and data infrastructure will be developed in accordance with the NWM NextGen modeling framework and its operational implementation to ensure full conformity and immediate transferability with NextGen’s modeling and data interfaces. The CHPS-HEFS system has been NWS’s operational forecast tool used by River Forecast Centers (RFCs) for making hydrologic forecasts across different time scales. Over the years, RFCs have identified several areas in the CHPS-HEFS system where improvement is needed. Given the structure and complexity of the CHPS-HEFS ensemble forecast system, we propose to leverage multiple sources of additional data to augment the CHPS-HEFS hydrologic ensemble predictions using machine learning (ML) based comprehensive data fusion approaches. These additional data will include but are not limited to recent/past observations of snowpack (e.g., SWE), streamflow (Q), terrestrial water storage (TWS), and gridded sea surface temperature (SST), if sufficient data records are available for training. We expect them to considerably improve the skills of the operational forecast system. The augmentation techniques will be tested on the NextGen modeling framework when its ensemble functionality is ready.