Authors: Patrick J. Clemins, Noah B. Beckage, Vamsi Dondeti, Muhammad Adil – University of Vermont
Title: Next Generation QPE Workflow for RFC Operations
Presentation Type: Lightning Talk
Abstract: The River Forecast Centers (RFCs) produce gridded quantitative precipitation estimates (QPEs) that are used as forcing data in several applications, including RFC hydrologic models and the National Water Model. RFC forecasters create this QPE product using the Multi-Sensor Precipitation Estimator (MPE) by combining precipitation estimates from several different sources, including Multi-Radar Multi-Sensor (MRMS), precipitation gauges, radar estimates, and satellite estimates. The MPE software has not been significantly updated in several decades and it is a time-consuming process for RFC forecasters to produce a quality product, especially in regions with poor radar and precipitation gauge coverage.
This project, a collaboration between the RFCs and the University of Vermont (UVM), will create modular data fusion tools that can be combined by RFC forecasters in a web-based tool to create precipitation estimate (NGPE) workflows. Some of these data fusion tools will utilize deep learning algorithms. This more sophisticated, model-driven framework of data fusion tools and configurable workflows will improve the efficiency and accuracy of the QPE by allowing RFC forecasters to build off prior workflows and utilize regional and local observed datasets more effectively. The modular, open-source design of the NGPE workflow will allow the broader hydrologic community to contribute new datasets and tools to the NGPE and create modules to format the output QPE for any hydrologic modeling framework they choose, including NextGen. Finally, increasing the accuracy of the QPE will lead to better hydrological modeling and prediction.