Authors: David Baude, Rebecca Diehl, Ijaz Ul Haq, Kristen Underwood, Alexander Prescott, and Beverley Wemple – University of Vermont; Belize Lane and Colin Phillips – Utah State University
Title: Back-end Design of the Next Generation QPE Workflow for RFC Operations pproaches Are Suitable for Flood Inundation Mapping
Presentation Type: Poster Presentation
Abstract: The Next Generation QPE (NGPE) workflow is a web-based platform that will allow River Forecast Center (RFC) forecasters to construct configurable precipitation estimation workflows by combining modular data fusion tools. This poster presents the back-end architecture that powers the platform, which is built on the Tethys Platform—a Python/Django framework for geoscientific web applications. The back-end is based on two primary components: data acquisition modules and standardized data classes. Data acquisition modules retrieve precipitation observations from external sources on demand, triggered by user interactions on the front-end through Tethys component hooks rather than scheduled jobs, ensuring that data is fetched only when needed.
To support the heterogeneous nature of precipitation data, the back-end employs two core data class abstractions: an xarray Dataset representation for gridded products such as radar, and a geopandas GeoDataFrame representation for point observations such as gauge networks, each tied to its native spatial structure. These abstractions encapsulate raw data and expose consistent interfaces for the front-end to render maps and apply user-driven operations such as merging, filtering, and aggregation. Initial data acquisition modules have been implemented for MADIS gauge networks and Multi-Radar Multi-Sensor (MRMS) radar products, with planned extensions to additional data products and the ingestion of AI model outputs as input data sources. A beta version of the platform is currently under development.