Achieving real-time, continental, building level, inundation forecasts using the National Water Model and Open Geospatial Data
Two thirds of U.S. federal disaster declarations include flooding as a cause, and the annual cost in the federal relief far exceeds any other natural disaster. Motivated by this, the U.S. is undergoing a paradigm shift in the way hydrologic prediction is achieved with efforts being led by the National Oceanic and Atmospheric Administrations Office of Water Prediction. In 2016, the National Flood Interoperability Experiment (NFIE) was proposed as a way of using National Water Model (NWM) streamflow predictions, along with preprocessed synthetic rating curves (SRCs) and a normalized digital elevation model to generate inundation forecasts maps. To date, there have been challenges in producing these maps and making them publicly accessible within the 1 or 3 hour forecast cycle of the NWM. Here we demonstrate a data science approach that by-passes the explicit creation of inundation maps to produce continental scale building-level flood forecasts. This process can be queried at a local or regional scale allowing forecasting to be a national effort but impact assessment a computationally distributed process. We focus on four key points that make this possible which include (1) the development of a geospatial risk inventory from open data efforts (2) improvements to the way SRCs are generated and represented, (3) modifications to the way NWM output is structured and managed, and (4) how these pieces come together to describe the evolving flood conditions in the country. We will demonstrate this approach for the state of North Carolina where inundation forecasts for over 15 million building and road locations can be queried in seconds to minutes using a real-time National Water Model 18 hour forecast. This system is being built out for CONUS as part of the Urban Flooding Open Knowledge Network in which the results can be more readily shared as linked data resources that paint a more insightful, and interactive understanding of forecasted flood conditions.
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021