A vision for the development and implementation of the Warn on Forecast concept

Pamela Heinselman

  1 Dec, 10:30 am, in 2155

Deterministic modeling approaches to forecasting severe weather presume a single solution for the evolution of storms and the environments in which they form. Similarly, deterministic severe weather warnings assume severe weather impacts are equally likely within the warning’s spatial and temporal extent. This approach fails to represent forecast uncertainty that we know exists, owing to the chaotic nature of the atmosphere, imperfect observations, imperfect prediction models, and other causes. Advancements in data assimilation, ensemble modeling, high-performance-computing, and probabilistic forecast and verification methods provide the means and opportunity to better represent these uncertainties within severe weather prediction systems.

The Warn-on-Forecast project at the National Oceanic and Atmospheric Administration (NOAA) National Severe Storms Laboratory in Norman, OK, USA, aims to improve severe weather forecasts, warnings, and decision support for high-impact events (e.g., tornadoes, hail, wind, and flooding) by leading convective-scale research and development activities that enable a new paradigm where convection-allowing, ensemble model forecasts become a key resource for NOAA National Weather Service (NWS) Watch and Warning operations. The result of these efforts is the development of a prototype Warn-on-Forecast system called the NSSL Ensemble Warn-on-Forecast System for ensembles (NEWS-e).

NEWS-e is a frequently updated, regional-scale, on-demand convection-permitting ensemble analysis and prediction system, nested with an hourly convection-allowing ensemble forecast system. The 2017 version of this system assimilates radar, satellite, and surface data every 15 minutes, and generates new probabilistic 3-hour and 1.5-hour forecasts at the top and bottom of each hour, respectively, at grid spacing O(~3 km). This multiscale data assimilation system uses the advanced research version of WRF, version 3.8+ (ARW) to produce storm-scale ensemble analyses and forecasts. Details of the system configuration will be provided within the presentation.

NEWS-e seeks to improve 0–3-h predictions of individual convective storms and mesoscale aspects of convection that provide enhanced probabilistic forecast guidance. Forecast swaths produced by this system, such as probability of simulated reflectivity > 40 dBZ and ensemble 90th percentile values of accumulated rainfall, 2–5-km updraft helicity, and 0–2 km vertical vorticity, are expected to revolutionize forecasters’ ability to anticipate not only storm location, mode, intensity, but also high-impact threats and their impacts on society. Toward this end, this prototype system is being tested in real time during peak severe weather season in the U.S. Following each storm season, findings from rigorous quantitative and qualitative case study evaluations are used to direct enhancements of the system. Example cases and the associated forecast verification will be shown.