Examining the Use of the NSSL Experimental Warn-on-Forecast System for Ensembles for the Prediction of Severe Storms Through Short-Term Forecast Outlooks During the 2018 Spring Forecasting Experiment

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 extreme rainfall). By leading convective-scale research and development activities, this project will enable a new paradigm in which convection-allowing, ensemble model forecasts become a key resource for NOAA National Weather Service (NWS) watch and warning operations. Through these activities, an experimental Warn-on-Forecast system called the NSSL Experimental Warn-on-Forecast System for ensembles (NEWS-e) has been developed.

NEWS-e is a frequently-updating, regional-scale, on-demand convection-allowing ensemble analysis and prediction system, nested with an hourly convection-allowing ensemble forecast system. The 2018 version of this system assimilates radar and satellite data every 15 minutes, surface data every hour, and generates probabilistic 6-hour, 5-hour, 4-hour, or 3-hour forecasts at the top and 30-minute mark of each hour at a grid spacing of 3 km. The NEWS-e background is the experimental HRRR-ensemble (HRRRE) provided by the NOAA ESRL/Global Systems Division. An 18-hour ensemble forecast is launched from the 1200 UTC HRRRE analysis to provide boundary conditions for the NEWS-e system for the period 1800 UTC Day 1 – 0600 UTC Day 2 and a 1-hour ensemble forecast is launched from the 1700 UTC HRRRE analysis to provide initial conditions for the NEWS-e system at 1800 UTC.

NEWS-e guidance was used to create severe weather outlooks between the watch and warning time scales within NOAA’s 2018 Hazardous Weather Testbed (HWT) Spring Forecasting Experiment. Participants were separated into two groups based on self-rated forecast experience, and a third group was comprised of a single forecaster with extensive operational experience. The primary goals of this experiment were to: 1) explore how the three groups used short-term ensemble forecast guidance from NEWS-e to produce a series of 1-hour severe weather outlooks, 2) evaluate participant perceptions of the NEWS-e forecasts, the submitted outlooks, and verification of the outlooks, and 3) gain insights into participants’ product usage during the outlook activity. First, “practically perfect” outlooks generated from local storm reports were used to subjectively and objectively verify the 1-hour outlooks. Next, participant perceptions were analyzed through surveys issued after each set of outlook submissions during the activity as well as after discussion of the “practically perfect” verification the following morning. Lastly, information on each group’s interactions with the NEWS-e web interface was collected and summarized to give a deeper understanding of participants’ information consumption while producing short-term total severe outlooks. This presentation will share preliminary results from the analysis of these three areas of study.