5.7 Probabilistic Hazard Information (PHI): Highlighting the Benefits via New Verification Techniques for FACETs

Friday, 12 June 2015: 2:45 PM
303 (Raleigh Convention Center)
Gregory J. Stumpf, CIMMS/Univ. of Oklahoma and NOAA/NWS Meteorological Development Laboratory, Norman, OK; and C. D. Karstens and L. P. Rothfusz
Manuscript (1.3 MB)

NWS severe thunderstorm and tornado warnings are issued today as deterministic polygons that cover a two-dimensional area of expected threat of severe weather within a time period usually from 30 to 60 minutes. As part of the Forecasting A Continuum of Environmental Threats (FACETs) initiative, new methods of delivering severe thunderstorm and tornado forecast and warning information are being investigated. A concept known as Probabilistic Hazard Information (PHI) has been under development and testing at the Hazardous Weather Testbed since 2008. In this concept, customized severe weather warning products can be derived from rapidly-updating probability grids that are produced by forecasters.

One advantage of PHI is that forecasters can communicate threat information prior to the thresholds that today's warnings are issued, both in time and space. For example, probabilistic information about threats that have yet to reach an intensity which would satisfy a forecaster's threshold decision to warn is made available to users who desire that information. In addition, information about a forecasted threat well downstream from the current threat location, well beyond the forward edge of any warning polygon, can also be conveyed. In each of these situations, a lower probabilistic threshold can be used to create custom warnings for users who are more at risk. In addition, the times and locations of higher confidence threats (e.g., tornadoes currently reported on the ground) can also be conveyed through the probabilistic grids.

A new gridded verification technique is used to show how warning accuracy is affected using varying probabilistic thresholds. The technique will be illustrated with the 27 April 2011 tornado outbreak event over the southeast United States. Using human-generated locations of mesocyclones and tornadoes at each radar volume scan interval, swaths of strike probabilities using the Thunderstorm Environment Strike Probability Algorithm (THESPA; Dance et al. 2010) technique are generated. These grids are enhanced using both forecaster-modified confidence values for each storm, as well as SPC probabilistic tornado outlooks. Innovative geospatial warning verification techniques, which include new measures such as false-alarm area and location-specific lead and departure time, will be employed to highlight the benefits of PHI-derived custom warnings.

Supplementary URL: http://tinyurl.com/ewp-thoughts

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner