NWC REU 2010
May 25 - July 30



Photo of author

Evaluation of NWS Storm-Based Warnings Using Gridded Products

Todd Ferebee, Kiel Ortega, and Kevin Scharfenberg


What is already known:

  • Hail reports available through NWS Storm Data are temporally and spatially sparse, making it difficult to describe the entire event.
  • Only one report of large hail is needed to verify an entire Severe Thunderstorm Warning area
  • Data from multiple radars can be merged together to create time-accumulated algorithm output, such as the maximum expected hail size "swaths."

What this study adds:

  • Dense ground reports of hail (and no hail) are matched up with gridded radar-based hail algorithm output for evaluation.
  • Several radar-based algorithms are identified as good performers to aid in hail warning operations and event evaluation.
  • The radar-based hail algorithms perform better for storms identified as supercells when compared to multicells.


In 2008, the National Weather Service began issuing storm-based polygon warnings instead of county warnings. Only one severe hail, wind, or tornado report is needed to verify an entire warning polygon. Few severe weather reports in the warning, and in turn for the storm which prompted the warning, makes difficult to determine the spatial extent of severe weather for a particular storm. Since 2006, the Severe Hazards Analysis and Verification Experiment (SHAVE) has been collecting severe weather reports at temporal and spatial resolutions much higher than those available in Storm Data. The National Severe Storms Laboratory (NSSL) produces several severe weather products, such as reflectivities at different isotherms and estimated hail size, on a grid for the entire contiguous United States. These grids could provide for synthetic verification of severe weather especially for the spatial extent of severe weather. This study will investigate how well the grids perform in determining where severe hail fell by using high resolution SHAVE reports. Discussed for applications of such grids for warning verification and improvement will also be included.

Full Paper [PDF]