87th AMS Annual Meeting

Tuesday, 16 January 2007: 9:15 AM
An assessment of hazard warning verification from a social perspective
206A (Henry B. Gonzalez Convention Center)
Steven Stewart, University of Arizona, Tucson, AZ; and K. Barjenbruch, K. Pennesi, J. Deo, S. A. Erickson, P. L. Heinselman, A. J. Ray, R. Low, R. A. Wolf, and A. Coles
Under what conditions can a hazard warning or severe weather forecast be considered successful? If a winter storm warning is issued and the snow comes 15 minutes after the warning expires, is that a valuable warning from the perspective of society? If a flash flood warning is issued for county A, but the flooding occurs in county B, is that warning successful?

The NWS bases verification or warning/forecast “success” on whether a warning is issued as well as the timing and duration of the warning. A warning is “verified” if the event occurs in that county during the time period in effect for the warning. If the event occurs in an adjacent county or in a different time period, the event is not verified. In contrast, social verification goes beyond the technical merits of the warning and provides information on the utility of forecasts and hazard warnings from users' perspectives. Warnings that are slightly inaccurate in terms of space or time may still be useful if they improve the well-being of society by inducing individuals to prepare for, avoid, or otherwise respond to hazard events that may have an impact .

We report the results of a pilot study to assess social verification of NWS forecasts. A cross-section of weather users representing casual users to emergency managers will enter their perceptions of the utility of forecast and null-event forecasts into a web-based log to during the Fall of 2006. Assessed products include hazard warnings, 2-14 day forecasts , and seasonal outlooks, with emphasis on hazard warnings. The results will be summarized using quantitative and qualitative assessment techniques and compared with official NWS verification results.

Supplementary URL: