On the importance of post-event surveys in assessing tornado occurrences

In less than one year, three significant severe weather outbreaks affected Oklahoma: 13 June 1998, 04 October 1998, and 03 May 1999. Owing to the level of resources available in central Oklahoma, it has been possible to take the initial, raw storm reports and subject them to a careful review and revision to an extent that goes well beyond the typicial review of storm reports. There are several sources of error and confusion in the initial reports; for example, multiple reports of the same tornado, reports giving the location of the spotter rather than that of the tornado, spotters underestimating the distance to the tornado, and reports giving the time of the report rather than that of the tornado, to name just a few. Using diverse information sources, such as WSR-88D radar data, raw storm reports, aerial surveys (when available), storm chaser logs/videos, television coverage, and house-by-house engineering surveys (again, when available), it is possible to refine the initial reports and obtain a significantly improved perception of the actual events, although it is likely that the result is still not perfect. Given that storm reports routinely do not receive this amount of attention, and that those storm reports eventually find their way into the national severe storm database (Storm Data), it is important for many reasons (e.g., forecast verification and scientific review of the meteorology of severe storm cases) that those reports be as accurate as possible. The passing of Dr. Fujita has left a national void that we have yet to fill, and this work suggests something of the tremendous negative impact on our science and its application resulting from that void.