| dc.description.abstract | Tornado climatologies represent an important tool for understanding the genesis, behavior, and dissipation of tornados and tornadic storms. The current tornado record is imperfect, but nevertheless useful. I undertook a study to construct a tornado climatology for the state of Alabama in order to test theories about the influence of surface heterogeneities on tornadogenesis as well as human-caused bias. Support was found for the notion that tornados may occur atop higher elevations more frequently in Alabama, but other orographic effects were more difficult to discern. The most strongly correlated variable tested was road proximity, suggesting that accessibility to surveyors has an outsized influence on where tornadogenesis points are recorded. In an effort to explore additional ways of detecting tornadic damage, a second study was undertaken to explore the utility of Sentinel-2 derived disturbance index imagery. The disturbance index was shown to be positively correlated with damage intensity across all land cover types. Actual values of disturbance index for a given damage intensity were highly variable, even within a land cover classification. The lower threshold of detectability was somewhere between higher end EF1 and lower end EF2 events. While this does not represent an improvement in detectability threshold over previous studies, the methodology presented, in conjunction with the use of Sentinel-2, has several advantages such as only requiring a single post-event image for analysis and increased spatial detail in the output compared to previous studies. The results were summarized across land cover type and damage intensity and predictive performance was much better in the types of forested areas for which the DI was designed. Although this is a limitation overall, the strengths of the methodology counterbalance the weaknesses of traditional ground-based survey methods. | en_US |
