Date of Award


Document Type

Open Access Thesis



First Advisor

Susan Cutter


Throughout the United States, tornadoes frequently occur throughout the entire year. With each tornado there is a tornado warning that the National Weather Service (NWS) issues with a goal of protecting life and property. By using social media, these messages quickly reach the public. By analyzing Twitter data, this study aims to gain a spatiotemporal understanding of tweets, including when and where they most frequently occur. Most tweets occur within the warning time (temporal) and inside the warning polygon (spatial). To gain a better understanding of the information the tweet contains, a content analysis shows key warning characteristics such as hazard, location, guidance, time and the source of information (Mileti & Peek, 2000; Mileti & Sutton, 2009) that are present or absent. Findings suggest that many warnings disseminated through Twitter contain variations of these characteristics, however most do not contain all five key characteristics. There is also extensive variation in portraying the information, such as varying colors for warning polygons and lack of protective action suggestions. With many discrepancies present in the findings of this research, the meteorological community needs a uniform approach to warning, limiting confusion by the user and milling time. Future work would need to consist of social scientists and meteorologists to better understand the magnitude that these discrepancies occur.


© 2018, Raelene C. Campbell

Included in

Geography Commons