The Rhetorical Storm of Uncertainty in Severe Weather Communication

Uncertainty is a concept that crosses disciplines, appearing in both social and physical phenomenon. In meteorology, uncertainty stems from, among other things, a lack of understanding of atmospheric processes and lack of observations; this uncertainty manifests in forecasts. However, despite being a common, accepted aspect of research and practice, uncertainty does not have a set typology, especially in the weather sector. The communication of forecast uncertainty has an important link to society, as the safety and livelihood of the public can often depend on the clarity of a forecast in severe weather. Interpreting uncertainty quickly and usefully in a rapidly onset hazardous events, such as severe weather, is crucial. The emergence of social media enhances the interest in studying the communication of uncertainty in a severe weather context because it provides a time-stamped view into the dialogue and processing of new warning and watch information in written and pictorial forms. The March 19, 2018, set of 6 tornadoes in the Birmingham CWA and surrounding areas offers a unique setting to study what uncertainty expressions are used by different official providers of weather forecast information. To conduct this research, a list of forecast providers’ Twitter handles was developed based on their proximity to the threat area, established careers near the weather sector (i.e. from weather forecasting offices, media broadcasters, and other related officials), and national media presence (i.e., for individuals who report on extreme weather events). A list of n=384 forecast providers was established, and all tweets they originated were collected from March 16 through March 21. A quantitative content analysis of this tweet dataset was conducted to examine relationships between elements of uncertainty, such as verbal and numerical expressions. These appear using qualifying phrases and ranges, while confidence is shown through the use of expressions that are more assured. Additionally, conflict between confident and uncertain expressions within a tweet were examined for a relationship. These manifestations of uncertainty and confidence were analyzed across five common “targets” that were determined a priori that often were associated with uncertainty; the targets include timing, location, intensity, social impact, and hazard. Quantitatively identifying and analyzing qualitative expressions of uncertainty and confidence can characterize the prevalence of such phrasing in forecast messages, as well as identify when they co-occur and have the potential to be misunderstood by their audience. Bringing together the fields of science communication and meteorology within this study affords the opportunity to examine severe weather risk communication across different sources, and can be applied to further work regarding and improving severe storm warning dissemination.