Capturing human response to Winter Storm Frankie based on X (formerly known as Twitter) data.

Abstract

This study delves into how people responded to Winter Storm Frankie in the United States based on X (formerly known as Twitter®) data according to a multitude of regions, periods, sociodemographic characteristics, census regions, and geographical scales. This study finds that people actively respond to natural disasters on X during the winter storm week. Specifically, the highest number of keywords during the winter storm week is 1.6 times greater than the second-highest number of keywords during the prewinter storm week. Second, the spatial distribution of tweets exhibits significant fluctuations across different periods. For instance, in the prewinter storm week, more tweets are posted in the West region, while in the winter storm week, the Northeast region experiences a higher volume of uploads. Third, regional variables exert a substantial influence on the number of tweets. For instance, Ohio and Montana demonstrate higher elasticity than Pennsylvania. Fourth, many sociodemographic variables, such as gender, age, education, and income, are associated with individual responses. For example, a 1 percent increase in males corresponds to a 0.01 percent increase in tweets.