Authors: Bin Zhang*, University of Iowa, Caglar Koylu, the University of Iowa, Kathleen Oberlin, Grinnell College, Barbara Eckstein, the University of Iowa, Hoeyun Kwon, the University of Iowa
Topics: Human-Environment Geography, Climatology and Meteorology, Geographic Information Science and Systems
Keywords: Extreme weathers, Climate Changes, Social Media
Session Type: Paper
Start / End Time: 8:00 AM / 9:40 AM
Room: Madison A, Marriott, Mezzanine Level
Presentation File: No File Uploaded
In recent decades, changes in climate increasingly produce extremes in weather such as high and low temperatures, heavy rain and snow, and frequent weather events such as tornadoes, blizzards and hurricanes. Although political polarization significantly influences people’s sharing of thoughts about climate change in social media, individuals continue to express their emotions and feelings about extreme weather regardless of their political view. In this paper, we evaluate two main hypotheses about people’s perception and expression of extremes of weather. We first hypothesize that the extremes of weather observed in recent years increases people’s awareness and expressions about weather. Second, due to the variation in climate between different geographical regions as well as political polarization at the state level, people’s perception and expression of weather extremes may exhibit geographic variation. To evaluate these hypotheses, we analyze a large collection of geo-located Twitter corpus between December 2012 and December 2018. Recent work examining the relationship between weather and people’s mood extensively draws upon Twitter data. Contributing to and extending this body of work, we analyze people’s attitude and expressions at multiple scales including national, regional and local, and across a large time span of six years. Using probabilistic topic modeling, we first identify tweets about climate change and extreme weather, and extract semantic, temporal and geographic variability among these tweets which could potentially be explained by regional climate and political polarization. We also overlay extreme weather tweets with temperature and precipitation data to understand the relationship between extreme weather and micro-blogging behavior.