Authors:
Amber Silver and Jean Andrey
In the summer of 2013, Environment Canada adopted impact statements and call-to-action statements in their weather products, including special weather statements and weather watches/warnings. As part of these updated warning products, Environment Canada asks citizens to share information on severe weather through e-mail or by Twitter with the use of provincial weather hashtags (e.g., #ONstorm, #ABstorm, #SKstorm).
Objective:
To examine how provincial hashtags are being used to share information, by whom, and whether this usage is a reliable indicator of public attention to severe weather.
Methods:
All tweets containing provincial weather hashtags were gathered from 8 July 2014 to 7 October 2014 (n=110,950) using Zapier, a web-based application automation service. Sample characteristics, including total number of tweets, retweets, and users, were determined for each province. Users were then classified as either ‘regular users or ‘high volume users’, based on the number of tweets they contributed during the study period. Next, all weather watches and weather warnings for Ontario (n=1008) were obtained from Environment Canada. A linear regression was conducted using the total number of tweets (including original tweets and retweets) and the number of weather watches/warnings per day for the study period.
Results:
A total of 33,555 users contributed 110,950 tweets during the study period. Of these tweets, 38,418 were original tweets and 72,532 were retweets. High volume users accounted for only 4.1% of the total sample, yet contributed 58.5% of the total tweets. This disproportionately high amount of activity was unexpected, and merited deeper investigation into the characteristics of regular users versus high volume users. Based on qualitative coding of users’ Twitter profiles, it was determined that the majority of high volume users were professionals, including meteorologists, media outlets, and weather-related professionals, whereas the majority of regular users were individuals.
The results of the linear regressions between total number of tweets and weather watches/warnings were similarly surprising. Previous research suggests that the number of tweets is associated with the issuance of weather watches and weather(Ripberger, Jenkins-Smith, Silva, Carlson, & Henderson, 2014). However, the association between the variables in this present study was weak with weather warnings (R2=0.17, p-value = 0.00) slightly more closely associated with tweets than weather watches (R2=0.04, p-value = 0.05). Taken together, the results of this research suggest that high volume users of provincial hashtags tend to be weather professionals who discuss upcoming severe weather independently of Environment Canada’s official watches and warnings. This raises questions about the extent that Twitter activity is indicative of public attention (versus ‘professional’ attention) to severe weather.
Future Work:
The next steps are to assess how the findings of this research relate to weather-related tweets published without provincial weather hashtags. To do so, all tweets from a future severe weather event will be pulled using weather keywords (e.g., tornado, city_name, storm). Then, similar analysis will be conducted on this broader tweet-stream to explore questions relating to public versus professional attention to severe weather.
References:
Ripberger, J. T., Jenkins-Smith, H. C., Silva, C. L., Carlson, D. E., & Henderson, M. (2014). Social Media and Severe Weather: Do Tweets Provide a Valid Indicator of Public Attention to Severe Weather Risk Communication? Weather, Climate, and Society, 6(4), 520–530.