Social Media Engagement With Extreme Weather in Northern Climates: Natural Language Processing for Multi‐Directional Communication

Meteorological information, conveyed through official alerts, is a daily presence in people's lives, aimed at enhancing preparedness and situational awareness. Such information can be found across multiple platforms, including television, radio, print publications and social media and is often mediated by third‐party broadcasters, social media personalities and others. Whereas the needs of forecasters and emergency managers in providing this information and managing weather events have been well‐studied in the literature; less attention has been paid to how the public engages with and acts upon such expert‐driven and real‐time weather information. Our study utilises Natural Language Processing (NLP) to examine public attention to the communication of disruptive weather events through Specific Area Message Encoding (SAME) and officially generated hashtag alerts during winter events in Canada to determine which characteristics make such information useful and timely for the public. We investigate the engagement of both the lay public and experts with this weather information through novel techniques designed to better understand public perceptions and values related to weather forecasts and inform the body of literature aimed at aiding emergency managers in the management of such crises. Utilising a bottom‐up approach to classification, we posit a new perspective on information stewardship in dynamic weather crises in places such as Northern climates, one that allows for a multi‐directional rapid flow of interaction and accounts for the gap between authority and expertise when considering actual weather‐related disruptions.