Bridging the gap in flood risk communication: a comparative study of community and organizational social media posts using natural language processing

Abstract

Floods are prevalent disasters in the United States (U.S.), with escalating risks due to climate change-induced factors like rising sea levels and erratic rainfall patterns. While organizational efforts to mitigate the social and economic impacts of floods exist, there is a continued need for innovative approaches to flood management, particularly in flood risk communication. Currently, flood risk communication remains largely one-directional (i.e., communication comes from organizational entities such as emergency services or the media and is not necessarily informed or updated by community need, which can limit preparedness and response capabilities). Prior studies have found that social media platforms offer valuable opportunities for more interactive and real-time dialogue during disasters, including flood events. Additionally, there is an established body of research exploring the communication relationships between communities and organizational entities, respectively; highlighting opportunities to better develop pathways for two-way communications. Our study builds upon this literature by examining the alignment between community and organizational messages on social media during a series of flooding events affecting nine U.S. states from July 2022 to August 2022. Specifically, the study uses Natural Language Processing (NLP) to compare the content, sentiment, and emotion of messages from each entity to assess whether organizational communications reflect public needs and effectively foster two-way communication. Results reveal key temporal and spatial trends in message frequency and topic focus across communities and organizational entities, from which we develop five targeted recommendations designed to support the execution of more effective two-way flood communication. Future work will provide further insights into tailoring communication strategies to more diverse populations and circumstances.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 8 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.