Effective communication between citizens and government is crucial for climate change preparedness and adaptation. Local authorities increasingly utilize social media platforms like X (formerly Twitter) for disseminating information. However, research on their communication strategies and citizen engagement is limited. This study addresses this gap by examining the social media presence of climate authorities in five major Brazilian cities—São Paulo, Rio de Janeiro, Belo Horizonte, Porto Alegre, and Belém—covering a large population and substantial social media following. The research questions are: 1) What publication patterns emerge from local government authorities' social media communication? and 2) How do citizens respond to these publications? The study employs Social Media Presence and Human Engagement frameworks to analyze a year's worth of data (10,200 publications and 5,500 replies) to provide evidence-based insights for improving communication strategies and fostering citizen engagement in climate action.

Existing literature highlights the importance of citizen-government communication in climate adaptation and preparedness. Social media platforms offer cost-effective tools for disseminating weather and climate information, bypassing traditional media. Previous research has used social media data to analyze public perceptions and concerns regarding climate change, often focusing on specific events or hashtags. However, studies on the nature of routine citizen-authority communication in the long term are scarce. This study builds upon conceptual frameworks of Social Media Presence (focused on publication content and strategies) and Human Engagement (analyzing citizen behavior and participation), using behavioral modeling and topic modeling approaches to gain a comprehensive understanding of citizen-authority interaction in the context of climate communication.

This study employs a mixed-methods approach combining quantitative and qualitative analyses. To characterize the authorities' social media presence, the researchers analyzed the content and periodicity of their posts. This included identifying communication strategies (single-loop vs. double-loop, active vs. passive), analyzing mention networks to understand who the authorities were interacting with, and using a clustering approach (K-means with Jaccard Similarity Measure) to identify recurring publication patterns. For citizen engagement, quantitative analysis focused on metrics like reaction time, activity ratio, relative activity duration, and probability of activity duration to distinguish between short-term (transient) and long-term (regular) engagement. Qualitative analysis involved topic modeling (Latent Dirichlet Allocation or LDA) to identify themes within citizens' replies, distinguishing between pre-defined topics (like mentions of "climate change") and emerging topics. The study collected data from the X platform API for the five selected Brazilian cities over a one-year period, including all public posts by authorities and replies by citizens. Data pre-processing involved natural language processing techniques, including stemming and lemmatization. The selection of cities considered population size, location across different Brazilian regions, diverse climates, experiences with extreme weather events, and the authorities' active presence on X. Meteorological warning data from INMET (Instituto Nacional de Meteorologia) was used to contextualize the findings.

The study reveals several key findings regarding the social media presence of climate authorities and citizen engagement. Firstly, authorities predominantly employed a passive, one-way communication strategy, primarily disseminating information through periodic posts with little interaction or response to citizen replies. Publications generally focused on meteorological alerts, forecasts, and weather updates, with standardized patterns identified in some cities. Mention networks showed that authorities primarily interacted with other governmental agencies rather than directly engaging citizens. Secondly, citizen engagement was largely short-term (transient). While citizens responded quickly to authorities' posts, often providing corrections, additions, and updates, their engagement rarely extended beyond a single interaction. Long-term engagement was rare and concentrated among a small group of highly active citizens. Topic modeling revealed that citizen replies often focused on local-level concerns, correcting information inaccuracies, providing updates on current conditions in their areas, and occasionally using the platform for unrelated activities such as publicizing donation drives. Importantly, citizens rarely connected local weather events to broader climate change phenomena. Analysis of meteorological warnings from INMET revealed variations in the types and frequency of extreme weather events across the five cities, but these variations did not significantly correlate with differences in citizen engagement patterns.

The findings highlight the limitations of current social media platforms for effective climate change communication. Authorities’ preference for one-way communication and the predominantly short-term nature of citizen engagement suggests a need for alternative or supplemental approaches. While social media effectively facilitates rapid dissemination of information, its limitations in fostering long-term engagement and deeper discussions about climate change are evident. The emphasis on local concerns in citizen replies underscores the need for geographically targeted information dissemination. The infrequent connection of local weather events to global climate change could be addressed through improved communication strategies that contextualize local events within the larger picture of climate change. The study's results also provide important insights for the design of communication platforms for local governments, suggesting the need for systems that support geographic targeting, enable efficient information updates, and foster community participation and feedback.

This study provides valuable insights into citizen-authority communication on social media regarding climate change. The findings highlight the need for more effective strategies to leverage social media for raising climate awareness, improving citizen preparedness, and building trust in government authorities. Future research could explore alternative communication channels and strategies, such as incorporating interactive features or utilizing other technologies, to enhance engagement and facilitate deeper discussions about climate change. Further investigation into the relationship between information credibility, citizen trust, and preparedness behaviors is warranted.

This study focuses on a specific set of five Brazilian cities and one social media platform (X). The findings may not be generalizable to other contexts, cultures, or platforms. The reliance on publicly available data may also introduce biases, as not all citizens use social media or express their views publicly. Furthermore, the study's focus on citizen replies to authorities may overlook other forms of engagement, such as retweets, shares, or discussions happening outside the direct interaction with the official accounts. The study is observational, thus correlation does not imply causation.