Analyzing Socio-Geographic Disparities Through Geolocated Social Media Content During Hurricane Ida

The goal of this study is to comprehensively analyze social and geographical disparities during Hurricane Ida. Specifically, this study aims to understand how individuals’ sentiment evolved prior, during, and after Hurricane Ida. The analysis covers the discussion of various topics related to the hurricane on social media. Additionally, the study sought to evaluate the level of hazard awareness and preparedness in different geographic areas affected by Hurricane Ida. The study focused on the Continental United States (CONUS), with a particular emphasis on severely affected areas in New York and New Jersey (NYNJ). The data collection period spanned from August 24, 2021, to December 31, 2021, encompassing the duration of Hurricane Ida's impact on the region.

The study utilized geolocated social media content, primarily from Twitter, during Hurricane Ida to analyze social and geographical disparities. In addition to sentiment analysis and topic modeling, this study introduced the Hazard Risk Awareness (HRA) index as an innovative metric for evaluating communities' preparedness for hurricane-related hazards. Sentiment analysis was conducted to capture emotional responses, topic modeling to categorize discussions into themes, and the HRA Index to assess local hazard awareness. These methods aimed to reveal how individuals and communities reacted emotionally, what topics dominated discussions, and the disparities in hazard perception and readiness across different regions affected by the hurricane.

Results show dynamic fluctuations in sentiment on both a daily and hourly basis, revealing the ever-changing emotional landscape of affected communities. Our analysis extended beyond temporal considerations and encompassed a spatial dimension, revealing notable variations in sentiment across different counties during the course of the hurricane. Notably, these variations underscored the nuanced nature of emotional responses, reflecting distinct local influences on sentiment patterns.

Within the extensive corpus of the processed tweets, a substantial portion emerged as reservoirs of crucial information. These tweets played a pivotal role in disseminating essential details related to the disaster, including reports of damage, injuries, warnings, and real-time updates. Our findings indicate that a significant 37.84% of classified tweets were dedicated to conveying such vital information. New York County exhibited the highest HRA index, reaching 0.88, signifying a commendable level of hazard awareness.

The amalgamation of our findings underscores the potential utility of sentiment analysis, topic modeling, and the HRA index in the context of Twitter data during Hurricane Ida. These analytical tools provide a quantifiable glimpse into the emotional states of affected populations and the prevailing topics of discourse during crises. As evidenced by our results, this information empowers emergency responders and organizations with actionable insights, enabling them to tailor their interventions to align more effectively with the evolving needs of the affected populace. Moreover, the implications of this research extend beyond immediate applications. valuable guidance for future disaster preparedness initiatives and the refinement of response protocols. This study ultimately highlights the pivotal role of data-driven insights in navigating the intricate interplay between human emotions, information dissemination, and effective disaster management.