San Francisco Bay area community cohesion and resilience: Two case studies

In this submission, the authors develop an innovative approach to measuring community resilience by mathematical analysis of its members' social-media microblogs. The approach involves applying machine-learning and graph-analytic techniques to infer social cohesion, which is later used as the state variable by which resilience is measured. We analyze community cohesion and its dynamics during two natural disasters that hit San Francisco Bay Area with an interval of only two years - the wildfires of 2020 and the torrential rainstorms during the water year of 2022/23. The backdrop of the wildfires was characterized by the first year of the COVID pandemic, with all the uncertainty, deficit of personal protective equipment (PPE), loss of jobs, social-justice protests, and Presidential elections. For the rainstorms, the backdrop consisted of the Omicron variant of COVID, structural damage due to heavy rains and winds, and midterm elections. Bay Area economy too was in a different state during the wildfires than it was during the rainstorms. In this submission, we measure the community resilience based on the dynamics of Bay Area recovering from these events. We propose novel metrics for community cohesion and investigate the mechanisms by which emotions, local economy, weather, and air quality affects community cohesion. We also explore whether community resilience is influenced by these mechanisms. Specifically, we analyze the mediating role played by emotions in the community cohesion and resilience processes.