DescriptionThis research investigates how to leverage multi-temporal geolocated social media data from Meta-Facebook in order to provide dynamic popualtion estimations and assess the spatio-temporal variations of vulnerable populations exposed to climate-related risks. Assessing populations exposed to climate change impacts traditionally rely upon estimations from census surveys, but these only provide a static picture of risk since census surveys are often undertaken and released over long time periods and thus cannot be updated on a regular basis. Building upon the case of typhoon events in the Philippines and advanced spatial analytical methods, we address the selection bias of social media data and further analyse the ways Facebook user densities vary across three key time periods, namely daily, weekly, and seasonal exposure. Results show how changes in population exposure combined with varying levels of social vulnerably can increase the size of population at risk at specific times periods and places. When comparing daytime with nightime periods, large to medium-size cities present an increase of population density combined with lower levels of vulnerability, while municipalities located in peripheric areas show a decrease of population counts coupled with higher levels of vulnerability. An opposite trend, however, is observed during the weekends and more particularly during holiday periods with a decrease of population densities in cities. While caution must be taken regarding the representativeness and availability of social media data, our findings remain valuable for guiding disaster management strategies and support climate resilient pathways in complementarity with other traditional datasets and practices.
|10 nov. 2023
|Titre de l'événement
|Quetelet 2023 Seminar: Unconventional Data Sources forPopulation Studies: Opportunities and Challenges