COMBINING SOCIAL MEDIA AND AUTHORITATIVE DATA FOR CRISIS MAPPING: A CASE STUDY OF A WILDFIRE REACHING CROATIAN CITY OF SPLIT
Keywords: Geosocial data mining, VGI, Social media, Wildfire, Croatia
Abstract. Due to climate changes, wildfire breakouts get more frequent and difficult to control. In the mid-July 2017, the wildfire spread from wildland to the city of Split, the second-largest city in Croatia. This unpredictable spread almost caused the collapse of emergency response systems. Fortunately, a greater tragedy was avoided with the composure of the responsible services and the help of the citizens. The citizens helped in extinguishing the fire and timely provided the significant amount of disaster-related information on different platforms and through social media. In this paper, we address the problem of identifying useful Volunteered Geographic Information (VGI) and georeferenced social media, for improving situation awareness while the wildfire was reaching the Croatian city of Split. Additionally, we combine social media with other external data sources (e. g. Sentinel-2 satellite images) and authoritative data (e.g. Croatian National Protection and Rescue Directorate official data and Public Fire Department of Split data) to establish the geographical relations between the wildfire phenomena and social media messages. In this manner, we seek to leverage the existing knowledge and data about the spatiotemporal characteristics of the Split wildfire in order to improve the identification of useful information from georeferenced social media with other integrated data sources that can be valuable for improving situation awareness in wildfire events.