Semantic analysis of social network site data for flood mapping and assessment

Flood assessments are increasingly vital and challenging due to the rapid urbanisation of cities and climate change phenomena. Methods for flood assessments have been actively improved to help mitigate the impact of flooding, thus reducing the risk. The emergence of social media has provided crowdsourced data that can potentially provide immediate and valuable information on flood situations. Therefore, this study uses crowdsourced data from social network sites (SNS) to assess a 2020 flood event in Kuala Lumpur. An exploratory data analysis method has been applied to reveal the statistical pattern of Twitter data to assess flood risk. Relevant tweets were analysed further by retrospective analysis in reviewing the text and media (image and video) contents against official flood reports for validation. The result shows that news channel accounts played a crucial role in disseminating information during a flood event, while tweets with images and videos can show accurate flood levels and locations of flooding. A multi-weighted index system was applied to generate a flood zone heat map. Therefore, it can be concluded that social network services are an effective means of disseminating valuable and nearly real-time information, which can assist both the general public and authorities during emergency response efforts.