Author(s): Joao Pita Costa; Gerald A. Corzo Perez; Inna Novalija; Luis Rei; Matej Senozetnik; I. Casals Del Busto

Linked Author(s):

Keywords: Disaster Management; Hydrological Events; Machine Learning; Public Perception; Sentiment Analysis

Abstract: Recent trends indicate a significant rise in hydrological extremes, challenging conventional management and forecasting methods. This study introduces an innovative approach leveraging machine learning and social media analytics, focusing on X (rebranded from Twitter), to improve the detection and management of hydrological events. By analysing geotagged tweets and incorporating news data, our machine-learning model identifies and categorises information related to hydrological extremes. This integration allows for enhanced real-time monitoring and sentiment analysis, providing insights into public perception and the effectiveness of response strategies. We employ Granger causality to establish predictive links between social media content and hydrological indicators, enabling preemptive measures and potentially reducing event impacts. Our comprehensive approach addresses the physical dimensions of extreme weather events and captures the public's emotional responses, offering a holistic view of disaster management. Significant findings from our research demonstrate the potential of combining machine learning with social media data to advance our understanding and management of water-related disasters. This strategy represents a significant step forward in environmental research and disaster response, harnessing the power of digital platforms and big data analytics. The study's outcomes suggest that this method can significantly contribute to hydrology and socio-hydrology, offering a novel perspective on integrating technology and social insights for better water management practices.

DOI: https://doi.org/10.3850/iahr-hic2483430201-278

Year: 2024