Analyzing Public Response to Hydrological Stress through Machine Learning and Social Media Sentiment: Evidence from Andros, Faroe, Mauritius and Samoa

Understanding how hydrological conditions influence public sentiment toward climate and environmental issues is essential for effective policy-making and communication strategies. This study adopts a co-creation approach by integrating hydrological data with insights from social media, engaging multiple stakeholders in the process of knowledge generation. Utilizing a multi-year dataset, we analyze daily weather parameters—specifically focusing on temperature and precipitation—alongside social media comments pertaining to environmental discussions.

Sentiment analysis methods, including both VADER and transformer-based machine learning models, are employed to identify and quantify negative sentiments within these comments. Additionally, time series analysis techniques such as Error-Trend-Seasonality (ETS) decomposition and LSTM neural networks are applied to forecast climatic conditions and assess their impact on sentiment patterns over time. This allows us to examine how adverse hydrological conditions, such as increased precipitation or extreme weather events, heighten negative public sentiment regarding climate issues.

Sentiment analysis methods are employed to identify and quantify negative sentiments within these comments, allowing us to examine patterns over time. By incorporating public perceptions expressed on social media, we co-create a more comprehensive understanding of how hydrological phenomena impact society.

Preliminary results indicate a significant association between adverse hydrological conditions, such as increased precipitation or extreme weather events, and heightened negative public sentiment regarding climate issues. By exploring this relationship, we aim to uncover how changes in weather impact public perceptions and attitudes toward the environment, facilitating mutual learning between scientists and the public.

This research bridges hydrological sciences and social media analytics, contributing to an interdisciplinary and participatory understanding of the societal impacts of hydrological phenomena. The insights gained will inform policymakers and stakeholders, aiding in the co-development of proactive communication strategies and interventions that address public concerns related to climate and weather. Through this collaborative approach, we demonstrate how integrating diverse knowledge systems can enhance water resources management and environmental decision-making.

Keywords: Hydrology, Public Sentiment, Climate Change, Social Media Analysis, Environmental Communication

Presentation: 2024 - Water and Surrounding Sentiment: Evidence from Andros for Greece Summer Symposium,
Greece-Qatar, https://arcg.is/11afjP