Using social media for rapid assessment of emotional impacts of volcanic hazards
Evgenia Ilyinskaya 1, Vésteinn Snæbjarnarson2,3, Hanne Krage Carlsen4, Björn Oddsson5
Affiliations: 1School of Earth and Environment, University of Leeds, Leeds, UK; 2Miðeind ehf, Reykjavík, Iceland; 3Department of Computer Science, University of Copenhagen; 4Department of Public Health and Community Medicine, Institute of Medicine, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden; 5Department of Civil Protection and Emergency Management, National Commissioner of the Icelandic Police, Reykjavik, Iceland
Presentation type: Talk
Presentation time: Monday 11:15 - 11:30, Room R380
Programme No: 7.2.4
Abstract
Social media posts have been previously used for assessing geophysical aspects of volcanic and seismic events, for example, locating earthquakes; and science communication. They have also been utilized for crisis and risk communication, rapid assessment of material damage, and aiding recovery efforts after several disasters including e.g. the Great Tohoku earthquake. Here we demonstrate how social media can be used as a cost-effective tool for gauging public discourse and the potential emotional impact of volcanic events in near-real time1. We used natural language processing (NLP) of Twitter posts to analyse the sentiments expressed in relation to volcanic-seismic unrest, and a subsequent fissure eruption in Iceland 2019-2021. We show that although these volcanic hazards were of small size and caused negligible material damage, they were associated with a measurable change in expressed emotions in the local populations. The seismic unrest was associated with predominantly negative sentiments (positive-to-negative sentiment ratio 1:1.3), but the eruption with predominantly positive (positive-to-negative sentiment ratio 1.4:1). Incorporating sentiment analysis of crowd-sourced information, such as social media posts, into local risk management has the potential for immediate and longer-term benefits. While our method does not provide direct measures of the mental health state and impacts, and is not intended to replace more formal investigations, it may be used to quickly gauge whether communities are under stress and may require additional surveying and/or resources. 1Ilyinskaya et al 2024 Brief communication: Small-scale geohazards cause significant and highly variable impacts on emotions, Nat. Hazards Earth Syst. Sci., 24, https://doi.org/10.5194/nhess-24-3115-2024.