Exploratory analysis and visualization of global earthquake and volcanic eruption data

• Main Author: Xu, Bihua
• Other Authors: Frederique Elise Oggier
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Earth and Environmental Sciences; Earthquake; Volcanic eruption; VEI; Prediction; Visualization
• Online Access: https://hdl.handle.net/10356/181232
• Institution: Nanyang Technological University

This study investigates earthquakes, volcanic eruptions, and their potential interactions to better understand natural disaster patterns for improved early warnings. Using data from GVP, USGS, and ISC, along with confirmed earthquake-volcano triggered pairs from Legrand (2022), we enriched the dataset for deeper exploration. Visualization techniques such as 3D global maps, time series plots, and density maps were utilized to reveal important patterns and characteristics related to earthquakes and volcanic activities. Machine learning models, including Random Forest, SVR, and GBM, were used to predict earthquake magnitude, volcanic VEI, event time difference, and distance. Results showed the enriched dataset improved VEI prediction accuracy to 79%, while magnitude prediction had strong variance explanation. However, time difference and difference prediction remained challenging, with only moderate improvements after feature enrichment. Feature importance analysis highlighted distance, time difference, and earthquake magnitude as critical factors in various prediction tasks. This study provides an in-depth assessment of key parameters that potentially influence the triggering relationships between earthquake-eruption pairs. The findings identify the parameters that significantly impact these triggering mechanisms, offering valuable insights to improve forecasting of future events where seismic activity may induce volcanic eruptions.