Climate Change and Environmental Studies Progress Report

Abstract

This research group report presents three complementary studies integrating statistical modeling and environmental data analysis.

(1) Personalized Functional PCA:
We develop a PerFPCA framework that decomposes functional variability into shared and individual-specific components, improving interpretability and predictive accuracy. Simulations and an application to Argo float data show that PerFPCA outperforms traditional FPCA for heterogeneous ocean regions.

(2) Groundwater–Seismic Interaction:
We investigate how seasonal groundwater level fluctuations influence localized seismic activity in southwestern Taiwan. The results suggest that hydrological loading modulates crustal stress and affects earthquake occurrence patterns.

(3) Regime-specific DAS Modeling:
Machine learning models trained on distributed acoustic sensing (DAS) data with regime-specific information achieve better P/S-phase identification than pooled models, demonstrating the importance of incorporating environmental context.

Together, these studies highlight the synergy of statistical innovation and geophysical understanding for analyzing complex environmental systems.

Please click here for participating the talk online.