ASSESSING THE VULNERABILITY OF PUBLIC WATER SYSTEMS TO DROUGHTS THROUGH INTEGRATED HYDROLOGICAL MODELING, REMOTE SENSING, AND SOCIAL SENSING

Author

Fisayo Akindele Akinde, Southern Illinois University CarbondaleFollow

Date of Award

8-1-2025

Degree Name

Master of Science

Department

Geography and Environmental Resources

First Advisor

Li, Ruopu

Abstract

Droughts rank among the most complex natural hazards due to their unpredictable onset, evolution, and cessation. It affects hydrological systems, ecosystems, and communities across different spatial and temporal scales. As climate change profoundly threatens water security, hydrologic models become a valuable tool for understanding the physical and social impacts of droughts and planning effective coping strategies. One of the critical factors that affects the performance of hydrological models at watershed scales is digital terrain data, also known as digital elevation models (DEMs). This research evaluated various DEM-related parameters (i.e., data sources, spatial scales, filtering, and flow barriers) on the Soil & Water Assessment Tool (SWAT) model’s goodness-of-fit in two gaging stations (Casey Fork and Rasey Creek). The results show that hydrologically enforced LiDAR-derived DEMs produced the lowest bias at the Casey Fork Creek for both monthly and daily models, but not at the Rasey Creek. Our results reveal that the coarser-resolution USGS 30m DEM outperformed the finer-resolution DEMs during the monthly calibration and validation phases at the Casey Creek gaging station, as indicated by the Nash-Sutcliffe Efficiency (NSE) and coefficient of determination (R²).Beyond evaluating model parameters, this study aimed to assess the vulnerability of Public Water Systems (PWS) to drought by integrating hydrological modeling, remote sensing, and social sensing. To achieve this, an interdisciplinary approach was adopted by applying a statistical model such as the Autoregressive Distributed Lag (ARDL) model to explore the dynamic relationships among hydro-meteorological indices, remote sensing indicators, and social sensing. Social media analysis of X (formerly known as Twitter) posts revealed that the frequency of drought-related tweets shows a significant correlation with short-term meteorological drought indices, such as SPEI. In contrast, vegetation indices, like EVI, were linked to delayed social responses. Public sentiment revealed a nuanced emotional response that was more context-dependent and indirect, showing a weak correlation with drought indicators. By integrating physical modeling with hydro-meteorological drought indices and social media data, our study offers a comprehensive framework for understanding drought dynamics and supporting more adaptive, community-informed water resource management.