ID: 1515
Presenting Author: Sewoong Chung
Session: 660 - Using Digital Tools to Enhance Transparency & Collaboration during Impact Assessments
Status: pending
A hybrid framework combining AI and mathematical models predicts turbidity from dam and its ecological stress on downstream river, supporting adaptive reservoir management under climate extr
Climate-driven extremes are intensifying soil erosion and turbidity problems in rivers and reservoirs across Korea, a monsoon-affected region. Unlike rivers where turbidity naturally dissipates within days, dam reservoirs store turbid water for hydropower generation and water supply, releasing turbid water downstream for extended periods. This prolonged turbidity imposes ecological stress on aquatic organisms, including fish and benthic macroinvertebrates. To address this challenge, we developed a digital, AI-enhanced impact assessment framework that integrates multiple models into a seamless decision-support system. The framework couples: a deep learning–based ensemble watershed model (DEWMOST) and an LSTM-supported two-dimensional turbidity model (CE-QUAL-W2). Calibration and validation using long-term datasets from the Soyanggang Dam—the most turbidity-prone reservoir in Korea—demonstrated that this hybrid system significantly outperforms conventional methods in predicting turbidity dynamics. Applications to the 2006 turbidity event and synthetic extreme scenarios revealed ecological stress index values exceeding critical thresholds, indicating lethal consequences for aquatic ecosystems. Beyond predictive accuracy, the digital platform provides an interactive tool for proactive turbidity forecasting, ecological risk assessment, and adaptive reservoir management. The findings highlight both the urgency of implementing soil erosion control and selective withdrawal strategies, and the potential of digital technologies to transform impact assessment under climate extremes.
Dr. Chung, Professor at Chungbuk National University since 2003, has 30 years experience in water research. He specializes in hydrodynamic and water quality modeling for sustainable water
Coauthor 1: Sungjin Kim
Coauthor 2: Dongmin Kim