Enhancing Long-Term River Flow Prediction for Effective Water Resource Management under Intensifying Drought Risks and Climate Change

The intensification of climate change has exacerbated the frequency and severity of extreme hydrological events, particularly droughts, posing critical challenges to global water resource management. The Zhuoshui River Basin, as a vital water supply region in Taiwan, has recently faced increasing extremes in rainfall and drought, highlighting the urgent need for effective management strategies. To address these challenges, this study develops a deep learning-based model for long-term monthly river flow prediction, emphasizing its significance in supporting water resource management and decision-making under worsening drought conditions.

Using historical hydrological data, the model was trained and optimized with input variables such as rainfall, evapotranspiration, and groundwater levels to explore their interactions with river flow and assess their influence on predictive performance. Future climate scenarios provided by the IPCC AR6 (Sixth Assessment Report) were employed to project river flow and groundwater levels over the next 80 years, offering insights into potential drought risks.

By combining the predicted river flow and groundwater levels with established drought assessment indices, the study quantifies drought severity and provides a scientific foundation for developing sustainable water resource management strategies in the Zhuoshui River Basin under the impact of climate change.

Keywords: Long-term streamflow forecasting, Deep learning, Drought Risk, Climate Change