Trend analysis of dam inflow data using the Trend Accuracy Index and the Potential-Evapotranspiration Correction Factor

Korea is facing growing challenges in water-resources management because climate change is increasing the variability of rainfall and runoff. Accurately identifying station-specific trends in rainfall and dam inflow in advance would help maximize the benefits of efficient water-resources plans and hydraulic structure design. This study performed cluster and trend analyses for past (2000–2019) and future (2021–2050, 2051–2100) periods under the SSP2-4.5 and SSP3-7.0 climate change scenarios, using rainfall data from 101 weather stations and inflow data for Hapcheon Dam in the Nakdong River basin, Korea. In the trend analysis based on the modified Mann-Kendall test, only monthly, seasonal, and annual series that passed all three homogeneity tests were used. To simulate future dam inflows, we adopted a Four-Tank Model augmented with the Trend Accuracy Index (TAI) and a Potential-Evapotranspiration Correction Factor (PET-CF). The model parameters were simultaneously optimized by a genetic algorithm, enabling the simulated dam inflows to replicate observed trend. K-means + + clustering with three clusters revealed distinct rainfall characteristics associated with location and elevation of the station. Trend analyses showed diverse monthly, seasonal, and annual tendencies, and the trends of rainfall and dam inflow were not always consistent in either the historical or future periods. When both TAI and PET-CF were applied, the Four-Tank Model reproduced dam-inflow trends with the highest concordance (76.4%) compared to observations. The identified rainfall clusters and trend relationships provide a practical basis for developing adaptive water-resources strategies that account for climate-driven variability.