Advancing the Reliability of Future Hydrological Projections in a Snow-Dominated Alpine Watershed: Integrating Uncertainty Decomposition and CycleGAN Bias Correction

Abstract

Given the sensitivity of snow to climate change and its critical role in the hydrological cycle of alpine regions, it is essential to reduce biases in meteorological forces for driving hydrological models. This study, taking the Manas River Basin (MRB) in Xinjiang China as the test bed, aims to quantify the uncertainties in hydrometeorological variables from the 24 NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP-CMIP6) simulations and further reduce these biases using a Cycle-Consistent Generative Adversarial Network (CycleGAN). The bias-corrected CMIP6 data are then used to drive the Soil and Water Assessment Tool model calibrated with both runoff and snow water equivalent (SWE) through a dual-objective approach for future projections. The results indicate that: (a) Model uncertainty brought by different climate models is the primary source of uncertainty in the original CMIP6 outputs. CycleGAN demonstrates substantial effectiveness in reducing model uncertainty; (b) Most subbasins of the MRB will experience absolute SWE reduction in the future, with changes varying significantly across elevation bands, decreasing to 30%–60% of baseline levels by the end of the century; (c) The runoff in the MRB has an increasing trend in the future, with projected increases ranging from 1.34% under SSP126 to 24.56% under SSP585. As the rain-to-snow ratio rises and snowmelt shifts earlier, low flows will increase during the dry period, elevating spring flood risks. These findings provide crucial insights for future management of water resources in snow-dominated watersheds.