A calibration framework for distributed hydrological models considering spatiotemporal parameter variations

Abstract

In urbanized watersheds, climate change and human activities significantly impact runoff, yet traditional hydrological models cannot dynamically adjust parameters based on land use changes, and calibration methods fail to capture hydrological processes under all flow conditions accurately. This study addresses these issues by first parallelizing the chaotic particle swarm genetic algorithm (CPSGA) and successfully applying it to calibrating distributed hydrological models. Secondly, considering the rapid land use changes in urbanized watersheds, the HBV distributed hydrological model was improved according to the distribution of hydrological corresponding units (HRUs) to achieve spatiotemporal parameter variation, overcoming the limitations of traditional models in long-term calibration due to land use changes. Lastly, we established a time-segmented spatiotemporal parameter variation calibration framework that considers the effects of human regulation and climate change, effectively capturing the inter-annual and intra-annual variations in hydrological processes, thereby improving model performance across different periods. The above methods were applied to the Shaying River Basin and validated, and the results show that the parallel CPSGA could enhance model calibration accuracy and speed. The model performance with a time-segmented spatiotemporal parameter variation calibration framework is significantly improved under different flow conditions. The suggested method in this study is an effective tool for simulating discharge that changes over time in a dynamic environment.