Scale effects of physically based TOPKAPI model in reservoir inflow flood forecasting for ungauged basins

Abstract

Study region: Yuecheng Reservoir area, Haihe River Basin, China. Study focus: After the construction and water storage of the reservoir, the conditions for runoff generation and routing method near the dam site have changed significantly. Certain complex, scale-related issues remain insufficiently understood, limiting the application of physically based distributed hydrological models in this area. We explore the application of physically based, distributed hydrological models, with particular emphasis on the scale effects of the TOPKAPI model in rainfall-runoff forecasting. A three-step investigative framework based on geomorphological theories of hydrological response is proposed. First, fractal and geomorphological theories are applied to assess scale dependency in distributed data inputs. Then, intensive multi-scale modeling is conducted across resolutions from 2000 m to 100 m to understand how scale influences model performance. Lastly, a scale extrapolation method is proposed and validated for broader application. New hydrological insights: The results reveal a strong correlation between geomorphological features and scale, with variations in grid cell size affecting the statistical characteristics of underlying surface data. Multi-scale modeling confirms that scale impacts the performance of hydrological models, with the proposed scale extrapolation method proving highly adaptable across scales. This approach enables the transfer of model parameters from gauged to ungauged areas, providing a reliable foundation for flood forecasting in ungauged basins and supporting improved flood risk management for reservoir catchments in data-scarce regions.