Towards simplification of hydrologic modeling: identification of dominant processes

Summary

This study applied parameter sensitivity analysis to a spatially distributed hydrologic model across the conterminous United States to identify dominant hydrologic processes and their associated sensitive parameters. Using the Fourier amplitude sensitivity test on 110,000 spatial modelling units, the authors determined which of 35 calibration parameters most influence model performance for different hydrologic processes. The findings indicate that apparent model complexity can be substantially reduced by focusing on processes with sensitive parameters, and that different hydrologic processes require different numbers of parameters for adequate simulation.

UK applicability

The methodology and findings are relevant to UK hydrologic modelling and water resources management, though the specific parameter sensitivities identified for the continental United States may require recalibration for British catchments with different climate, geology, and land cover characteristics. The principle of identifying dominant processes to simplify model calibration is transferable to UK environmental monitoring and flood forecasting systems.

Key measures

Parameter sensitivity values for 35 PRMS calibration parameters computed using Fourier amplitude sensitivity test; sensitivity summarised by hydrologic process (snowmelt, surface runoff, infiltration, soil moisture, evapotranspiration, interflow, baseflow, runoff) and model performance statistic (mean, coefficient of variation, autoregressive lag 1)

Outcomes reported

The study used parameter sensitivity analysis on 110,000 spatial modelling units across the conterminous United States to identify which hydrologic processes are most sensitive to model parameters and which parameters most influence model performance. Results demonstrate that model complexity can be reduced by focusing calibration efforts on processes associated with sensitive parameters.

Topic tags