An assessment of rainfall-runoff modeling methodology

Abstract

This study reports model performance calculations for three event-based rainfall-runoff models on both real and synthetic data sets. The models include a regression model, a unit hydrograph model and a quasi-physically based model. The real data sets are for small upland catchments from the Washita River Experimental Watershed, Oklahoma; the Mahantango Creek Experimental Watershed, Pennsylvania; and the Hubbard Brook Experimental Forest, New Hampshire. The synthetic data sets are generated with a stochastic-conceptual rainfall-runoff simulator. Model performance is assessed for a verification period that is carefully distinguished from the calibration period. Performance assessment was carried out both in forecasting mode and in prediction mode. The results on the real data sets show surprisingly poor forecasting efficiencies for all models on all data sets. The unit hydrograph model and the quasi-physically based model have little forecasting power; the regression model is marginally better. The performance of the models in prediction mode is better. The regression model and the unit hydrograph model showed acceptable predictive power, but the quasi-physically based model produced acceptable predictions on only one of the three catchments. The performance of the regression and unit hydrograph models, in both forecasting and prediction modes, for synthetic data is much better than for the real catchments. The performance of the quasi-physically based model on a synthetic data set is surprisingly poor. Supplemental data gathered from the Oklahoma catchment was used for a spatial variability analysis of steady-state infiltration rates. These data were then used to re-excite the quasi-physically based model; the new information resulted in improved model performance. The concept of space-time tradeoffs across the hydrologic data sets of competing models is introduced and tested. Results show the existence of space-time tradeoffs within model data sets but not across model data sets. It is the belief of the author that the primary barrier to successful application of physically based models in the field lies in the scale problems that are associated with the unmeasurable spatial variability of rainfall and soil hydraulic properties. The fact that simpler, less data intensive models provided as good or better predictions than a physically based model is food for thought. The model evaluation and space-time tradeoff experiments reported in this study are conceptually linked to data-worth analysis, network-design, and model-choice criteria for future studies.