Open Collections

Abstract

A suite of three underlying rainfall-runoff modeling techniques is applied to two data sets and the results used to compare model efficiencies for selected events. Linear regression, unit hydrograph, and quasi-physically based models make up the modeling suite. The two data sets come from a 7.2 KM subwatershed (MCW) near Klingerstown, Pennsylvania and a 0.096 KM2 subwatershed (R-5) near Chickasha, Oklahoma.-Individual model efficiencies are determined on the basis of a sums of squares criterion. These efficiencies are surprisingly poor. Results indicate that the most informative independent linear regression variables for MCW and R-5 are volume of rainfall and average rainfall intensity respectively. There is a general improvement in correlation coefficients and regression model efficiencies for both MCW and R-5 with increases in the number of selected events. The unit hydrograph and quasi-physically based models exhibited predictive prowess only for the R-5 events. The unit hydrograph technique is found to be strongly dependent upon an accurate estimate of spatially-variable excess rainfall. The efficiency of the physically-based, deterministic, distributed model was found to deteriorate drastically with increases in basin size due to the lumping of spatially-variable soil hydraulic properties. Based on this work a definitively superior rainfall-runoff modeling technique is not suggested. Limitations of each of the three models and the efficiency criterion used for their evaluation are discussed. This work provides the foundation for a subsequent investigation to be carried out by the author, to determine if space-time tradeoffs exist across data sets of various rainfall-runoff modeling techniques. Future research will focus on the concept of data-worth and the question of model choice.