Comparison of conceptually based and regression rainfall-runoff models, Denver Metropolitan area, Colorado, and potential applications in urban areas

Abstract

Multievent, conceptually based models and a single-event, multiple linear-regression model for estimating storm-runoff quantity and quality from urban areas were calibrated and verified for four small (57 to 167 acres) basins in the Denver metropolitan area, Colorado. The basins represented different land-use types light commercial, single-family housing, and multifamily housing. Both types of models were calibrated using the same data set for each basin. A comparison was made between the storm-runoff volume, peak flow, and storm-runoff loads of seven water-quality constituents simulated by each of the models using identical verification data sets. The multievent, conceptually based models studied were the U.S. Geological Survey's Distributed Routing Rainfall-Runoff Model Version II (DR3M-II) (a runoff-quantity model designed for urban areas), and a multievent, urban runoff-quality model, DRsM-QUAL. Water-quality constituents modeled were chemical oxygen demand, total suspended solids, total nitrogen, total phosphorus, total lead, total manganese, and total zinc. A multiple linear-regression analysis, based both on log-transformed and untransformed data, was made. A separate regression equation was developed for each runoff characteristic or water-quality constituent, and all the regression equations were basin specific. The two types of models produced comparable results for most runoff characteristics and water-quality constituents in the basins studied. However, development and implementation of multievent, conceptually based models are more costly and time consuming than development and implementation of single-event, multiple linear-regression models.