Predictive Modeling of Storm-Water Runoff Quantity and Quality for a Large Urban Watershed

A predictive model for storm-water runoff was implemented on a GIS platform based on the unit area loading method and Browne’s empirical relation for soil characteristics for the Upper Ballona Creek Watershed in Los Angeles. The heterogeneity of the watershed was quantified by dividing it into many small subareas and applying lumped parameters for each. Characterization of total pollutant load by land-use types to total loads was achieved through zeroth-order regularization and limited memory Broyden–Fletcher–Goldfarb–Shanno bound constrained optimization techniques. Relative form was used in the objective function to compensate for strong contributions of high magnitude variables. Model predictions showed reasonable agreement with pollutant loadings, using Zn as an example, measured at the mass emission site at watershed mouth. The predicted runoff volumes using the developed quantity model were in good agreement with the data and had R2 of 0.86. The RMS error of the quality model was 9kg , which is low compared to the mean discharge of 77kg∕event .