Variable source area modeling in urban areas

A variable source area model called TOPMODEL was modified to incorporate urban areas in both the topographic index and the mechanism of surface runoff generation. The revised model was applied to a small catchment in the Hamilton-Wentworth area. A Geographic Information System was utilized to delineate the Ancaster Creek Catchment and produce Digital Elevation Models (DEM). FORTRAN programs in conjunction with the DEM were used to compute topographic indices that explicitly accounted for urban areas. Additional FORTRAN programs were modified to determine flow rates from this semi-urbanized catchment. Model testing was conducted in three phases: (1) a study of aggregation and scale effects using a partially revised model (TOPURBAN v. 1) on three separate time periods; (2) a fully revised model (TOPURBAN v. 2) tested on six time periods; (3) comparisons between a conventional Hortonian flow model and TOPURBAN v. 2.

Ten different DEM sizes were used to compute topographic indices and TOPURBAN v. 1 was applied during the snow-free period of a wet, dry and median year. Four separate parameters were optimized. The results indicated that three of the four calibrated parameters varied only slightly with DEM size but the hydraulic conductivity increased with increasing grid cell size. Twenty-one statistics were computed to evaluate TOPURBAN v. 1's performance for predicting flow with regard to DEM size. The statistics indicated that the DEM size of 50 m produces the most optimum conditions for the best TOPURBAN v. 1 performance. TOPURBAN v. 1 was later modified to create TOPURBAN v. 2 which better accounted for the storage in urban areas that was not considered in version 1. The new version added an additional calibration parameter and was tested on 6 data series in total. TOPURBAN v. 2 increased Nash and Sutcliffe Efficiency by anywhere from 2% to 8%. This increase proves that TOPURBAN v. 2 is a suitable model for combining urban areas into a hillslope hydrology scheme.

TOPURBAN v.2 was compared to the conventional Hortonian-Flow model QualHYMO, on all 6 continuous series, on two single events, and on the regional storm. TOPURBAN v. 2 out-performed QualHYMO in the single event tests and in the continuous series simulations by anywhere from 20%–80% in Nash and Sutcliffe Efficiency. Estimates of peak flow rate for the regional storm were within roughly 35% of the estimate provided by MacLaren Plansearch Ltd. in the Spencer Creek Watershed Study that included the Ancaster Creek Catchment.