An Analysis of the influence of the field level hydraulic property approximations in stormwater modeling

Abstract

This research analyzes the sensitivity of field estimated watershed parameters and their influence to the results of watershed mathematical modelling using US EPA SWMM 5. The approximations done in channel physical parameter measurements and the catchment intermittent storages modelling were evaluated in this study. The literature study on the state of the art stormwater modelling reveals that EPA SWMM 5 is capable of handling complex watershed behaviours and can meet the objectives of this study. The Karasnagala watershed, located in Gampaha district in the western province of Sri Lanka, was used as the study area, considering the availability of gauged data and evaluates the applicability of EPA SWMM 5 to regional watershed. Event based modelling approach is closer to science and it has a capability of better representation of watershed behaviour. So the event based approach was followed in this study. Event separation used 3 days minimum inter event time (MIT), which was derived from Unit Hydrograph theory and 0 mm/day minimum inter event depth (MED) as the threshold rainfall cut-off value of the event analysis. Twenty five rainfall-runoff events for model calibration and 25 rainfallrunoff events for model verification were separated from the 1971 to 1982 period of measured data at the Karasnagala gauging station. The baseflow separation from the observed hydrographs was required to compare the EPA SWMM 5 derived direct runoff hydrograph with the observed streamflow hydrograph. Systematic approach was followed to identify the generation and the behaviour of the baseflow and then identified the advantages and the disadvantages of baseflow separation methods used in the research studies. The evaluation of baseflow separation methods with literature support and according to the objectives of this study; concluded that die graphical baseflow separation methods have the top priority. The concave method was selected for baseflow separation for this study.