Powells Creek has a catchment area of approximately 8.1 square kilometres to Homebush Bay Drive and 9 square kilometres to its junction with the Parramatta River in Homebush Bay. Saleyards Creek is the major tributary of Powells Creek and has a catchment area of 3.2 square kilometres, joining it immediately upstream of Homebush Bay Drive. Powells Creek has a catchment area of approximately 4.9 square kilometres to Homebush Bay Drive.

Flooding in both creeks has occurred several times in the past, but only limited records are available. A number of drainage studies have been previously undertaken to assess the capacity of the drainage network and to investigate solutions to the flood problem. This report details the hydrologic and hydraulic investigations which have been undertaken to determine the design flood levels along the main drainage lines.

The approach adopted for the hydrologic and hydraulic modelling of the catchment was influenced by the quantity and quality of the data available and accepted practice. The ILSAX hydrologic model and the HEC-RAS hydraulic model were considered to be the most appropriate for the task. The ILSAX model was set up to cover the entire catchment of both creeks; the MEG-RAS model was structured to model the main drainage lines within the floodplain.

Calibration of the hydrologic and hydraulic models was undertaken using the available historical data. The calibration was limited by the availability of streamflow and rainfall data from the University of New South Wales (UNSW) and recorded historical flood levels. Overall the calibration of the models was considered satisfactory and the models are therefore appropriate for use in design.

Design rainfall intensities were obtained from the Bureau of Meteorology and design temporal patterns from Australian Rainfall and Runoff, 1987. These were input into the hydrologic model to determine the flows required for the hydraulic model. Design flood levels were then obtained using the hydraulic model.

The design flood levels obtained at the Elva Street stream gouge (operated by the UNSW) were compared to the results obtained from flood frequency analysis of the gouge records and a very good match was obtained Sensitivity analyses were undertaken to test the robustness of the adopted design flows.

The accuracy of design flood levels at any location depends upon a number of factors including the availability of nearby reliable historical flood data, the quality of the survey data assumptions made in determining the downstream boundary conditions and to a lesser extent upon the accuracy of the design rainfall intensities. Due to the paucity of historical flood data, and the difficulties in determining flow paths in a densely built urban area, the accuracy of the design levels determined in this study is considered to be within ±0.4 m.

It is recommended that flood data should be collected, surveyed, photographed and recorded immediately following all future floods. This data could then be used to review the design flood levels presented in this report and thereby reduce the error band around the design levels. The quantity and quality of available peak height data can be significantly improved with installation of flood monitoning equipment. The most cost-effective approach is to install a series of maximum height recorders and prepare and implement a post-flood evaluation program. This will ensure that the data are accurately recorded and analysed. A more comprehensive calibration of the hydrologic model should be undertaken as data becomes available from the UNSW stream and rainfall gauges.

The findings of the study are provided in the Appendices of this report as a tabulation of design levels, flows and velocities along the main drainage lines.