Reliable assessment of flood impact: an urban stormwater case study of a City of Sydney catchment

The flood risk in many urban catchments is poorly understood. Legacy stormwater infrastructure is often substandard and anticipated climate change induced sea level rise and increased rainfall intensity will typically exacerbate present risk. In a Department of Climate Change and Energy Efficiency (DCCEE) funded collaboration between Geoscience Australia (GA) and the City of Sydney (CoS), the impacts on the Alexandra Canal catchment have been studied. This work has built upon detailed flood hazard analyses by Cardno commissioned by the CoS and has entailed the development of exposure and vulnerability information. Significantly, the case study has highlighted the value of robust exposure attributes and vulnerability models in the development of flood risk knowledge. The paper describes how vulnerability knowledge developed following the 2011 Brisbane floods to include key building types found in the inner suburb of Sydney. It also describes the systematic field capture of building exposure information in the catchment area and its categorisation into 19 generic building types. The assessment of ground floor heights using the Field Data Analysis Tool (FiDAT) developed at Geoscience Australia is also presented. The selected hazard scenario was a 100 year ARI event with 20% increased rainfall intensity accompanied by a 0.55m sea level rise in Botany Bay. The impact from the selected scenario was assessed in terms of monetary loss for four combinations of vulnerability model suite (GA and NSW Government) and floor height attribution method (assumed 0.15m uniformly and evaluated from LiDAR and street view imagery). It was observed that the total loss is higher in the case of assumed floor heights compared to FiDAT processed floor heights as the former failed to capture increased floor heights for newer construction. However, the loss is lower when only two vulnerability models developed by NSW Government are applied for the entire building stock in the region as two models could not reliably represent the whole building stock.