A study on initial and continuing losses for design flood estimation in New South Wales

![CDATA[Flood estimation is often required in the design and safety assessment of water infrastructure. Rainfall runoff model is often used in design flood estimation, which needs various inputs on rainfall and catchment characteristics such as rainfall duration, rainfall intensity, rainfall temporal patterns and losses. Loss is defined as the amount of precipitation in a rainfall event that does not appear as direct surface runoff at the stream gauge. The initial loss-continuing loss is the most commonly adopted conceptual loss model in Australia. Currently there is inadequate information on design loss values in Australia which is a major weakness in Australian flood hydrology. The currently recommended design losses in Australian Rainfall and Runoff (ARR) are not compatible with the design rainfall information and are likely to result in significant bias in flood estimation. This can result in either under or over-design of water infrastructure; both of these have important economic significance. Also, the currently recommended ARR design loss values ignore its probabilistic nature. The objective of this paper is to derive improved initial and continuing loss values using data from selected catchments in New South Wales (NSW). This intends to derive mean and median loss values as well as stochastic losses which can be applied with both the Design Event Approach and Joint Probability Approach/Monte Carlo simulation technique to design flood estimation. The method used to estimate continuing loss involves the use of a water balance equation involving a rainfall and runoff event. A total of 253 rainfall runoff events are selected from five NSW catchments. From the analyses, the median initial loss has been found to be 17 mm which is closer to the lower limit of the ARR recommended value of 10-35 mm. The median continuing loss value has been found to be 0.94 mm/h which is 62% lower than the ARR recommended value of 2.5 mm/h. The resulting design floods from the observed loss values are likely to be higher than those obtained from the ARR recommended loss values. It has been found that the observed initial loss and continuing loss values show a wide variability from storm to storm and catchment to catchment. It is thus more logical to adopt stochastic losses in design flood estimation rather than a fixed value of loss. It has been found that two-parameter Gamma distribution can be used to describe the observed initial loss and continuing loss distributions. The stochastic losses described by the Gamma distribution can be used in design flood estimation by adopting Joint Probability Approach/Monte Carlo simulation technique. The Gamma distribution to be used for this purpose is specified by a mean value of 22 mm and standard deviation of 19 mm for initial loss and a mean value of 1.20 mm/h and standard deviation of 1.04 mm/h for continuing loss.]]