Optimum number of storms required to derive site mean concentrations at urban catchments

摘要

A reiterative analysis was applied to determine the optimum number of storms required to generate site mean concentrations (SMC), using total phosphorus data from 17 urban catchments. For each analysed catchment, event mean concentration data were randomly placed into various calibration set sizes, ranging from 1 to N (where N was equal to the total number of storms available at the given catchment). Geometric mean estimates of SMCs associated with each calibration set size were then calculated, and verified using all available data from the catchment of interest. This process was repeated 10,000 times for each catchment. Average errors associated with each sample size were then plotted and used to estimate the optimum number of storms required to derive SMCs at each catchment. The optimum was derived by evaluating the balance between cost and uncertainty, whereby the minimum number of storms producing a relatively accurate estimate of SMC was accepted. Overall, it was found that between five and seven storm events were sufficient. In addition, it was deduced that sampling only six storm events would be approximately 40% cheaper than sampling 12 events.