Performing a sampling strategy based on 10 random grab samples, 24-h, 48-h or 7-d sampling periods or weekly grab samples for 10 consecutive weeks yielded an estimate of the EF with a relative error between maximum −39% and 48% (Table 2). If the number of random 24-h, 48-hr or 7-d sampling periods was increased, the relative error decreased (Fig. 7, Fig. 8 and Fig. 9). The estimate of the EF, based on 20 random 24-h sampling periods, had relative errors in a 95% confidence interval between −15% and 15% (dataset 1) and between −11% and 17% (dataset 2). For the 48-h sampling strategy, the estimate of the EF, based on 20 random 48-h sampling periods, had relative errors in a 95% confidence interval between −12% and 14.8% (dataset 1) and between −10% and 18% (dataset 2). Sampling 20 random 7-d periods reduced the relative errors further to −6% and 7% (dataset 1) and −4% and 11% (dataset 2).