Abstract:
To obtain reliable at-site flood quantile estimates from a small sample of observations is a trivial exercise. To over come this problem, attempts to (i) use Probability Weighted Moment (PWM) method of parameter estimation and (ii) to increase the length of database through the incorporation of discontinuous extraordinary events by making the new sample a censored one have been in vogue. Though the use of PWM method has generally been accepted as a means of reduction of bias in the estimates, the usefulness of the later is yet to be established. This study, through a computer simulation, tries to bring out the worth of incorporation of the extraordinary events, particularly while such events as well as the observed sample are likely to have been corrupted with varying degrees of measurement error. It is found that incorporation of extraordinary event(s) yielded reliable estimates of quantiles, particularly with small sample sizes, even if the extraordinary event and the sample had measurement error up to 30 and 10% respectively.
