Determination of design hydrologic characteristics with sampling uncertainty considerations

A common problem in hydrologic frequency analysis is that the sample size is often too small to adequately represent the statistical property of an entire population. Uncertainties can thus be introduced during the determination of design hydrologic characteristics. In this paper, a new method based on Bootstrap is used to estimate the influence of sample uncertainties on the design characteristics. Compared to conventional approaches, the new method provides not only point estimates and interval estimation for the design characteristics, but also a quantitative evaluation of uncertainties associated with hydrological frequency analysis. In addition, combing with the Bootstrap technique, three schemes are established for assessing the effectiveness of the new method when using different parameter estimation approaches. The schemes are the moment method, the weight function method and the linear-moment method. The Nantong annual rainfall data from 1970 to 2011 is used to validate the new method. The result shows that with the new method, the selection of different parameter estimation approaches will have an insignificant impact on the determination of design hydrologic characteristics. This is from the point of view of the expected value, 90% confidence intervals and the final design characteristics. The new method can also deal with the limitation of the B value.