การแจกแจงความน่าจะเป็นของฝนรายวันสูงสุดในประเทศไทย

Abstract

Abstract

The objective of this paper is to perform the frequency analysis of extreme daily rainfalls in Thailand. The frequency analysis applies the linear moment diagrams of corresponding precipitation data for 121 stations in Thailand. The moment diagram method is more convenient than the widely used goodness-of-fit test because it is able to compare the frequency analysis results by only 3 graphs. Further, the linear moments are almost unbiased statistics. Results of comparing the fit of 2-parameter distributions (lognormal: LN2, Gamma, and Gumbel) and 3-parameter ones (LN3, Generalized Extreme Value: GEV, Generalized Logistic: GLOG, Pearson type 3: P3, and Log P3: LP3) with the maximum precipitation data have shown that the LN2 and GEV models are the true 2- and 3-parameter distributions of the rainfalls respectively. To conform to the principle of parsimony, the LN2 model is recommended when the available rainfall data of a considered site are limited (< 20 years). Otherwise, the GEV model is suggested.