小時風速頻率分佈的曲線擬合－極大似然法

Curve Fitting Frequency Distribution of Hourly Wind Speed Data by the Maximum Likelihood Method

王金慶(Jin-Chihng Wang)

《大氣科學》 13卷1期 (1986/03) Pp. 77-88

摘要

本文以二個參數、三個參數、四個參數對數常態分佈函數以及二個參數、三個參數、四個參數韋伯分佈函數作為不考慮風向的風速頻率理論分佈函數，以極大似然法推估頻率分佈函數的參數值，並以桃園中正國際機場地面風速觀測資料作為個案研究，以五種適合度統計量及Theil氏不等式係數作為適合度準則，結果顯示四個參數韋伯分佈函數最適於作為桃園濱海地區全年與冬半年地面風速頻率的理論分佈函數，而夏半年則以四個參數對數常態分佈函數較適合。

並列摘要

A computer program has been developed for fitting frequency distribution of hourly wind speed data. The program fits the following continuous distribution: two-parameter, three-parameter, four-parameter lognormal and two-parameter, three-parameter, four-parameter Weibull. The parameters of each distribution are estimated by the maximum likelihood method. Data from Tao-Yuan/Chung Cheng Airport are used to demonstrate the use of the above models to correlate wind speed data. Although the chi-square test leads to the rejection of the hypothesis that wind data are lognormal or Weibull distributed, the four-parameter lognormal and four-parameter Weibull appear to fit the data well enough for engineering calculations.