区域频率分析在黑龙江省基本雪压估算中的应用

Application of regional frequency analysis in estimation of ground snow load in Heilongjiang province

在极值分析中,样本量不足会导致小样本效应,这将影响分析结果,需采取适当措施予以减弱。以黑龙江省83个气象台站测得的积雪数据为例,探讨了区域频率分析方法在基本雪压估算中的应用。为此,介绍了区域频率分析的具体步骤,而后采用蒙特卡洛模拟技术,以模拟计算结果的均方根误差与偏差分布情况为基准,将区域频率分析方法与两种常用方法进行了对比。模拟结果表明,区域频率分析方法能有效提高基本雪压估算结果的准确性和稳定性,当样本数量较小时,其优势更加明显。最后给出了3种不同方案估算得到的黑龙江省基本雪压分布图,发现我国规范建议的估算方法偏于保守。

Small sample size effect affects the accuracy of the extreme value analysis results. Various techniques could be applied to reduce this effect,including the regional frequency analysis（ RFA）. An application of RFA to estimate the ground snow load using annual maximum snow depth records from 83 meteorological stations in Heilongjiang province was presented. The estimated extreme values were compared with those obtained from on-site analysis.Moreover,Monte-Carlo technique was employed to assess the adequacy of the RFA versus the on-site analysis to estimate the extreme snow depth or snow load. Judging based on root-mean-square-error and the bias obtained from the simulation analysis,it is concluded that the RFA is preferred,especially when the sample size is relatively small. The snow load contour maps are developed for the province by using return period values of annual maximum snow load estimated by three different approaches,indicating that the approach recommended by the Chinese design code is relatively conservative.