利用短序列高密度站资料推算暴雨重现期方法研究及应用

Study and application of method for estimating rainstorm return period from short-series high-density station data

中文关键词:  空间抽样合成法  百分位合成样本  年最大日降水  重现期推算

英文关键词:Spatial bootstrap synthesis method, Percentile of synthetic sequence, Annual maximum daily rainfall, Return period estimation

中文摘要:

      [目的]暴雨重现期是城市排水防涝设计的重要基础，通常基于长年代观测数据进行推算。但在无降水观测或观测时间较短的情况下，如何进行重现期推算和暴雨强度评估，是目前亟需解决的科学问题。[资料和方法]本文基于重庆市近11年高密度降水观测资料，建立了逐年日降水极值抽样数据集，以“空间换时间”的思想，对日降水极值样本进行空间抽样，通过与国家站长序列观测数据（＞60年）进行交叉检验，构建试验区目标点最佳百分位合成序列样本集，该方法简称为空间抽样合成法（SBS）。[结果]通过重庆地区34个测站长年代观测资料，采用SBS与邻近点替换、Cressman空间插值、年多个样法等推算的重现期降水量同“真值”进行对比检验，就平均而言，SBS的相对误差最小，其中含目标点样本的SBS相对误差最小为7.2%，临近点替代法相对误差最大为13.2%，表明SBS可以较好地用于我国复杂地形的重庆地区，利用短序列高密度站资料构建无长序列观测资料目标点处的长序列极值降水样本，从而开展概率拟合优选及暴雨重现期推算。在上述方法验证基础上，实现重庆地区2062个高密度气象观测站多年（50 a）一遇重现期降水量推算，提高了日尺度极端降水的精细化水平，结果能更好反映山地地形对降水的影响。[结论]SBS可以充分利用短序列高密度雨量站观测资料，实现区域内任意目标点重现期降水量推算，应用前景广泛。

英文摘要:

      [Purpose] The rainstorm return period is an important basis for urban drainage and flood control design, which is usually calculated based on long-term observation data. However, in the absence or short periods of precipitation observations, how to calculate return periods and evaluate storm intensity is an important scientific problem that needs to be solved urgently. [Data and methods] Based on the high-density precipitation observation data in Chongqing over the past 11 years, we established the yearly maximum daily rainfall data set. With the idea of "space trade for time", the daily precipitation samples are bootstrapped and using cross-examination of the national stations with long-term observation data (more than 60 years) to select the optimal percentile synthetic sample set of the target point, which is defined as the spatial bootstrap synthesis method (donated as SBS). [Results] Comparing the return period rainfall of 34 stations with long-term observation data in Chongqing by original sequence calculation and other various methods, on average, the relative error of the SBS is smaller than the other three methods including the nearest station replacement, Cressman interpolation and annual multi-sampling method. Among them, the SBS containing target point samples has the smallest relative error of 7.2%, and the nearest station replacement method has the largest relative error of 13.2%. This indicates that the SBS can be used well in Chongqing, a complex terrain area of China, to construct long-sequence extreme precipitation samples by making use of short-series high-density stations surrounding target point, which use to fit the probability distribution function and calculate the return period of precipitation. On this basis, the 50a return period of precipitation of 2062 high-density meteorological observation stations in Chongqing are calculated, which improve the fine level of daily extreme precipitation and better reflect the influence of mountainous terrain. [Conclusions] The SBS can make full use of the observation data of short-series high-density rainfall gauge stations to estimate the return period of precipitation at any target point in the region, and has a wide application prospect.