Abstract:

Rainstorm patterns and extreme rainfall return periods are important meteorological parameters for cities response to rainstorm and flood disasters. Based on the 60-year minute-by-minute precipitation data from the national meteorological stations in Xining, the fuzzy recognition method and statistical analysis were used to classify and analyze the characteristics of rainfall processes in Xining. The Gumbel distribution, Pearson-III distribution, and generalized extreme value distribution were applied to fit the probability distributions of the annual maximum rainfall extremes for 11 different durations. The goodness of fit for each distribution was evaluated, and the return periods of rainfall extremes for each duration were estimated. The results show that, from 1954 to 2022, there were a total of 95 rainstorm events in Xining, with cumulative rainfall ranging from 25 to 40 mm. The number of rainstorm days showed significant interannual variation, with an overall increasing trend. The rainstorm pattern was predominantly unimodal, accounting for 85.3% of total rainstorm events, which mainly occurred from July to September. In the unimodal pattern, the peak rainfall was primarily concentrated in the early to mid-phase of the rainfall process. The high rainfall intensity occurred from 20:00 to 09:00 the following day, while the high-frequency rainfall periods were between 23:00 and 09:00 the following day, although these two periods did not fully coincide. The probability distributions of rainfall extreme values for different durations obeyed the Gumbel, Pearson-III, and generalized extreme value distributions well, and the Pearson type III distribution had the highest fitting accuracy. The maximum hourly rainfall for Type I events reached 43.7 mm, exceeding the 100-year return period; for Type II events, the maximum hourly rainfall was 38.1 mm, exceeding the 50-year return period; and for Type V events, the maximum hourly rainfall was close to 20.0 mm, approaching the 20-year return period. The maximum hourly rainfall for other patterns ranged from the 2-year to the 10-year return period. The identification of rainstorm patterns and extreme rainfall return periods in Xining can provide valuable scientific data for urban flood prevention and outdoor drainage system design.

Key words: Xining, temporal pattern of torrential rain, short-duration heavy rainfall, return period

摘要： 暴雨雨型和极端强降水重现期是城市应对暴雨洪涝灾害的重要气象参数。利用西宁市国家基本气象站近60 a逐分钟降水资料，采用模糊识别法和数理统计法对西宁市暴雨过程进行雨型分类和特征分析，并利用耿贝尔分布、皮尔逊Ⅲ型分布和广义极值分布对11种不同历时年最大降雨极值的概率分布进行拟合，评价拟合优度，并推算各历时降雨极值重现期。结果表明：1954—2022年西宁市共出现暴雨95场，过程累计雨量为25~40 mm，暴雨日数年际差异较大，整体呈增加趋势。暴雨雨型以单峰型为主，占总暴雨场次的85.3%，主要出现在7—9月。单峰雨型中，峰值多位于降水过程的前中期；雨量高值时段为20：00—次日09：00，高频率时段为23：00—次日09：00，两者并未完全重合。不同历时降雨极值的概率分布均较好地服从耿贝尔分布、皮尔逊Ⅲ型分布和广义极值分布，其中皮尔逊Ⅲ型分布的拟合精度最高。Ⅰ类雨型中小时降雨量最大达43.7 mm，超过100 a一遇的水平，Ⅱ类雨型的最大小时降雨量达38.1 mm，超过50 a一遇水平，Ⅴ类雨型的最大小时降雨量接近20 a一遇水平，其他雨型的最大小时降雨量在2 a一遇到10 a一遇水平。通过确定西宁市暴雨雨型和雨量极值的重现期，可以为城市内涝防治和室外排水设计提供科学依据。

关键词: 西宁, 暴雨雨型, 短历时降雨, 重现期