考虑成洪源头差异性造成的洪水样本不一致,根据成洪暴雨的主要类型构建分季序列,进而提高洪水频率分析结果的合理性。以武江上游坪石站为例,将月最大瞬时流量分为枯期(11-翌年2月)、锋面雨集中期(3-6月)和台风雨集中期(7-10月)3组分季序列。通过对比频率结果,分析各分季序列变化对洪水频率的影响,以及基于年最大、分季序列进行洪水频率计算的差异。结果表明:1武江流域上游洪水年内季度差异性显著,枯期均值(295 m^3/s)及对年最大构成比重(2.22%)皆为最小;锋面雨期均值(1 111 m^3/s)及对年最大构成比重(77.78%)最大;台风雨期波动性最大。2分季序列中含有上升趋势,造成同一量级洪水对应重现期增大。其中,锋面雨期变化对洪水频率影响最大,但影响有减小趋势;台风雨期变化造成洪水设计值两极化,且影响有增加趋势。3非一致性频率计算所得同频洪水量级比传统更大,基于分季序列所得设计值则比年最大高,且随重现期增大而增加,百年一遇洪水量级高出约1 000 m^3/s。
Changes of flood series vary among different seasons. This research developed a flood frequency analysis method based on non-stationary seasonal maximum peak flow. Taking upstream of Wujiang River( UWR) as our study area,seasonal extremes were drawn according to three seasons: dry season( DS) from Nov. to Feb.,season of frontal rain( SFR) from Mar. to Jun. and season of typhoon rain( STR) from Jul to Oct. The influences of different seasons on the results of flood frequency analysis were also studied. The results show that: 1 Statistical characteristics of different seasons vary in UWR. Mean and contribution to AMS( annual maximum series)( 2. 22%) of DS are both the lowest while that of SFR are the highest. Standard deviation of STR is the highest. 2 All the seasonal series( SS) have ascending trends. Change of SFR causes the most influence on design flood,but the influence is descending. Variation of STR expands the deviation between design floods of different return periods,and the influence is increasing. 3 Design flood from non-stationary flood frequency analysis were larger than the one from traditional method. The one based on SS is larger than AMS,and the difference between increases along with the return period,which is more than 1000 m~3/ s when return period is 100 years.