淮河流域极值降雨频率估计时空分布一致性检验和纠正

Test and Adjustment of Spatiotemporal Consistency on Quantile Estimates of Extreme Precipitation in Huaihe River Basin

由于地面雨量站点空间不均及站点资料长度有限、相邻时段和相邻一致区的最优分布函数选择不同,造成降雨频率估计值出现交叉和空间不连续的现象,时空不一致的问题尚没有得到合理的解决。为了获取更加完善合理、时空一致的频率估计值,利用水文气象分区线性矩法计算出的频率估计值,通过误差分摊和二次空间插值及平滑技术对时空分布一致性进行检验和纠正,消除分布曲线的交叉和边界梯度的问题。研究结果表明:通过误差分摊纠正后,消除了交叉异常曲线,各个时段的频率估计值相互制约更加合理;通过空间二次插值纠正后,边界梯度和误差得到明显改善,各站点的频率估计值的经验频率与理论概率更接近,空间分布也更合理。多时段不同重现期下的频率估计值空间分布总体趋势一致,最大值主要分布在淮河流域的东北沂蒙山区,最小值分布在流域的北部平原地区,与站点的实际观测相一致,可为工程设计暴雨和洪水提供基础资料和重要水文依据。

Quantile estimates of extreme precipitation show an intersection of distribution or a spatial discontinuation due to the lack of rainfall sites with spare spatial distribution as well as the short data length and different optimum distribution between ad-jacent duration and homogeneous region.The spatiotemporal inconsistency still didn't get an effective solution.In order to get more normative and spatiotemporal consistent quantile estimates,quantile estimates were computed by utilizing hydrometeorological regional L-moments method in the first section of this study,and were adjusted by spatial interpolation and distributing the surplus of the ratio of the longer duration vs.the shorter duration to eliminate the cross and discontinuation.The results show that the anomalous estimates are removed and values of mutual restriction are more reasonable by distributing the surplus of the ratio,and that adjust-ed quantiles exhibit better reliability expressed in relative error of the empirical frequency to the theoretical exceedance probability of the data,and a better spatial smoothing pattern shown as well by using spatial interpolation.The spatial patterns of quantiles are similar to different durations and return periods.High values were found in the northeast mountainous area of basin and low valueswere in the north plain area,which is in the agreement with the observed precipitation extremes.It will provide important hydrologic basis for designing rainstorm and flood of water engineering and resources.