基于Hurst系数与Bartels检验的水文变异联合分析方法

Joint Analysis Method for Hydrological Variation Based on Hurst Coefficient and Bartels Test

针对Hurst系数法对水文序列周期性及相依性诊断灵敏性弱的问题,新增Bartels检验法予以增强.采用蒙特卡洛方法生成试验序列,依次通过Hurst系数法和Bartels检验法得到了各自的检测性能:Hurst系数法对于序列的趋势性及跳跃性检测效果较强,Bartels检验法对于序列的周期性及相依性检测效果较强.提出了基于Hurst系数与Bartels检验的水文变异联合分级方法:两种方法测出均无变异则可确定为整体无变异;至少一种方法测出强变异则可确定为整体强变异;其它情况一律视为弱变异.对长江中上游重要控制站点寸滩、宜昌和汉口站年径流序列进行变异分析,结果表明寸滩站、宜昌站和汉口站年径流变异显著,变异程度分别为强变异、强变异和弱变异,变异形式可能为周期或相依变异.

Bartels test is added to enhance diagnostic sensitivity of periodic or dependent hydrological series for the weak diagnostic sensitivity of Hurst coefficient method for them. Monte Carlo method is used to generate testing sequence, and the test performance is acquired through Hurst coefficient method and Bartels test .respectively： Hurst coefficient method is superior for the series with trend or jump variation and Bartels test gets better for the series with periodicity or dependency variation. Joint classification method for Hydrological variation based on Hurst coefficient and Bartels test was proposed ： When both of the two tests detect no variation, we can conclude that the whole has no variation;at least one test detects strong variation corresponding to the whole with strong variation;others are regarded as weak variation. The variation of annual runoff series at Cuntan, Yichang and Hankou station which are the important control sites of middle and upper reaches of Yangtze River was analyzed. The results showed that the series of the above-mentioned three stations were significantly inconsistent. The variation degree was strong, strong and weak, respectively, and the variation form could be periodic or dependent.