中文摘要：

基于近似熵复杂性测度方法分析泾河流域张家山、杨家坪和雨落坪水文站1956-2000年的天然、实测径流复杂度.结果表明：1）径流复杂度具有人类活动影响指示特性：杨家坪和张家山站实测径流复杂度大于天然径流的,杨家坪站径流复杂度变化可能与上游的巴家嘴水库的运用方式有关,而雨落坪站实测径流复杂度小于天然径流的,可能与上游大规模的水土保持有一定关系,流域降水量的变化对径流复杂度有重要影响;2）径流复杂度具有自相似特性与随机波动特性：杨家坪和雨落坪站径流复杂度的自相似性强于张家山站的,张家山站径流复杂度的随机波动性较强,与降水量关系密切;3）径流复杂度与降水量、径流量之间存在着反向变化特性：当降水量和径流量增加时,径流复杂度减少,反之则增大;4）径流复杂度具有径流峰值异常特性：径流峰值出现前后,径流复杂度显著下降,之后又逐渐上升.基于近似熵方法检测出的径流复杂度多维特性,可为径流系统特征分析提供参考.

英文摘要：

Based on the complexity measuring approach for approximate entropy, the complexity of observed and natural runoffs from the Yangjiaping, Yuluoping and Zhangjiashan hydrologic stations in 1956-2000 in the Jinghe Watershed were analyzed and the results demonstrated that： 1） The runoff complexity had the indicator characteristics of human impact. The complexity of the observed ruoffs was bigger than that of the natural runoffs for the Yangjiaping and Zhangjiashan stations. The complexity of the observed runoff from the Yangjiaping station was possibly in relation to the operation mode of the Bajiazui reservoir, whereas the complexity of the observed runoff from the Yuluoping station was smaller than that of its natural runoff, which was probably relative to the macro scale of the soil and water conservation activities at the upper watershed. The changes in watershed precipitation have had important influence on the runoff complexity; 2） The runoff complexity showed self-similarity and stochastic fluctuation characteristics. The self-similarity of the runoff complexity from the Yangjiaping and Yuluoping stations was larger than that of the Zhangjiashan station, with the latter having a stronger stochastic fluctuation characteristic of runoff complexity. The characteristics had a close relation to the watershed precipitation; 3） There was an inverse changing characteristic between complexity and runoff and precipitation. When the runoff and the precipitation increased, the complexity decreased, otherwise it would increase; 4） The complexity degree of runoff showed abnormal characteristics of runoff peaks. The complexity degree significantly decreased before and after runoff peaks occurred, and then gradually increased when the peaks disappeared. The approximate entropy approach has the ability of accurately indentifing the multi-dimension characteristics of the runoff complexity, thus offering a useful tool for analyzing the characteristics of a runoff system.