内蒙古高原典型草原内陆河流域径流的时序演变特征

高原内陆河在我国半干旱地区社会经济发展的区域水资源开发利用过程中起到了重要的支撑作用,认识其演变特征及进行模拟对于流域的可持续发展具有重要的意义。以内蒙古高原典型草原内陆河流域——巴拉格尔河流域为研究对象,基于1959-2015年的水文、气象和社会经济等数据,采用双累积曲线法、累积距平法、改进的M-K趋势检验法和小波变换法对流域径流序列的变化特征进行了解析,并通过细化降水特征因素影响和以社会经济指标量化人类活动影响,运用统计相关检验法筛选径流时序演变模拟的主要输入要素,以改进的BP神经网络实现了水文年、季尺度下径流的时序演变模拟。结果表明:巴拉格尔河流域径流量在年、季尺度下呈现显著减少趋势;径流变化的突变年份为1998年,在这之前受到气候变化为主的影响,而在这之后为加入人类活动的双重影响;年、枯水季径流序列均存在6a和30a周期,而丰水季径流在分析序列内无明显周期变化;降水指标中的降水量、降水天数和降水次数,气候因素中的蒸发和相对湿度,以及人类活动的所有指标是与径流显著相关的影响因素;在无法实现分布式水文模拟时,改进的BP神经网络可以较好地模拟该流域径流的时序演变过程。

Evolution characteristics of runoff in the typical grassland inland river basin of Inner Mongolia Plateau

Inland river plays an important role in the social and economic development and regional water resources utilization in semi-arid plateau areas of China. To recognize and simulate its evolution characteristics is very significant for the basin''s sustainable development. In this paper, Balag erriver basin was selected as the study object . Based on the hydrological, meteorological, and socio-economic data in 1959-2015, we analyzed the characteristics of the runoff series by the double mass curve method, the cumulative departure method, the improved M-K trend test, and the wavelet transform method. We used statistical correlation test to select the main input factors for runoff evolution simulation, and realized the simulation of annual and seasonal runoff series evolution by BP artificial net work. The results showed that the runoff decreased significantly at the annual and seasonal scale. The abrupt change of runoff occurred in 1998, before which the runoff was mainly influenced by climate change, while after which by both human activities and climate change. There was a 6-year period and a 30-year period for the annual and dryseason runoff, yet the wet-season runoff showed no obvious periodic change. The impact factors significantly correlated with runoff included precipitation quantity , wet days, number of precipitation events, evaporation, relative humidity , and all indicators of human activity. The improved BP neural network can well simulate the evolution of runoff in this basin when the distributed hydrological simulation cannot be realized.