黑河上游高寒山区集水面积阈值确定方法探讨

针对我国高寒山区河流的最佳集水面积阈值确定的问题, 选取黑河上游山区为研究区, 利用SWAT(soil and water assessment tool)模型的流域离散模块(watershed delineator)基于DEM提取河网, 探讨河网总长度与集水面积阈值关系曲线法和适度指数法在高寒山区的适用性. 结果表明: 两种方法所对应的最佳集水面积阈值相差较大, 所提取的河网难以反映河流真实情况, 效果较差, 主要原因是上述方法仅考虑流域面积、地形和几何特征的影响, 缺乏对降水和其他下垫面因子的综合考虑. 相较而言, 利用蓝线河网推求最佳集水面积的适度指数法的效果较好. 在高寒山区进行河网提取时, 应综合考虑影响河网发育的各个因素, 在流域分区的基础上, 通过不同集水面积阈值实验, 获取更高精度的数字河网, 改善分布式水文模型的空间离散效果.

How to determine drainage area threshold in alpine regions of upper reaches of Heihe River

Determination of drainage area threshold plays an important role in extracting drainage network based on DEM and affects the topological and geometric characteristics of the digital river system. To investigate the drainage area threshold of the catchments in alpine regions, eight watersheds in the upper reaches of Heihe River are selected as the case study areas, watershed delineator of the soil and water assessment tool is used to extract drainage network based on DEM. Two methods for identifying drainage area threshold are discussed and compared, including the relation between drainage area threshold and the total length of drainage network method and fitness index method. The optional drainage area thresholds obtained from these two methods greatly differ each other, and the drainage network extracted under the thresholds cannot describe the real pattern of watersheds, because these two methods mainly controlled by area, topography and geometrical characteristic of the basin, and do not take account of precipitation, underlying surface condition and so on. By contrast, the fitness index method is better, as the optional drainage area threshold is identified by using actual drainage network in this method. To obtain high precision digital drainage network and improve the spatial discretization of the distributed hydrological model, the individual factors should be taken into comprehensive consideration and the watershed should be divided into sub-regions, then different drainage area thresholds are set in the sub-regions, respectively.