主成分聚类法在辽宁无资料地区河流洪水模拟中的应用

Application of principal component cluster method in river flood simulation for region without data in Liaoning Province

针对辽宁省新建中小河流无实测和调查径流资料,难以建立预报方案的难题,提出主成分聚类方法,将有资料地区水文站点参数进行移植,用于无径流资料地区的洪水模拟。结果表明:主成分聚类方法如综合考虑下垫面类型、水文气候特性,参数移植更为可靠。通过与参考站实测洪水对比,参数移植后洪水模拟的确定性系数可达0.70以上,洪峰模拟相对误差均值低于20%,总体预报精度可达乙级标准。成果可为其他省份无资料区域中小河流预报方案构建提供参考和借鉴价值。

Some middle and small rivers in Liaoning province have no observed data or survey runoff data and are difficult to build forecast scheme.In the view of this problem,the paper provides the principal component cluster method,made hydrological stations parameter transferring from region with data to the region without runoff data for flood simulation.The results show that the principal component cluster method could comprehensively consider the type of underlying surface and the hydrological climate characteristics and make a more reliable parameter transfer.Compared with the observed flood of the referential station,after the parameter transfer,the deterministic coefficient of the flood simulation could reach to 0.70 or more,the relative error average of the simulated flood peak is below 20%,the total forecast accuracy could reach to B-class standard.The results could provide reference for forecast scheme building of middle and small rivers in region without data in other provinces.