摘要
降水空间分布信息是进行区域水资源管理的基础,同时也是区域防洪减灾中所需的重要信息。采用地质统计学方法克里金方法,建立了区域降水插值模型。并以黄河故县以上集水区域为例,进行多年平均月降水插值实验。由于研究区域为地形起伏的山区,因此降水空间分析中又建立了融合地形信息的人工神经网络模型。此外还选取泰森多边形法、距离平方反比法、线性回归法,采用交叉验证法对各方法的插值精度进行对比分析。研究显示,克里金方法和融合地形信息的人工神经网络模型插值精度较高。。
The spatial distribution of precipitation is very important for integration management of water resourceand flood prevention and reduction of a specified area. In this research, Geostatistics method （ Kriging method） is adopted and two regional precipitation interpolation model is set up, which are based on ordinary Kriging method and BP artificial neural network （ANN） method respectively. A precipitation interpolation experiment is taken to test the model in Guxian drainage area. Guxian drainage area lies in down stream of Yellow river basin belong to a mountain area. Many other interpolation methods such as Thiessen polygon method, inverse square distance method, and linear regression method are also used here to compare with the two methods. It is showed that Kriging method and BP-ANN method, which not only consider the spatial relationship of raingages but also can integrate other information such as topography information （ DEM）, are more accurate than other methods. Analysis of spatial variability for precipitation indicates that the precipitation distribution in Guxian drainage area is related. But spatial variance of July, August and September is very high.
出处
《灌溉排水学报》
CSCD
北大核心
2006年第2期34-38,共5页
Journal of Irrigation and Drainage
基金
国家重点基础研究发展计划("973"计划)
关键词
山区
降水空间分布
数字高程模型
克里金方法
BP神经网络
mountain area
spatial distribution of precipitation
DEM
kriging method
BP artificial neural network
