文献类型：期刊文献

中文题名：基于地形和主风向效应模拟山区降水空间分布

英文题名：Estimation of precipitation using altitude and prevailing wind direction effect index in Mountainous region

作者：孙鹏森[1] 刘世荣[1] 李崇巍[2]

第一作者：孙鹏森

机构：[1]中国林业科学研究院森林生态环境与保护研究所;[2]北京师范大学生命科学学院

年份：2004

卷号：24

期号：9

起止页码：1910-1915

中文期刊名：生态学报

外文期刊名：Acta Ecologica Sinica

收录：CSTPCD;;Scopus;北大核心:【北大核心2000】；CSCD:【CSCD2011_2012】；

基金：国家科技部重大基础研究计划资助项目 ( 2 0 0 2 CB1115 0 4) ;国家科技部杰出青年基金资助项目 ( 3 0 12 5 0 3 6) ;国家林业局森林生态环境重点实验室基金资助项目~~

语种：中文

中文关键词：岷江上游;降水量;主风向效应指数;插值;归一化交叉检验

外文关键词：upper stream of Minjiang; precipitation; prevailing wind-direction effect index; interpolation; generalised cross validation

分类号：P338.1;S715

摘要：在 ANUSPL IN和 GIS空间分析技术的支持下 ,采用岷江流域内及周边地区共计 5 1个雨量站的 1988～ 2 0 0 2年各月连续观测数据 ,模拟产生岷江上游面积达 2 2 919km2的流域范围内月平均降水量的空间分布栅格。为了体现当地季风方向和坡向之间的耦合效应 ,建立主风向效应指数 (PWEI) ,并从 DEM中提取海拔高度形成两个协变量 ,以雨量站的大地坐标位置作为独立变量。降水的模拟采用样条平滑技术 ,利用降水量值和 4个变量的统计关系拟合产生样条表面 ,并进而结合 DEM和 PWEI栅格产生空间分辨率达 5 0 0 m的降水量栅格。依据归一化交叉检验值 (GCV)确定平滑参数 ,并通过多次诊断运行实现平滑降噪 ,提高预测精度。统计结果表明 ,月平均降水量的预测误差变动在 15 %～ 4 2 %之间 ,是现有雨量站分布条件所能实现的较好的结果。雨季 (5～ 9月份 )的预测误差远小于旱季 ,表明东南季风对迎风坡面有明显的致雨效应 ,并因 PWEI的运用提高了模拟精度 ;旱季 PWEI效果不明显 ,降水分配主要依赖地形。和单纯利用海拔高度一个变量相比 ,增加 PWEI可使全年平均预测误差降低 3.0 %左右。
Spatial explicit data of precipitation contribute to water resource management as well as large-scale eco-hydrological processes modeling. In this paper, spatial interpolation model ANUSPLIN and Geographic Information System techniques were employed to simulate the spatial distribution of mean monthly precipitation in the mountainous region. Upper Stream of Minjiang (USM), where this study was carried out, is a critical area in Southwest China. USM locates between the east Tibet plateau and Sichuan basin, covers 22919km^2 watershed area. Precipitation data were acquired through 51 rain observation gauges for the period of 1988~2002. Two independent variables and two covariates were used in simulating surfaces of monthly precipitation. Independent variables x, y are position information of the gauges in a projected coordinate system of Beijing 1954. Topographic descriptor DEM and Prevailing Wind-direction Effect Index (PWEI) were extracted as covariates. The latter were established by linking the interacting effects of aspect and prevailing wind direction (i.e. wind direction of monsoon). All variables were calculated and discretized as raster surfaces with 500 m resolution. Thin plate smoothing spline surfaces for monthly precipitation were fitted on the basis the gauged precipitation points, which enable the regular gridded precipitation data set to be produced by coupling spline surfaces with DEM and PWEI. Generalised Cross Validation (GCV) is applied to determine smoothing parameter that can be used to minimize the prediction error. Mean monthly precipitation varied significantly throughout the year. The highest mean monthly precipitation 174.05mm occurred in August and the lowest 7.98mm occurred in December. Spatial distribution patterns of precipitation appeared to be high topographic relevant. The higher precipitation occurred in the southeast upslope that exposed to the prevailing moisture-bearing wind, and the lower occurred in the north plateau and in the middle dry valley. Scatterplots of observed versus predicted precipitation showed overestimate in the upper end of the range and underestimate in the lower end of the range in dry season (October to April). In contrast, wet season (May to September) appeared to be well predicted. Narrow range of the value in May and June indicate there were no remarkable annual change of precipitation at the beginning of wet season. Statistics showed the prediction errors of monthly precipitation vary between 15.02%~42.06%. The prediction errors in wet season varies between 15.02%~21.04% and dry season 17.71%~42.06%. Lower prediction errors in wet season indicated stronger effect of prevailing wind-direction of monsoon than dry season. Better performance occurred when simulating has been done with PWEI than without it. Annual mean prediction error can be reduced by 3.0 percent when applying PWEI.