基于K最近邻模型的青藏高原CMORPH日降水数据的订正研究  

Correction of CMORPH Daily Precipitation Data over the Qinghai-Tibetan Plateau with K-Nearest Neighbor Model

作　　者：王玉丹[1,2] 南卓铜[3] 陈浩[4] 吴小波[1] 

机构地区：[1]中国科学院寒区旱区环境与工程研究所,甘肃兰州730000 [2]中国科学院大学,北京100049 [3]南京师范大学地理科学学院,江苏南京210023 [4]宝鸡文理学院,陕西宝鸡721013

出　　处：《遥感技术与应用》2016年第3期607-616,共10页Remote Sensing Technology and Application

基　　金：国家自然科学基金面上项目(41471059);宝鸡文理学院博士启动费项目(ZK16065)

摘　　要：青藏高原的降水数据主要由遥感产品和多源观测数据融合产生,由于青藏高原的观测站点分布稀疏不均,遥感数据误差较大,因此常用的CMORPH(Climate Prediction Center Morphing Technique)等降水数据集精度有限。通过K最近邻(K-Nearest Neighbor,简称KNN)模型,可以建立环境(海拔、坡度、坡向、植被)、气象因子(气温、湿度、风速)和日降水量的关系,从而订正青藏高原的CMORPH日降水数据集,提高数据精度。对CMORPH日降水数据的误差分析表明,采用KNN模型订正后的CMORPH降水数据优于原始数据和采用PDF(Probability Density Function Matching Method)法订正的CMORPH数据,且空间分布较好地符合青藏高原的降水分布特征。Precipitation data of the Qinghai-Tibetan Plateau（QTP）are generally fused from multiple source remote sensing products and observation data.While the meteorological observations on the QTP are scarcely and unevenly distributed,the commonly used precipitation datasets,such as CMORPH（Climate Prediction Center Morphing Technique）bear fairly large errors.In this paper the K-Nearest Neighbor（KNN）model was applied for correcting CMORPH daily precipitation over the QTP by establishing the relationship between daily precipitation and environmental,such as elevation,slope,aspect,and vegetation,and meteorological factors such as air temperature,humidity,and wind speed.The results show that the KNN-corrected CMORPH precipitation is more accurate than both the original CMORPH precipitation and the PDF-corrected results which were processed with a probability density function matching method and are available for downloading on the official Web site of Chinese Meteorological Administration.Examination of typical regions shows the KNN-corrected results well represent the characteristics of precipitation distribution over the QTP.

关 键 词：K最近邻模型 降水数据 CMORPH 青藏高原 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]