引入NDVI改进SRTM DEM的新方法  

A New Method of Introducing NDVI to Improve SRTM DEM

作　　者：刘焕军[2,3] 马雨阳 杨昊轩 姜芸[2] 巩超 吕航 Liu Huanjun;Ma Yuyang;Yang Haoxuan;Jiang Yun;Gong Chao;Lü Hang(School of Geography and Information Engineering,China University of Geosciences,Wuhan 430073,China;College of Public Administration and law,Northeast Agricultural University,Harbin 150030,China;Northeast Institute of Geography and Agricultural Ecology,Chinese Academy of Sciences,Changchun 130012,China;Jilin Provincial soil and Fertilizer Station,Changchun 130033,China)

机构地区：[1]中国地质大学(武汉)地理与信息工程学院,湖北武汉430074 [2]东北农业大学公共管理与法学院,黑龙江哈尔滨150030 [3]中国科学院东北地理与农业生态研究所,吉林长春130012 [4]吉林省土壤肥料总站,吉林长春130033

出　　处：《遥感技术与应用》2021年第5期1083-1091,共9页Remote Sensing Technology and Application

基　　金：国家自然科学基金项目(41671438);东北农业大学“学术骨干”项目。

摘　　要：精准/智慧农业需要田块尺度高精度地形数据,而现有耕地范围地形测绘数据不能满足需求。为获得高精度数字高程模型(DEM),以SRTM DEM为基础进行改进,获取2016年6~9月SPOT 6多光谱数据,测量实际高程,将实际高程通过克里金空间插值获得分辨率为6 m的DEM;将SRTM DEM格网的栅格值和植物生长季节的归一化植被指数(Normalized Difference Vegetation Index,NDVI)作为输入量,建立逐步多元回归和BP神经网络模型,以实测的地面高程和无人机获取的DEM数据作为验证,与空间插值方法和资源三号获取DEM进行对比。结果表明:①引入生长季节NDVI的线性回归模型的精度达到96.0%,RMSE为1.12;BP神经网络模型精度达到98.7%,RMSE降为0.86;②生长季节NDVI的时空变化是坡度、坡位等地形因子作用的结果;③利用BP神经网络改进的SRTM DEM获得更高空间分辨率的DEM精度高于资源三号生成的DEM,与无人机DEM的空间趋势相似。可为田间变量管理、精准管理分区、土壤分类与精细制图等提供支持。Precision agriculture and intelligent agriculture need high precision terrain factor data of field scale,but the existing topographic mapping data of arable land can not meet the demand.In order to establish a high spatial resolution Digital Elevation Model(DEM),obtained SPOT-6 multispectral data in June,July,August and September 2016,SRTM DEM and the actual elevation of the study area were measured.High resolution DEM(6 m)data are obtained by Kriging spatial interpolation based on the measured elevation.Taking the grid value of SRTM DEM and Normalized Difference Vegetation Index(NDVI)as inputs,the multiple linear regression and BP neural network reconstruction model are established,The verification is based on the measured ground elevation and the DEM data obtained by UAV.Compared with the UAV DEM and ZY-3 DEM.The results show that:①The accuracy of the linear regression model introduced into NDVI time series is 96%.The RMSE is 1.12;The accuracy of BP neural network model is as high as 98.7%,and RMSE is reduced to 0.86.②The temporal and spatial variation of NDVI in growing season is the result of topographic factors such as slope and slope position.③The improved SRTM DEM based on BP neural network achieves higher spatial resolution,which is similar to the spatial trend of UAV DEM,and its accuracy is higher than that of ZY-3 DEM.It can provide data support for field variable management,precise management zoning,soil classification and fine mapping.

关 键 词：SRTM DEM BP神经网络 NDVI 多元线性回归 克里金插值 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]