基于多源遥感数据的大豆叶面积指数估测精度对比

发布时间：2018-05-18 15:57

本文选题：多源遥感数据 + 无人机　；参考：《应用生态学报》2016年01期

【摘要】：近年来遥感技术的革新促使遥感源越来越丰富.为分析多源遥感数据的叶面积指数(LAI)估测精度,本文以大豆为研究对象,利用比值植被指数(RVI)、归一化植被指数(NDVI)、土壤调整植被指数(SAVI)、差值植被指数(DVI)、三角植被指数(TVI)5种植被指数,结合地面实测LAI构建经验回归模型,比较3类遥感数据(地面高光谱数据、无人机多光谱影像以及高分一号WFV影像)对大豆LAI的估测能力,并从传感器几何位置和光谱响应特性以及像元空间分辨率三方面分析讨论了3类遥感数据的LAI反演差异.结果表明:地面高光谱数据模型和无人机多光谱数据模型都可以准确预测大豆LAI(在α=0.01显著水平下,R~2均0.69,RMSE均0.40);地面高光谱RVI对数模型的LAI预测能力优于无人机多光谱NDVI线性模型,但两者差异不大(E_A相差0.3%,R~2相差0.04,RMSE相差0.006);高分一号WFV数据模型对研究区内大豆LAI的预测效果不理想(R~20.30,RMSE0.70).针对星、机、地三类遥感信息源,地面高光谱数据在反演LAI方面较传统多光谱数据有优势但不突出;16 m空间分辨率的高分一号WFV影像无法满足田块尺度作物长势监测的需求;在保证获得高精度大豆LAI预测值和高工作效率的前提条件下,基于无人机遥感的农情信息获取技术不失为一种最佳试验方案.在当今可用遥感信息源越来越多的情况下,农业无人机遥感信息可成为指导田块精细尺度作物管理的重要依据,为精准农业研究提供更科学准确的信息.
[Abstract]:In recent years, the innovation of remote sensing technology makes remote sensing sources more and more abundant. In order to analyze the estimation accuracy of leaf area index (Lai) of multi-source remote sensing data, soybean was studied in this paper. Using ratio vegetation index (RVI), normalized vegetation index (NDV), soil adjusted vegetation index (LAI), difference vegetation index (DVI), triangular vegetation index (TVB) 5, and LAI measured on the ground, an empirical regression model was established. The ability of estimating soybean LAI from three kinds of remote sensing data (ground hyperspectral data, UAV multispectral image and high-fractionated 1 WFV image) was compared. The LAI inversion differences of three kinds of remote sensing data are analyzed and discussed from three aspects: geometric position, spectral response characteristics and pixel spatial resolution. The results showed that both the ground hyperspectral data model and the UAV multispectral data model could accurately predict soybean Lai (0. 40 渭 g / L, 0. 69% RMSE), and the LAI prediction ability of ground hyperspectral RVI logarithm model was better than that of UAV multispectral NDVI linear model. However, there was no significant difference between the two groups. The difference between the two groups was 0.04% and 0.04%, and the prediction effect of the WFV data model No. 1 on soybean LAI in the study area was not satisfactory (0.70%). For satellite, computer and ground remote sensing information sources, the ground hyperspectral data is superior to the traditional multi-spectral data in retrieving LAI data, but it can not meet the needs of crop growth monitoring in field scale. Under the precondition of obtaining high precision soybean LAI prediction value and high working efficiency, the technology of agricultural information acquisition based on UAV remote sensing is the best test scheme. With more and more remote sensing information sources available, the remote sensing information of agricultural UAV can become an important basis for guiding crop management in fine scale, and provide more scientific and accurate information for precision agriculture research.
【作者单位】：河南理工大学测绘与国土信息工程学院;国家农业信息化工程技术研究中心;农业部农业信息技术重点实验室;
【基金】：国家自然科学基金项目(41271345) 北京市自然科学基金项目(4141001)资助~~
【分类号】：S127;S565.1

【参考文献】