基于无人机高光谱数据的多类型混合作物LAI反演及尺度效应分析
发布时间：2018-10-26 20:33
【摘要】：叶面积指数(Leaf Area Index,LAI)作为表征不同作物生长状况的基本参数,是农业精细化管理及农田生态系统建模的关键。我国农田作物种植比较离散,受地表空间结构非均一性和反演模型非线性等因素影响,不同尺度遥感数据估算的作物LAI存在一定的差异,即农田作物LAI的遥感反演普遍存在尺度效应问题。以包头遥感综合验证场农业示范区为研究区,利用无人机高光谱数据结合PROSPECT+SAIL模型构建典型农作物区多类型作物的查找表(Look-Up-Table,LUT)反演农田LAI,研究查找表用于玉米、马铃薯、向日葵、瓜地等不同作物LAI反演的适用性和精度;通过无人机高光谱数据聚合获得多尺度遥感数据源,结合Taylor展开理论和计算几何模型,提出了一种既考虑类间差异又考虑类内异质性的尺度转换模型,定量描述多种作物混合的非均一地表LAI反演过程中的尺度效应特征。结果表明:基于分类和参数敏感性分析的LUT方法能很好地应用于包头典型农作物区多类型混合作物LAI反演,总估算精度为相关系数R~2=0.82、均方根误差RMSE=0.43m~2/m~2。随着反演尺度的增加,作物类间差异造成的反演偏差明显高于类内异质性,利用本文所提出的尺度转换模型均能较好纠正低分辨率LAI反演的尺度效应问题。
[Abstract]:Leaf area index (Leaf Area Index,LAI), as the basic parameter of different crop growth status, is the key to agricultural fine management and farmland ecosystem modeling. Crop cultivation in China is relatively discrete and affected by the heterogeneity of surface spatial structure and nonlinear inversion model. There are some differences in crop LAI estimation based on remote sensing data of different scales in China. In other words, the scale effect problem exists in the remote sensing inversion of crop LAI. Taking the agricultural demonstration area of Baotou remote sensing comprehensive verification farm as the research area, using UAV hyperspectral data combined with PROSPECT SAIL model to construct the lookup table (Look-Up-Table,LUT) of multi-type crops in typical crop area to invert farmland LAI,. The applicability and accuracy of LAI inversion of corn, potato, sunflower, sunflower and melon field were studied. Based on the Taylor expansion theory and computational geometry model, a scale conversion model considering both intra-class heterogeneity and inter-class difference is proposed by aggregation of hyperspectral data from UAV to multi-scale remote sensing data source. The scale effect characteristics of heterogeneous surface LAI inversion with multiple crop mixtures are quantitatively described. The results show that the LUT method based on classification and parameter sensitivity analysis can be well applied to the LAI inversion of mixed crops in Baotou typical crop area. The total accuracy of the estimation is the correlation coefficient RG20.82, and the root mean square error (RMSE=0.43m~2/m~2.). With the increase of inversion scale, the inversion deviation caused by crop difference is obviously higher than that of intra-class heterogeneity. Using the scale conversion model proposed in this paper, the scale effect of low-resolution LAI inversion can be well corrected.
【作者单位】： 中国科学院定量遥感信息技术重点实验室中国科学院光电研究院;中国科学院大学;
【基金】：国家863计划项目“遥感载荷性能与数据质量检测技术”(2013AA122102)
【分类号】：S127;TP79
本文编号：2296870
[Abstract]:Leaf area index (Leaf Area Index,LAI), as the basic parameter of different crop growth status, is the key to agricultural fine management and farmland ecosystem modeling. Crop cultivation in China is relatively discrete and affected by the heterogeneity of surface spatial structure and nonlinear inversion model. There are some differences in crop LAI estimation based on remote sensing data of different scales in China. In other words, the scale effect problem exists in the remote sensing inversion of crop LAI. Taking the agricultural demonstration area of Baotou remote sensing comprehensive verification farm as the research area, using UAV hyperspectral data combined with PROSPECT SAIL model to construct the lookup table (Look-Up-Table,LUT) of multi-type crops in typical crop area to invert farmland LAI,. The applicability and accuracy of LAI inversion of corn, potato, sunflower, sunflower and melon field were studied. Based on the Taylor expansion theory and computational geometry model, a scale conversion model considering both intra-class heterogeneity and inter-class difference is proposed by aggregation of hyperspectral data from UAV to multi-scale remote sensing data source. The scale effect characteristics of heterogeneous surface LAI inversion with multiple crop mixtures are quantitatively described. The results show that the LUT method based on classification and parameter sensitivity analysis can be well applied to the LAI inversion of mixed crops in Baotou typical crop area. The total accuracy of the estimation is the correlation coefficient RG20.82, and the root mean square error (RMSE=0.43m~2/m~2.). With the increase of inversion scale, the inversion deviation caused by crop difference is obviously higher than that of intra-class heterogeneity. Using the scale conversion model proposed in this paper, the scale effect of low-resolution LAI inversion can be well corrected.
【作者单位】： 中国科学院定量遥感信息技术重点实验室中国科学院光电研究院;中国科学院大学;
【基金】：国家863计划项目“遥感载荷性能与数据质量检测技术”(2013AA122102)
【分类号】：S127;TP79
【相似文献】
相关期刊论文 前4条
1 汪小钦;叶炜;江洪;;基于光谱归一化的阔叶林LAI遥感估算模型适用性分析[J];福州大学学报(自然科学版);2011年05期
2 彭哠,张树文;基于NDVI与LAI的水稻生长状况研究[J];东北测绘;2002年04期
3 崔要奎;赵开广;范闻捷;徐希孺;;机载Lidar数据的农作物覆盖度及LAI反演[J];遥感学报;2011年06期
4 ;[J];;年期
相关硕士学位论文 前1条
1 邵田田;玉米叶绿素和LAI的高光谱遥感模型研究[D];中国科学院研究生院(东北地理与农业生态研究所);2012年
，本文编号：2296870
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2296870.html
