摘要
光学遥感影像可以快速提取大面积玉米冠层信息,但无法提供冠层垂直结构信息,导致反演玉米叶面积指数(LAI)时存在无法表达植被冠层内部叶片贡献而使反演LAI偏低的问题;地基激光雷达能够获取玉米冠层的高精度三维结构信息,但是每次只能在有限样区内获取。结合这两种技术的优势,利用将激光雷达数据体素化的方式,通过冠层分析法提取高精度的冠层结构信息;利用Landsat8光学影像获得大面积玉米冠层反射率,与得到的冠层结构信息进行回归分析,从而反演得到大面积的玉米冠层精确LAI结果。研究结果表明,归一化植被指数(NDVI)与激光点云计算的LAI相关性最强,相关系数R2=0.8086,均方根误差(RMSE)为0.1230,比值值被指数(RVI)相关性最差,R2=0.7079,RMSE为0.1520,通过实测值验证分析,三种模型的平均相对误差均小于10%,模型的可信度较高。
Optical spectral remote sensing images can be used to extract corn canopy structure information rapidly in a large area. However, it cannot provide vertical canopy structure information, which leads to underestimated leaf area index （LAI） result. Terrestrial laser scanning can provide high precision 3D structure information of corn canopy, but only in the limited sampling area. Therefore, these two technologies are combined to extract high precision canopy structure through canopy analysis by using terrestrial laser scanning data voxelization method. Reflectance of large area of corn canopy using Landsat8 optical images is obtained, and accurate corn canopy LAI results are got through regression analysis of canopy structure information of voxel-based canopy profiling. The results show that LAI has the strongest correlation with the normalized difference vegetation index （NDVI）, the correlation coefficient R2=0.8086, the root mean square error （RMSE） is 0.1230, and the correlation between LAI and ratio vegetation index （RVI） is the worst, R2=0.7079, RMSE is 0.1520. Based on the validation analysis of the measured values, the average relative error of the three models is lower than 10%, and the credibility of the three models is relatively high.
出处
《中国激光》
EI
CAS
CSCD
北大核心
2015年第11期238-244,共7页
Chinese Journal of Lasers
基金
国家自然科学基金(41371327)
