Study On LAI Extraction Based On Landsat Image

Taking Yixing City of Jiangsu Province as the study area, the statistic regression models are established between LAI (Leaf Area Index) measured in field and the different forms of Vegetation Index (VI) derived from Landsat TM imagery data. Based on the R and R2 values of the models, the performance and the limitations of the models are explored, and the influences from soil and atmosphere on the models are investigated.The study has led to the following findings and conclusions:(1) The multivariate linear regression can be considered as the best model to describe the correlation between LAI and VI. The equation with R=0.930 and R2=0.864 is as follows:(2) The soil and atmosphere have obvious impacts on the relationship between LAI and VI. However, Atmospheric correction on the data of the remotely sensed imagery can effectively improve the LAI-VI correlation and SAVI can partly remove the influence of soil on the relationship. Although SARVI is designed to reduce the image distortion which is resulted from soil and atmosphere effects, the LAI-SARVI relationship performs not so ideal because of the complicated relations between soil and atmosphere.(3) Compared with the simple linear regression model the cubic regression model and multivariate linear regression model are more accurate in estimating LAI and, however, these models will lead the simple LAI-VI relations to acomplicated ones.(4) Because of the parameters used in the models are dedicated to the special environment of the study area and due to the limitation of LAI sampling in quantity, the regressing models provided in this study can not be universally applied elsewhere.