Comparative Studies On Leaf Area Index Retrieval Of Urban Vegetation Based On LandsatTM And SPOT5 Images

Urban green space is a central part to maintain urban ecosystem function stability,it has the function of maintaining regional carbon balance, regulating the microclimate, purifing air, conservating water and protecting biodiversity. And leaf area index is the key parameter of seting up various ecosystem function index models, for this reason the research on the methods of measuring leaf area index is an important aspect of urban ecology research. With the continuous development of satellite remote sensing technology, remote sensing inversion method has been the only way to aquire the distribution of leaf area index in a large area.In this research we choose the area within the third ring road in Wuhan as the research area. We obtain the basic information through sample investigation, use the leaf area index parameter model set up by the senior to calculate the actual leaf area index, and take use of the SPOT5 and LandsatTM images in the same period to extract the DN value and various vegetation index of the corresponding points. By correlation analysis and regression analysis, we evaluat and optimize the best equations. Finally the images of leaf area index distribution in Wuhan is made out.The main research results are as follows:(1) Most of the DN values of TM and SPOT image show a sigenificant correlation with LAI. Gnerally they have a negative correlation. No matter we use TM image or SPOT image, the correlation coefficients for LAI and the red band’s DN value are the highest among all the bands. The optimal regression model of LAI and the DN value of TM image is Y=-0.065Band3-0.124Band6+23.298 (R2=0.237, Forecast accuracy= 58.557%). And the optimal regression model of SPOT image is Y= 0.031Bandi-0.043Band2+2.929 (R2=0.408, Forecast accuracy=79.426%).(2) The determination coefficients of LAI-VIS regression model are higher the LAI-DN value regression model. The the optimal regression model of LAI and TM images vegetation index is Y=9.789 NDVI-DVI 0.001+0.001, the determination coefficient R2 is 0.237, the model prediction accuracy is 70.324%. The optimal regression model of LAI and SPOT images vegetation index is Y=-0.093x3+1.281x2-4.224x+5.631, the determination coefficient R2 is 0.657 and the model prediction accuracy is 88.454%.(3) The regression model of SPOT image is better than TM image. The best regeression model is Y=-0.093RVI3+1.281RVI2-4.224RVI+5.631 (R2=0.657, Forecast accuracy is 88.454%) by using the SPOT image.(4) In this study, the leaf area index of remote sensing inversion data is usually lower than the the actual leaf area index of corresponding point on the graph. Compared with TM image the result of using the SPOT image can be closer to actual LAI.(5) The LAI inversion graph can effectively response the urban spatial distribution characteristics of leaf area index in Wuhan. The high LAI areas are characterized by patches of scattered distribution within the third ring road in Wuhan.(6) Mixed pixels of remote sensing image have a great influence on the inversion result of LAI. The error of mixed pixels is obviously higher than than that of pure pixels, and the error of TM images’s mixed pixel is higher than that of SPOT image.