Study On Leaf Area Index Based On Multi-source Remote Sensing Data

Vegetation is an important part of the earth ecological system, and has a great significance to human existence, climate change and social development. As one of the most basic parameter of vegetation canopy, LAI(Leaf Area Index) not only directly affects the photosynthesis, transpiration and population vegetation growth, but also it is a key parameter in many models. The dynamic changes of LAI are also closely linked with the regional climate characteristics. Remote sensing technology is widely used to provide powerful tools for the LAI measurement and application.In this paper, taking winter wheat ground canopy spectral in Beijing suburb and combing with the PROSAIL radiative transfer model, we made a systematic study on the LAI estimation about vegetation index saturation, soil background effects, vegetation spectra features and best estimation methods. Different satellite images were also used to study winter wheat in regional area with proposed methods. The main research contents and results are as follows:(1)The present commonly used vegetation index is also affected by soil background influence and saturation problems. According to the radiative transfer model and the spectral data, LAI=3was confirmed as the more appropriate segment point. The best vegetation index segment combination OSAVI (LAI≤3)+TGDVI (LAI>3) partly overcame soil factors and the saturation problem. The joint inversion results were significantly superior to the single vegetation index retrieval accuracy.(2)Analysing the moisture indices importance for LAI estimation, visible-near infrared vegetation indices were multiplied with moisture indices to construct new vegetation indices. The PROSAIL model simulation and measured data demonstrated that this method was feasible. Particularly, multiplying sLAIDI*to construct new vegetation indces, estimation results were improved significantly and the equation models were also more stable.(3)By analyzing the characteristics of winter wheat canopy reflectance spectra, the best optimal variables were selected to construct three data sets from the spectral characteristic bands, spectral characteristic positions and vegetation indices. Stepwise regression, principal component regression and partial least squares regression were used to estimate LAI for discussing and analyzing the optimum estimation scheme in three data sets. Partial least squares regression estimated effect and stability were the best method, and decision coefficient(R2) was over0.8. (4)Proposed the segmentation method and new vegetation indices also have better results in Landsat5TM image for LAI estimation. A comparative analysis of different atmospheric correction methods was used to study FY-3A/MERSI remote sensing data. Based on the TM and MODIS data, the quality of MERSI data was analyzed and finally estimated to LAI. Although MERSI data quality and the retrieval accuracy is not well, but the data distribution and LAI inversion results overall trend were consistent with the results of TM and MODIS data.