Inversion Of Masson Pine Forest LAI By Muti-Angle Remote Sensing On The Western Part Of Fujian

Remote sensing inversion technology,which based on the establishment of functional relation between the sensor parameters and the geographical parameters,can be used to estimate the values of geographical parameters or biological parameters in large scale.However,compared with the traditional satellite nadir point image which is not enough to infer the special structure of the target site,multi-angle remote sensing has an increase of angle information that can get the BRDF of vegetation canopy.Therefore,multi-angle remote sensing will probably have a potential opportunity to improve the precision of inversion of vertical structure parameters.The research site is located in Hetian,Changting County,the west of Fujian Province.The study is using multi-angle remote sensing information with hyperspectral data CHRIS,from European Space Agency,to inverse leaf area index of masson pine forest in this paper.In the hyperspectral remote sensing,it is necessary to extract the feature bands useful with the construction of vegetation index Thus three feature bands,which can represent green,red and infrared spectrums,were selected from the 18 bands of CHRIS high spectrum based on principal component analysis and optimum index factor method.Using these feature bands to build RVI,NDVI,SAVI and NHDVI regression model by fitting experimental LAI data,this study aims to get the maps of LAI and make precision evaluation on inversion results.Main conclusions include the following:1.After the pre-processing treatment for the CHRIS,the study using PCA method to transform the original spectral space into the PCA space.Results show that the first two main components accounted for 99.91%of all the main components of the band eigenvalues,and indicate that the three bands in the original spectral space with the highest contribution to the first two main components are the 18th band,the 5th band and the 3th band in 0°,+36°,36°and+55° images.(the results of-55° image in infrared band have slightly different for 17th band with the other images);Extraction three feature bands from eighteen bands of the original spectral space by calculate its optimal index factor,it can come to a conclusion that the results of extracting OIF is consistent with the results of PCA in the nadir point image.While the highest OIF value of the three bands is the combination of the 18th band,the 4th band and the 3th band in other four images.Above all,the results of two methods were slightly different.2.Analyzing the Fitting process is conducted on LAI and f ratio vegetation index,normalized difference vegetation index,and soil adjusted vegetation index in feature bands of the CHRIS data.It turned out that,the maximum correlation of LAI with RVI,NDVI,SAVI happened in nadir point image.Their R2 are 0.6672,0.7341,and0.7341 successively.The goodness of fit gradually decrease while the observation angle deviate from the nadir direction,probably due to diverse vegetation coverage in different viewing directions and the effect of soil background brightness.Minimum correlation between LAI and VI present to the+55°image,their R2 are 0.1735,0.2097 and 0.2098 respectively.We extract its 4th band as infrared band,and 3th band as a red band,in order to construct NDVI based on French high resolution image Preiades.Then make regression analysis between the NDVI and the ground measured LAI.Compared with CHRIS nadir image,it can be found that the spatial resolution parameters of image have significant influence on the NDVI-LAI goodness of fit.Preiades'NDVI-LAI correlation coefficient square is 0.801,higher than that of CHRIS.3.Appling regression model to get five VI-LAI distribution maps of LAI.Then take 12 measured LAI datas to make precision assessment on inversion results.The assessment shows that SAVI-LAI's R2 is 0.7238,and RMSE is 0.2151,probably due to that SAVI import a vegetation coverage factor which can reduce the noise from soil background.Studies indicate that SAVI-LAI's inversion precision is superior to the others.4.Using four scale geometrical optics model to simulate anisotropy distribute regulation of the NDVI of forest canopy.In this study,we construct a muti-angular vegetation index NHDVI(normalized thermal scotoma vegetation index),which combined spectrum information with angular information.Results show that the NHDVI's goodness of fit is significantly higher than SAVI,valued in 0.8272.And the RMSE of NHDVI inversion results is the lowest among four indexs,valued in 0.12.To sum up,angular information is important to improving the retrieval accuracy to LAI.