Research On Forest Parameters Retrieval Based On Lidar Data And Multi-angle Remote Sensing Model

Optical remote sensing has made great achievements on forest resources survey.The forest classification thematic map and forest coverage map were got successfully,which makes human have a certain understanding on earth forest two-dimensional information.However,because the forest was seen as a three-dimensional objective,the multi-dimensions and multi-source remote sensing data need to be used to accurately obtain the three-dimensional structure information of forest.Currently,remote sensing has achieved a great of development in spatial resolution,time resolution and spectral resolution,and a certain number of sensors have the ability of multi-angle observation.At the same time,as a new type of remote sensing detection sensors,the Li DAR has also been widely attention and used.But there are many works to do in understanding of basic theoretical system,perfect and carry out the applications.In order to make full use of the Li DAR and multi-angle remote sensing technology in processing of forest structural parameters inversion,the physical law of interaction between electromagnetic wave and vegetation composition need to been understand,and to build mathematical physics model for quantitative description,ultimately the forest structural parameters were extracted from the multi-source remote sensing data accurately.In this paper,the method of bidirectional reflectance distribution function simulation and laser radar waveform simulation were researched based on the physical theory of remote sensing.The relationship between different forest structure and two kinds of remote sensing signals was found to build multi-angle vegetation index and to optimize the Li DAR matrix for improving the accuracy of forest parameters inversion,which improve the utilization efficiency of own satellite data and the ability of automatic monitoring of forest resources in China.The research that how to retrieve forest parameters using multi-angle and Li DAR data based on physical model will be discussed.Firstly,according to the laws of the electromagnetic wave propa gation in the atmospheric homogeneous medium,the geometric structure s of forest vegetation were parameterized.The interaction law between electromagnetic and forest composition was found to understand how to build the extinction coefficient and scattering coefficient of blade—foliage—canopy.The Monte Carlo simulation was used to calculate the forest coverage and leaf area volume density function in case of forest canopy overlapping,and then the multi-angle model was finished.The relationship between laser waveform and forest canopy structure was found using the canopy gap to build the Li DAR model of forest canopy,and then the laser waveform was simulated according to different forest scene.Secondly,the application scope of each model was researched based on the theory of multi-angle remote sensing model.The advantages and disadvantages of radiative transfer model and geometric optics model were summarized,and the airborne high spatial resolution hyperspectral data was used to test that the radiative transfer model was suitable for leaf area index retrieval.The look up table of forest biomass was built using the multi-angle model and forest growth model,and was used to found the relationship between band,angle and forest biomass.The angles around the hot–spot and dark-spot were sensitive to forest biomass,and then the sensitive angles and bands were found to build the multi-angle vegetation index for forest biomass retrieval using the CHRIS data and ground measured biomass.Finally,the laser waveform was corrected according the law of vegetation Li DAR function and then was used to get the real forest structure information.The biomass index that suitable for sample scale biomass retrieval was built based on the formula of signal tree biomass.The biomass index can make full use of the forest stand density,tree height and tree species from the corrected laser waveform to improve the retrieval precision of forest biomass,which was test using the LVIS and ground measured biomass.The forest biomass of Howland study area was retrieved combining Li DAR and CHRIS multi-angle data,the result shows that model was suitable for the regional forest biomass inversion,which was composed of a number of training sample from Li DAR,the sensitive angle and band from optical multi-angle data.