Remote Sensing Retrieval Method Of Forest Carbon Storage In Beijing

In recent years,the smog and climate warming in Beijing has become more and more serious,which cause people's attention.Forest can effectively alleviate the above phenomenon by absorbing greenhouse gases and PM2.5 dust.In order to understand the dynamic changes of carbon storage in China better,it is necessary to estimate the forest carbon storage in Beijing accurately and efficiently.This thesis is taken the Landsat-8 and GF-1 remote sensing images in Beijing as the original data,at the same time extracted 80 and 50 factors from them.150 samples also were selected as the field data.The carbon samples were calculated by using the nine-tree polygonal method.The two sets of imaging model factor and carbon storage were modeled by principal component analysis and multiple linear regressions respectively.The optimal model was selected in the four groups to estimate the background carbon storage.We use the remote sensing image and the field plots to estimate the forest carbon reserves in Beijing.This kind of method saves a lot of material resources and financial resources,and provides reference for the development of forest management plan and the realization of Internet+precision forestry.The main contents and conclusions of the article are as follows:(1)Model factor acquisition:Landsat-8 remote sensing images have 7 available spectral information,but the resolution is low.However,GF-1 remote sensing image have four spectral information,but the resolution is only 8 meters.In this paper,the Landsat-8 satellite remote sensing image and the GF-1 satellite remote sensing image are used to extract the spectral information.After pretreatment,comparing the two sets of images in the texture of each band,the terrain factor,the vegetation index,the single band and some bands gray scale to each other.There are 80 model factors of Landsat-8 image and GF-1 image has 50 model factors.The model factors are used to fit subsequent models.(2)Field sample biomass acquisition:We choose Nine tree polygon sample area instead of traditional one field individual measurement to calculate average diameter at breast height,average height,volume,biomass and carbon storage.Then through the random sampling of 18 plots as the detection data,we calculate the plot of the various forest factors according to traditional methods.After precision analysis,the relative risk of each forest survey factor between 8.80%?13.44%,the correlation coefficient is between 0.624 and 0.927.The laboratory team also compiled the calculation model of the nine-tree polygonal pattern into software,and the input survey method was automatically obtained by inputting the basic measurement information section,and the sampling observation was further facilitated.The use of nine tree polygonal method to investigate the forest factor can meet the needs of the second category of investigation.The method can replace the traditional method of wood to obtain the forest survey factor,which is suitable for observing the factors of the plot,ensuring that the precision can greatly reduce the amount of field labor.(3)Multiple linear regression models can correctly know the situation about the correlation degree of each factor and carbon storage model and the fitting situation of regression model through the analysis of dependent variable and independent variable.The advantages that spectral information is rich and more factor model is extracted can be taken a full use.The method of principal component analysis can reduce the multicollinearity by reducing the dimension of data,which solves the problem of multiple regression analysis to a certain extent.At the same time,we choose the two most commonly used models to fit the carbon storage with many model factors the information.The accuracy of the four sets of models is,Landsat-8 image principal component analysis:79.66%,Landsat-8 image multiple linear regression:81.72%,GF-1 image principal component analysis:75.01%,GF-1 image multiple linear regression:75.93%.The best model factor is that the Landsat-8 image fitted with multiple regression analysis,the accuracy is 81.72%,the correlation coefficient is 0.82,and finally the model factor is chosen to estimate the forest carbon reserves in Beijing.(4)Estimation of carbon storage:Using the selected optimal model to estimate the forest carbon storage in Beijing.The forest carbon storages in Beijing are 22.788Tg.According to our estimated carbon storage data,Beijing's carbon stocks were divided into four.The distribution of carbon stocks in Beijing is analyzed.It is concluded that Beijing's carbon reserves are concentrated,mainly in the north and southwest of Beijing,and the urban parks have sporadic little carbon reserves.