Estimation Study Of Aboveground Biomass In Daxing’anling Forest In Inner Mongolia Based On Sentinel 2 Data

In the context of global climate change,accurate estimation of forest biomass to assess the carbon budget of terrestrial biota is essential.Based on Sentinel-2 remote sensing images and 2018 continuous inventory data of forest resources,this thesis constructs an above-ground biomass(AGB)estimation model based on Sentinel-2 remote sensing images and 2018 continuous inventory data of forest resources.Firstly,the sample survey data were processed,and the sample AGB was calculated according to the allokinetic growth equation of the tree species.Secondly,the Sentinel-2 remote sensing image and elevation data were preprocessed,and the required texture features,vegetation index,band reflectance and terrain factors were extracted.Finally,through correlation analysis,factors with high correlation with AGB are selected to form the optimal feature set.Then,the random forest and support vector machine are used to construct the AGB model,and the prediction accuracy of the model is compared to select the optimal model.At the same time,the influence of red-edge band,texture feature and terrain factors of DEM data on the AGB estimation model in optical remote sensing data is explored.The main conclusions are as follows:1.There are large differences in the correlation between different variable features and AGB.Among the spectral variables,the highest correlation was found for the red-edge band B6,with a correlation coefficient of 0.191,while the highest correlation was found for the vegetation index,with a correlation coefficient of 0.371.Among the texture features,the highest correlation was found for the mean texture feature obtained from the B2 band,with a correlation coefficient of 0.187.It was also found that the terrain factor had a positive correlation with AGB,with the highest importance of slope,with a correlation coefficient of 0.147.The correlation coefficient was 0.147.2.By comparing the random forest and support vector machine algorithms,the coefficients of determination of the random forest and support vector machine models are0.30 and 0.16,respectively,and the root mean square errors are 29.74 t/hm~2and 32.64t/hm~2,respectively,and the optimal model of the random forest algorithm is obtained.The AGB estimation model obtained by the random forest algorithm based on the red-edge band has a coefficient of determination of 0.33 and a root mean square error of 29.13 t/hm~2,and the AGB estimation model obtained by adding texture features has a coefficient of determination of 0.35 and a root mean square error of 28.78 t/hm~2.The AGB estimation model obtained by adding topographic factors has a coefficient of determination of 0.35and a root-mean-square error of 28.66 t/hm~2,indicating that adding red-edge bands,texture features and topographic factors will improve the estimation accuracy of the model.Using the optimal model,it can be seen that the above-ground average value of the forest in the Daxinganling is 96.31 t/hm~2.