| Forest biomass,as a key indicator for weighing forest structure and changes,can be used to analyze various aspects of the overall performance of forest ecosystems.Inversion studies of forest biomass play an important role in forest carbon cycling capacity,resource management and its utilization.The combination of optical and radar data,both of which are rich in feature information,can complement each other’s strengths and weaknesses to obtain a more complete spatial structure and information of the forest and further improve the fitting effect and inversion accuracy of the AGB estimation model of Liupan Mountain forest.In this paper,the forest area of Liupan Mountain is selected as the research area of the paper,and the optical data is selected as Landsat 8 OLI and the LIDAR data is selected as ALOS-2 PALSAR-2 with Lband,and the forest inventory data is combined with the forest inventory data as the data source for the study of the inverse model of forest AGB in Liupan Mountain,and the relevant feature variables of the two selected data sources are extracted,analyzed,and screened to construct the inverse model of forest AGB based on classical and improved algorithms.The inverse model of forest AGB in Liupan Mountain is constructed based on classical and improved algorithms.The main research contents of the paper are as follows:1.Extract the corresponding characteristic parameters from Landsat 8 OLI optical remote sensing image and ALOS-2 PALSAR-2 lidar image,such as texture feature,vegetation index,greenness component,etc,and use the importance analysis of characteristic variables based on random forest after preliminarily using SPSS software to analyze and select the features with high correlation with Forest Aboveground Biomass,Further select the characteristics with high correlation with Forest Aboveground Biomass for the implementation of subsequent biomass inversion model.2.After comparing the advantages and disadvantages of commonly used inversion algorithms,three classical algorithms of support vector regression(S VR),random forest(RF)and BP neural network were selected to establish the forest AGB inversion model in the study area of Liupan Mountains,and through the comparison of the three inversion models,a new forest above-ground biomass inversion model was constructed by optimizing the BP neural network model with the atomic optimization algorithm(ASO)-By comparing and evaluating the accuracy of four biomass inversion models,the ASO-BP inversion model with the highest accuracy was selected to be more applicable to the above-ground biomass inversion of the Liupan Mountain forest,and the above-ground biomass estimation and analysis of the Liupan Mountain forest was completed.The ASO-BP inversion model was selected as the most accurate model for the above-ground biomass inversion of the Liupan Mountain forest.3.The most suitable and optimal forest AGB inversion model obtained by comparison was used to estimate the biomass of Liupan Mountain,and the spatial distribution map of biomass in Liupan Mountain was obtained.The spatial distribution characteristics of forest AGB in Liupan Mountain were analyzed and summarized in terms of global autocorrelation,local autocorrelation and trend surface for the biomass estimation results in the study area,and the ecological impact,development and importance of Liupan Mountain on the counties where it is located,so that the relevant departments can effectively manage the forest resources.By analyzing the effects of different algorithmic estimation models,different sets of image variables and their characteristic variables on the estimation of forest biomass,we obtained an above-ground biomass estimation model that is more applicable to the forest area of Liupan Mountain,which provides an important scientific research basis and support for the ecological environment development planning of the counties and districts,forest resource management and allocation of the relevant forestry departments,etc. |
PDF Full Text Request