Tree Species Identification And Forest Parameters Estimation Using Very High Spatial Resolution Optical Imagery

Forest resources monitoring is the key link of forest resource management.What forestparameters inversion quality relates to the effectiveness and accuracy of forest resources monitoringis the important research field in modern forestry monitoring.High spatial resolution images have rich texture and more prominent space structure information,which can reflect the forest canopy information better.It is vital significant for the species identification and forest parameters extraction.This study based on the QuickBird and Worldview-2 high resolution image,according state-owned forest farms in Jiangle as an experimental zone,which discusses the high resolution images of classification model and the forest parametersestimation model on the small and medium scale from the scope of the independent variables such as terrain,segmentation algorithm,classification methods and model methods.First of all,using gray level co-occurrence matrix for extracting the texture feature information and combining with effective angle information and vegetation index information establish forest tree species identification model and its parameters estimation model.In addition,this paper discusses the NDVI band texture for the role of forest tree species identification and parameter inversion.The main research results content include:1.Based on Chinese fir plantation ground data,the paper analyzes the slope to the factor of Chinese fir plantation structure rule.The influence of diameter of the same forest age of Chinese fir plantation in different slope upward rule is put in bigger difference,and the diameter of the structure and laws of a single sample and overall samples that there was a significant difference between should slope to the effect analysis.2.Based on the QuickBird high-resolution image data construct a kind of based on information of the image object homogeneity and heterogeneity of random forest classification method,and use the WorldView-2 data verified the method.Studied the vegetation index information,the NDVI band implied image texture information and effective information on forest classification and the impact of forest parameters inversion can establish high precision of tree species identification classification model.Forest tree species identification model of the optimal accuracy was 80%-84%.3.The image data based on high precision forest parameter inversion model method is established.Whose correlation coefficient based on the QuickBird and WorldView-2 two high resolution image data to establish forest volume estimate multivariate regression model is 0.72.Stand DBH estimated multiple regression model correlation coefficient is 0.78.Forest tree height estimated multiple regression model correlation coefficient is 0.88.The above estimate models can response the relationship better betweenforest parameters and remote sensing factors.