Recognition of species composition and age classes of forest stands using spectral and textural features using high resolution satellite images

Abstract

The task of using multispectral satellite images of high spatial resolution WorldView-2 to determine the species and age composition of tree stands is considered. The solution of this task has great importance for the implementation of the program of the State Forest Inventory on the territory of the Russian Federation and for solution of a number of issues of strategic planning of forest management. The site of the Valuisky forest area of the Belgorod region is considered as a test area. The article shows the features of the joint use of spectral and textural features during the classification to solve the investigated problem. For thematic processing of high-resolution satellite images, a modified decoding method is used. A distinctive feature of the proposed algorithm is the ability to use more informative features while maintaining stability. Ground survey data are used to build training samples and validate results. The presented analysis of the results of thematic processing allows us to conclude that the proposed method can be used to determine the species composition and age classes of tree stands from satellite images. The obtained results can help automate the work on updating forest management data.The task of using multispectral satellite images of high spatial resolution WorldView-2 to determine the species and age composition of tree stands is considered. The solution of this task has great importance for the implementation of the program of the State Forest Inventory on the territory of the Russian Federation and for solution of a number of issues of strategic planning of forest management. The site of the Valuisky forest area of the Belgorod region is considered as a test area. The article shows the features of the joint use of spectral and textural features during the classification to solve the investigated problem. For thematic processing of high-resolution satellite images, a modified decoding method is used. A distinctive feature of the proposed algorithm is the ability to use more informative features while maintaining stability. Ground survey data are used to build training samples and validate results. The presented analysis of the results of thematic processing allows us to conclude th...