Estimates of changes of structural parameters of forest ecosystems in decoding high resolution satellite images

• Published in: Nature Conservation Research: Заповедная наука Vol. 1; no. 1; pp. 98 - 107
• Main Authors: Rozhkov, Yuri F., Kondakova, Maria Yu
• Format: Journal Article
• Language: English; Russian
• Published: Fund for Support and Development of Protected Areas "Bear Land" 01.01.2016
• Subjects: classification ISODATA; interpretation of satellite images; the symmetry of pixel distribution
• ISSN: 2500-008X; 2500-008X
• DOI: 10.24189/ncr.2016.008

Aim of this study was to assess the possibility of using of the parameter of symmetry of pixel distribution in the forest condition monitoring. Multispectral satellite imagery and their fragments of high and medium resolution (Landsat TM/ЕТМ+, Aster, Spot, IRS), which have been made in 1995–2011, were processed in two stages. At the first stage, uncontrolled classification has been carried out using the method ISODATA (Iterative Self-Organizing Data Analysis Technigue). At the second stage, parameter of symmetry of pixel distribution was calculated. The results of classification were divided into two halves. Classes with lower optical density of the reflected light were concentrated in the upper half while classes with higher optical density of the reflected light were concentrated in the bottom half. The prospects for using the parameter symmetry of pixel distribution aim to assess the degree of forest disturbance after the fire impact was demonstrated. Disturbed forest areas a have larger sum of pixels in the bottom half of the classification results compared with the upper half. In contrast, undisturbed forest areas have a larger or equal sum of pixels in the upper half of the classification results compared with the bottom half. The prospects for using the parameter symmetry of pixel distribution in monitoring of seasonal changes of forest status were demonstrated. Comparison of two forest fragments with dominance of larch and Siberian pine showed that during the autumn months (September, October) after needle fall and leaf fall, there is a sharp decrease of the parameter "symmetry of pixel distribution" within the fragment with dominance of larch due to the increase of the proportion of pixels with high optical density. Seasonal changes in the parameter of symmetry of pixels distribution were less pronounced for the forest fragment with dominance of Siberian pine. We considered the prospect to use the parameter of symmetry of pixel distribution in the long-term monitoring of forest ecosystems status ( for example of the forest restoration process after fire impact). The rate of forest recovery was determined within the burned fragment with an area of 6.98 sq. km