Health of Vegetation in the Area of Mass Outbreaks of Siberian Moth Based on Satellite Data

• Published in: Contemporary problems of ecology 2019-12, Vol.12 (7), p.743-752
• Main Authors: Knyazeva, S. V., Koroleva, N. V., Eidlina, S. P., Sochilova, E. N.
• Format: Article
• Language: English
• Subjects: Algorithms; Barren lands; Biomedical and Life Sciences; Classification; Coefficient of variation; Coniferous forests; Damage; Dead wood; Defoliation; Ecology; Fires; Grasses; Image classification; Infestation; Landsat; Landsat satellites; Life Sciences; Mathematical analysis; Outbreaks; Pest outbreaks; Pests; Reforestation; Regeneration; Regrowth; Remote sensing; Satellite imagery; Spatial discrimination; Spatial resolution; Territory; Undergrowth; Vegetation; Vegetation cover
• ISSN: 1995-4255; 1995-4263
• DOI: 10.1134/S1995425519070114

This paper presents the results of analyzing the state and dynamics of damaged vegetation from satellite images of high and ultrahigh spatial resolution. This study was conducted based on the example of the site of coniferous forests in the Lower Angara region (Krasnoyarsk krai), where a large outbreak of the Siberian moth took place in 1944–1995. The remote assessment of the state of dark coniferous forests revealed the trends of the SWVI (or NDMI) and NDVI indices that characterize long-term changes in the vegetation cover over the period 2000–2018. The SWVI index is the most informative indicator: a sharp decrease in average values and increase in the coefficient of variation of the index are noted for dead and severely damaged wood stands (crown defoliation of more than 75%). The area of dead forests was calculated according to the difference images of the indices (ΔSWVI) with the threshold criterion lσ (the standard deviation). In 2000, the area of forests that died under the impact of the Siberian moth was approximately 19 200 ha. Alter two major fires in 2004 and 2011 and as a result of destructive factors combination, the area of dead forests increased up to 20 400 ha by 2017–2018. Reforestation within the boundaries of dead stands was estimated from the classification of Landsat images (June 20, 2017; June 23, 2018) by the Random Forest algorithm using the selection of templates from detailed Resurs-P images with a spatial resolution of 1 m (Geoton-LI—July 22, 2015 and December 3, 2018), which were taken during different seasons. The classification proved to be highly reliable (Kappa index is more than 0.9). The areas of classified deciduous and coniferous stands, deciduous and mixed stands with mainly coniferous regrowth, grass–shrub vegetation, and barren soil were calculated. Natural regeneration of mainly coniferous undergrowth occurred in 17% of the damaged area, and deciduous regrowth occurred in approximately 10% of the area 23 years after damage by pests. The area damaged by the moth affected reforestation: the larger the area of the outbreak, the higher the share of open lands with grass and shrub vegetation (it accounts for more than half of the area for the large outbreak and approximately 45% for smaller outbreaks). Regrowth was found in proximity of patches of stands and deadwood in the sites unaffected by large fires. Frequent fires in the territory of moth infestation limit the process of reforestation; therefore, most of the vegeta