Informativeness of the Spectral and Morphometric Characteristics of the Canopy-Gap Structure Based on Remote Sensing Data

The differences in morphometric features of the canopy-gap structure of the three dominant forest types in the Valuevsky Forest Park were investigated with high-resolution and detailed-resolution remote sensing data. The forest-community groups (deciduous forest with a predominance of lime, deciduous forest with a predominance of birch or aspen, and coniferous forest with a predominance of spruce or pine) were classified according to the random forest method based on Sentinel-2/MSI multispectral satellite images. The better classificaton accuracy was 0.96 (k = 0.88). The Sentinel-2 data were used to create a layer of segments: spectrally homogeneous forest parcels. The forest gaps were obtained via cluster analysis with Resurs-P1 Geoton panchromatic data and visual interpretation of the clusters. Eight morphometric parameters were calculated for each gap. The differences were analyzed at the segment level (the Mann-Whitney U-test) and for all gap sets of each forest-community group (the Kruskal-Wallis H-test). The highest U-test values for the average values of morphometric features at the level of forest-community segments were obtained for the gap area (U = 24), gap perimeter (U = 19.3), gap-shape complexity index (U = 19.0), and the ratio of the perimeter to the gap area (U = 18.7). The highest values of the H-test at the level of individual gaps were calculated for the fractal dimension of the gap (H = 2229.2), the ratio of the perimeter to gap area (H = 2064.9), and the gap area (H = 1718.4). Analysis of the calculated and published data made it possible to find the possible reasons for the differences in the gap structure and gap parameters of coniferous, small-leaved, and lime communities of the model territory.