No 3 (2024)

Algorithms based on convolutional neural networks are the most efficient for semantic segmentation of images, including segmentation of forest cover disturbances from satellite images. In this study, we consider the applicability of various modifications of the U-net architecture of convolutional neural network for recognizing logged, burnt and windthrow areas in forests from multi-temporal and multi-seasonal Sentinel-2 satellite images. The assessment was carried out on three test sites that differ significantly in the characteristics of forest stands and forest management. The highest accuracy (average F-measure of 0.59) was obtained from the U-net model, while the models that showed the best results during training (Attention U-Net and MobilNetv2 U-Net) did not improve segmentation of independent data. The resulting accuracy estimates are close to those previously published for forests with a substantial proportion of selective logged areas. Characteristics of logged areas and windthrows, namely their area and type are the main factor determining the accuracy of semantic segmentation. Substantial differences were also revealed between the images taken in different seasons of the year, with the maximum segmentation accuracy based on winter pairs of images. According to summertime and different-season pairs of images, the area of forest disturbances is substantially underestimated. Forest species composition has a less significant effect, although for two of the three test sites, the maximum accuracy was observed in dark coniferous forests, and the minimum in deciduous forests. There was no statistically significant effect of slope lighting factor calculated from digital elevation model on segmentation accuracy based for winter pairs of images. The accuracy of segmentation of burnt areas, which was assessed using the example of 14 large forest fires in 2021-2022, is unsatisfactory, which is probably due to the varying degrees of damage to the forest cover in the burnt areas.

The paper presents a model of microwave emission from mouth regions of Arctic rivers taking into account radiometer pixel contamination by land. Modeling of seasonal and interannual dynamics of brightness temperature of different regions of the Yenisei Bay is performed on the example of MIRAS radiometer data from the SMOS satellite. The necessity of considering the coastal zone in brightness temperature modeling in the studied regions is shown. Comparison of the model calculations with SMOS L1C data has shown a good agreement. Analysis of model calculations and satellite data has allowed us to determine the location of the fresh and salt water mixing zone in the Yenisei Bay during the ice period, provided that the model takes into account the coastal zone captured by the radiometer pixel.

In recent years, extensive human activities have had a profound impact on the estuaries and coastal areas of Vietnam, most notably in coastal erosion and accretion. This paper used the Landsat multi-temporal data for the period 1988–2022 to assess coastline change in Thanh Hoa province (North Central Vietnam). Water indices calculated from Landsat imagery data, including NDWI, ANDWI, MNDWI, AWEInsh, AWEIsh, and BandWet, are used to extract surface water areas and then vectorize and overlay to estimate shoreline variability. The Otsu thresholding method is used to classify “water surface” and “land objects” and then evaluate the accuracy using the Kappa coefficient. The obtained results show that the ANDWI index has the highest accuracy in extracting the water body of the study area, in which the value of the Kappa coefficient reaches 0.95 compared to 0.91, 0.92, 0.93, 0.92 and 0.92 at using NDWI, MNDWI, AWEInsh, AWEIsh and BandWet indicies. Boundary vectorization and vector image overlays were performed to assess shoreline variability and map shoreline dynamics. The results obtained show that in the northern part of the coastal zone of Thanh Hoa province there is active accretion (increment) of the coastline. The average accretion rate was 150 m/year, the maximum rate was 457 m/year. In contrast, on the southern coast of Thanh Hoa province, coastline erosion predominates with a maximum rate of 38 m/year and an average rate of about 10 m/year.

The results of studies of the structure and dynamics of water in the zone of a large-scale cyclonic gyre in the northern half of the Sea of Japan are presented, where satellite IR images annually in the autumn-winter period most clearly show two areas of low temperatures, separated by the influx of warm Tsushima waters from Japan. The location of these thermal structures coincides with the location of the western and northern cyclonic gyres, which are inextricably linked with deep upwelling. During the autumn-winter periods 2019-2021 it has been established that deep upwelling in the northwestern part of the Sea of Japan extends from the bottom to the surface layer, focusing along the axial line passing through the Pervenets Rise and the Bersenev and Vasilkovsky ridges in the area of 42° N. between 132°E and 135.5° E. The western cyclonic gyre, located in the western part of the large-scale cyclonic gyre in the region of the considered deep upwelling, is a large topographic eddy. In the northern part of the large-scale cyclonic gyre, deep upwelling is confined to the continental clone, and the small northern cyclonic gyre is also located there. It is assumed that in the autumn-winter period the interaction of anticyclones that form vortex belts with cyclonic gyres leads to an increase in deep circulation. The peculiarity of the variability of the speed of deep currents - an increase from October to March, is probably due to the nature of the development of vertical and transverse horizontal circulation in the system of cyclonic gyres - vortex belts as a result of the intensification of deep upwelling with increased winds from the northern directions in winter.

The desertification processes that are dynamically occurring in the north-east of the Stavropol Territory are closely interrelated with climatic conditions, which necessitates the rapid mapping of qualitative changes in the state of open sands in accordance with various scenarios. The mapping was carried out on the basis of the technique of chorodynamic (diachronic) mapping for two periods (July-August 2022 and June-July 2023), data on the monthly dynamics of open sand areas, precipitation and dust storms were used to select time slices. The mapping was carried out visually on the basis of multispectral satellite images of Landsat for each month. In 2022, 14 dust storms were registered, in 2023 none were registered, and days with critical wind speeds were accompanied by heavy precipitation, in connection with which a significant decrease in the visible areas of open sands was revealed as a result of overgrowth. Based on the available data, using overlay operations, schemes of qualitative transitions in the test area (part of the Levokumsky district) corresponding to qualitative transitions in the “dry” and “wet” scenarios were compiled. Despite the progressive overgrowth of the sands, stimulated by the absence of dust storms and abundant precipitation in 2023, starting from areas with the lowest sediment capacity, the restoration of the original plant communities without carrying out phytomeliorative measures is extremely unlikely due to high anthropogenic load and seed desertification, as a result of which low-value annual crops grow on open sands (weed solyanka, kumarchik, nightshade horned, etc.).