Geoinformation modeling of permafrost landscapes of North-Eastern Siberia
Résumé
The landscape-indicative approach makes it possible to determine permafrost landscapes based on the identification of two physiognomic indicators-variables of relief and vegetation, as well as stratigraphic-genetic sediment complexes. For permafrost characteristics of landscapes, multilevel combinations of environmental variables (criteria) for identifying landscapes are used. All this is the result of the analysis and processing of a huge array of field data in various landscapes of Yakutia (Fedorov, 2022). One of the main methods for obtaining these variables is geoinformation modeling as a methodological solution based on the construction and use of models of spatial objects, their relationships, and the dynamics of processes using GIS tools (Zhurkin et al., 2012). The aim of our study is to develop a geoinformation modeling technique for studying mountainous permafrost landscapes using the Orulgan ridge as the study case. The Orulgan Ridge is characterized by a meridional orographic structure and the presence of all varieties of mountain permafrost landscapes in the Arctic, which makes it a reasonable choice among the other mountain regions in terms of coverage and diversity. Cloud calculator platforms such as Google Earth Engine(GEE) are becoming the main tool for land cover-based landscape indication (DeLancey et al, 2019). The core of geoinformation modeling is the data of multitemporal multi-zone satellite images and a digital elevation model, the synthesis and processing of which make it possible to carry out landscape indication. The methodological workflow for obtaining data on the spatial structure of permafrost landscapes consists of the sequential compilation and synthesis of vegetation cover data, using the supervised phenology-based classification of the time series of Sentinel 2 MSI and Landsat 8 OLI data. Landform classification is performed on the basis of GIS-based terrain analysis by the ASTER GDEM scenes. The characteristics of the relief and genetic deposits were compiled in the analysis of the relief by morphometric variables - slope and topographic position. Integrated machine learning algorithms successfully handle vegetation mapping, and the interface is user-friendly. When classifying vegetation, a flexible approach is used to map geobotanical units, since their spatial distribution in mountainous areas is very different in river valleys and on slopes. The resulting environmental variables made it possible to categorize 8 classes of mountain permafrost landscapes. One of the significant advantages of using GEE is the versatility of the developed script for processing multi-temporal multi-zone Earth remote sensing data. This advantage allows it to be used in other regional objects, which can contribute to the mechanism for introducing permafrost-landscape research to improve the system of territories for traditional nature management, design of road infrastructure, settlements, etc.
Origine | Accord explicite pour ce dépôt |
---|---|
Licence |