Small-scale clustering of the Russian part of Arctic by periodicity type of the topographic patterns

Homogeneous topography pattern - can be an indicator of similar Earth's surfaces genesis and age. It is difficult to automatically formally describe these features and map the terrain. To describe the Earth's surface periodicity, we developing spectral terrain characteristics (STC). Their calculation consists of the following: a sliding window of different sizes decomposes the DEM into a Fourier row from which it is extracted:  1) amplitude of the main harmonic wave; 2) its length; 3) dispersion of heights given by 5% of the most important waves in relation to the general dispersion of heights; 4) general direction of oscillations of the height field; 5) unidirectionality / expression of this direction, etc. Areas with similar values of these parameters have visually homogeneous topographic pattern. We have calculated the above mentioned and some more complicated parameters for the whole territory of the Russian Arctic on a shallow scale: according to GMTED 2010 30" (1000 m per cell) on moving windows with sizes from 40 to 100 km and with the step of 10 km. Fifty-six raster models of SRC were obtained - 8 parameters at 7 scales each. Using them, a map of topographic dissection types in the Russian part of the Arctic was created with the help of self-organizing Kohonen neural networks and subsequent hierarchical clustering of individual neurons. 10 clusters have been identified related to geostructural, geological and geomorphological differences. 

This study was funded by the Russian Science Foundation, project no. 19-77-10036.