Geoinformation analysis with the construction of a neural network model for predicting location of archaeological monuments in the landscapes of South-Eastern Altai.

Abstract

This article identifies landscape patterns of the location of archaeological sites in the South- East Altai using GIS and machine learning. The information database of archaeological sites of South-Eastern Altai was created based on literary sources, the authors’ own field research. Schemes of distribution of archaeological sites according to some landscape features: absolute height; position in relation to watercourses; slopes; exposition; solar radiation intensity in June; Intensities of solar radiation for December was created on the basis of geoinformation analysis of the relief of the territory and the available archaeological data. The obtained statistical regularities of the distribution of archaeological sites in the landscapes of South-Eastern Altai were the basis for the creation and verification of the algorithm of a machine-learning model — a neural network. Based on the results, a forecast map of the location of archaeological sites was created. The greatest probability of discovering new archaeological objects could be provided by the following landscape parameters, no further than 500-600 m from the river with a slope steepness of up to 4 degrees, with high intensity and summer (June) solar radiation and with an exposure of slopes: southern, south-eastern and western. Unexplored archaeological sites in South-East Altai are most likely located along river valleys in the middle and lower reaches, at the confluence of rivers, along the periphery of intermountain basins, or on wide flat areas of above-floodplain terraces of river valleys. The data obtained make it possible to assess the contribution of landscape features to the spatial distribution of religious buildings of ancient peoples and provide opportunities for the search for new archaeological sites.