MACHINE LEARNING METHODS FOR ANALYSING THREE-DIMENSIONAL SPATIAL DATA IN KAZAKHSTAN’S LAND USE PLANNING

Abstract

Contemporary machine learning (ML) techniques offer robust instruments for the processing and analysis of extensive spatial and climatic datasets, essential for sustainable land-use planning. This article examines the capabilities of machine learning and neural network methodologies for the analysis of three-dimensional geospatial data in Kazakhstan, specifically on urban development in Alatau City, one of the city's rapidly expanding regions. We examine the use of open geospatial information, such as Copernicus satellite imagery, ERA5 climate reanalysis, and QGIS spatial databases, to produce high-resolution 3D models of urban areas. The research delineates the use of neural networks, including multilayer perceptrons (MLP) and convolutional neural networks (CNN), for land use classification, urban growth prediction, and evaluation of land suitability for residential and infrastructure development. Additionally, we emphasize the significance of machine learning in amalgamating terrain, vegetation, and climate data to facilitate decision-making in land use planning. The analysis indicates that ML-based approaches can significantly enhance the efficiency, adaptability, and sustainability of urban development initiatives in Kazakhstan, facilitating the shift towards data-driven territorial management.