Forecasting changes in landcover types using statistical and neural network models
Abstract
The study presents the results of a comprehensive analysis and forecasting of land cover type transitions along with their interpretation in the context of land degradation challenges in the Nukus district of Karakalpakstan. The research is based on the integration of Earth remote sensing data, climatic and socio-economic information, as well as statistical and neural network modeling approaches.At the initial stage, an analysis of the dynamics of pasturelands, croplands, built-up areas and desert territories over the period 2001-2022 was conducted using data from the Trends.Earth module. The results indicate that the key feature of land cover transformation in the study area during this period is a decline in agricultural lands accompanied by an expansion of artificial territories.For short-term forecasting (up to 2028), an ARIMA model was applied, enabling the incorporation of time series autocorrelation and structural shifts in indicator dynamics. The obtained results indicate a further intensification of urbanization pressure on agricultural lands; a continued decline in cropland and pasture areas is observed alongside persistently high growth rates of artificial territories.Long-term forecasting of land cover changes was performed using spatial modeling based on Cellular Automata and Artificial Neural Networks (CA-ANN), implemented in the MOLUSCE module. The model incorporated the influence of a combination of natural and anthropogenic factors, demonstrating high accuracy (Cohen’s kappa up to 0.95).A comparative analysis of the results from statistical and spatial modeling demonstrated the consistency of the projected trends, including a reduction in cropland and pasture areas, an expansion of artificial territories and the relative stability of desert landscapes.Received: 06/05/2026
Accepted: 07/20/2026
Accepted date: 11/20/2026
Keywords: ARIMA; CA-ANN; MOLUSCE; sustainable land use; Karakalpakstan; remote sensing; land degradation neutrality
DOI: 10.55959/MSU0137-0944-17-2026-81-4-121-131
Available in the on-line version with: 18.11.2026
-
To cite this article:

This work is licensed under a Сreative Commons Atribiution - NonCommercial 4.0 International (CC BY-NC 4.0)

