An integrated approach of machine learning methods coupled with cellular automation for monitoring and forecasting of land use and land cover

Abstract

The expeditious urban development is transforming the contemporary features of Land Use Land Cover (LULC) globally. The investigation aims to estimate the past, and potential future LULC changes in one of the semi-arid regions of Central India. This research has designed four schemes based on two machine learning algorithms: Maximum Likelihood Classifier (MLC) and Random Forest Classifier (RFC). The MLC and RFC were applied on the multi-spectral Landsat imagery to identify previous land use trends and land cover patterns between 2016 and 2022. The logistic regression (LR) and artificial neural networks (ANN) machine learning (ML) techniques were integrated into the CA model in QGIS. The previous patterns of LULC were employed in the Hybrid model (LR-CA & ANN-CA) to simulate changes in LULC for the future years (2028 and 2040). From the analysis and interpretation, it was observed that MLC with the ANN-CA model more precise technique to identify LULC features and predict changes for future years. This comprehensive and robust LULC modeling offers special spatially explicit statistics, vital for earth system analysis and understanding the complex interactions between human activities and the environment. This work develops a methodology to forecast the LULC changes through four models by uniquely integrated supervised classification in machine learning techniques. This study provides a robust framework for understanding and forecasting land use and land cover changes, which can aid in sustainable urban planning in similar regions globally.