Measuring Land Cover Changes in a Mining Area in Mexico Using Remote Sensing and Machine Learning

Mining generates various alterations to the environment, affecting flora, fauna, soil, and air quality. To contribute to solving this problem, this study proposes a methodology to identify the best algorithm and data combination for measuring land cover (LC) changes induced by open-pit mining in Mexico. The methodology uses remote sensing (RS) techniques with multi-temporal Landsat 5 and 8 satellite imagery and supervised LC classification with remote sensing and machine learning (ML) algorithms. The results showed that the spectral angle mapping (SAM) algorithm and the combination of bands 6, 5, and 4 yielded the best results, with an accuracy of 85.16% and a Kappa coefficient of 0.79. Land cover (LC) change measurements revealed an increase in water body surface area of 556.83 ha, mining cover of 1729.35 ha, infrastructure of 2.61 ha, and bare soil of 1488.15 ha, while also showing a loss of soil of 2372.49 ha, scrubland of 1444.59 ha, and vegetation of 9.45 ha. The use of supervised classification of multi-temporal satellite imagery allowed for the measurement of land cover alterations. These alterations highlight the need for sustainable management strategies, environmental restoration, and the importance of continued monitoring for informed decision-making. It is recommended to explore variations in classification categories, band combinations, spectral indices, and techniques such as deep learning to improve the accuracy of LC classification.