Comparative Analysis of Vegetation Change Detection Using the Normalized Difference Vegetation Index and Support Vector Machine Algorithm Based on Landsat-8 Data: A Case Study of Bamyan City
DOI:
https://doi.org/10.68924/fd6e0y54Keywords:
Bamyan City, Changes, Landsat-8, NDVI, SVM, VegetationAbstract
Remote sensing techniques are used in issues such as making maps of large areas, estimating crops, getting suitable areas for cultivation, checking changes and making correct decisions, etc. In this research, the vegetation change of Bamyan city was investigated using Landsat-8 satellite images in 2015 and 2023 in two ways. In this research, after performing radiometric and atmospheric corrections, the spatial resolution of the images was improved from 30 meters to 15 meters by the Pan-Sharpening algorithm, then, from the Normalized Difference Vegetation Index (NDVI) and the classification of the Support Vector Machine (SVM) algorithm, vegetation and its changes were investigated. Classified images were improved using morphological operators, and finally, by performing the difference method of two images, the amount and manner of changes were investigated and validated. Kappa coefficient in the months of May, June, and July in 2015 for the NDVI index is 0.89, 0.91, 0.90, respectively and for the SVM algorithm; 0.87, 0.91, 0.89 and in 2023, for the NDVI index 0.87, 0.88, 0.89 and SVM algorithm respectively; 0.85, 0.84, and 0.86 were obtained. This shows that the NDVI index has a better performance compared to the SVM algorithm for obtaining vegetation cover. Vegetation level in the interval of 8 years, in the months of May, June, and July, based on the NDVI index (0.06%), (0.1%), (0.3%), respectively, and based on the SVM algorithm (0.04%), (0.09%), (0.31%) increased, respectively. This research showed that satellite data has special importance and value for evaluating and investigating the process of changes.