نوع مقاله : مقاله پژوهشی
نویسندگان
گروه علوم خاک دانشکده کشاورزی، دانشگاه زنجان
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
This research aims to evaluate the abilities of four non-parametric machine learning algorithms, including Support Vector Machine (SVM), Random Forest (RF), Classification and Regression Tree (CART), and Minimum Distance (MD), to produce LULC maps. Utilizing multi-temporal data from Sentinel-2 and Landsat-8 sensors, the investigation was conducted within the Google Earth Engine (GEE) framework. The outcomes underscore the superior reliability of Sentinel-2 data compared to Landsat-8 data across all classifiers. The SVM classifier, with an overall accuracy of 92.9% and 92.2% for Sentinel-2 and Landsat-8 images, respectively, provided the best performance compared to other classifiers. The results pertaining to the identification of LULC alterations during the study duration, employing the optimal classifier (SVM), revealed an expansion in the expanse of olive groves, rice paddies, and built-up areas, alongside a contraction in water bodies and barren lands. The evaluation of the implications of LULC variations on Land Surface Temperature (LST) manifested that augmenting vegetation cover corresponded with diminished LST values within the study area. This shift led to LST values ranging from 36.48 to 21.8 Celsius in 2019, which evolved to 33.84 and 19.67 Celsius in 2023. The research concludes that the combination of high-spatial-resolution satellite data and the SVM algorithm presents an accurate and efficient approach for generating LULC maps and assessing environmental transformations.
کلیدواژهها [English]
)