Comparative Analysis of Supervised and Unsupervised Classification Methods for Sentinel-2 Imagery

The proposed article aims to apply and compare supervised and unsupervised classification methods on Sentinel-2 imagery, employing the Normalized Difference Water Index (NDWI) to detect the coastline. The study is carried out within a Geographic Information System (GIS) framework, enabling efficient spatial data processing and analysis. The primary objective of the research is to evaluate the effectiveness of two classification methods: Maximum Likelihood Classification (MLC), a supervised method, and K-Means clustering, an unsupervised method. While MLC relies on prior knowledge and training sites to classify image pixels, K-Means automatically groups pixels into clusters without prior training. The first step is to achieve NDWI from the Sentinel-2 dataset, a widely used index for water body detection. This index enhances the contrast between water and land, making it easier to identify water bodies. The extracted NDWI data is then subjected to both MLC and K-Means classification techniques. The results indicate that both classification methods are valid and effective for identifying water bodies using NDWI. However, the K-Means clustering method demonstrated superior performance in terms of accuracy and consistency. This suggests that the unsupervised approach may be more suitable for applications where prior training data is unavailable or limited.