AI Summary & Key Takeaways
AI InsightResults show that overall accuracy was 94.26% for Sentinel-2 while it was 85.04% for Landsat-8, similarly, the Kappa coefficient was calculated 91.7% and 78.3% for Sentinel-2 and Landsat-8, respectively.
📌 3 Key Takeaways
Abstract
In recent years, Kabul city's rapid urbanization has adversely affected the urban land cover, such as surface water bodies and croplands. Surface water resources are threatened due to overpopulation in the city either qualitatively or quantitatively, also croplands are being lost with the development of urbanization activities through the city. To monitor and assess surface changes accurately, we classified the city area using satellite images of both Landsat-8 and Sentinel-2 and compared both of their findings. The Support Vector Machine classifier was applied to multi-senor data to classify four different land categories using the same training sites and samples with the same period. All the procedures were conducted in Google Earth Engine (GEE) cloud platform. The surface reflectance bands of both satellites were used for classification. Confusion matrixes were created using the same reference points for Sentinel-2 and Landsat-8 classification to compare the results and determine the best approach for classification of land cover. Results show that overall accuracy was 94.26% for Sentinel-2 while it was 85.04% for Landsat-8, similarly, the Kappa coefficient was calculated 91.7% and 78.3% for Sentinel-2 and Landsat-8, respectively.