SIF: Satellite Image Fusion for Deforestation Analysis in the Amazon Using S-1 and S-2 Data for LULC Applications

Deforestation is the purpose of converting forest into land and reforestation compared to deforestation is very low. That’s why closely and accurately deforestation monitoring using Sentinel-1 and Sentinel-2 satellite images for better vision is required. This paper proposes an effective image fusion technique that combines S-1/2 data to improve the deforested areas. Based on review, Optical and SAR image fusion produces high-resolution images for better deforestation monitoring. To enhance the S-1/2 images, preprocessing is needed as per requirements and then, collocation between the two different types of images to mitigate the image registration problem, and after that, apply an image fusion machine learning approach, PCA-Wavelet. As per analysis, PCA helps to maintain spatial resolution, and Wavelet helps to preserve spectral resolution, gives better-fused images compared to other techniques. As per results, 2019 S-2 preprocessed collocated image enhances 42.2508 km2 deforested area, S-1 preprocessed collocated image enhances 23.7918 km2 deforested area, and after fusion of the 2019 S-1/2 images, it enhances 16.5335 km2 deforested area. Similarly, the 2023 S-2 preprocessed collocated image enhances 49.2216 km2 deforested area, S-1 preprocessed collocated image enhances 23.8459 km2 deforested area after fusion of the 2023 S-1/2 images, enhancing 35.9185 km2 deforested area. These improvements show that combining data sources gives a clearer and more reliable picture of forest loss over time. The overall paper objective is to apply effective techniques for image fusion of Brazil's Amazon Forest and analyze the difference between collocated image pixels and fused image pixels for accurate analysis of deforested area.