Analyzing the performance of Landsat-8, Sentinel-2 satellite and their images fusion for detection and evaluation of wheat yellow rust

Evaluation of wheat yellow rust was done in a part of Khuzestan for two beginning and peak stages of disease, using the methods based on NDVI, GNDVI and MSI, based on classification using images of Sentinel-2 and Landsat-8, and based on images fusion. Disease was detected using visible and infrared bands. Then Sentinel-2 and Landsat-8 images were classified and disease was detected in them, finally, disease was detected by fusion of the images of two satellites. Also, accuracy of each of the mentioned methods was calculated and compared to provide the best technique for detecting disease in similar areas. Two first scenarios shows that Landsat-8 image classification for disease beginning stage with RMSE equal to 0.845 and Sentinel-2 image classification for disease peak stage with RMSE equal to 0.845, have had the best results. Comparison of vegetation indexes showed that NDVI in disease beginning (Sentinel-2 image with RMSE equal to 0.959) and disease peak (Landsat-8 image with RMSE equal to 0.959) had the best result. Comparison of all the implemented scenarios in this research shows that classification based on fusion of Landsat-8 and Sentinel-2 images was the best for classification in both beginning and peak stages of disease. IHS fusion method had the best results for disease classification in beginning stage with RMSE equal to 0.809, while GST fusion was the best in peak stage with RMSE equal to 0.793. Present research results confirm a high capability of satellite images fusion for improving classification results of wheat yellow rust disease.