Comparison of MODIS and OLI image downscaling methods for industrial dust detection

One of the problems of most airborne and space-based sensors is the lack of high spatial, radiometric and temporal resolution, due to the high technical and sensor design costs. On the other hand, the identification and monitoring of the factors in natural ecosystems, such as water, soil, and atmosphere requires high spatial, radiometric and temporal resolution. Therefore, it is necessary using merge methods for integrating two or more spatial, radiometric and temporal resolution. Aerosols, especially dust of mines and industries, are part of the contaminate particles that are important in identifying them. Aerosol monitoring requires high spatial, radiometric and temporal resolution sensor, which is practically impossible in a sensor. For this purpose, it is possible to merge images with a high radiometric resolution like Modis and high spatial images like Landsat. One of the most popular indicators for dust detection is the NDDI index, which is obtained using SWIR (2.1μm) and blue (0.47 μm) wavelengths. In this research, we used several merging algorithms, including Bovery, Gram-Shcmidt, STARFM, ESTARFM, wavelet, PBIM, SIFM and HPF to integrate Modis and Landsat image data of 8 July 2016, and then provided NDDI index maps. The results of the evaluation showed that the best method was STARFM, ESTARFM, and PBIM with correlation coefficient (R2) of 0.88, 0.91, and 0.99, respectively with Landsat image and 0.51, 0.5, 0.57 with Modis image. The mean squared error (RMSE) for all three methods was negligible: 0.02, 0.400, and 0.055 respectively, with the original Landsat images and 0.004, 0.6 and 0.1 with the main images of Modis. Therefore, the STARFM, ESTARFM and PBIM methods could be used to merge Modis and Landsat images to extract data with high precision.