Radiometric Enhancement of Landsat 8 OLI Imagery Using Coastal/Aerosol Band

Abstract

The presence of atmospheric particles in multispectral imageries such as Landsat 8 OLI can reduce the visual acuity of the imageries. The most ideal method to reduce the existence of atmospheric particles in the imagery, as well as to enhance the visual appearance of the imagery, is to employ atmospheric corrections. However, atmospheric corrections are a very complex process. Besides, sometimes the results don't have an impact visually. There are many other methods to enhance imagery radiometrically, either by stretching the pixel value, shifting the histogram, or reducing the presence of clouds. This research aims to develop practical formulations to enhance the spectral value of the Landsat 8 OLI imagery bands, by reducing the presence of aerosol particles using the C/A band. Several regression models were involved in the construction process of these formulations. The accuracy assessment was performed using the Pearson correlation coefficient and RMSE, using the USGS Landsat 8 OLI TOC imagery as a comparison. The results showed that the radiometric imagery enhancement using the C/A band gave satisfactory results. Apart from providing a significant visual sharpness increase, for the exponential model with parameters, the average Pearson correlation coefficient is 0.96, with an RMSE value of 0.04, relative to the USGS Landsat 8 OLI TOC product. For a more practical model, we can omit the parameters in the exponential model. The results that will be obtained are still quite accurate. Furthermore, we can implement this enhancement model directly on digital numbers.