Comparative analysis between methods of atmospheric correction in images Operational Land Imager sensor (OLI) / Landsat 8

Abstract

The process of atmospheric correction is essential for processing of satellite images, it aims to minimize the influence of factors that make up the Earth's atmosphere. Because the images of the OLI/Landsat 8 sensor are a recent product, the functions/packages for the atmospheric correction of said sensor have not been effectively implemented in most software digital image processing. Therefore, the study aimed to perform a comparative analysis of three methods of atmospheric correction applied to images of the OLI/Landsat 8 sensor. The Dark Object Substract (DOS) and methods linear calibration scale data consist of radiation at the top of the atmosphere by means of radiometric coefficients given in "metadata" product file. The scene used for application of the methods was obtained from the database of the USGS. For the first method, the ldrisi Selva® was used, while for the second linear calibration equations were processed in Matlab® and ENVI® software. A comparison was made from the curves of spectral signature of selected targets on the processed images. The method of weighted linear calibration performed in ENVI® demonstrated a better validation of the data before comparing the reflectance values of the targets. It is noteworthy that the validation of the proposed methods take into account only the spectral curves available in the literature, a breakthrough in this study in future work would be to consider the reflectance of targets from data obtained by a spectroradiometer.