ATMOSPHERIC CORRECTIONS OF OLI LANDSAT IMAGES WITH HORIZONTAL VISIBILITY DATA FROM FIELD OBSERVATIONS AND AIRPORT

Resumo

The accurate use of satellite images for mapping and environmental monitoring requires the image transformation to ground reflectance through atmospheric correction. However, it is a challenge to obtain the horizontal visibility, which is used by the atmospheric correction models to estimate the aerosol optical depth. The aim of this paper is to present the comparison of atmospheric correction of OLI Landsat 8 images using horizontal visibility from field observation and from airport data. OLI images were acquired from four dates, 02/26/2014, 02/10/2014, 10/11/2015 and 04/20/2016. Field work was conducted at the same time of satellite overpass and horizontal visibility was obtained by observing targets at different distances and recording the maximum distance at which targets could be identifiable by visual inspection yielding values from 12 to 17 km. For comparison the horizontal visibility was also downloaded from METAR database for the Galeão airport, which were up to 10km. Atmospheric correction was carried out for the two sources of horizontal visibility using the Atmcor4OLI program, adapted through the 6S code. These two methods were compared through graphs and a statistical test from samples of four targets using apparent and surface reflectance. The results show that the atmospheric correction is paramount to analyze the spectral response of targets as the atmosphere interferes with the spectral characteristics of the targets from the visible to the mid infrared. In the visible the additive effects predominate while in the near and mid infrared the subtractive effects dominate. The visibility of the airport and from field observation yielded surface reflectance values which were different by the test of means at 1% and not significant at the 5%, as the field observations were not much higher than 10 km. It is concluded that an accurate source of horizontal visibility is key for obtaining correct surface reflectance values, mainly when field observation at the time of satellite overpass was not possible.