Atmospheric correction methods that will be applied to a WorldView-2 image
ISBN 978-85-88783-11-9
Authors
1Santos, T.; 2Antunes, M.; 3Keidel, G.; 4Sousa, G.; 5Seoane, J.
1UFRJ Email: talita.calaca@gmail.com
2UFRRJ Email: mauroantunes@ufrrj.br
3UFRJ Email: gabrielkeidel@hotmail.com
4UFRRJ Email: gustavobond@gmail.com
5UFRJ Email: cainho@geologia.ufrj.br
Abstract
Remotely sensed images are captured at a large distance from the Earth’s surface. Consequently, the electromagnetic radiation travels a long distance in the atmosphere before reaching the sensor. Depending on the wavelength and atmospheric conditions, the radiation that reaches the sensor can be changed. Generally, the atmosphere intereference is the major cause of degradation of original images, often compromising their analysis and interpretation. The main effects of the atmosphere on remote sensing imagery are reducing the contrast between the targets, the possibility of detecting very small targets and not to distinguish targets that present very close reflectance (Rose, 2009). The intensity of the atmospheric influence on the images depends on the wavelength, which is different from band to band, and correcting its effects by using mathematical models. We will present three atmospheric correction methods that will be applied to a WorldView-2 image: the DOS method, the model AtmCorWV2 and model Atcor2. The empirical method Dark Object Subtraction - DOS (Chaves, 1988) consists on the correction of atmospheric scattering in which the atmospheric interference is estimated directly from the digital numbers (DN) of the satellite image, ignoring atmospheric absorption. This method assumes that is a high probability that dark targets (pixels) exists in the images and that they should present a very low ND in. This methodology is based on information collected from the image itself, thus the is no need to enter informations about the atmospheric conditions at the time of image capture (Chavez., 1988; Sanchez et al, 2011). The second method used was the AtmCorWV2 model, which was adapted from the 6S model for use in WorldView-2.satellite images. The model is based on the 6S radiative transfer and it considers the atmospheric conditionsat the time of image capture to calculate the the needed flows to obtain the bidirectional reflectance of the surface (Antunes et al., 2012). The 6S model (Second Simulation of Satellite Signal in the Solar Spectrum) was developed by Vermote et al. (1997) and was originally implemented for the simulation of radiances that reaches the sensor within bands from the solar spectrum, between 250 to 4000 nm, and therefore can be used to establish the characteristics of the sensors. From the direct and diffuse irradiance that reaches the surface and from the reflectance its possible to obtain the target’s radiance. Through radiative transfer is obtained the radiance that reaches the sensor and also obtain the apparent reflectance based on the irradiance that reaches the top of the atmosphere . The third applied model consists on the ATCOR (Atmospheric and Topographic Correction for Satellite Imagery)technique, which needs to be fed by the atmospheric conditions at the time of acquisition of imaging The main purpose is to simulate the main effects on the spectral response from the targets that are resulting from the absorption and scattering by gas molecules and aerosols according to the Horizontal visibility, eliminating light scattering and thus representing an important step for the evaluation of surface features (Richter, 2000). Such a technique is based on a radiative MODTRAN type transfer model (Moderate Resolution Atmospheric Transmission) algorithm developed by AFRL (Air Force Research Labs) in collaboration with (SSI) (Spectral Sciences Inc). The ATCOR2 module was embedded in ERDAS IMAGINE 2011 program. The objective of this study is to evaluate which type of atmospheric correction is the most effective for Multispectral sensor images from the WorldView-2 satellite. For this the values of surface reflectance (of the corrected images by DOS and Atmcor4WV2 ATCOR2 methods) and the apparent reflectance (reflectance at the satellite level) from three target samples, and then these values were compared with the expected target’s response as published in the literature. The results showed that models using the radiative transfer are more effective in correcting images f from the Multispectral Sensor WorldView-2 satellite.
Keywords
atmospheric correction; WorldView-2; radiative transfer