DISTORTIONS CAUSED BY THE LACK OF RADIOMETRIC CALIBRATION AND ATMOSPHERIC EFFECTS ON THE ANALYSIS OF ORBITAL REMOTE SENSING DATA

Abstract

This paper investigates the distortions in information extracted from remote sensing data when radiometric calibration and atmospheric effects correction are not carried out. For the comparison, we used pixel values from raw data in Digital Numbers (DN, or grey tones), from the top of the atmosphere reflectance (TOA) uncorrected data, and from the surface reflectance data (SR), gathered from selected targets (such as burned crop, native vegetation, rivers and bare soil) in a multispectral image acquired by the Landsat 7 satellite, row 226, path 071, depicting the Cuiabá city in MT state, on 28/09/2002. Next, we compared Normalized Difference Vegetation Index calculated from the uncalibrated data  and from the surface reflectances (atmosphere corrected) so that we could verify the errors when using raw data. The analysis show the importance of the atmospheric correction prior the analysis when dealing with orbital sensors, given that the spectral signatures of the targets were severely drifted from the ones expected according to the specialized literature.