This paper presents absolute radiometric calibration coefficients (gains) that explain the relationship between the digital number (DN) and at-sensor radiance for the multispectral camera (MSC) on Korea's first high-resolution satellite (KOMPSAT-2). Absolute radiometric calibration was performed using a reflectance-based method. In addition, the suitability of vicarious results from radiance- and reflectance-based validations was analyzed with reference to IKONOS and QuickBird images. The latter are spectrally similar to KOMPSAT-2 images and have been validated in a large number of studies. For all bands, the R2 values of fitted lines for the gain ranged from 0.82 to 0.94, representing an improvement compared to previous findings for the KOMPSAT-2 MSC. To analyze the suitability of the vicarious results, same-pixel at-sensor radiances across different spectral bands were compared. In all bands, except the red band of QuickBird, the at-sensor radiances of KOMPSAT-2 MSC were highly correlated with those of IKONOS and QuickBird. In addition, same-pixel comparisons of reflectance across different spectral bands showed that the slopes of the least-squares lines for each band were similar to the results of the radiance comparison. The standard deviation among top of atmosphere (TOA) reflectances was within 0.019 for all bands. To calculate the tasseled cap transformation (TCT) coefficients for the KOMPSAT-2 MSC, an empirical method was applied using radiometric normalization. The results were similar to those obtained using the TCT coefficients for IKONOS and QuickBird in the brightness, greenness, and wetness components. The TCT images showed similar patterns. The absolute radiometric calibration coefficients presented here appear to be a good standard for maintaining the optical quality of the KOMPSAT-2 MSC, for which prelaunch, on-board, and vicarious calibration data are lacking.
The purpose of this paper is to ensure the reliability of the standard images from KOMPSAT-2 and the various kinds of
thematic information additionally generated by the images by means of empirical relative radiometric correction. For this
purpose, it examines the similarity between wave-length bands of KOMPSAT-2 and Landsat-7 ETM+, gathers their
images which are taken within 5 days earlier to perform the ortho-rectification and the radiometric correction. And it
performs the empirical relative radio correction in which a linear correlation is derived from radiance and then it is
applied to the images from KOMPSAT-2. As a result, the wave-length similarity of the blue band is low: 0.52; that in
other bands shows high rate of concordance: 0.82 or higher. And the RGB images expressed in radiance show little
change between before and after correction. According to the calculation of the index of vegetation by using reflectivity,
the correlation of the NDVI images in January 2008 increases from 0.004 before correction to 0.8964 after correction;
that in November 2008 increases from 0.1717 before correction to 0.7964 after correction; the correlation of the SR in
January 2008 images increases from 0.0357 before correction to 0.8876 after correction; that in November 2008
increases from 0.0003 to 0.8083. Thus, it is judged that using the images from Landsat-7 ETM+ for the relative
radiometric correction of the images from KOMPSAT-2 can contribute to ensuring their reliability.
Much effort has been made in the radiometric calibration of the ocean scanning multispectral imager (OSMI) since after the successful launch of KOMPSAT-1 in 1999. A series of calibration coefficients for OSMI detectors were obtained in collaboration with the NASA Sensor Intercomparison and Merger for Biological and Interdisciplinary (SIMBIOS) project office. In this study, we compare the OSMI level-2 products (e.g., chlorophyll-a concentration) calculated from the NASA cross-calibration coefficients with the SeaWiFS counterparts. Sample study areas are some of diagonostic data sites recommended by the SIMBIOS working group. We will present the preliminary results of this comparative study.