Inter-slot radiometric discrepancy correction (IRDC) for GOCI ocean colour products

ABSTRACT

The inter-slot radiometric discrepancy (IRD) causes serious inconsistency in geostationary ocean colour imager (GOCI) radiometric products across the neighbouring slots. An accurate IRD correction is essential for generating operational ocean colour products, but it remains a challenging task because of its dynamic variation irrespective of sensor’s optical mechanism. The IRD is caused due to the variation in time, location, viewing geometry, and electrical sensitivity of the sensor, which result in a radiometric discontinuity in the radiance distribution between the slot images observed at the top of atmosphere (TOA). Atmospheric correction of erroneous TOA radiance (L_t, Level-1 product) leads to noisy and degraded remote-sensing reflectance (R_rs, Level-2 product) products that limit ocean colour work. To improve the quality of Level-2 products (such as R_rs and chlorophyll), radiometric bias caused due to the IRD error in Level-1 product level needs to be removed. In the present study, a practical method for correcting the IRD effects is developed and tested on several GOCI radiometric products. The method is applied to TOA radiance data from all the slots based on a variable gain factor (${\mathrm{\Delta }}_s$). The gain factors for every slot are calculated with linear statistical and Lagrangian interpolation methods and applied consistently across the slot images provided by the GOCI sensor. In the absence of in situ data for these chosen images, GOCI results are compared with the corresponding MODIS-Aqua data. The proposed method is found to be effective in terms of reducing the radiance discontinuity for different slot pairs of the GOCI image, with mean relative differences of less than 0.1% and 2% for the Level-1 and Level-2 products, respectively. Despite its successful application to GOCI data, this method can be extended to similar sensors which rely on the slot-wise reception of ocean colour information over a large area.

Acknowledgements

This research was partly supported by the Department of Science and Technology (DST) through the Networked Project on Big Data Analytics-Hyperspectral Data (OEC1617130DSTXPSHA). We acknowledge the KIOST/KOSC for providing the GOCI data and supporting the work and the OBPG of NASA-GSFC for the support of the SeaDAS Software. We sincerely thank Dr. Jim Gower, Editor, International Journal of Remote Sensing, and the two anonymous reviewers for their valuable comments and suggestions which greatly helped to improve the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Department of Science and Technology, Ministry of Science and Technology [OEC1617130DSTXPSHA].