고속의 구름 이동을 고려한 GOCI 해색영상에서의 구름영향 픽셀 인지

Abstract

The study aims to establish a set of pixel-wise tests to identify those cloud-affected pixels in the GOCI Level-2 data generation process, in view of producing high-quality Level-2/Level-3 data. To this end, we propose a set of pixel-wise spectral consistency tests over satellite reflectances 1) in three visible bands, 2) in 660 and 680 nm bands, and 3) in 745 and 865 nm bands. The performance of the proposed test set is evaluated in terms of near-cloud statistical behavior on chlorophyll a (chl-a) imagery, by comparing statistics in “very near-cloud pixels” and “next near-cloud pixels”. The test set results in statistically more stable near-cloud field in the sense that the average (std. dev.) of very near-cloud chl-a records closer value, typically within 5% (10%) range, to the next near-cloud chl-a field.pixel-wise spectral consistency tests over satellite reflectances 1) in three visible bands, 2) in 660 and 680 nm bands, and 3) in 745 and 865 nm bands. The performance of the proposed test set is evaluated in terms of near-cloud statistical behavior on chlorophyll a (chl-a) imagery, by comparing statistics in “very near-cloud pixels” and “next near-cloud pixels”. The test set results in statistically more stable near-cloud field in the sense that the average (std. dev.) of very near-cloud chl-a records closer value, typically within 5% (10%) range, to the next near-cloud chl-a field.