Evaluation of chlorophyll retrievals from Geostationary Ocean Color Imager (GOCI) for the North-East Asian region

Abstract

Estimation of chlorophyll concentration in the marine biosphere has been the central topic of ocean color remote sensing since its advent. While various algorithms were proposed in the literature so far and tested for oceanic waters of diverse constituent composition, an independent algorithm evaluation is needed for local ocean waters that have dynamic variation in optically active water constituents such as colored dissolved organic matters (CDOM) and suspended particulate matter (SPM). This paper evaluates the performance of chlorophyll algorithms for Geostationary Ocean Color Imager (GOCI) radiometric data, using in situ measurements collected at 491 stations around Korea Peninsula during 2010-2014 from which there were 130 match-ups with GOCI data. For the evaluation in areas with high variation in SPM, water samples were first classified into three levels of SPM, and then the coefficients of candidate algorithms were newly derived for the turbidity cases using the in situ and GOCI remote sensing reflectance (R-rs) data. Functional forms of traditional band ratio algorithms (e.g. OC algorithms (O&apos;Reilly et al., 1998) and Tassan&apos;s algorithm (Tassan, 1994)), fluorescence line height algorithm, and near-infrared-to-red band ratio approach were tested. The evaluation results for the coincident in situ pairs of R-rs and chlorophyll measurements showed that the mean uncertainty was <35% with the correlation around 0.8 by using the 00 with turbidity consideration (OCT) and Tassan&apos;s algorithm with turbidity dependent coefficients (Tassan-TD). For the GOCI match-ups, the mean uncertainty for all turbidity levels was around 35% with correlation around 0.65, when OCT and Tassan-TD were used. (C) 2016 Elsevier Inc. All rights reserved.