Adaptability Analysis of Various Versions of GDPS Based on QA Score for GOCI Data Processing in the Yellow Sea

Spectral remote-reflectance sensing (R (A)) is significant to the retrieval of ocean bio-optical properties from ocean color. R (A) is defined as the ratio of water-leaving radiance to the downward irradiance just above the water surface. About 90% of the total signal received by ocean color sensor is contributed by the atmosphere, while only less than 10% is contributed by ocean water. Therefore, the process of atmospheric correction is an essential part of ocean color remote sensing to get accurate remote-sensing reflectance. Based on a large number of high-quality on-site hyper-spectral remote-reflectance sensing data, a QA (Quality Assurance) evaluation system for R data has been established to diagnose problematic or possibly wrong R spectra by calculating R QA scores. GOCI (Geostationary Ocean Color Imager) is the main sensor on the world's first geostationary satellite COMS (Communication Ocean and Meteorological Satellite), launched by the Korea Ocean Satellite Center (KOSC). Its high observation frequency (8 observation survey Data/day) makes it possible to monitor daily changes in biogeochemical parameters. KOSC developed GDPS (GOCI Data Processing System) software for GOCI data processing, within which atmospheric correction algorithm is integrated. The versions GDPS1. 1, GDPS1. 2, GDPS1. 3, GDPS1. 4, GDPS1. 4. 1 and GDPS2. 0 have been published and provided for free. In this paper, a QA Score evaluation system was applied to evaluate the quality of GOCI remote-sensing reflectance product processed by GDPS1. 2, GDPS1. 3, GDPS1. 4. 1 and GDPS2. 0 in the China Yellow Sea. It showed that the amount of valid R data from GDPS1. 2 was significantly less than that either from GDPS1. 3, GDPS1. 4. 1 or GDPS2. 0. The QA score of R data from GDPS2. 0 was lower than that of GDPS1. 2, GDPS1. 3 or GDPS1. 4. 1. It makes sense that the QA score of R from GDPS1. 3 and GDPS1. 4. 1 are the same. Because compared to GDP1. 3 only software modularization is optimized, and some minor problems are fixed in GDPS1. 4. Based on our results, when applying GOCI R data to the Yellow Sea, it is suggested that if R ratio is the first- order parameter (i. e., retrieving chl-a concentration) and there is no requirement of valid data amount, atmospheric correction codes in GDPS 1. 2 can be selected to used to get R. If Rrs data at a certain wavelength is concerned, GDPS2. 0 is more suitable for processing GOCI data.