GOCI-II Ocean Color Algorithm Development Environment

Abstract

In 2019, Geostationary Ocean Color Imager&#8211 II (GOCI-II), the follow-up satellite sensor of GOCI, will be launched. GOCI-II will observe 2 more ocean color images (8 times for GOCI) a day for 5 more spectral bands (8 bands for GOCI) than GOCI. Also, the ground sample distance (GSD) of GOCI-II will be enhanced as 250 m (500 m for GOCI) and the number of its products will be increased to 26 (13 for GOCI). Consequentially, the amount of GOCI-II data will be exponentially increased due to significantly improving the remote-sensing capability of GOCI-II. For producing ocean color information from GOCI-II data, the new ocean color algorithms will be needed because the assigned spectral bands for GOCI-II are different from GOCI. For GOCI-II data processing, however, the existing ocean color algorithm development approach with no regard for the high-performance data processing will not be applied because very large data sets would be processed and distributed in real time. In this study, therefore, we developed a software development environment for facilitating the implementation of GOCI-II ocean color algorithm using the high-performance computing technology. The proposed environment supports various parallelism techniques such as open multi-processing (OpenMP), open computing language (OpenCL), and message passing interface (MPI). Also, there are 4 kinds of servers (i.e., data processing server, build server, account n GOCI. Also, the ground sample distance (GSD) of GOCI-II will be enhanced as 250 m (500 m for GOCI) and the number of its products will be increased to 26 (13 for GOCI). Consequentially, the amount of GOCI-II data will be exponentially increased due to significantly improving the remote-sensing capability of GOCI-II. For producing ocean color information from GOCI-II data, the new ocean color algorithms will be needed because the assigned spectral bands for GOCI-II are different from GOCI. For GOCI-II data processing, however, the existing ocean color algorithm development approach with no regard for the high-performance data processing will not be applied because very large data sets would be processed and distributed in real time. In this study, therefore, we developed a software development environment for facilitating the implementation of GOCI-II ocean color algorithm using the high-performance computing technology. The proposed environment supports various parallelism techniques such as open multi-processing (OpenMP), open computing language (OpenCL), and message passing interface (MPI). Also, there are 4 kinds of servers (i.e., data processing server, build server, account