STATUS AND IMPROVEMENT OF PRECISE GEOMETRIC CORRECTION ALGORITHM OF GEOSTATIONARY OCEAN COLOR IMAGER DATA

Abstract

The geostationary ocean color imager (GOCI), one of the three payloads of COMS which hasbeen successfully launched in 26 June 2010, is currently in operation by Korea Ocean SatelliteCentre (KOSC) at Korea Institute of Ocean Science & Technology (KIOST) to observe an oceancolouraround the Korean Peninsula. This paper introduces the geometric correction proceduresof GOCI data and discusses the direction of the geometric correction techniques to be applied tothe GOCI-II which will be launched in 2019. Currently, GOCI provides L1B data with 1 to 2pixel positional accuracy through four steps: image navigation, landmark determination,navigation update and image resampling. Image navigation is the process of establishing an initialsensor model using metadata provided by the vendor. The landmark determination is the processof estimating the errors included in the initial sensor model through matching between the GOCIdata and the reference data, and the navigation update is of establishing a precise sensor modelby applying the estimated error to the image navigation. Image resampling generates an L1Bproduct by rearranging the grids using the precise sensor model. More advanced technologies arerequired for the geometric correction of GOCI-II with higher specification. For example, starimage processing technology to improve sensor model accuracy and development of newmatching technology for landmarchnology (KIOST) to observe an oceancolouraround the Korean Peninsula. This paper introduces the geometric correction proceduresof GOCI data and discusses the direction of the geometric correction techniques to be applied tothe GOCI-II which will be launched in 2019. Currently, GOCI provides L1B data with 1 to 2pixel positional accuracy through four steps: image navigation, landmark determination,navigation update and image resampling. Image navigation is the process of establishing an initialsensor model using metadata provided by the vendor. The landmark determination is the processof estimating the errors included in the initial sensor model through matching between the GOCIdata and the reference data, and the navigation update is of establishing a precise sensor modelby applying the estimated error to the image navigation. Image resampling generates an L1Bproduct by rearranging the grids using the precise sensor model. More advanced technologies arerequired for the geometric correction of GOCI-II with higher specification. For example, starimage processing technology to improve sensor model accuracy and development of newmatching technology for landmar