The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLII-4
https://doi.org/10.5194/isprs-archives-XLII-4-587-2018
https://doi.org/10.5194/isprs-archives-XLII-4-587-2018
19 Sep 2018
 | 19 Sep 2018

COMPARISON OF GEOMETRIC CORRECTION SCHEMES FOR GEOSTATIONARY OCEAN COLOR IMAGER SLOTS WITHOUT GCPS

J. Son, H. Kim, and T. Kim

Keywords: GOCI, Geometric correction, Satellite image, GCP, Sensor Model

Abstract. Currently, the geometric correction process of GOCI (Geostationary Ocean Color Imager) image is performed by matching slot images against shorelines and utilizing the matching results as GCPs (Ground Control Point). However, there are several GOCI slots without shorelines and for such slots acquiring GCPs is not easy. The purpose of this paper is to compare several alternative geometric correction schemes applicable to the slots without GCPs. We analyzed three schemes. The first scheme is to apply the correction angle of the same slot in the most recent dataset. The second scheme is to apply correction angle of the previous slot in the current dataset. And the last scheme is to apply correction angle of the slot with the largest number of GCPs in the same time dataset. Overall process for comparing the quality of the three geometric correction schemes consisted of the following steps. Firstly, using ephemeris metadata of GOCI Level 1A, we established initial sensor model, which defines geometry relationship between ground coordinate system and image coordinate system of a GOCI image. And then, by matching edge detected from GOCI slot images and shoreline landmark chips, we obtained GCPs. Using these GCPs, we calculated correction angle of each slot. After then, through the three schemes, we conducted precision sensor modeling. Among three schemes, geometric correction applying the previous slot correction angle showed the best quality. The average RMSE of this scheme was about 1.4 km, which was quite close to geometric correction quality applying correction angles from GCPs.