Estimation of Ocean Surface Currents from Maximum Cross Correlation applied to GOCI geostationary satellite remote sensing data over the Tsushima (Korea) Straits

Abstract

Attempts to automatically estimate surface current velocities from satellite-derived thermal or visible imagery face the limitations of data occlusion due to cloud cover, the complex evolution of features and the degradation of their surface signature. The Geostationary Ocean Color Imager (GOCI) provides a chance to reappraise such techniques due to its multi-year record of hourly high-resolution visible spectrum data. Here we present the results of applying a Maximum Cross Correlation (MCC) technique to GOCI data. Using a combination of simulated and real data we derive suitable processing parameters and examine the robustness of different satellite products, those being water-leaving radiance and chlorophyll concentration. These estimates of surface currents are evaluated using High Frequency (HF) radar systems located in the Tsushima (Korea) Strait. We show the performance of the MCC approach varies depending on the amount of missing data and the presence of strong optical contrasts. Using simulated data it was found that patchy cloud cover occupying 25% of the image pair reduces the number of vectors by 20% compared to using perfect images. Root mean square errors between the MCC and HF radar velocities are of the order of 20 cm s−1. Performance varies depending on the wavelength of the data with the blue-green products out-performing the red and near infra-red products. Application of MCC to GOCI chlorophyll data results in similar performance to radiances in the blue-green bands. The technique has been demonstrated using specific examples of an eddy feature and tidal induced features in the region. This article is protected by copyright. All rights reserved.