Restoration of Wintertime Ocean Color Remote Sensing Products for the High-Latitude Oceans of the Southern Hemisphere

Satellite ocean color products have been widely used to monitor spatiotemporal variations in marine ecological environments from regional to global oceans. However, current satellite ocean color products fail to provide effective records during the winter in high-latitude oceans, limiting understanding of the marine ecological environment during the winter season. In this study, we proposed an atmospheric correction model, namely, the Neural Network Atmospheric Correction algorithm for the Southern Hemisphere (NN-AC-SH), and recover winter satellite ocean color products for the high-latitude oceans of the Southern Hemisphere from 2003 to 2020. The accuracy of the NN-AC-SH model was verified based on the in situ data from the Aerosol Robotic Network-Ocean Color (AERONET-OC). The results indicate that the NN-AC-SH model performed better than the traditional near-infrared (NIR) iterative atmospheric correction algorithm, e.g., in the 443, 488, and 531 nm bands, the relative deviations of the NN-AC-SH model were 23.11%, 20.96%, and 23.14%, respectively, while the values of the NIR model were 30.72%, 22.85%, and 24.81%, respectively. Under the observation condition of a high solar zenith angle (SZA), the NN-AC-SH model performed better than the NIR model in terms of the amount of effective data (94 versus 78 data points) and model inversion accuracy (a relative deviation of 32.83% versus 43.60%). Moreover, in situ chlorophyll concentration data from the NASA bio-Optical Marine Algorithm Dataset (NOMAD) and Chinese Antarctic Research Expedition (CHINARE) were used to verify the accuracy of the restored chlorophyll concentration products, and the results reveal that the NN-AC-SH model resulted in more effective records of higher accuracy than those obtained with the NASA-distributed chlorophyll concentration products. Overall, for the first time, this study recovered long time-series (2003-2020) ocean color products for the high-latitude oceans of the Southern Hemisphere (>= 50 degrees S) in the winter, which could provide unique satellite products for investigating the marine ecological environment of the Antarctic and sub-Antarctic oceans in the winter.

Automated summary

This study proposes a neural network atmospheric correction algorithm to recover satellite ocean color products for the high-latitude oceans of the Southern Hemisphere during winter. The algorithm outperforms traditional methods and provides more effective records of higher accuracy. The recovered products can be used to investigate the marine ecological environment of the Antarctic and sub-Antarctic oceans during winter.