Abstract:

Recent rapid warming of the Arctic has promoted extension of the Arctic navigation period. However, the reliability of spatial information regarding the edge of the sea ice area, provided by remote sensing products, is critical for safe operation of Arctic shipping and polar research vessels. We compared and analyzed nine sea ice concentration (SIC) products, and used high-resolution synthetic aperture radar (SAR) data to evaluate their accuracy in determining the sea ice edge area. Results showed that the largest daily deviations from the mean values of the nine products occurred during the summer navigable window (June–August). Average deviations of ASI products with a passive microwave high-frequency algorithm and OSI SAF products with an integrated algorithm were 0.59% and −0.65%, respectively. The greatest inconsistency in spatial distribution among the nine products was found in the sea ice marginal area. In 2018, the difference between the daily maximum and minimum sea ice extent of the nine products ranged from 2.51 × 10⁶ to 6.26 × 10⁶ km². The SAR data were used for comparison with the nine products in the ice margin area. In the closed pack sea ice marginal area, apart from the three products based on the ASI and NT2 algorithms, the other products overestimated (underestimated) the low (high) SIC area. The ASI and OSI SAF products based on AMSR2 sensors produced the highest correlation with the SAR data, with R² values of 0.85 and 0.82, respectively, and root mean square errors (RMSEs) of 12.19% and 9.70%, respectively. However, in the drift sea ice edge area, all products underestimated SIC; the OSI SAF product had the highest R² value (0.78) and lowest RMSE (6.19%). Of the nine products tested, the analysis indicated that OSI SAF products based on AMSR2 sensors could better represent the sea ice margin and low SIC areas in Arctic navigation channels.

Key words: sea ice concentration, Northwest Passage, ice edge, products comparison