Lin Huang

As Arctic sea ice rapidly declines, accurate daily sea ice concentration (SIC) forecasts become crucial for Arctic research and operations. Emerging deep learning (DL) forecast models have the advantage of consuming fewer computational resources compared with numerical forecast systems. However, DL forecast models face challenges of lacking sea ice domain knowledge, low forecast accuracy in the marginal ice zone (MIZ), and poor explainability. This study proposes an explainable DL model, SIFNet, for daily Arctic SIC forecasts, which considers the domain knowledge of teleconnections and lagged correlations by integrating the convolutional block attention module (CBAM) and temporal feature convolution module (TFCM). To improve SIFNet’s accuracy in the MIZ, an MIZ-weighted loss function is used for fine-tuning, leading to a notable 1.36% reduction in the mean absolute error (MAE) of MIZ forecasts (MIZ MAE). In the January 2020–December 2023 test scenario, forecasting SIC for the future 49 days using the past 21 days’ data, SIFNet demonstrates forecast accuracies with the MAE of 4.69%, binary accuracy (BACC) of 95.16%, structural similarity index (SSIM) of 95.13%, and MIZ MAE of 18.22%. Compared with the other two DL models and nine numerical forecast systems, SIFNet’s forecasts achieve higher accuracies. Using layerwise relevance propagation (LRP) technique, the LRP- $\vert z\vert $ rule is constructed to improve SIFNet’s explainability. During Arctic minimum sea ice extent (SIE) periods, SIFNet forecasts confirm upward and downward surface solar radiation as predictors for minimum SIE. Similarly, over the past 30 years, precipitation phase is considered a potentially significant factor influencing SIC forecasts.

With global warming, the decrease in sea ice creates favorable conditions for Arctic activities. Sea surface temperature (SST) is not only an important driven factor of sea ice concentration (SIC) changes but also an important medium of the ocean–atmosphere interaction. However, the response of sea surface temperature to Arctic sea ice varies in different sea areas. Using the optimal interpolated SST data from the National Centers for Environmental Information (NCEI) and SIC data from the University of Bremen, the temporal and spatial characteristics of SST and SIC in the Arctic above 60°N and their relationship are studied, and the melting and freezing time of sea ice are calculated, which is particularly important for the prediction of Arctic shipping and sea ice. The results show that (1) the highest and lowest monthly mean Arctic SST occur in August and March, respectively, while those of SIC are in March and September. The maximum trends of SST and SIC changes are in autumn, which are +0.01 °C/year and −0.45%/year, respectively. (2) There is a significant negative correlation between the Arctic SST and SIC with a correlation coefficient of −0.82. (3) The sea ice break-up occurs on Day of the Year (DoY) 143 and freeze-up occurs on DoY 296 in the Arctic. The melting and freezing processes lasted for 27 days and 14 days, respectively. (4) The Kara Sea showed the strongest trend of sea ice melting at −1.22 d/year, followed by the Laptev Sea at −1.17 d/year. The delay trend of sea ice freezing was the most significant in the Kara Sea +1.75 d/year, followed by the Laptev Sea +1.70 d/year. In the Arctic, the trend toward earlier melting of sea ice is smaller than the trend toward later freezing.