Improvement of Arctic Sea Ice Type Concentration Retrieval From Microwave Remote Sensing Data

Sea ice types, such as first-year ice and multi-year ice, can be discriminated based on radiometric and scattering signatures. However, changes in ice surface properties caused by processes like ice surface deformation and melt-refreeze events, can lead to significant misclassification. To address this issue, this study proposes an improved Sea Ice Type Concentration (SITC) Algorithm from Microwave observations, named SITCAM. It is built on a previous method, ECICE, but adopts a new cost function weighted by the separation efficiencies of microwave observations and a new pre-classification scheme to account for the bimodal distributions of ice types in microwave observations. Results show that SITCAM can effectively retrieve Arctic SITC distributions even under surface change conditions, showing a better agreement with reference data including a sea ice age product, sea ice type samples, and SAR images compared to the original method ECICE. The retrieval accuracy for first-year ice and multi-year ice samples increases to 92% and 97%, respectively (increasing by 16% and 5%); the average absolute error in multi-year ice concentration compared to the SAR-based interpretation results reduces from 15% to 4%. Furthermore, SITCAM performs well with various combinations of microwave data, such as solely radiometer data, demonstrating its potential to date SITC back to the 1970s.