Evaluation of Sea Ice Concentration Data Using Dual-Polarized Ratio Algorithm in Comparison With Other Satellite Passive Microwave Sea Ice Concentration Data Sets and Ship-Based Visual Observations

Abstract

The dual-polarized ratio (DPR) algorithm is a new algorithm that enable calculation of Arctic sea ice concentration from the 36.5-GHz channel of the sensor Advanced Microwave Scanning Radiometer for EOS/Advanced Microwave Scanning Radiometer 2 (AMSR-E/AMSR2). In this paper, we demonstrate results that the sea ice concentration data using DPR algorithm (DPR-AMSR) are evaluated and compared with other eight Arctic sea ice concentration data products with respect to differences in sea ice concentration, sea ice area, and sea ice extent. On a pan-Arctic scale, the evaluation results are mostly very similar between DPR-AMSR and the bootstrap algorithm from AMSR-E/AMSR2 (BT-AMSR), the bootstrap algorithm from SSM/I or SSMIS (BT-SSMI), the ARTIST Sea Ice algorithm from AMSR-E/AMSR2 (ASI-AMSR), and the enhanced NASA Team algorithm from AMSR-E/AMSR2 (NT2-AMSR). Among of these products, the differences in sea ice concentration agree within ±5%. However, European Space Agency Climate Change Initiative algorithm from AMSR-E/AMSR2 (SICCI-AMSR), the European Organisation for the Exploitation of Meteorological Satellites Ocean and Sea Ice Satellite Application Facility from SSM/I or SSMIS (OSI-SSMI), the ARTIST Sea Ice algorithm from SSM/I or SSMIS (ASI-SSMI), and the NASA Team algorithm from SSM/I or SSMIS (NT1-SSMI) are all lower than DPR-AMSR at sea ice edge. And NT1-SSMI had the largest negative difference, which was lower than -15% or even 20%.The difference of sea ice area was consistently within ±0.5 million km2 between DPR-AMSR and BT-AMSR, BT-SSMI, ASI-AMSR, and NT2-AMSR in all years. The smallest difference was with BT-SSMI (less than 0.1 million km2), whereas the largest difference was with NT1-SSMI (up to 1.5 million km2). In comparisons of sea ice extent, BT-AMSR, NT1-SSMI, and NT2-AMSR estimates were consistent with that of DPR-AMSR and were within ±0.5 million km2. However, differences exceeded 0.5 million km2 between DPR-AMSR and the other data sets. When ship-based visual observation (OBS) values ranged from 85% to 100%, the difference between DPR-AMSR and OBS was less than 1%. There were relatively large differences between DPR-AMSR and OBS when OBS values were less than 85% or were recorded during the summer, although those differences were also within 10%.