Comparison of SSM/I and AMSR-E Sea Ice Concentrations With ASPeCt Ship Observations Around Antarctica

Abstract

We compare passive microwave (PM)-derived sea ice concentrations (SIC) with more than 21 600 ship-based observations (OBS) of SIC acquired around Antarctica. PM SIC are derived from SSM/I-SSMIS and AMSR-E measurements in 1991-2009 and 2002-2010, respectively, with the ARTIST Sea Ice (ASI), Comiso Bootstrap (BST), NASA-Team (NT), enhanced NASA-Team (NT2), and EUMETSAT OSI-SAF (OSI) algorithm. We compare correlation coefficients (CC), RMSDs, and biases, separately for SSM/I-SSMIS data for algorithms ASI, BST, OSI, and NT, and for AMSR-E data for algorithms ASI, BST, and NT2. With OBS SIC and PM SIC being on fundamentally different spatiotemporal scales, we develop a new colocation approach using daily-average along-ship-track SIC values. CC between OBS SIC and PM SIC agree within their uncertainty for all algorithms and sensors. Year-round CC values are around 0.85 (AMSR-E) and 0.82 (SSM/I); CC values are similar during summer, but drop significantly during winter. Year-round RMSD values range from 13% (BST and OSI) to 17% (NT) for SSM/I and from 12% (BST) to 16% (NT2) for AMSR-E. RMSD values are similar during summer, but decrease for winter (BST: 8% for AMSR-E, 10% for SSM/I). For AMSR-E, biases are below 0.5% for BST and ASI, but between 5% (winter) and 9% (summer) for NT2. For SSM/I, biases are smaller during summer, -0.7% for BST to -7.8% for NT, than winter, -3.6% for BST to -13.9% for NT. Overall, best agreement between OBS and PM SIC is found for BST.