Two CryoSat-2 sea ice thickness products derived with independent algorithms are used to initialize a coupled ice-ocean modeling system in which a series of reanalysis studies are performed for the period of March 15, 2014-September 30, 2015. Comparisons against moored upward looking sonar, drifting ice mass balance buoy, and NASA Operation IceBridge ice thickness data show that the modeling system exhibits greatly reduced bias using the satellite-derived ice thickness data versus the

more »

... l model run without these data. The model initialized with CryoSat-2 ice thickness exhibits skill in simulating ice thickness from the initial period to up to 6 months. We find that the largest improvements in ice thickness occur over multi-year ice. Based on the data periods examined here, we find that for the 18-month study period, when compared with upward looking sonar measurements, the CryoSat-2 reanalyses show significant improvement in bias (0.47-0.75) and RMSE (0.89-1.04) versus the control run without these data (1.44 and 1.60, respectively). An ice drift comparison reveals little change in ice velocity statistics for the Pan Arctic region; however some improvement is seen during the summer/autumn months in 2014 for the Bering/Beaufort/Chukchi and Greenland/Norwegian Seas. These promising results suggest that such a technique should be used to reinitialize operational sea ice modeling systems. Published by Elsevier Ltd on behalf of COSPAR. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).