Combining CryoSat-2 Data with Ocean-Sea-Ice Models to Improve the Understanding of Arctic Sea Ice Thickness

Sea ice plays a major role in the global climate system. Its presence governs the radiation budget and determines how much heat and momentum can be transferred between ocean and atmosphere. In order to monitor sea ice and analyse how it has changed over the recent past, different satellite observations are used. Sea ice area is monitored by passive microwave and optical sensors. To make accurate estimates of the sea ice volume, sea ice thickness (SIT) estimates are also needed. The monitoring and estimation of SIT from satellite observations is an active area of research. This thesis contributes to this research by developing sea ice model-based parameterizations for satellite radar observations and an assimilation framework for these observations, with the aim of improving our understanding of Arctic SIT. The work presented here concentrates on CryoSat-2 freeboard (FB) observations.

When estimating SIT from space, only it’s FB, can be observed. To convert FB to SIT, assumptions about sea ice density, snow thickness, snow density and water density are made. These assumptions are known to introduce errors, and the work presented here introduces a model-based approach to estimating them. The main novelty is the derivation of a sea ice density and the use of ocean model water density. Further, this thesis introduces a Kalman Filter-based approach to assimilate FB to improve SIT. One challenge in retrieving CryoSat-2 FB is to know where the radar signal is reflected. CryoSat-2 carries a Ku-band radar, which prior to launch was assumed to penetrate the snow and to be reflected at the ice snow interface. Later studies found that this is not the case and that the scattering horizon varies depending on the region and season. Comparing the SIT from a classically-derived SIT product to the SIT-derived from the FB assimilation framework, presented here, shows that the assimilation gives better results in two regions where independent SIT observations were available. Finally, this thesis analyses the sea ice model used to derive the FB and the influence different model parameterizations have on the SIT, sea ice concentration and snow thickness. The form drag parameterisation is found to have the largest influence on the sea ice volume and snow thickness. More work is needed to determine which form drag parameterisation is best suited for the model FB parametrization.

There is a reception after the defense at 15:30 at DMI, Sankt Kjelds Plads 11, 2100 København Ø.