Estimation of Arctic Sea Ice Freeboard and Thickness Using CryoSat-2

Tuesday, 16 December 2014
Sanggyun Lee, Jungho Im, Jin-woo Kim, Miae Kim and Minso Shin, Ulsan National Institute of Science and Technology, Ulsan, South Korea
Arctic sea ice is one of the significant components of the global climate system as it plays a significant role in driving global ocean circulation. Sea ice extent has constantly declined since 1980s. Arctic sea ice thickness has also been diminishing along with the decreasing sea ice extent. Because extent and thickness, two main characteristics of sea ice, are important indicators of the polar response to on-going climate change. Sea ice thickness has been measured with numerous field techniques such as surface drilling and deploying buoys. These techniques provide sparse and discontinuous data in spatiotemporal domain. Spaceborne radar and laser altimeters can overcome these limitations and have been used to estimate sea ice thickness. Ice Cloud and land Elevation Satellite (ICEsat), a laser altimeter provided data to detect polar area elevation change between 2003 and 2009. CryoSat-2 launched with Synthetic Aperture Radar (SAR)/Interferometric Radar Altimeter (SIRAL) in April 2010 can provide data to estimate time-series of Arctic sea ice thickness.

In this study, Arctic sea ice freeboard and thickness between 2011 and 2014 were estimated using CryoSat-2 SAR and SARIn mode data that have sea ice surface height relative to the reference ellipsoid WGS84. In order to estimate sea ice thickness, freeboard, i.e., elevation difference between the top of sea ice surface should be calculated. Freeboard can be estimated through detecting leads. We proposed a novel lead detection approach. CryoSat-2 profiles such as pulse peakiness, backscatter sigma-0, stack standard deviation, skewness and kurtosis were examined to distinguish leads from sea ice. Near-real time cloud-free MODIS images corresponding to CryoSat-2 data measured were used to visually identify leads. Rule-based machine learning approaches such as See5.0 and random forest were used to identify leads. The proposed lead detection approach better distinguished leads from sea ice than the existing approaches. With the freeboard height calculated using the lead detection approach, sea ice thickness was finally estimated using the Archimedes’ buoyancy principle. The estimated sea ice freeboard and thickness were validated using ESA airborne Ku-band interferometric radar and Airborne Electromagnetic (AEM) data.