Sea ice motion from low-resolution satellite sensors: An alternative method and its validation in the Arctic

The retrieval of sea ice motion with the Maximum Cross‐Correlation (MCC) method from low‐resolution (10–15 km) spaceborne imaging sensors is challenged by a dominating quantization noise as the time span of displacement vectors is shortened. To allow investigating shorter displacements from these instruments, we introduce an alternative sea ice motion tracking algorithm that builds on the MCC method but relies on a continuous optimization step for computing the motion vector. The prime effect of this method is to effectively dampen the quantization noise, an artifact of the MCC. It allows for retrieving spatially smooth 48 h sea ice motion vector fields in the Arctic. Strategies to detect and correct erroneous vectors as well as to optimally merge several polarization channels of a given instrument are also described. A test processing chain is implemented and run with several active and passive microwave imagers (Advanced Microwave Scanning Radiometer‐EOS (AMSR‐E), Special Sensor Microwave Imager, and Advanced Scatterometer) during three Arctic autumn, winter, and spring seasons. Ice motion vectors are collocated to and compared with GPS positions of in situ drifters. Error statistics are shown to be ranging from 2.5 to 4.5 km (standard deviation for components of the vectors) depending on the sensor, without significant bias. We discuss the relative contribution of measurement and representativeness errors by analyzing monthly validation statistics. The 37 GHz channels of the AMSR‐E instrument allow for the best validation statistics. The operational low‐resolution sea ice drift product of the EUMETSAT OSI SAF (European Organisation for the Exploitation of Meteorological Satellites Ocean and Sea Ice Satellite Application Facility) is based on the algorithms presented in this paper.