Arctic ocean ice thickness: Modes of variability and the best locations from which to monitor them

Abstract

Model simulations of Arctic sea ice and ocean systems are used to determine the major spatial and temporal modes of variability in the ice thickness. A coupled ice-ocean model is forced with daily NCEP-NCAR reanalysis surface air pressure and surface air temperature fields for the period 1951-2003 with the analysis of the results performed for the 51-yr period 1953-2003. Ice concentration data and ice velocity data (beginning in 1979) are assimilated to further constrain the simulations to match the observed conditions. The simulated ice thins over the study period with the area of greatest thinning in a band from the Laptev Sea across the Pole to Fram Strait. The thinning rate is greatest since 1988. The major spatial modes of variability were determined with empirical orthogonal functions (EOFs) for the ice thickness within the Arctic Ocean. The first three EOFs account for 30%, 18%, and 15%, respectively, of the annual mean ice thickness variance. The first EOF is a nearly basinwide pattern, and the next two are orthogonal lateral modes. Because of the nonstationary nature of the ice thickness time series, significant changes in the modes are found if a shorter period is analyzed. The second and third principal components are well correlated with the Arctic Oscillation. The model results are also used to simulate an observation system and to then determine optimal mooring locations to monitor the basinwide mean ice thickness as well as the spatial and temporal patterns represented in the EOF analysis. The nonstationary aspect of the ice thickness limits the strength of the conclusions that can be drawn. © 2006 American Meteorological Society.