HR: 0800h
AN: C21B-1106    [Abstracts]
TI: Antarctic Sea Ice-Atmosphere Interactions: A Self-organizing Map-based Perspective
AU: * Reusch, D B
EM: dbr@geosc.psu.edu
AF: EMS EESI, 517 Deike Building The Pennsylvania State University, University Park, PA 16802 United States
AB: Interactions between the ocean, sea ice and the atmosphere are a significant component of the dynamic nature of the Earth's climate system. Self-organizing maps (SOMs), an analysis tool from the field of artificial neural networks, have been used to study variability in Antarctic sea ice extent and the West Antarctic atmospheric circulation, plus the relationship and interactions between these two systems. Self-organizing maps enable unsupervised classification of large, multivariate/multidimensional data sets, e.g., time series of the atmospheric circulation or sea-ice extent, into a fixed number of distinct generalized states or modes, organized spatially as a two-dimensional grid, that are representative of the input data. When applied to atmospheric data, the analysis yields a nonlinear classification of the continuum of atmospheric conditions. In contrast to principal component analysis, SOMs do not force orthogonality or require subjective rotations to produce interpretable patterns. Twenty four years (1973-96) of monthly sea ice extent data (10 deg longitude bands; Simmonds and Jacka, 1995) were analyzed with a 30-node SOM. The resulting set of generalized patterns concisely captures the spatial and temporal variability in this data. An example of the former is variability in the longitudinal region of greatest extent. The SOM patterns readily show that there are multiple spatial patterns corresponding to "greatest extent conditions". Temporal variability is examined by creating frequency maps (i.e., which patterns occur most often) by month. With the annual cycle still in the data, the monthly frequency maps show a cycle moving from least extent, through expansion to greatest extent and back through retreat. When plotted in "SOM space", month-to-month transitions occur at different rates of change, suggesting that there is variability in the rate of change in extent at different times of the year, e.g., retreat in January is faster than November. Twenty five years (1977-2001) of monthly 500 mb temperature and pressure data (from the ECMWF 40-year reanalysis, ERA-40) from a region centered on the Antarctic Peninsula were analyzed independently for a separate SOMs-based study. Dominant SOM temperature patterns include the expected summer warmth and winter cold, plus "dipoles" of warm Atlantic (Pacific) and cold Pacific (Atlantic) sectors (with corresponding pressure patterns). Temporally, there is the expected annual progression from warmth, through cooling and back to warmth, with no particularly predominant patterns in many of the monthly frequency maps when the full record is used. Stratifying by high/low values of the Southern Oscillation Index (SOI) suggests that the spatial patterns of cooling and warming may be related to conditions in the tropical Pacific: in a low SOI year (1987), cooling and warming both begin in the Atlantic sector, with the opposite true in a high SOI year (1989). Further study of this aspect is planned. In addition to direct comparisons of the SOM analysis results from each study, a joint SOM analysis will be done on the combined data sets, exploiting the flexibility and power of this technique. We anticipate additional useful insights into the joint variability and relationships between Antarctic sea ice and the overlying atmosphere through this expanded analysis.
DE: 0724 Ice cores (4932)
DE: 0750 Sea ice (4540)
DE: 1610 Atmosphere (0315, 0325)
DE: 3305 Climate change and variability (1616, 1635, 3309, 4215, 4513)
DE: 9310 Antarctica (4207)
SC: Cryosphere [C]
MN: Fall Meeting 2005