| The sensitivity of sea ice to the temperature of the overlying air implies that observed trends in Arctic ice conditions may also indicate general climatic changes.; An historical record of Arctic imagery from orbiting passive microwave sensors starting from 1972 provides us with an excellent data source for studying climate change and variability. In this thesis, nine years of sea ice concentration data derived from Scanning Multichannel Microwave Radiometer (SMMR) imagery are analyzed.; Four hypertemporal methods--image trend analysis, principal components analysis, time series analysis, and temporal mixture analysis--are developed within the context of modelling change and variability in a 108 monthly image sequence (nine year period) of Arctic sea ice concentrations.; The analysis results are combined to form a new spatial-temporal description of Arctic sea ice concentrations which is data driven. A number of regions, principally the Barents, Bering, Chukchi, and Labrador Seas, and the Sea of Okhotsk, are shown to have a highly variable ice cover suggesting that they are easily affected by shifts in atmospheric conditions. Thus, the ice cover in these regions may be the first to show changes in global climate conditions. Indeed, the image trend analysis revealed statistically significant decreases in Arctic sea ice concentrations in the Barents and Okhotsk Seas. The problem in using these areas as climatic indicators, however, is the difficulty in detecting a definite climate signal from the noisy data. Alternatively, Baffin and Hudson Bays, and the Arctic, Beaufort, and East Siberian Seas all have relatively constant ice cover. A climate shift would show up very clearly in the sea ice cover in these areas. Because of the relative persistence of the sea ice here, however, several years may pass before any changes become evident.; Temporal mixture analysis is an effective tool for identifying changes between groups of temporal sequences (e.g., one year time series) and a reference sequence (e.g., a long-term normal). Further, these experiments revealed that the ocean tends to remain completely ice-free or ice-covered during the year; it seldom remains partially ice-covered for more than one month, except during transition to/from complete coverage. There must be forces at work which cause sea ice to resist a change of state (freezing or thawing) well beyond the point at which the atmospheric conditions dictate. Once a certain threshold has been passed, the state change occurs rapidly and totally as the system regains equilibrium.; While the other hypertemporal analysis techniques discussed here provide information about an overall process, they are generally unable to provide feedback on the internal nature of the process and any significant events which occur during the sequence. Principal components analysis is a tool which does provide this information. A spatially coherent pattern of large ice anomalies throughout the Arctic in the year following the strong 1982-83 El Nino event is clearly identified in the fourth component analyzed here. (Abstract shortened by UMI.)... |
