|
Activity Number:
|
275
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #309392 |
|
Title:
|
State-Space Discrimination and Clustering of Atmospheric Time Series Data Based on Kullback Information Measures
|
|
Author(s):
|
Joseph Cavanaugh*+ and Thomas Bengtsson
|
|
Companies:
|
The University of Iowa and Alcatel-Lucent
|
|
Address:
|
Dept of Biostatistics C22-GH, Iowa City, IA, 52242-1009,
|
|
Keywords:
|
atmospheric science ; classification ; geostatistics ; state-space process
|
|
Abstract:
|
We consider the problem of identifying geographically homogeneous regions based on similarities in the temporal dynamics of weather patterns. Two disparity measures are proposed and applied to cluster time series of observed monthly temperatures from locations across the state of Colorado. The two measures are based on state-space models, where the monthly temperature anomaly dynamics and seasonal component variation are structurally represented in the latent process. Application of these measures produces clusters consistent with known atmospheric flow structures. Importantly, our procedures allow for discrimination and clustering according to the individual, unobserved structural components of a state-space process.
|
