HR: 1330h
AN: A52C-0811 [PDF]
TI: The Sensitivity of Atmospheric Trajectory Cluster Analysis Results to Clustering Methods Using
Trajectories to the PICO-NARE Station
AU: * Owen, R C
EM: rcowen@mtu.edu
AF: Department of Civil and Environmental Engineering
Michigan Technological University, 1400 Townsend Dr, Houghton, MI 49931 United States
AU: Honrath, R E
EM: reh@mtu.edu
AF: Department of Civil and Environmental Engineering
Michigan Technological University, 1400 Townsend Dr, Houghton, MI 49931 United States
AU: Merrill, J
EM: jmerrill@boreas.gso.uri.edu
AF: Center for Atmospheric Chemistry Studies
Graduate School of Oceanography, University of Rhode Island, Narragansett, RI 02882 United States
AB:
The use of cluster analysis to group atmospheric trajectories according to similar flow paths has become a common tool in
atmospheric studies. Many methods are available to conduct a cluster analysis. However, the dependence of the resulting
clusters upon the specific clustering method chosen has not been fully characterized. Specifically, the use of hierarchical
versus non-hierarchical clustering algorithms has received little focus. This study presents the results of two cluster
analyses: one using the hierarchical clustering algorithm average linkage, and one using the non-hierarchical clustering
algorithm k-means. These results demonstrate the sensitivity of this cluster analysis to the use of a hierarchical method
versus a non-hierarchical method. In addition, this study analyzes methods for dealing with the vertical component of
trajectories during the clustering process. The analyses were performed using a 40-year set of trajectories to the PICO-NARE
station, located atop Pico Mountain in the Azores Islands in the central North Atlantic.
DE: 3309 Climatology (1620)
DE: 3399 General or miscellaneous
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting