Hierarchical clustering of atmospheric soundings

✍ Scribed by Marina Živkovíć

• Publisher: John Wiley and Sons
• Year: 1995
• Tongue: English
• Weight: 974 KB
• Volume: 15
• Category: Article
• ISSN: 0899-8418
• DOI: 10.1002/joc.3370151004

✦ Synopsis

Abstract

This is a procedural paper that compares six commonly used hierarchical clustering techniques applied to atmospheric soundings. The following techniques are compared: single‐, complete‐, and two average‐linkage techniques, the Ward's technique, and one centroid type technique. Vertical profiling of the atmosphere is based on common soundings sampled on synoptic scales over the Northern Hemisphere. At each analysis point, the atmospheric profile is represented by an ‘atmospheric state vector’ consisting of surface pressure, total column (precipitable) water, and temperature, wind and relative humidity values at 12 discrete vertical levels.

An evaluation of the techniques is performed by comparing the mean state vectors of final clusters. Four techniques produce comparable results, with the largest differences between the techniques appearing for the surface pressure component of the cluster state vector. Also, large differences are found in the temperature and wind vector components at the 850 hPa pressure level, and the smallest differences are found for the relative humidity components. The results support findings of studies on airmass typing on local or regional scales: the average‐linkage method based on group average provides the most distinct and homogeneous clusters.