Author(s)

G. Ibarra, L. Alonso, M. Navazo, A. Elias & G. Gangoiti

Abstract

Cluster analysis application to automatic air pollution dispersive pattern recognition in complex terrain areas G. Ibarra/ L. Alonso^ M. Navazo/ A. Elias& & G. Gangoiti" ^Dpto. I.N. y Mecdnica de Fluidos Alda Urkijo s/n. 4 SO 13 Bilbao, Spain ABSTRACT Application of cluster analysis to meteorological parameters has made possible to recognize different air pollution situations as they are perceived by a monitoring network, thereafter assigning a concentration field to each standard situation. Those dispersive patterns are associated to specific SO% concentration fields and had been previously identified by experimental dispersion studies. Concurrently, an algorithm has been developed which assigns each day of data to a specific standard situation with on-line meteorological data from the network. The algorithm designed for this purpose is based on the definition of «distance between one day (case) and an ideal dispersive pattern» within a 240 dimensions hyperspace.

Keywords