Combining point correlation maps with self-organizing maps to investigate atmospheric teleconnection patterns in climate model data

Hunt, F.K.; Hirschi, Joel; Sinha, Bablu. 2012 Combining point correlation maps with self-organizing maps to investigate atmospheric teleconnection patterns in climate model data. [Poster] In: IUGG Conference on Mathematical Geophysics, Edinburgh, 18-22 Jun 2012. (Unpublished)

Before downloading, please read NORA policies.

Preview

PDF
IUGG_poster_draft9.pdf - Presentation
Download (1MB) | Preview

Abstract/Summary

A new method for identifying teleconnection patterns in gridded climate data is presented. Point correlation maps constructed from NCEP/NCAR reanalysis sea level pressure (SLP) for the period 01.1984-12.2005 are used to train a self-organizing map (SOM), which topologically orders the patterns and provides a measure of frequency of pattern occurrence. Well known patterns can be identified within the SOM, such as the NAO, ENSO and the PNA, however the flexibility of the SOM allows these patterns to be viewed as part of a continuum of patterns, each identifiable as a variation within a defined teleconnection pattern. As the SOM is a non-linear method, asymmetries between patterns generated from opposite centres of action are revealed. Clustering the SOM patterns identifies the regions of the SOM corresponding to different teleconnection types by classifying similar patterns together. This retains the continuum of patterns, but allows generalization and characterization of the teleconnections present in the data. The patterns identified by the SOM can be used to evaluate the teleconnections in climate model SLP data. Point correlation maps are determined for the model data and compared to the SOM. By matching each of the NCEP/NCAR correlation maps and each of the model correlation maps with their most similar pattern on the SOM, discrepancies between the datasets are revealed. Additionally, the base points corresponding to the correlation maps for each teleconnection show the regions important to their existence. Differences in the location of the base points between NCEP/NCAR and the models provide insight into the biases underlying the model deviations from reality. The method can be extended to investigate other variables, for example the SOM can be trained using both SLP and geopotential height to investigate the 3D structure of teleconnections, while the location of the base points of the correlation maps for certain patterns can be used to assess the impact of teleconnections, such as rainfall and temperature patterns.

Item Type:	Publication - Conference Item (Poster)
Programmes:	NOC Programmes
Date made live:	23 Jan 2013 17:22 +0 (UTC)
URI:	https://nora.nerc.ac.uk/id/eprint/447539

Item Type:

Publication - Conference Item (Poster)

Programmes:

NOC Programmes

Date made live:

23 Jan 2013 17:22 +0 (UTC)

URI:

https://nora.nerc.ac.uk/id/eprint/447539