Seasonality of circulation in southern Africa using the Kohonen self-organising map

A technique employing the classification capabilities of the Kohonen self-organising map (SOM) is introduced into the body of computer-based techniques available to synoptic climatology. The SOM is one of many types of artificial neural networks (ANN) and is capable of unsupervised learning or non-linear classification. Components of the SOM are introduced and an application is then illustrated using observed daily sea level pressure (SLP) from the Australian Southern Hemisphere data set. To put the technique in the context of global climate change studies, a further example using simulated SLP from the GENESIS version 1.02 General Circulation Model (GCM) is illustrated, with the emphasis on the ability of the technique to highlight differences in seasonality between data sets. The SOM is found to be a robust technique for deducing the modes of variability of map patterns within a circulation data set, allowing variability to be expressed in terms of inter and intra-annual variability. The SOM is also found to be useful for comparing circulation data sets and finds particular application in the context of global climate change studies.

Grantor dc:publisher.institution

Department of Environmental and Geographical Science

Year dc:date.issued

1997

Author dc:creator

• Main, Jeremy P L

Advisor dc:contributor.advisor

• Hewitson, Bruce

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/13889

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/13889