NONLINEAR EFFECTS OF THE SOUTHERN OSCILLATION IN THE NEW-ZEALAND REGION

Authors
Citation
Ab. Mullan, NONLINEAR EFFECTS OF THE SOUTHERN OSCILLATION IN THE NEW-ZEALAND REGION, Australian meteorological magazine, 45(2), 1996, pp. 83-99
Citations number
40
Categorie Soggetti
Metereology & Atmospheric Sciences
ISSN journal
00049743
Volume
45
Issue
2
Year of publication
1996
Pages
83 - 99
Database
ISI
SICI code
0004-9743(1996)45:2<83:NEOTSO>2.0.ZU;2-D
Abstract
The Southern Oscillation has a significant effect on New Zealand weath er, but the pattern is a complicated one. While there is a general ten dency for more frequent cold southwesterlies during the negative (El N ino) phase of the oscillation, and warm moist northerly or northeaster ly airstreams over the country during the positive (La Nina) phase, th ere is a pronounced seasonal variation to this pattern. In this study we concentrate on the linearity of the climatic response to positive a nd negative phases of the Southern Oscillation, and show that even wit hin a season it is not always valid to assume La Nina conditions will be equal and opposite to El Nino. In particular, during the southern s pring and summer, the description of circulation variations over a sub stantial part of the Australia-New Zealand region studied is significa ntly improved by assuming a bilinear response to the Southern Oscillat ion rather than a linear relationship. The most significant improvemen t occurs in the representation of the spring westerlies. An examinatio n of temperature and precipitation data for New Zealand also shows evi dence of non-linearity that can be related to the circulation non-line arities. The results support the idea that statistical predictions der ived using all Southern Oscillation data are likely to be more applica ble to describing negative extremes than positive extremes. At the neg ative extreme, the linear correlation model usually predicts the same pattern of circulation anomalies as the bilinear, although the amplitu de will differ. However, at a positive extreme, the bilinear model som etimes predicts an anomaly of the opposite sign to that from a linear model. The negative skewness of the commonly used Southern Oscillation Index time series means that negative events may be expected to contr ibute more to a linear correlation than positive events.