Forecasting Adelaide's maximum temperature using principal components and multiple linear regression

Fields of MSL pressure, surface temperature and low cloud cover are used as predictors to forecast Adelaide maximum temperatures. Although this is an exercise in multiple linear regression, multivariate procedures are used to minimise difficulties arising from large numbers of correlated predictors. Thus there are principal component analyses on fields of meteorological variables with illustrations of their meteorological application in a specific problem. The results indicate that a non-linear model is necessary. It is also noted that the method has a more general application, because fields of variables are used and stations in the settled areas of South Australia and western Victoria are within its field of application.