The Utility of Surface and Upper AIr Data in Synoptic Climatological Specification of Surface Climatic Variables

The relative merits of applying 1000 and 500 hPa map classifications to specifying departures in daily and monthly climatic elements have been examined for a network of 82 stations over New Zealand. The map classification technique was based on eigenanalysis of twice daily geopotential height fields at 2·5° resolution, followed by cluster analysis. Five significant EOFs were obtained at each level and subsequent cluster analysis resulted in stable patterns of seven and ten synoptic classes at the 1000 and 500 hPa levels respectively. The application of EOF analysis and cluster analysis to the combined data from both levels led in each case to equivalent-barotropic patterns, so that it was not possible to distinguish between varying vertical structures associated with the same low-level circulation. On a daily basis, the ten 500 hPa classes were better able to distinguish between maximum and minimum temperatures and daily precipitation, but comparable results with the seven 1000 hPa map types were obtained for the duration of bright sunshine and daily wind run. The difference in skill was at least partly related to the number of synoptic classes at each level. Monthly mean departures for all variables over 1980–1993 were estimated by regression from the relative frequencies of each synoptic class, from the weighted daily mean departure patterns for each class, and from the mean EOF values for the month. In general the first two methods showed similar skill, with estimates based on the 500 hPa data slightly superior to those based on 1000 hPa analyses, but combinations of the predictors at both levels gave the best results. Direct estimates from the monthly mean EOFs proved better overall than those based on daily map classifications, limiting the value of the stochastic weather generator approach to the down–scaling of New Zealand’s regional climate.