Verification of General Circulation Models Applied to the Hamburg University GCM. Part I: Test of Individual Climate States

Abstract

One objective of general circulation models is to simulate, e.g., a ?January? which is not distinguishable from observed Januaries. A strategy to verify an individual simulated state is proposed. Its main elements are: data compression by means of EOFS, performance of a multivariate parametric test, and a subsequent univariate analysis. The suggested technique is applied to four January simulations perforated with the Hamburg University GCM. The meteorological parameters treated are the zonally averaged January mean of the geopotential itself and of the intensity of transient and stationary eddies of geopotential height at 300, 500 and 8 50 mb. The comparison is based on daily observations from 15 Januaries (1967?81). It turns out that the midlatitudinal meridional gradient of geopotential height is significantly overestimated at all levels. The intensity of the transient eddies is significantly overestimated at 850 mb at practically all latitudes and at 300 and 500 mb at midlatitudes.