EVALUATION OF THE GFDL GCM CLIMATE VARIABILITY .2. STOCHASTIC MODELING AND LATITUDE-TEMPORAL FIELDS

This paper statistically analyzes and compares the variability of the observed and simulated (by the Geophysical Fluid Dynamic Laboratory ge neral circulation model) northern hemisphere monthly surface air tempe rature latitude-temporal fields. The objects used for the analysis are as follows: multivariate autoregressive and linear regression model p arameters, the diffusion equation coefficients, and two-dimensional sp atial-temporal spectral and correlation characteristics. The estimates reveal a significant latitudinal trend of all second moments. A compa rative study of the estimates demonstrates a significant distinction b etween the character of the meridional circulations of the observed an d simulated systems. The methodology and results presented make it cle ar that the variability validation problems can be formalized by apply ing standard statistical techniques of multivariate modeling and multi dimensional spectral and correlation analysis to the data averaged spa tially and temporally.