Response of a global atmospheric circulation model to spatio-temporal stochastic forcing: ensemble statistics

Abstract. The response of a simplified global atmospheric circulation model (PUMA) to spatiotemporal stochastic forcing is analyzed using the statistical measures originally developed for ensemble forecast evaluation. The nontrivial effects of time and length correlations of the stochastic forcing on the ensemble scores (e.g. spread and 'error') are studied. A maximum for these scores is observed to occur for specific values of the correlation time. The effects of multiplicative and additive contributions of the correlated noise are analyzed in terms of the noise and PUMA parameters.