Stochastic Models to Represent the Temporal Variability of Zonal Average Cloudiness

Abstract

Seasonal autoregressive and integrated moving average models are employed to describe the contemporary variability of different time series. In all, zonal/monthly records of five variables, produced and archived by the Nimbus 7 Cloud Data Processing team and International Satellite Cloud Climatology Project, are analyzed. The main goal is to improve our (statistical) understanding on the zonal mean variability of cloud amount separating the regular and noise parts. Two simple models have been applied depending on the variability of initial series. Adequacy of the applied models is tested by means of portmanteau test. Both datasets show good agreement describing an asymmetric distribution of belts with different temporal variability. A large pan of the variance (more than 80%) can be explained by means of simple models in certain belts corresponding to the areas with significant annual cycle of cloud amount. These zones are located in distinct about 10° wide belts in the Tropics and Northern Hemisphere. Between them, slightly narrower belts with insignificant annual cycle are situated. Quantitative parameters obtained through representation by means of stochastic models, describing such a structure, can be used to test the validity of cloud amount parameterization schemes in physically based models.