Predictability and randomization of surface air temperature chaotic dynamics

Alexander Bershadskii

The surface air temperature daily records at the land-based locations with different climate conditions (from the Arctic to Patagonia) have been studied on the daily to intraseasonal time scales (low-frequency annual and seasonal variations have been removed by subtracting a wavelet regression from the daily records). It is shown that the power spectra of the daily time series exhibit a universal behavior corresponding to distributed chaos dominated by the buoyancy-inertial and buoyancy-diffusive mechanisms. Global average temperature fluctuations (land-based data) and the tropical Pacific sea surface temperature fluctuations (El Nino/La Nina phenomenon) have also been considered in this context. It is shown that the practical smooth predictability for the surface air temperature dynamics is possible at least up to the fundamental (pumping) period of the distributed chaos.