Privalsky V., Yushkov V. ENSO influence upon global temperature in nature and in CMIP5 simulations // Atmosph. Sci. Lett. 2014, Volume 16, Issue 3, pp. 240–245. DOI: 10.1002/asl2.548

Abstract. Statistical properties of observed and CMIP5-simulated bivariate time series ‘annual global surface temperature (AGST) – sea surface temperature in the Niño area 3.4 (SST3.4)’ are analyzed in the time and frequency domains. Both observed and most simulated data show that AGST is explicitly affected by SST3.4 but not vice versa. Though the AGST spectral density is low at intermediate frequencies, most CMIP5 models reproduce its behavior and show the high coherence observed in nature inside that frequency band. However, CMIP5 models show a dependence between AGST and SST3.4 at frequencies below 0.1 year −1 , which is not found in nature.