Improved complete ensemble empirical mode decompositions with adaptive noise of global, hemispherical and tropical temperature anomalies, 1850–2021

Abstract

ICEEMDAN, a variant of Empirical Mode Decomposition (EMD), is used to extract temperature cycles with periods from half a year to multiple decades from the HadCRUT5 global temperature anomaly data. The residual indicates an overall warming trend. The analysis is repeated for the Southern and Northern Hemispheres as well as the Tropics, defined as areas lying at or below 30 degrees of latitude. Multiannual cycles explain the apparently anomalous pause in global warming starting around 2000. The previously identified multidecadal cycle is found to be the most energetic and to account for recent global warming acceleration, beginning around 1993. This cycle’s amplitude is found to be more variable than by previous work. Moreover, this variability varies by latitude. Sea ice loss acceleration is proposed as an explanation for global warming acceleration.