The Warming Hiatus, Natural Variability and Thermal Ocean Structure

Abstract:

Long before the recent concern with the warming hiatus, Ghil and Vautard (1991, Nature) stated at the end of their abstract that “The oscillatory components [in global temperature time series] have combined (peak-to-peak) amplitudes of 0.2°C, and therefore limit our ability to predict whether the inferred secular warming of 0.005°C/yr will continue.” Present capabilities of the advanced spectral methods introduced into the global warming problem by that paper permit us now to consider oscillatory aspects of natural variability in much greater detail.

In a multivariate analysis of the upper-ocean thermal structure, we examine properties of the recent long-term changes and of the naturally occurring global-climate fluctuations on interannual-to-interdecadal time scales. M. Ghil and associates (Ghil and Vautard 1991; Plaut et al. 1995, Science; Ghil et al. 2002, Rev. Geophys.), among others, have argued that this natural variability has some regularity embedded into it. Although the existence of such regularity on the interannual time scale is fairly well established by now, evidence for similar regularity on decadal and interdecadal time scales is more difficult to establish, due to the shortness of instrumental temperature data.

To identify spatio-temporal patterns, we rely on the method of multichannel singular spectrum analysis [M-SSA; see Ghil et al. (2002) for a review] and on its recent improvements that help separate distinct patterns (Groth and Ghil 2011, Phys. Rev. E; Groth and Ghil 2015, J. Climate). Results on the temperature field from the Simple Ocean Data Assimilation (SODA) reanalysis (Carton and Giese 2008, Mon. Wea. Rev.; Giese and Ray 2011, J. Geophys. Res.) will be shown and contrasted with results on the HadCRUT surface temperature dataset (Morice et al. 2012, J. Geophys. Res.). We will focus, in particular, on the robustness of the geographical distribution of long-term changes in both data sets and discuss the significance of superimposed natural regularities in the traditional context of red noise (Hasselmann 1976, Tellus A). Finally, we will discuss the extent to which these regularities could have contributed to the recent hiatus in the long-term changes, as predicted by Ghil and Vautard (1991).