Intrinsic Pink-Noise Multidecadal Global Climate Dynamics Mode

Understanding multidecadal variability is an essential goal of climate dynamics. For example, the recent phenomenon referred to as the “global warming hiatus" may reflect a coupling to an intrinsic, preindustrial, multidecadal variability process. Here, using a multifractal time-series method, we demonstrate that 42 data sets of 79 proxies with global coverage exhibit pink-noise characteristics on multidecadal timescales. To quantify the persistence of this behavior, we examine high-resolution ice core and speleothem data to find pink noise in both pre- and postindustrial periods. We examine the spatial structure with an empirical orthogonal function analysis of the monthly averaged surface temperature from 1901 to 2012. The first mode clearly shows the distribution of ocean heat flux sinks located in the eastern Pacific and the Southern Ocean and has pink-noise characteristics on a multidecadal timescale. We hypothesize that this pink-noise multidecadal spatial mode may resonate with externally driven greenhouse gas forcing, driving large-scale climate processes.

Figure 1

Spatial distribution of the shortest timescale (in years) at which pink-noise behavior appears in the GISS data set. This transition takes place on multidecadal timescales nearly everywhere. The red color denotes locations that do not show pink-noise characteristics on timescales up to 65 yr (half of the total length of the data set), with the most prominent feature being in the tropical eastern Pacific. White regions show locations where continuous data are absent.

Figure 2

Spatial distribution of the values of $d(\tau )$ represented by one-point correlation maps and an EOF analysis using the GISS monthly averaged surface temperature from 1901 to 2012. Centered in the eastern Pacific at ${120}^{\prime }\mathrm{W}$, ${20}^{\prime }\mathrm{N}$, we calculate the correlation between the $d(\tau )$ at this position and that at any other position (a). The spatial distribution of the correlation is nearly identical to the dipole mode called the PDO [1]. The newly constructed index (red), the normalized value of $d(\tau ){|}_{120^{\prime }\mathrm{W},20^{\prime }\mathrm{N}}-d(\tau ){|}_{180^{\prime }\mathrm{E},40^{\prime }\mathrm{N}}$, is compared with the traditional normalized PDO index (blue), which shows an excellent match. A similar one-point correlation map is constructed based on the geographic position at ${50}^{\prime }\mathrm{W}$, ${38}^{\prime }\mathrm{N}$ and is shown in (b). This map is very similar to the SST pattern in the negative state of the NAO [19], as shown by the correlation between $d(\tau ){|}_{{50}^{\prime }\mathrm{W},{38}^{\prime }\mathrm{N}}-d(\tau ){|}_{{40}^{\prime }\mathrm{W},{50}^{\prime }\mathrm{N}}$ (red) and the normalized NAO index (blue). The EOF analysis is applied to the values of $d(\tau )$, with the leading mode explaining 21% of the total variance, as shown in (c), along with the PC. This mode connects the major PDO region in the eastern Pacific to the Southern Ocean through a continuous same-sign region, as distinguished from the other areas. The time series of the principle component of the mode is analyzed using MFTWDFA (d). At lower frequencies, the variability of $d(\tau )$ parallels pink noise (red dashed line, $\beta =1$), with a crossover time of $\approx 15\mathrm{yr}$.

Figure 3

The initial (a),(c) and final (b),(d) timescales exhibiting pink-noise dynamics in the paleoclimate data across the globe [20], where (a) and (b) [(c) and (d)] show the analysis for the complete data set (after removing data from 1850 to present), to enable us to distinguish between natural climate variability and anthropogenic forcing. There are no discernible differences between (a),(b) and (c),(d), implying that pink-noise dynamics are an internal characteristic of Earth’s climate system.

Figure 4

Fluctuation functions $F_2(s)$ for paleoclimate proxy data (see the inset and Table I in Ref. [20]) spanning at least the past 80 000 yr. The blue stars show the crossover timescale of $\sim 1470\mathrm{yr}$, associated with Dansgaard-Oeschger events. For reference, the black diamond denotes the 100 000-yr Milankovitch eccentricity cycle, the timescale of Late Pleistocene glaciations. The black line has a slope of unity ($\beta =1$).