Croll revisited: Why is the northern hemisphere warmer than the southern hemisphere?

Abstract

The question of why, in the annual-mean, the northern hemisphere (NH) is warmer than the southern hemisphere (SH) is addressed, revisiting an 1870 paper by James Croll. We first show that ocean is warmer than land in general which, acting alone, would make the SH, with greater ocean fraction, warmer. Croll was aware of this and thought it was caused by greater specific humidity and greenhouse trapping over ocean than over land. However, for any given temperature, it is shown that greenhouse trapping is actually greater over land. Instead, oceans are warmer than land because of the smaller surface albedo. However, hemispheric differences in planetary albedo are negligible because the impact of differences in land-sea fraction are offset by the SH ocean and land reflecting more than their NH counterparts. In the absence of a role for albedo differences it is shown that, in agreement with Croll, northward cross-equatorial ocean heat transport (X-OHT) is critical for the warmer NH. This is examined in a simple box model based on the energy budget of each hemisphere. The hemispheric difference forced by X-OHT is enhanced by the positive water vapor-greenhouse feedback, and is partly compensated by the southward atmospheric energy transport. Due to uncertainties in the ocean data, a range of X-OHT is considered. A X-OHT of larger than 0.5 PW is needed to explain the warmer NH solely by X-OHT. For smaller X-OHT, a larger basic state greenhouse trapping in the NH, conceived as imposed by continental geometry, needs to be imposed. Numerical experiments with a GCM coupled to a slab ocean provide evidence that X-OHT is fundamentally important in determining the hemispheric differences in temperature. Therefore, despite some modifications to his theory, analysis of modern data confirms Croll’s 140-year-old theory that the warmer NH is partly because of northward X-OHT.

Appendix

Figure 16 compares the annual-mean inter-hemispheric temperature difference $(\Delta T)$ using the Jones et al. (1999) data and NCEP/NCAR reanalysis data. The two data sets show surprisingly similar trend. It appears that over the twentieth-century, $\Delta T$ has been influenced by the Atlantic Multidecadal Oscillation (AMO, Ottera et al. 2010), with large $\Delta T$ in the mid $20^{th}$ century related to the positive (warm) phase of the AMO and a smaller $\Delta T$ in the late twentieth-century when the AMO was negative. $\Delta T$ would also have been influenced by industrial aerosols which preferentially cooled the NH until pollution controls became effective in the 1970s onward and it will also be influenced by more rapid warming of the NH in response to rising greenhouse gases. However, as Mann and Emanuel (2006) claim, aerosol impacts and the AMO may not be entirely independent and, in general, the $\Delta T$ history will be a combined effect of all these processes.

Fig. 16

The annual-mean inter-hemispheric temperature difference $(\Delta T=T_N-T_S)$ in °C from 1850 to 2010 using the Jones et al. (1999) data (solid) and that from 1949 to 2010 using NCEP/NCAR reanalysis (dashed)

For the entire available period of each data set, $\Delta T=1.30\pm 0.11$ in the Jones et al. (1999) data and 1.27 ± 0.17 in NCEP/NCAR. In particular, for the same period from 1979 to 2010, $\Delta T=1.34\pm 0.15$ in the Jones et al. (1999) data and 1.25 ± 0.16 in NCEP/NCAR. Although the Jones et al. (1999) data exhibits slightly higher $\Delta T$, the similarity of trends confirms the robustness. Hence, the figures are produced using finer resolution NCEP/NCAR, but the values in the text show the mean and the standard deviation from the two data sets.