Have I found it? Likely not. Since the average altitude of the land mass is 680 meters above sea level and sea level likes to change, there is likely not a meaningful global surface temperature that everyone could agree with. I personally would like a better estimate of some surface temperature, so I made my own using NOAA's Extended Reconstructed Sea Surface Temperature and the Berkeley Earth Surface Temperature project data to date. BEST is putting together a combined Land and Ocean data set, but I just could not wait.
Using the Reynold's Optimally Interpolated SST data version 2, so hopefully this version is actually optimum with a baseline from November 1981 to March 2013 I "adjusted" the ERSST data from 1880 to present for 90S to the equator and for the equator to 90N. Of course as you approach the 90s, data quality gets worse. I haven't bothered with error bars, but I came up with some numbers. I did the same thing with the ER land data using a 1951 to 1980 baseline that matched the BEST absolute temperature estimate for the hemisphere land masses.
The reason I did this nonsense is for this;
The reason that is so interesting is that the radiant forcing down or "back radiation" at the upper troposphere to turbopause is the same for both hemispheres. Corriolis effect tend to isolate the hemispheres somewhat in the atmospheric circulation but forces ocean heat flow north from the Antarctic Circumpolar Current. There is mechanical pump action messing with the oh so sickeningly sweet ideal radiant "greenhouse effect".
You may notice, if you are a savvy climate fan, that the end swoop starting at about 1985 just happens to correspond with the shift in "global" diurnal temperature trends.
I just wanted to get this posted since the BEST combined Land and Ocean is supposedly due out any day now and I figure this might be interesting to compare with their results.