I started trying to reconstruct a global temperature graph. I found that land based measurements like CRUTEM were reading too high and identified this as massive warming in Northern hemisphere land based measurements.
Sea surface measurements like HADSST gave longer term trends that were more consistent with satellite data and other metrics. However, they did not have the same year to year changes as the satellite data.
So, I proposed to take the fast changes from CRUTEM and add it to the longer term trend of HADSST as I hope the following graph shows:
The top green is CRUTEM (land) – which leads the pack showing excessive warming. The lower green (toward the end) is CRUTEM less any longer term trends. The blue line is the sea temperature from HADSST with short term trends removed.
Purple shows UAH6 satellite and Red shows my reconstructed metric which simply adds long term trends from the sea measurements to short term trends from land.
This is clearly a good fit. “HASELER” is a bit low in the 1998 El Nino year and a bit too high in the 2016 El Nino year. But generally HASELER reconstruction follows UAH6 well.
Except … it shouldn’t!!
I’m happy that the sea temperatures representing 6/7 of the globe might represent the vast majority of global temperatures and so the sea long term trend will be close to that of the global temperature. But why on earth does the short term changes in global temperature reflect only the land-based measurements. It should only be 1/7 of the changes.
This suggests to me that most of the short term changes in global temperatures are affected by what happens over land and that the atmospheric transfer into the sea is relatively poor.
This suggests that over short periods the air flowing over the land, determines the temperature (or is determined in temperature by) the atmospheric temperature. But that what determines long term temperatures is the sea (or what is determined by long term temperatures).