How much does Earth temperature change in one year?

Not a discussion of how warm it is or how it changes over long time. Simply, from one year to the next, how much can it realistically change?

There are two different (I’d like to say complementary) ways to measure it. One is compiling data from land-surface and sea-surface measurements with thermometers across the globe. That is done by HADCRUT and several other groups. All those present similar results. HADCRUT starts in 1850 and others’ starts vary. Those who doubt surface-T compilations refer to ‘adjustments’ that might vary over the long term; not from one year to the next.

UAH and RSS use observations of ultraviolet emission from oxygen from several satellites, starting in 1978 December. They extract lower-troposphere temperatures (related to surface T) in slightly different ways. Here I only consider UAH because it is more widely called upon by both sides of the climate debate.

Set all that aside and think about Earth. Big. About 1/4 land area, and about 3/4 ocean area with more thermal inertia. Ice also has thermal inertia but most of it is in poorly-lit areas. From one year to the next, solar input does not vary much. Even if infrared-absorbing gases change, they don’t change much in a year. Cloud cover (reflectivity) can change promptly and other factors can do as well.

Putting all that together, I appeal to ‘common sense’ and assert that from one year to the next, global T cannot change a lot. Perhaps you agree. In any case, if it did, we’d probably detect it in other ways.

Now to those data. Throughout HADCRUT record the largest year-to-year (YTY) increase was 0.31 oC and largest decrease was 0.27. Those were rare and 71% of YTY were from -0.1 to +0.1 oC. UAH, covering much less time, had YTY changes from -0.5 to 0.49 oC, with only 50% within -0.1 to +0.1 oC. HADCRUT during UAH era had YTY changes from -0.23 to 0.21 oC and 72% in the 0.1 oC range.

Too many numbers eh? Same data can (perhaps more kindly) be expressed graphically:



If you are not much on squiggles I’d like to say that UAH is very different from these others. But whether one prefers numbers or squiggles, it remains that UAH temperatures show more YTY variability than HADCRUT (or any other) surface-T compilation. What may we conclude?

Not, I think, that surface-T compilations are ‘adjusted’ to limit YTY variations. No one has suggested that. UAH (etc.) satellite retrievals are much more responsive to ENSO, and no one from the satellite-T side has ever suggested why that may be so; thus why more-variable satellite-T is the better measure of global earth-surface T.

Thermometers are where they are and obviously not everywhere. Satellites look down everywhere; that is their greatest strength. But the latter have much more YTY variability, and their champions remain silent on this. I guess that vertical mixing of troposphere differs during extreme ENSO years, and because higher troposphere is so cold, satellite-T retrievals can get made wrong sometimes.

==
Stepping back, since 1970, HADCRUT and other surface-T compilations show +T of about 0.17 oC per decade. UAH since 1979 shows +T of about 0.12 oC per decade. I don’t consider those vastly different. It’s a problem that satellite-T is short, more variable, and (because of the latter) has been invoked to claim that over a short, carefully selected interval that warming has stopped.

We want best global T that can be known. Perhaps facing YTY variability amongst the records can help with that.