What the Flux? What Anomaly?

j. ferguson in a comment asked about temperature versus energy and energy versus energy considerations in methods.  Since radiant energy flux has a 4th power relationship to temperature and all energy transfer is not radiant, comparisons can be complicated.  Almost all of the estimates of change require a small range of change or there will be errors that increase with the range.

The chart above is the new Hadley Climate Research Unit version 4 with the C average being the simple average of the NH and SH anomaly scaled to 14 C, the rough "average" surface temperature.  The Wm-2 average goes like this.  The NH and SH time series are converted to Wm-2 with 10 K difference between the two.  The NH scaled to an average of 19C and the SH scaled to an average of 9C, then convert both to Wm-2, average the Wm-2 and convert back to C degrees.  The ten degrees difference produces the 0.14 mean error in this quick example.  The use of anomalies is very useful but there is some error that can be produced if the differences are large.  Nothing new, not shown is a 2 degree difference that produced an average error of ~0.01 C error.

This is an Amundsen-Scott temperature series comparing the normal anomaly to the shifted 0C reference anomaly.  Colder temperature have less energy change per degree than warmer temperatures.    Since the change in temperature is only an indication of the the change in energy with careful bookkeeping, there is potential for small, but significant errors in any averaging since a small, ~1% change in energy is all that is expected.

TheAirVent with Jeff Id, along with some other blog "skeptic" publish a paper, O'Donnell et.al 2010, that noted that improper averaging or interpolation methods produce a large amount of potential error.  Since the Antarctic has the fewest surface stations per area and the greatest temperature change per degree of latitude, the range of uncertainty is extremely large when compared to the rest of the globe.

The Berkeley Earth Surface Temperature project has re-estimated the Antarctic region using a Kriging method of interpolation which has many advantages over other methods attempted in the past, but it would still have the latitudinal variation issue which, with limited data, is unavoidable.  Even with plenty of data, averaging or Kriging across latitudes will produce smearing of temperatures.

Since all the data is in temperature and the response to energy change is the large portion of the problem, there is unavoidable uncertainty.

This is the main reason that I focus on simple static models, and the moist air portion of the globe.  With higher average energy/temperature they would produce lower average error.  There is still too much uncertainty to have a eureka moment, but the simpler model does produce interesting results.

For example:  Using the simple model, an increase in atmospheric forcing would produce a decrease in Antarctic average temperature.  Some climate scientist are hell bent on "proving" warming using the Antarctic when cooling is actually a "proof" of increased GHG forcing according to my simple model.

Who is right will be pretty obvious in a few years.