Sorry … I pulled this post because the USHCN adjustments are fairly linear from 1979-2011, and there is no reason to expect that they will change the RSQ values in any meaningful way.
I will revisit this topic later when I have more meaningful stats.
 | gelcarrion0t on “You Can’t Hide Yo… |
| gelcarrion0t on Doubling Energy Costs |
| gelcarrion0t on Australia Permanent Drought Up… |
| gelcarrion0t on “New computer modeling… |
 | Morgan Wright on Climate Destroying Shrimp |
 | Christopher Simpson on 1896 : Japan Hit By Earthquake… |
 | energywise on What’s At Stake? |
 | Robert Cherba on Too Hot To Live |
| gelcarrion0t on Too Hot To Live |
 | Nicholas McGinley on “The world began to end… |
An interesting plot that begs for more information. Is it possible to put a time basis into this plot some how either by color spots or using a third axis? If your other plots about adjustment over time are correct,, the inclusion of a time element in the scatter of the corrected data might be most interesting.
I like Sean’s suggestion.
Also, I’m getting the raw data’s slope at ~1.10 and the adjusted data at ~1.22. So the adjusted data has a larger slope AND a worse R2? If anything, larger slopes with equivalent noise would give a higher R2.
Also, what people often forget or forget to mention is that the RSS increase should actually be HIGHER than land increases according to AGW theory (CAGW or non-C AGW…either one). Therefore, both slopes as you plotted them should be <1 according to the theory.
-Scott
The larger slope is because adjusted data cools the past and warms the present – so they get a larger spread along the y-axis. Try flipping the axes and see what you get ;^)
A perfect correlation would have a slope of 1.0
Neither R-square value is very good (much less excellent) for anyone wanting to claim solid physical understanding of both measurements (surface and satellite), and 0.77 can hardly be taken to be much worse than 0.80; they both tell me the satellite vs. surface measurements are simply not as well-correlated as I would want to see them (with an R-square greater than 0.95, say), in order to say they are both measuring the same physical reality, without extraneous errors in one or both measurement procedures. And I know the surface data (assuming this is USHCN) is fraudulently adjusted according to the level of atmospheric carbon dioxide, which is the real smoking gun (and the R-square for that is 0.974, in my short analysis, which is what an excellent correlation really looks like). In fact, considering the fraudulent nature of the surface measurement adjustments, I am surprised the second graph above is not much worse, unless the carbon dioxide level follows the temperature, globally, even over short periods of time — the seasonality in the carbon dioxide level, between winter and summer, is apparent in the Mauna Loa record, the graph of which you can see at the above link. This is just my two cents, for anyone who happens by and wants a second opinion from a competent physicist with a fair amount of experience (over the past 40 years) in applied data analysis.
The point is that adjusted data does not correlate as well as raw data.
Satellite data is measured at 14,000 feet. It isn’t the same physical phenomenon. We expect larger swings in satellite data than surface data.
There’s no statistically significant difference between the two results.