By request, I’ve computed the estimated influence of el Niño, volcanic aerosols, and solar variations on both surface temperature data and satellite data.
The adjustment for el Niño is more sophisticated now, allowing for both a prompt and a longer-delayed response, and allowing for a seasonal effect on the el Niño influence. I’ve also done this for more data sets, specifically adding the Cowtan & Way and the Berkeley data sets. Hence I’ve done so for the following data sets:
For the surface temperature data sets, it now begins in 1951.
The attached file contains time, plus four entries for each data set: the original data, the model incorporating the given factors, the residuals from the model, and “adjsted” data removing the influence of el Niño, volcanic aerosols, and solar variations.
The data are here as an ASCII text file, but it’s named “.xls” (an Excel file) because wordpress won’t allow me to upload a text file. Just change the file extension from “.xls” to “.txt” and you should be fine.
Just to illustrate, I’ll graph the results for NASA (surface) and RSS (satellite) data. We’ll start with NASA monthly data:
Here are the annual averages of same:
Here’s the satellite data from RSS TTT v4:
And here are annual averages of same:
This blog is made possible by readers like you; join others by donating at Peaseblossom’s Closet.
Open Mind 
Tamino, you rock!
Tamino, do you have any plans to publish this analysis? It would be great to have as a reference.
thank you, Tamino
very, very cool!
We see a pretty straight line with a quite narrow noise band.
approximately 1.85 K/century slope.
Thank you very much Tamino for this pretty & helpful exercise.
You make me dream of a more or less strange monthly time series starting e.g. with january 1979, where each month has a grid of say 2.5° resolution like UAH’s gridded data, giving 72 x 144 cells.
Each cell in a grid is a column of pressure levels like has the IGRA dataset, from the surface up to e.g. 30 hPa.
Lastly, each level in a cell column contains a percentage coefficient to be applied to the appropriate time series you want to extract these influences off.
P.S. Am I the one and only person asking for solely removing the volcano influence out of temperature data? For me climate layman, they are the only really exogenous factor: I intuitively identifiy ENSO and TSI as being fully integrated in the climate system, whereas volcanos of course influence it but are not directly a part of.
[Response: Yes, you’re the only one who has asked for that. Whether one regards ENSO/Solar as exogenous or not, the purpose of removing them is to better isolate the greenhouse-gas influence.]
A very impressive series of close correlations.
What is the climate sensitivity upon which the models are based?
It is not a physical model. It is just a statistical one. Short term fluctuation are removed by statistical attribution.
A while ago, (https://tamino.wordpress.com/2015/09/24/exogenous-redux/#comment-90654) you said you would look at using part of the data set for training and seeing how well it fitted the last 15 years. Have you done that? I am curious as to how robust the parameter fit is especially with more physically reasonable model for El Nino and seasonal effects. If the statistical model is a good match for underlying physics, then I would expect this to be pretty good and also a contrast to various mathturbation exercises like fitting to planetary motion.
[Response: More work … but I guess that’s what I do.]