RealClimate recently hosted a guest post by Barry Bickmore about some of Christopher Monckton’s claims. It included this passage about Monckton’s claim that CO2 concentration isn’t rising exponentially, but only linearly:
In other words, the slope keeps getting larger in an exponential trend, but stays the same in a linear trend. Monckton is right that you can do that sort of statistical test, but Tamino actually applied Monckton’s test to the Mauna Loa observatory CO2 data since about 1968 and found that the 10-year slope in the data has been pretty continuously rising, including over the last several years.
The reference is to this post in which I produced this graph:
Monckton himself commented on the RealClimate post, saying this:
It is suggested that we did the test incorrectly, because a climate-extremist performed a similar test on the Mauna Loa CO2 concentration dataset and came up with a different result. However, as our detractors ought to have realized, the Mauna Loa dataset, taken from a single location intermittently perturbed by regional volcanic activity, is not the same dataset as the NOAA global dataset that we used. Accordingly, we are unimpressed by their reliance upon an entirely different dataset.
Wow! He used a different dataset! That changes everything!!
Or does it? Let’s do the same analysis using the NOAA global CO2 dataset Monckton refers to. You can get it here.
There’s nothing new here except the dataset. Computing linear trend rates over moving 10-year intervals gives this result:
It’s clear. The rate of CO2 increase has gone up during the time span covered by the NOAA global data set. Including recently. But wait — is that result significant? Let’s add (2-sigma) error bars to the results:
Why yes, the increase is significant.
One wonders, why would you use the NOAA dataset when the Mauna Loa data set covers a longer time span? Maybe the two datasets give meaningfully different results? Let’s compare:
Nope. Not really different. Let’s add the error bars.
Nope. Not really different.
Referring to the Mauna Loa data, Monckton said it was “taken from a single location intermittently perturbed by regional volcanic activity.” Now, if you’re familiar with CO2 concentration data (which you should be if you’re going to pontificate on the subject), you know that the regional differences are much smaller than the growth over time. You’d also know that the Mauna Loa atmospheric observatory takes great pains to avoid contamination of their data by regional volcanic activity. Heck, if you really knew what you were talking about you’d already know that the Mauna Loa data is a very high-quality dataset, that it agrees extremely well with worldwide averages, and that it can be used as a suitable estimate of the global average.
But it looks like Chris Monckton doesn’t know these things. I guess he doesn’t really know this subject very well.
By the way. If you want to know whether or not the growth in atmospheric CO2 is exponential, there’s an easier way. Log-transform the data. Here’s a plot of log(CO2) concentration using the NOAA global data:
Let’s fit a straight line by linear regression:
It sure looks like log(CO2) has increased faster than linear, i.e., that CO2 concentration has increased faster than exponential. We can test this by fitting a quadratic curve to the residuals from our linear fit:
Sho’nuff. CO2 has increased faster than exponential. Even using the shorter NOAA global dataset. And yes, the result is statisically significant.