He is all over me claiming he can do a better analysis of the highest TMax data than I have done here. The dispute is over what is happening in Canada with the highest TMax of the year. These guys simply cannot accept that it is either flat or dropping, which should not be happening if AGW is correct. So of course, the faithful on that site are all over me trying to claim I didn't do the analysis correctly. I can't possible know what I'm doing with this data because I'm not a PhD in climatology, and the best of all, my analysis is too "simple" to show any "proper" trends. Only this Chris is capable of doing that.
They wanted the raw data so Chris could do his magic and prove me wrong. He chose the station, Station 2973, which I provided the raw data for him to download.
On his site which is supposed to show how wrong I am, he gets this graph:
Very difficult to read and the reason is it difficult to read is because this is not a time series. What he did was to take the top 100 highest TMax temps and plotted them. Highest in his mind was anything 34C and over. Hence he missed 66 years out this time series where the highest TMax is below 34C. He has years in this plot that have more than one point. What gives with this nonsense? This is "proper" analysis?
His graph should have been the same as mine for this station which I did properly as line graphs which is much easier to understand.
Chris hates this method of plotting that data, dismissing it as a "proper" method of plotting the data because using line graphs "fools the eye" into producing a trend that he claims is not there. Yeah, right Chris. More like it shows Tmax is cooling since the 1940's and you can't have that now can you. So he has to hide the decline by producing his scatter plot of just those records over 34C.
However, even his own graphs is not convincing enough for Chris to realize that he is confirming what I have seen, that TMax is dropping since 1940, he has to employ more "proper" analysis techniques. One is to do a polynomial trend on the data.
This is what he gets:
Notice this time it appears he used the correct data points, one for each year. Begs the question, why was the first one just the top 100 temps and not the highest time series?
Justifying this curve fit he says:
"The best-fit polynomial regression (cubic in this case) ... "
I had to question that. Why a third order polynomial? Why not two, or four?
His reply:
"Richard, a quadratic shows a less good fit than a cubic, hence I didn't use it. Sheesh"
Really? He did not supply anything to back up the claim, so of course one has to test that. So I did a scatter plot of the highest TMax, unlike Chris' first plot, I correctly used one point for each year (which appears to be his second plot), and then applied both polynomials.
Black is Chris' third order, while the blue is a quadratic (second order). But is running polynomials appropriate for this data? That depends on the R Squared. Chris' graph shows an R Squared of 13% for the third order trend, however my plot showed 17%. The second order trend has an R Squared of 10.1%.R Squared is the degree of fit of the data to the curve. A perfect fit is 100%. Thus both the 3rd and 2nd order polynomials are pathetic, and shows there is very little correlation between the curves and the actual data. Though his 3rd order is barely better, it cannot be counted on to predict future TMax points. Hardly surprising with only 100 points. The two are just as much an unrealistic depiction of what the highest temperature is doing.
However, assuming Chris' is the correct one, for the moment, what can one deduce from his curve of future temps? One test of how good a curve is is it's ability to predict the future, what R Squared is also about. Well, taking Chris' 3rd order curve and projecting into the future, we get this:
Oops, 50C by the next 50 years for a summer high temp? Anyone who believes that is possible needs their head examined. Question is, why did Chris not do a projection to see how it would fit? It would certainly be my first test to do on any curve. (BTW, the 2nd order drops way too low).
So Chris' curve is not a depiction of reality. It's a curve he chose because it looked to have an increase at the end. In other words, he chose it to try to "hide the decline". This is tantamount to fraud on Chris' part. Nice example of the faithful doing anything they can to keep the faith alive.
If the degree of swing in the highest TMax is a sine wave, undulating up then back down like a pendulum with a 50 year period, it certainly could still have a 17% correlation to a 3rd order polynomial because of the short length of the dataset. But be so only by co-incidence.
So Chris has not proved anything near that my original analysis is wrong. He used a completely erroneous method of plotting the data. If anything, he has confirmed I'm correct, but simply just does not have the wherewith all to admit it.
Contrary to what the AGW Faith claims, it's definitely NOT hotter today than in the 1940's. No matter how much Chris twists the data, runs curves through the plots, that fact does not change.
What the highest temperature is doing is dropping since the mid 1940s. And there is nothing Chris can do to change that, and Chris refuses to admit I'm right. Even to the point of dismissing my use of a line plot as a proper method of graphing the data. Even though AGW graphs of temperature anomalies use a time series line graph, Chris claims my use of the same type of graph for TMax is wrong. His claim is that only scatter plots correctly depict the data. Such is the state of AGW faithful hypocrisy.
The fact that Chris is attempting to discredit my analysis and a cooling summer means I have struck a cord that he is uncomfortable with. AGW should be making summers hotter! How can we make it look like it is getting hotter?!
The best part of all is when Chris linked to a reference:
Global observed changes in daily climate extremes of temperature and precipitation
https://bora.uib.no/bitstream/1956/1477/1/Stephenson.pdf
This is a study done on global temperature data to see what is happening in the extreme ends of the yearly temperature range. One of course they did was the plotting of the highest TMax for each year. Exactly what I did, and exactly what Chris claimed I should not do because it was inappropriate.
In that paper was this near the end:
...rather than viewing the world as getting hotter it might be more accurate to view it as getting less cold. (pg. 17)
Gee, where have I seen this before?
So what is Chris' reply to all this? He posts on that blog that he had a paper rejected and need the time to redo it, so would not be pursuing this analysis any further. So he is a scientist of some kind, involved in climate science I would suspect. Not surprising he would have a paper rejected if this is any indication of the quality of the work he does.
So, this is the quality of scientists we are producing today? Sad.
No comments:
Post a Comment