Dec 04 2011
Update: Reader Blog Lurker at CA discovered my dyslexia – and the fact I have the wrong email number. Fixed below, but the number is #3468 – end update
Update: Jeff Id at Air Vent wanted something pithy to summarize this post.. So since I am watching endless Harry Potters today I have decided to preview this post with a summary of the problem and trick covered at the end. Here was the challenge facing the IPCC team in 2006:
It is not right to ignore uncertainty, but expressing this merely in an arbitrary way (and as a total range as before) allows the uncertainty to swamp the magnitude of the changes through time.
Without doing something, the IPCC claims of historic warming would sink into the noise of the measurements and be forever determined a fraud. So Briffa comes up with a plan:
the comparison of past and recent temperature levels is not as influenced by the outlier estimates.
Mischief Managed! – end update
I have been reading the Climategate 2 emails, repeatedly stunned at how bad it really is in the climate ‘science’ backwater. Talk about your amateur scientific method!
Just this morning though I tripped over an email from Keith Briffa at CRU that likely is another smoking gun that will completely up-end the Hockey Stick and all similar claims of unprecedented warming in the last half century.
Before we get to the email (and I request assistance from people more skilled in the use of time series), I want to reiterate my primary case against the IPCC claims of historic warming. The actual temperature measurements from 1850-1960 are sparse and uneven (to be kind). Apparently for the entire southern hemisphere there are only a handful of continuous records to cover this HALF of the world. From 1960 to 1990 the sample size fills in globally and the consistency increases, but it really is only in the satellite era (1990 onward for good temps) that we get consistent global measurements of decent precision. 20 years is a very short time to make claims covering centuries of time – even if we were using this data.
The IPCC is not using this data it turns out. The modern (and most accurate) global temperature record (1960 to present) diverges from the primary temperature proxies used to assess current temps against historic temps (i.e., tree rings). Tree rings show “a decline” in temperature in many parts of the Northern Hemisphere in the last 50 years.. Therefore we do not see in the Hockey Stick calibrated tree ring temperature signals from the last 50 years compared to previous ages like the Medieval and Roman Warming periods.
The reason we do not see this is because if they did compare the modern era to the past using tree rings calibrated to the modern record there would be no historic warming today.
This is known as hiding the decline.
My beef all along is how no one has defined the integrated error in the IPCC global temperature value. Starting from the errors of the temperature reading itself, to the errors that combine (i.e., expand) by averaging and smearing these local measurements to 500×500 km grids over weeks, months and seasons, which then combine at the hemispheric and global level, we are looking at what has to be an error/precision level at the global values of +/- 2-5°C. And from this we find a 0.8° C warming???
The IPCC charts don’t show this formal error analysis. And now we might know why.
In email #3468 from 2006, Keith Briffa makes an astounding admission. He explains how he combined errors from multiple time series to ensure the modern temp numbers (already fudged by covering up the tree ring numbers with thermometer numbers the tree rings don’t agree with) do not fall inside the precision of the historic numbers. First he admits why he must fudge the data:
we are having trouble to express the real message of the reconstructions – being scientifically sound in representing uncertainty , while still getting the crux of the information across clearly. It is not right to ignore uncertainty, but expressing this merely in an arbitrary way (and as a total range as before) allows the uncertainty to swamp the magnitude of the changes through time.
Let this settle in for a moment. If the Team has to represent uncertainty as a total range (which is not just arbitrary, but the professional and scientific norm), then the uncertainty would swamp their claims and destroy the IPCC’s alarmist message. As I predicted, the uncertainty in the measurements combined with the uncertainty of deriving a temperature value from tree rings results in an error bar so wide we don’t know if we warmer or colder than the Medieval Period.
Briffa admits this behind the scenes in this email. But how to salvage the Hockey Stick? Briffa creates his own new math (in total violation of all professional rules of applied statistics):
By overlapping all reconstructions and giving a score of 2 to all areas within the 1 standard error range of the estimates for each reconstruction , and a score of 1 for the area between 1 and 2 standard errors, you build up a composite picture of the most likely or “concensus” path that temperatures took over the last 1200 years (note – now with a linear time axis).
This is kludge is arbitrary and a completely nonsensical way to combine uncertainties. Basically he throws out uncertainty that is unique to a time series. The only uncertainty that is allowed to be shown is where it overlaps with another series. So if one series has huge error bars (like tree rings) that portion that is outside the others is lost. Another way it falsifies the integrated picture is if one series is showing cool temps while 2 others are showing warm. The cool side of the error is now lost, only the overlap between the cool one and warm two shows up. Basically, he is filtering data AGAIN to select only what he wishes was reality. And he freely admits it:
This still shows the outlier ranges, preserving all the information, but you see the central most likely area well, and the comparison of past and recent temperature levels is not as influenced by the outlier estimates.
Good Lord, if you are not going to integrate the data that disagrees with your expectations, why even pretend to be doing statistics or science? Just pick your data and plot it and get it over with!
I leave it to those who do know statistics to describe how bad this really is, but it looks to be a complete crock.
Addendum: To be clear here, I am not sure what Briffa did (I do not have the picture) or if it made it into any report. I assume it did. Maybe not a defensible assumption – but it seems everything else in these emails has a tangible product out there so I will stick with it. And while I am not sure what Briffa did mechanically, I know why he did it since he says it himself. He could not use standard error ranges, so he concocted something else. I’ll leave it to others to determine the severity of this action – but in my mind it simply was another way to hide the mathematical truth that the hockey stick is all smoke and mirrors.