The scandal of fiddled global warming data

[EternalOptimist]

That reminds me of possibly the best statistics lesson I ever had, courtesy of the Psychology department at uni.

• Take a for/against poll of opinions on a couple of controversial subjects.
• Discuss those subjects
• Take another poll to see if anyone has changed their mind as a result of the discussions
• Apply some stats magic

The result was that 2 out of 6 girls changed their votes, and 2 out of 6 blokes changed their votes.

Common sense would say there was no difference between the sexes in the result.

Yet the stats produced by the lecturer claimed a significant difference in how each sex responded to the experiment.

We gave up at that point. Resistance (or indeed logic) was futile.

There are two main adjustments that are used. It is utterly gobsmacking

#1
they take Brillos great grand pappys 20c thermometer and look at the entire record for that thermometer (station)
then they look for 'break points'. Thats where the station suddenly reports a large step change.

Something serious must have occurred. Nobody asks WHAT has occurred, because its an automated process.

Then the automated process assumes that the latest temp is the correct one and applies the step change to ALL the previous measurements.

So that recorded 20c is overwritten.

Now for the real bummer. What causes the step changes ? Many of the historical stations were situated in areas free of shelter, for accuracy. They slowly get overgrwon and provide shelter and temperatures creep slowly opwards. When this gets cleared, temperatures go down sharply (break point)and the historic record is reduced. The growth comes back , the temps creep up, till the growth is cleared. Temps go down, the record is reduced. again.
The past gets cooler and cooler, giving a warming trend based entirely on automated adjustments.

Many stations were moved. Many were moved to airports or towns.
When the sceptics complain about Urban Heat Island, or jet exhausts, the are resited again.
result is as above. temps go down, record is adjusted down, trend goes up

Brillos great great grandpappys 20c is now showing up on the graph considerably lower.

[BrilloPad]

BP

let me ask you a quiz question.

blah blah blah

Can I ask the audience?

Really? Do we have to? Again?

You can just totally ignore these threads.

[pjclarke]

There are two main adjustments that are used. It is utterly gobsmacking

they take Brillos great grand pappys 20c thermometer and look at the entire record for that thermometer (station)
then they look for 'break points'. Thats where the station suddenly reports a large step change.

Something serious must have occurred. Nobody asks WHAT has occurred, because its an automated process.

Except that the Berkeley (BEST) methodology is different. They treat the time series before and after the break as two separate stations, no adjustments needed. Guess what? Negligible difference.

Then the automated process assumes that the latest temp is the correct one and applies the step change to ALL the previous measurements.

So that recorded 20c is overwritten.

No. The adjusted data and raw data are both still available, should you want the 20C it's still there and online here: ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/.

Now for the real bummer. What causes the step changes ? Many of the historical stations were situated in areas free of shelter, for accuracy. They slowly get overgrwon and provide shelter and temperatures creep slowly opwards. When this gets cleared, temperatures go down sharply (break point)and the historic record is reduced. The growth comes back , the temps creep up, till the growth is cleared. Temps go down, the record is reduced. again.

This may come as a shock, but not everything the amatuers at Watts' place write is the truth, the whole truth and nothing but the truth. Shelter may also provide shade, so the temperatures would go up. Also the homogenisation algorithm doesn't just look for step changes, it also looks for trends that diverge from the average for an area.

Many stations were moved. Many were moved to airports or towns.
When the sceptics complain about Urban Heat Island, or jet exhausts, the are resited again.
result is as above. temps go down, record is adjusted down, trend goes up

So the raw data is in need of adjustments?

The NASA GISS dataset only uses rural stations for trend analysis so UHI, jet exhausts etc are moot. It shows pretty much the same trend as the urban+rural dataset. UHI is only an issue if the amount of H increases over time, and there are methods in place to address this. Globally its a non-issue - no UHI in rural areas or over the seas. See The Berkeley paper on this (Judith Curry is a co-author) http://www.scitechnol.com/2327-4581/2327-4581-1-104.pdf

And then there's the small issue, that if the land based record has been deliberately biased low, why does it agree so well with measurements from satellites?

Which of these is 'biased'?




What is gobsmacking is that this became a non-issue some years ago, but the pseudo-sceptics keep on flogging this and a small stable of other dead horses ....

[speling bee]

You can just totally ignore these threads.

It's a bit like trying to ignore a learning disabled elderly gent singing Zip-a-Dee-Doo-Dah while rubbing his suspiciously stained trousers against your car front driver side window.

[BlasterBates]

oh dear just want to correct pj's mistake, he talked about GISS and then plotted the wrong temp. Easily done.

Looks like a divergence to me.

[pjclarke]

oh dear just want to correct pj's mistake, he talked about GISS and then plotted the wrong temp. Easily done.

Oh dear. In fact <Sigh> ...

These plots are expressed as an anomaly from a baseline, not absolute temperatures, and each series uses a different baseline. If you're going to plot them on the same graph you have to add or subtract an offset, otherwise its like comparing the heights of two people while one of them is standing on a box.

The author of the graph-plotting site explains:

The main temperature series we have on this site - HADCRUT3, GISTEMP, UAH and RSS - are all expressed as monthly temperature anomalies from a defined baseline period. This means that the average temperature for each similar month (all Januaries, all Februaries, etc.) is subtracted from the monthly value to remove any seasonal cycle, and (in theory) any difference between the absolute starting positions of the series.
Why 'in theory'? Well, the problem arises because the four series use three different baseline periods. Here are the baseline periods as reported by each source:


GISTEMP Jan 1951 - Dec 1980 (30 years)
HADCRUT3 Jan 1961 - Dec 1990 (30 years)
RSS Jan 1979 - Dec 1998 (20 years)
UAH Jan 1981 - Dec 2010 (30 years)


If you think about the different baseline periods, the reason for this [the need for an offset] is obvious . GISTEMP has the earliest baseline period, when temperatures were cooler, so its anomalies from this baseline are always higher. HADCRUT and RSS are somewhere in the middle, and UAH has the most recent, warmest baseline, so its anomalies are lowest now.
To make a fairer comparison between the series, we need to know what these offsets are, and we can then correct for them. As UAH is the lowest and newest baseline, let's use their baseline period - Jan 1981 to Dec 2010 - as the period of comparison. We will then calculate the average anomaly of each series during this period to get an approximate offset between them.
Conveniently (in fact, just for this purpose!), the data output for all four series for 1981-2011 will give us the mean values at the end of each data set (look for lines beginning "#Mean:"). Note that we quote 2011 as the end month, because it is "up to but not including", so the last month is Dec 2010.

Here are the means for 1981-2011 for each source, to 2 decimal places:

GISTEMP 0.35
HADCRUT3 0.26
RSS 0.10
UAH 0.00

Wood for Trees: Notes

So if you're going to compare RSS and GISS on an apples-to-apples basis, on the same plot, you need to offset by 0.25C, which means Blaster's graph should look like this



while the two series certainly zig and zag together there is a statistically insignificant difference in trend, circa 0.002C / decade, easily lost in the noise, but then RSS is also diverging from the other main UAH satellite series.

Yeah. Oh dear.

[EternalOptimist]

#2
The second way they do it.

They have two stations, but they are too far apart to give full coverage. In the absence of the third station, they must take the values from the other two and perform some sort of calculation.
These are calculated values, a critic would say fabricated, a fan would say whatever pj says next (i cant imagine)

All is well and good so far.

but when the calculated values are plotted alone, and compared with the real values alone, they rocket up into catestrophic warming.
A lot of the warming we are panicking about is actually fabricated

un be lieveable

[speling bee]

#2
The second way they do it.

They have two stations, but they are too far apart to give full coverage. In the absence of the third station, they must take the values from the other two and perform some sort of calculation.
These are calculated values, a critic would say fabricated, a fan would say whatever pj says next (i cant imagine)

All is well and good so far.

but when the calculated values are plotted alone, and compared with the real values alone, they rocket up into catestrophic warming.
A lot of the warming we are panicking about is actually fabricated

un be lieveable

Do us a favour and share the peer reviewed research study on this one.

[EternalOptimist]

Do us a favour and share the peer reviewed research study on this one.

you show me the peer reviewed support for their methodology

[speling bee]

you show me the peer reviewed support for their methodology

I haven't posted in favour of a methodology. I'm simply asking about your post.