Eat that GW believers!

[Jccarlton]

Interesting piece from American Thinker on how the AGW models do not fit the data:
http://www.americanthinker.com/2010/01/ ... ts_wh.html
It seems the climate cabal knew it too.

[MSimon]

But I know you guys don't trust any of the records. Without reason, imo, as we see here in D'Alio's case.

Since Hide the Decline I believe it all has to get sorted. And to say concern is unwarranted? Didn't Nixon say that about WaterGate? Yes he did. It took a year and a half to get that taken care of.

And as you well know I have offered to help you bring your concerns to the biggest sceptic blog. I have connections. And I intend to use them. One step at a time. Let us give them time to respond - a day or two more.

As I said. My first concern is honesty. My bias is second. If honesty shows my bias is incorrect I will change it.

[MSimon]

Interesting piece from American Thinker on how the AGW models do not fit the data:
http://www.americanthinker.com/2010/01/ ... ts_wh.html
It seems the climate cabal knew it too.

From the link:

Model results and observed temperature trends are in disagreement in most of the tropical troposphere, being separated by more than twice the uncertainty of the model mean.

In English, that says that the models could not be trusted. This news publicly enraged the gang led by Dr. Jones. They fired off more than 29 e-mails concerning this one paper. But the real story is that these findings did not surprise them. In one of the recently uncovered Climategate e-mails from Dr. Fred Pearce to Dr. Keith Briffa, dated the 13th of October, 1996, Dr. Pearce delivers the bad news that the data does not agree with the models.

The models' error was not, perhaps, too surprising. As Barnett points out, they do not include vital "forcing" mechanisms that alter temperature, such as solar cycles and volcanic eruptions. Nor can they yet mimic the strength of the largest year-on-year variability in the natural system, the El Nino oscillation in the Pacific Ocean, which has a global impact on climate.

This statement means that as far back as 1996, the Jones Gang knew that the GCMs were producing significant errors and problems. This resulted an inability to reconcile the forecasts with reality. They seemingly knew that specifically excluding solar and El Niño influences would cause the forecast to be untrustworthy. But apparently they wished to keep these problems a secret. So to accomplish this, they chose to deal with the problem in a surprising way, as the e-mail further states:

I believe I have brought up that point before. Dismissed by Josh with a wave of "talk to the hand". i.e. the models are mostly right. Why worry about their most important feature?

[Josh Cryer]

Jccarlton,

Something rotten here perhaps? That's sort of what we have been saying all along. That's what a lot of people have been saying.

Well, when you deal with large data sets and you want to get rid of bad values, rather than deleting those values entirely, you can insert random values or a "check" value (I use -1 usually, for non-signed data). If that's why GHCN is all in the 20s then it's a reasonable practice. It should be documented, though.

It's just one sentence. "USHCN values were randomized so they would be caught by quality control."

But I don't know that this is what happened, for sure. I'll get to the bottom of it. Certainly we know that the data isn't "real." And that anyone attempting to do a comparison between USHCN and GHCN (see: Macintyre) cannot claim that GHCN is "fixing the data" unless these arbitrary values are not caught by quality control.

I found on the NCDC site that daily values have better history than monthly values. For Central Park they go back to 1895. Monthly only goes back to 1945.



MSimon,

They write, referring to the troposphere discrepancy:

In English, that says that the models could not be trusted.

In science that says that the models and the data cannot be trusted.

We improve the data a little bit, and it looks like it matches the models. I gave you several links about this, but you dismissed them. We still need more data before we can form this "conclusion." Data on the upper troposphere is sorely inadequate.

Note, I don't know of any model that doesn't include El Nino. They're using dates from 1996 so I can believe that they possibly didn't almost 15 years ago. So I'll let that slide.

[MSimon]

OK. The models and data can't be trusted.

We are now standing on the same base. (Not too close fella).

And given HIDE THE DECLINE. Scientists can't be trusted.

http://market-ticker.denninger.net/arch ... reats.html

[MSimon]

I'm all for improving the data. Can I improve my bank account the same way?

[Josh Cryer]

I'm all for improving the data. Can I improve my bank account the same way?

If you want the overall bank account to have less money (improved data, so far, has resulted in lower numbers, again I concede that the more we know the lower the warming trend might be, but certainly not lower than what we have as the lower bound).

[MSimon]

I'm all for improving the data. Can I improve my bank account the same way?

If you want the overall bank account to have less money (improved data, so far, has resulted in lower numbers, again I concede that the more we know the lower the warming trend might be, but certainly not lower than what we have as the lower bound).

I was hoping to change my account by homogenization.

I want to be homogenized with Bill Gates and Warren Buffet.

[Josh Cryer]

I was hoping to change my account by homogenization.

I want to be homogenized with Bill Gates and Warren Buffet.

Capitalism claims that Bill Gates' and Warren Buffets' support your economy, so in a way that is what is done.

[MSimon]

I was hoping to change my account by homogenization.

I want to be homogenized with Bill Gates and Warren Buffet.

Capitalism claims that Bill Gates' and Warren Buffets' support your economy, so in a way that is what is done. ;)

Ah. So homogenization is a socialist plot. First you homogenize the temps. Then incomes.

[Josh Cryer]

I've done a comprehensive analysis of NCDC and USHCN / GHCN raw. The correlation is exquisite. If there is any "raw data cooking" it goes deeper than USHCN / GHCN, and would be at the NCDC level.

I have a slight problem with how NCDC rounds its mean values. However, any data producer is probably going to use min-max-tobs and do their own mathematical rounding.

NCDC rounds .5 up, so Tmean turns out a bit high (this does not affect NCDC meanTmean, however, since it is averaged over Tmin-Tmax). USHCN / GHCN does not appear to be wholly affected by this. I do think that NCDC should round to even, though, as that is more statistically sound, and data users who do not see that they are rounding up may make a mistake on small data sets. (Larger data sets will not be affected.)

There also do appear to be some station measurements which are not on NCDC CDO but appear in USHCN / GHCN (some stations only go back to 1910 in NCDC whereas in USHCN / GHCN they go back as far as 1895). I'll try to find where those earlier values are. I could be looking at the wrong data sets.

[BenTC]

Josh, I can't follow all the implications of what you're reporting, but just wanted to say I appreciate the effort you're putting in.

[Josh Cryer]

BenTC, thanks, the ultimate goal is to reproduce USHCN and GHCN raw from NCDC (where I have data, anyway, as I said, some stations don't go back as far from the CDO, it could simply be the way it is processed on the web form, I haven't checked).

It will all be open source.

BTW, I set up a site here: http://asseverated.emenace.com/

[MSimon]

I've done a comprehensive analysis of NCDC and USHCN / GHCN raw. The correlation is exquisite. If there is any "raw data cooking" it goes deeper than USHCN / GHCN, and would be at the NCDC level.

I have a slight problem with how NCDC rounds its mean values. However, any data producer is probably going to use min-max-tobs and do their own mathematical rounding.

NCDC rounds .5 up, so Tmean turns out a bit high (this does not affect NCDC meanTmean, however, since it is averaged over Tmin-Tmax). USHCN / GHCN does not appear to be wholly affected by this. I do think that NCDC should round to even, though, as that is more statistically sound, and data users who do not see that they are rounding up may make a mistake on small data sets. (Larger data sets will not be affected.)

There also do appear to be some station measurements which are not on NCDC CDO but appear in USHCN / GHCN (some stations only go back to 1910 in NCDC whereas in USHCN / GHCN they go back as far as 1895). I'll try to find where those earlier values are. I could be looking at the wrong data sets.

But is a Tmin-Tmax average really the average temperature? A time weighted average would be correct.

Hypothetical:

It is 40C for one hour and 20C for 23 hours (a hot wind blew in).

Is the average 30C? or about 20.8C?

I don't trust the ground data. And on top of that the farther back you go the less you know about reality. And with ground data there is the problem of samples. And homogenization to CREATE missing samples.

And in any case min/max has only been done since 1930 or so.

As far as I can tell the data collectors have never been QCed to prove that they are sufficiently accurate to achieve the accuracy claimed for the record.

On top of that we do not know how careful the recorders were for each station. Or the quality of their eye sight. Or how calibrated their eyeballs were for interpolation.

===

Problems with glass thermometers:

http://www.bipm.org/utils/common/pdf/it ... pter15.pdf

Calibration temperatures should be approximately equally spaced, their number depending upon the thermometer range, the accuracy required and the scale graduation interval. Calibration every 50 to 100 scale divisions should give an accuracy within one-half of one division. For linear interpolation between calibration points on a plot of scale correction versus temperature, the uncertainty is about one-half the largest discrepancy found by extrapolating each linear segment to the next calibration point. A thermometer calibration applies as long as the ice-point reading remains the same as during calibration. Subsequent changes in the ice-point reading will result from small changes in the glass of the thermometer bulb which affect its volume. Volume changes in the capillary are minimal by comparison and, as a result, changes in the ice-point reading of the thermometer (taken after not less than 3 days at room temperature) will be accompanied by similar changes in readings at each point along the scale. The ice point should be taken periodically, and scale corrections adjusted as necessary. If the ice point change is too large, the thermometer must be recalibrated. How large a change is "too large" depends upon the application and the type of thermometer . Parallax errors of many tenths of a division may easily occur in reading a liquid-in-glass thermometer. To avoid them one must view the thermometer from exactly the same angle as was used for calibration, virtually always perpendicular to the liquid column. One of the better ways of avoiding parallax and at the same time obtaining a precise reading is to view the thermometer through a telescope which is aligned at the proper angle, or through a magnifying lens attached to the thermometer. In the latter case, lack of parallax is indicated by a straight (as opposed to curved) image of the graduation mark.

Changes in bulb volume also occur because of both irreversible (secular) and reversible (temporary depression) structural changes in the glass that are influenced by time and heat treatment respectively. The secular change is almost always a slow contraction of the bulb, producing an increased thermometer reading. The rate of secular rise decreases with time but increases with exposure of the thermometer to high temperatures. The maximum secular rise is unlikely to exceed 0.1 °C over several years in well-constructed thermometers provided the glass is not heated beyond its exposure limit. The reversible changes appear as a hysteresis on thermal cycling. The bulb expands on exposure to high temperatures and does not return to its original volume immediately on cooling, resulting in a depression of the ice-point reading (and all other readings). Recovery may not be complete for 24 to 72 hours or, if the thermometer is cooled very slowly, no depression may occur. It is to monitor and correct for these changes in bulb volume that ice points are taken before and after calibration, as described earlier. In a good thermometer the temporary depression is small.

Depressing isn't it?

[Josh Cryer]

MSimon,

But is a Tmin-Tmax average really the average temperature? A time weighted average would be correct.

They have hourly going back to the 40s, and it appears that NCDC data is based upon it when those stations have that ability.

Obviously the further back the records go the more uncertainty there is (which is why it is amusing that so many of you trot out really old measurements as a state of fact).

Hypothetical:

It is 40C for one hour and 20C for 23 hours (a hot wind blew in).

Is the average 30C? or about 20.8C?

The monthly average gets rid of such biases. A hot wind isn't going to change the overall trend for a whole month, much less a whole year.

As far as I can tell the data collectors have never been QCed to prove that they are sufficiently accurate to achieve the accuracy claimed for the record.

Spend hours looking over the data. I only found the rounding problem between NCDC and USHCN. Other than that the data appears very cohesive.

If the data were broken in any way I would have noticed it.

On top of that we do not know how careful the recorders were for each station. Or the quality of their eye sight. Or how calibrated their eyeballs were for interpolation.

Doubt, doubt, doubt. At least now you're attacking the weather guys and not the digital archivers.

Problems with glass thermometers:

Good thing that the records only go to one decimal point.