"When my information changes, I change my opinion. What do you do, Sir?"

(The title of this post is a quote from John Maynard Keynes.) Today I want to look at different responses to new information about global warming.

I'll go first: In my archives I found a Usenet post of mine from 11 Aug 1988. In response to a suggestion that global warming was caused by waste heat from power plants, I wrote:

Waste heat does not contribute significantly to global warming. It is all (if it's really happening---we probably won't be sure until its too late) caused by the greenhouse effect. I agree with Brad---burning fossil fuels could well be more harmful to the environment than nuclear power.

Since I wrote that, enough new evidence has been gathered to overcome my scepticism. Man-made global warming is very real.

Next, Ken Parish. In the past, Parish has been sceptical about global warming, with some notable debates with John Quiggin. More recently, Parish has evaluated a test of the theory:

Daly argued that the marked global warming of the last 7 or 8 years was caused by the confluence of a Solar Maximum with 2 large El Nino events in 1998 and 2002, and not by human-generated CO2. Daly went on to say that this could be expected to reverse itself over the period 2003-2006 as we moved towards a Solar Minimum and (likely) La Nina. JQ and I both agreed that this presented a real test of the reality of CO2-induced global warming.

What is the result so far? Well, any objective observer could only conclude that there hasn't been any noticeable cooling at all to date. Have a look at the data for Global Mean Temperature compiled by GISS and kept on the NASA website. They show that average temperatures for the first 4 months of 2004 remained well above the long-term (1951-1980) average of 14°C, by amounts varying between 0.43 and 0.64°C. If Daly had been correct (i.e. that CO2-induced warming was a myth), you would have expected average temperatures to have fallen back to well below the long-term average by now, instead of remaining well above it.

Given the new information, Parish changed his opinion:

I now think we should ratify and seriously implement the Kyoto Protocol without delay (even though it's only a minimalist first step); and enact carbon taxes and an international carbon credits trading scheme designed to create price signals leading to quicker adoption of non-carbon-based energy sources.

Third, Louis Hissink. Hissink claims that the increase in surface temperatures is caused by the urban heat island effect. Following our lengthy discussion in the comments of another post, he finally looked at the GHCN data, broken down into urban, suburban and rural stations. He was able to remove the warming trend from the urban data by subtracting 1.25 degrees from the more recent measurements. After doing this, Hissink found that there was still a warming trend in the rural stations. Some people would then have been forced to form the conclusion that the warming trend was not caused by any urban heat island effect, but Hissink is made of sterner stuff. He writes:

In fact data I posted on Bizarre Science recently showed that for the weather stations in urban and suburban areas showed no rise in temperature for the last 50 years, and while a rise in rural temperatures were measured, we now know that this is a data artefact from poor sampling support, closing of stations, and a bias towards urban stations.

And on his blog:

There is something not quite right here. It is almost as if the data were designed to show there is no such thing as UHI, and that the increasing rural temperatures prove that global warming is happening.

One therefore concludes that there is something not right with the mean temperature data now classed as urban, suburban and rural. One is reminded of mal-odours in Denmark.

If the data contradicts his theories, well, something must be wrong with the data.

Fourth, anti-Kyoto activist John Humphreys. It used to be the case that satellite temperature records disagreed with surface measurements and showed almost no warming or even slight cooling. Global warming sceptics seized on this as evidence that warming was not happening. However, as more data as accumulated and improvements have made in the analysis, the satellite data has come to show warming. The only dispute between the scientists is about how much warming is occurring. Fu et al reckon that it is about 0.2° per decade (the same as the surface data), while Christy and Spencer calculate that the warming is only 0.08° per decade. However, Humphreys, apparently unaware of this, writes:

Satellite and weather balloon estimates of temperature over the past 25 years are consistent with each other and provide us with the best measures of temperature change in the earth's troposphere (the atmosphere within the first 10km of earth). They clearly show that temperatures have decreased 0.1 degrees Celsius since 1980.

His statement is, needless to say, incorrect. In the extensive comments to his post you can see the results of my attempts to persuade him to post a correction. Humphreys accused me over and over again of lying, called me "angry and bitter", wrote "you're too busy hating the enemy", and then denied that he had attacked me personally:

I checked your links... and then checked with some other experts. I found that there is continued uncertainty. My data was not wrong, and it never has been. I will indeed ignore your next link because (a) it probably says what your last ones said; and (b) you have shown you have no credibility or decency.

You have personally attacked others. I have not personally attacked you, and yet you lie once again and say that I have. I have pointed out several of your lies and you have not retracted one of them. You have not been taken out of context and you know it - another lie. Consistent with your character, I expect you'll respond to this post with more lies.

Humphreys steadfastly refused to admit that the satellite data showed warming, writing

as I've already said (and you seem to have ignored), I've checked with my source and it hasn't changed. It still shows a 0.1 degree decrease in temperatures since 1980.

His source? This page, which actually says:

The overall trend in the tropospheric data is near zero, being +0.04 C/decade through Feb 2002.

Now this statement is out of date (with the latest data the warming trend is +0.08), but it does not say that there has been a 0.1 degree decrease since 1980.

At least Hissink noticed that the data contradicted his theory, Humphreys didn't even get that far.

Fifth is our old friend Iain Murray. His tactic in the face of all the new data supporting the global warming theory is not Hissink's contention that the data must be wrong or Humphreys' steadfast refusal to acknowledge its existence. Instead, Murray goes on the attack and accuses the scientists of not changing their minds in the face of new data. In the National Review Online he writes:

Scientists change their minds when data contradicts their models---except in one area, the relatively new scientific discipline known as climatology.

What! The climatologists are the only phony scientists? What about all those biologists that refuse to recognize all the evidence that disproves evolution?

The fact that the satellite data now shows warming? Murray says that proves that those rotten climatologists have been cooking the books:

But atmospheric data from both satellites and weather balloons show only a trifling rise in temperature over the past couple of decades, while the surface temperature has been rising steadily. In 2000, a National Research Council study confirmed the data's discrepancy with the model.

The proper scientific response would be to reexamine the models and adjust them to fit reality. But that hasn't happened in climatology. Instead, there have been repeated attempts to manipulate the satellite data fit the models. Recently, a study published in the journal Nature tries to hammer the square peg of the satellite data into the round hole of the theory, using a method that satellite temperature experts John Christy and Roy Spencer of the University of Alabama at Huntsville had considered and rejected as incorrect in 1991.

Murray neglects to tell his readers that Christy and Spencer's method shows satellite warming of 0.08° per decade. He also has no evidence that Fu et al (the study published in Nature) set out to concoct a result. It's tough being a climate scientist. Get a result that Iain Murray doesn't like, and he'll accuse you of dishonest practices. As for the dispute between Fu and Spencer about how much warming the satellites show, I don't know enough about the science to know who is right and neither does Murray. However, Fu et al's work went through peer review and was published in Nature, while Spencer's criticism was published in Tech Central Station without any peer review, so if I had to bet, my money would be on Fu.

Murray finishes off with this:

Perhaps global warming theory is closer to religion than to science.

Where have I heard that sort of argument before? Oh yes, from the Institute for Creation Research: Evolution is Religion--Not Science.

More like this

For years now global warming skeptics have been using satellite measurements to argue that global warming isn't happening, For example (from 1998): Surface-based temperature records are too few in number and too unevenly spaced to generate accurate global temperature maps. Only 30 percent of the…
The Australian continues to display its contempt for science, scientists and the scientific method. They've published this piece of AGW denial by David Evans. Last time I looked at Evans he was saying that new evidence since 1999 had changed his mind about global warming, with this new evidence…
In my previous post I commented on the various responses of sceptics now that both that satellite and surface record show global warming is happening. (The map below shows the global warming trend for the troposphere from the satellite record.) Scott Church left a…
Last week Kyoto came into effect. Apparently that was the signal for columns by a whole bunch of pundits who have two features in common: 1. they are manifestly ill-equipped to understand the science and 2. they are utterly certain that there is no such thing as global warming. Our…

That sort of Argument by projection seems to be rife these days. Funny, in a way, except so many seem to take their arguments seriously. Anyway, I have followed (to the extent of my limited scientific ability) much of the GW debate and it seems fairly obvious that the consensus is that global warming is a reality, and that the consequences of global warming are misunderstood by almost everyone who pronounces on them . Randomly destabilizing our climate when it is in one of its rare moments of habitable equilibrium just seems like a bad idea. But that's just me.

By bigring55t (not verified) on 11 Aug 2004 #permalink

First, Tim doesn't do me the service of saying that my post was a parody. He knows I am not seriously trying to prove global cooling, but that I was making a broader point about the use of statistics. Therefore, to say: "At least Hissink noticed that the data contradicted his theory, Humphreys didn't even get that far" is not fair, because I am not actually pushing a global cooling "theory".

Second, as can be checked on the link Tim provides, my statistics were in fact correct. If I am correct on this point -- it will be interesting to see whether Tim's correction will be magnanomous or spiteful.

Third, since this post is about views evolving in light of information, Tim might have done me the service of mentioning that my views on this topic have changed as I've learned more. Indeed, I think it likely that there is global warming and that humans have some responsibility.

I wouldn't mind adding that, while Tim has painted me as the nasty one, Tim threw around his fair share of unfriendly comments. While I did have a go at him eventually, I was not trying to establish a bad relationship. Maybe we both could have taken a more constructive approach.

While your post was a parody, that cannot excuse your misrepresentation of the satellite data, since you specifically claimed "the statistics and arguments cited above are true". The intent of your parody seems to have been to slander climate scientists by implying that they cherry pick the data to support global warming.

Your statistics are not correct. Anyone who cares can look at the satellite data and see that it shows warming and not cooling. The page clearly states that there has been warming in the satellite record.

I'm glad to see your concession on the existence of anthrpogenic global warming. Correct your false statement about the satellite record and all will be forgiven.

In fact, you made unprovoked personal attacks quite early in the discussion. For example:

However, from what you and Ken have written it sounds like you are trying to deny the fact that particles in the atmosphere can have the effect of reducing temperatures. I believe you are aware that your comments can be read like this, and have intentionally not clarified yourself on this issue.

Irrespective of whether I explained the process correctly or not (it wouldn't be the first time) -- you know the particles effect exists and you and Ken seem to prefer to try and hide this fact. I accept that you know your science quite well, but you behaviour here raises serious questions about your objectivity.

And notice how you didn't just accuse me of arguing dishonestly, but Ken Miles as well.

I can't see how that is a personal attack. That entire paragraph is specifically focused on what you said and didn't say (and what that says about your objectivity), not what type of person you are. But I'm sorry if it offended you.

Importantly, my position on global warming is not a concession and I stronly object to that comment. I held my view before we started our discussion and you know that. Concession implies that I lose something by changing my mind. That would only be the case if I was wedded to a particular idea. That implication is unnecessary and wrong.

Finally - I've gone back to that website several times yesterday and today (and previously) and it is still showing the same thing. The statistics I quoted are correct. I know what the graph shows, I know what the data shows and I know what the page says... and it does not contradict what I originally said. I said that the temp is lower now than in 1980 -- and that's exactly what the data shows. I'm sorry for repeating myself, but I strongly suggest you take a closer look at the data and maybe correct the record.

Yoooo, Timmy boy.....Australia's BOM cannot even get three months in advance weather predictions right so how can we get 50 years out correct ??


The major scientific paper that attempted to prove global warming over a 600 year period got trashed because the quacks in charge doctored the data sets.......


And here is a good graphic of the before (doctored) data sets and the corrected data sets.


Where would you like to go next ??

By Shaun Bourke (not verified) on 12 Aug 2004 #permalink

Well, if you maintain that your comment was not a personal attack, you won't take it personally when I observe that your previous comment was thoroughly dishonest. Note that I am not saying what sort of person you are, just that you lied in your comment. But please don't take this as personal criticism and accept my apologies for any offence.

The satellite data shows warming since 1980. You can see this if you look at the statement on the page ("+0.04 C/decade through Feb 2002"), the graph on that page, or the table of numbers (which shows warming of +0.082 C/decade to Jun 2004). But you refuse to admit this, once more making your dishonest claim that the data shows cooling since 1980.

You amaze me Tim. The +0.04 that you mention is not between 1980 and now. If you look carefully at the data underlying the graph you will see that the temp in 1980 was higher than the most recent temperature. Exactly what I claimed. I can't work out why you continue to deny this fact, especially when it is so easy to check. The latest reading (June 2004) is -0.021. The equivalent reading from 1980 (June) was +0.119. The temp in 1980 got up to +0.204 and averaged +0.112. All of these numberse are higher than -0.021.

Your mock apology is unnecessary and unhelpful. I was attempting (unsuccessfully as it turns out) to move the discussion to friendlier ground. I also note that you didn't apologise for your "concession" comment.

The point of my suggestion that you comment was dishonest was to demonstrate the problem with your accusation of bad faith and your apology. If you think my comment was unfriendly and my apology was a mock one, then so the same applies to your comment and apology.

My comment about "concession" was intended as praise, but if you object I retract any suggestion that you have given any ground in this discussion.

You claimed that "They clearly show that temperatures have decreased 0.1 degrees Celsius since 1980." Now you try to tell us this (your word) "exactly" the same as saying that the temperature in June 2004 was less than that in June 1980. Perhaps you didn't notice, but June 2004 did not immediately follow June 1980. There were, oh, about 287 months in between. What those wacky scientists do is work out a trend so that those other observations are included. That is on the page you cited and it shows a warming trend of 0.082 degrees per decade. Not that you're likely to admit this.

And if you were comparing the temperature in 1980 with the temperature now, the comparison should be to compare a year with a year rather than a year with a month. Why didn't you tell us how the average temperature in 1980 compares with that in 2004? (And remember that 1980 was cherry-picked because it was unusually warm.)

Woooo there Timmy boy...... Part of your commentary in this thread talks about sat temps at certain levels in the atmosphere so why not open it up to other stratas of the atmosphere as some of the good Drs that work for NASA have....
Your inclusion of zeros as a substitute for non existent data into data sets is scientific fraud.
Since you wish to impune McKitrick due to an association with Tech Central it is therefore incumbent on you to impune all the climate stuff that eminates from the UN because the UN from Annin on down has systematically looted the Iraqi people of tens of billions of dollars through the 'Oil-for-Food-Scam' which clearly questions the overall integrity of the UN.
I for one would like to see at least a doubling of CO2 in the atmosphere because there are only benefits for all.....
The myth that you and your fellow travellers continue to propargate is consicely debunked by Dr Landscheidt....
It is interesting to note that many of the remains of older castles in Europe and Great Britian dating back over 700 years have no fireplaces or they are much later additions.
On the campus of the UNSW there used to be a department of Astrophysics in the Science School. You should take the time to wonder over there with a bible in hand and ask the head of Astrophysics to explain to you two astronomical observations quoted in the bible at Joshua 10:13 and Habakkuk 3:11. When framing your inquiry do it within the context that the observations were factual and accurate.
Since for some decades now the climate myths have alternated from cooling to warming and now I see a creeping back in of cooling it should be incumbent on you lot to at least get your stories straight before dumping fear on the greater population at large.

By Shaun Bourke (not verified) on 13 Aug 2004 #permalink

You seem to have missed the entire point of my post. I was not trying to prove a cooling trend. I was trying to show that it is possible to use correct statistics (and correct theory) to give misleading conclusions. You seem to agree with my point, but for some reason you think that I don't get my own point. You have spent a lot of time trying to get me to admit that my data didn't exist. Now that you realise that it does -- a small apology would be appreciated.

I'm perfectly happy to "admit" (again, biased language) that the available satelite data has shown a (moderate) warming trend. Why wouldn't I? You have no reason for saying that I would deny this and it is offensive. You seem determined to accuse me of all sorts of evils, but you don't seem to even understand what I'm trying to say.

I compared 1980 with the most recent measurement. I could have just as easily started at 1988 or 1991, or picked an end-point of 2000. Likewise, if I'd picked 1979 or 1985 and/or an end-point of 2002 I would have been wrong.

I know what yout point was regarding your mock apology. However, my apology was genuine and you responded with a mock apology which I thought was unnecessary. And now you're implying that if your apology was not genuine, then it's impossible for mine to have been genuine. That is a non sequitor.

The use of the word "concession" has some quite obvious implications. I accept that you may not have considered this previously and did not intend any offense, but I suggest that you consider it for future reference.

Tim and John - if I may interject, I think I know where each of you is coming from. Technically, both of you are right in

that the data do say what each of you are saying. But Tim is more right. Here's why.

The graph you are referring to is not of temperatures, but of temperature anomalies. There's a difference.

What is being shown is deviations from the 25 year average (1979-2004). John is right, in that if you look closely at the

far right end of the graph, there is a tiny slot where the data dip to about -0.21 deg. as he says, and in 1980 it does show

about 0.1 to 0.12 deg. For his -0.1 deg./decade cooling trend, he simply connected the dots between these 2 points - and his

post at libertarian.org.au was a parody, as he claimed. If I understand it right, by choosing the numbers he did, he was

trying to parody discussions of global warming by showing that anything can be proven from the data - just pick your starting

and ending points where you will within wildly fluctuating data and you can "prove" warming or cooling as you please. But

this argument is a misuse of temperature anomaly data. Tropospheric temperature trends, as characterized by NOAA/NASA POES's

(Polar Orbiting Environmental Satellites) varies widely and are not monotonic. They are sensitive to things like ENSO (El

Nino Southern Oscillation), volcanic eruptions (of which there have been at least two during the POES record), daily and

interannual seasonal fluctuations, and other transients. These comprise significant portions of the record for the last 25

years and must be filtered out to get a meaningful long-term trend. This is done not by picking 2 endpoints just

anywhere, but by integrating the entire anomaly curve over the satellite record period. For the data in the graph

used here, doing this leads to a 1979-2001 trend of +0.061 deg. C/decade (the 0.04 deg. K/decade trend mentioned with the

graph was obsolete as of 2003, and the web page is an older one - see Christy et al, 2003). The +0.08 deg. C/decade is the

most current for the 1979-2004 period, as Tim has repeatedly pointed out. Though technically, Mr. Humphreys' numbers are

correct, his "cooling trend", parody or not, can only be derived by treating this temperature anomaly data as

connect-the-dots stuff - which no one who understands where it comes from would do.

But there's more. POES satellite temperature data have been gathered since 1979 using Microwave Sounding Units (MSU's)

carried by these satellites. These are cross-track scanning radiometers that measure upwelling atmospheric radiation on 4

channels operating at 4 frequencies between 50 and 60 GHz, each of which is sensitive to a different layer of the atmosphere.

The relevant channels here are Channel 2 (53.74 GHz - which measures the middle troposphere with peak weighting at about 7

km altitude) and Channel 4 (57.95 GHz - which measures the lower stratosphere). The chart and data you are both referring to

come from the UAH team (University of Alabama, Huntsville - led by John Christy and Roy Spencer) and represent measurements

of MSU Channel TLT. This is a composite channel which focuses on the lower troposphere (peak signal at approx. 4 km

altitude) by differencing nadir (straight down) and off-nadir (side-looking) views. UAH's middle troposphere numbers for

1979-2001, taken directly from Channel 2 are +0.013 deg. K/decade.

But UAH is only one of 4 teams that have analyzed the MSU record. Other teams have avoided using UAH's TLT Channel because

it is subject to high sampling noise and surface emission noise (up to 20 percent over land and 10 percent over oceans) that

seriously degrade the quality of the signal. In addition to UAH, Prabhakara's team obtained a trend of +0.13 deg. K/decade

for the 1980-1999 record (Prabhakara et al, 2000) and the team from Remote Sensing Systems, Santa Rosa, CA (RSS) led by Carl

Mears, Frank Wentz, and Matthias Schabel, obtained a trend of +0.097 deg K/decade for 1979-2001 (Mears et al, 2003). With

few exceptions, global warming skeptics typically avoid discussing these results. Another analysis was done by Vinnikov and

Grody and published in Science last fall. But this analysis is completely different than the ones above and more open to

interpretation, so I won't consider it here.

The differences between these numbers can be traced to how these teams merged the records of different satellites during

their service life overlaps. The complete MSU record up to 2001 was built from some 11 satellites, each with uncertainties

and temperature offsets that differ from the others. These must be joined together, or "merged" into a single continuous

time series which minimizes the errors of each. Part of this process involves characterizing each satellite's hot target

calibration. MSU's calibrate their thermometers on the fly by comparing the temperature of deep space (2.73 deg K) with an

onboard "hot target" which is nothing more than a simple blackbody emitter at a known temperature monitored by 2 platinum

resistance thermocouples. This is how it knows what a deg K is from moment to moment. These hot targets are supposed to be

fixed in temperature, but in fact they drift somewhat as the satellite warms or cools in response to variations in how the

sun illuminates it. This is known as Instrument Body Effect, or IBE. This effect is characterized in merge calculations by

a hot target calibration coefficient (denoted by alpha) that relates the magnitude of the resulting error to the hot target

temperature drift.

The punchline is this - when UAH derived the numbers in the chart you both are referring to, they did so by omitting a few of

the satellite overlap records because they deemed them too short to be reliable - a decision which though it has its

rationale, is somewhat arbitrary. When they did this, they obtained a hot target calibration coefficient for NOAA-9 that is

over 3 times larger than that derived by the other teams and larger than their own figures for the coefficients of the other

satellites in the record. Now the hot targets are simple blackbody emitters and are all identical - there is no reason why

one should over 200 percent larger than the others, and when other analyses include all the data, they get a value for NOAA-9

that is similar to the others - as expected. This high NOAA-9 hot target coefficient stretches the definition of a deg K in

UAH's analysis and lowers their numbers below those of everyone else. Furthermore, when all satellite overlaps are included

in UAH's analysis, their middle troposphere trend shoots up to about 0.08 deg K/decade, approaching that of RSS (Mears et al,

2003). This proves that in fact, over 80 percent of the difference between the low trends given by UAH that you are both

referring to, and those of RSS and Prabhakara are due to their merge and hot target coefficient characterization methodology,

NOT a more accurate measurement of atmospheric temperatures. The remaining differences can be traced to differing methods of

removing diurnal cycle fluctuations (drifts in satellite local equatorial crossing times which create spurious waqrming or

cooling) and data smoothing methods (UAH use 60 to 120 day smoothing and RSS use 5-day, or pentad, smooting to allow

characterization of shorter satellite record overlaps).

As for radiosondes, or "weather balloons" as Mr. Humphreys calls them, they do not provide the independent support he claims.

The trends given above (+0.01 deg K/decade, +0.097 deg K/decade) are global average trends. These mask a large amount of

regional variation. The large majority of radiospnde stations worldwide since the 50's have emphasized land locations in the

northern hemisphere. Here, the UAH, RSS, and Prabhakara analyses agree quite well with each other, and the radiosondde data

vindicates ALL of them. The lower numbers of the UAH team result from lower trends in the tropics and southern hemisphere.

In these regions, the radiosonde data is quite sparse, and what there is typically displays variance that is as large as the

trend being measured. Also, sonde data must be massaged in a number of ways to make it comparable to the MSU record before a

meaningful comparison can be made (Sondes measure specific temperatures as a function of altitude whereas MSU's measure the

bulk temperature of atmospheric layers by Channel). How this is done is problematic, and often not independent of the MSU

data it is supposed to be an "independent" check of (NRC, 2000; Angell, 2003; Seidel et al, 2003).

Lastly, there is the work of Fu and his colleagues that you both refer to. Fu et al did not do an independent analysis of

the MSU record. What they did was demonstrate numerically that the MSU Channel 2 trends (middle troposphere), as derived by

the 3 teams mentioned above, have been contaminated by lower stratospheric trends. MSU Channel 2 is weighted most heavily

toward the middle troposphere as noted. But it receives a significant portion of its signal (up to 20 percent) from the

lower stratosphere. Now the lower stratosphere is known to have cooled dramatically over the last 25 years due to ozone

depletion (another man-made problem that global warming skeptics also try to rationalize away). Since MSU Channel 2 is

partially detecting this as well as the middle troposphere warming, it will under-represent that warming somewhat. Fu and

his colleagues used the Channel 4 stratosphere data, which is more uniform and better characterized, to derive an effective

Channel 2 weighting function - part real and part virtual. The real portion represents the lower to middle troposphere (the

850 to 300 HPa layer - HPa being hectapascals, as it is common to denote altitude with pressures in climate change science

parlance). The virtual part is a negative function which uses Channel 4 to derive the stratospheric contribution to the

Channel 2 trend and remove it, leaving only the pure lower to middle troposphere warming rate - which is what everyone is

after. When this was done, they found that the stratospheric cooling removed up to 0.08 deg K/decade from the real trend,

which bump the RSS numbers up to +0.18 deg K/decade and the Prabhakara numbers up to +0.21 deg K/decade. These numbers are

well within the reasonable range of those predicted by general circulation models of the atmosphere and oceans, and theory,

and essentially resolves the "cooling troposphere" dilemma of the 90's - to the consternation of global warming skeptics.

As for Spencer's criticism of this work at Tech Central Station, it was based on a complete misunderstanding of Fu's work.

Spencer claimed that Fu overestimated the warming because his MSU weighting function went negative above 100 HPa. Though he

rightly points out that a radiation signal can't be negative (so a weighting function can't be) he fails to notice that Fu's

weighting function above 100 HPa was virtual, NOT real - because it was designed to remove a stratospheric contribution that

was spurious to the troposphere trend, not characterize an actual one. That is why it went negative. Had Spencer read Fu's

paper more carefully he would have seen that Fu specifically stated that his function was not real, and he clearly

discriminated it from a real one, which must be positive everywhere (Fu et al, 2004, Pg. 56).

Thus, the real lower to middle tropospher trend appears to be around +0.10 to +0.13 deg K/decade uncorrected for

stratospheric cooling, and +0.18 to 0.21 deg K/decade with this accounted for - exactly as predicted by theory and the latest

models. Due to the problems I've described above, few people today believe UAH's lower and middle tropopshere trends. The

only exceptions to this are global warming skeptics, who prefer them because they yield the desired low warming that their

belief system requires. It is not surprising that they typically avoid discussing the work of RSS and other groups, and the

few times they do, they are dismissive of it without being particularly clear as to why. Outside of ultra-conservative and

industry special iterests, the claim of satellite measured atmospheric cooling is a colossal red herring. It is sad to see so

many cling to a sinking ship for reasons that are entirely ideological rather than scientific.


Angell J.K. 2003. Effect of exclusion of anomalous tropical stations on temperature trends from a 63-station radiosonce

network, and comparison with other analyses. J. Climate, No. 16, 2288-2295.

Christy, J.R., R.W. Spencer, W.B. Norris, and W.D. Braswell. 2003: Error estimates of Version 5.0 of MSU-AMSU bulk

atmospheric temperatures. J. Atmos. And Oc. Tech., No. 20, 613-629.

Fu, Q., C.M. Johanson, S.G. Warren, D.J. Seidel. 2004: Contribution of stratospheric cooling to satellite-inferred

tropospheric temperature trends. Nature, No. 429, (6987), 55-58.

Mears, C.A., Schabel, M.C., Wentz, F.J. 2003: A Reanalysis of the MSU Channel 2 Tropospheric Temperature Record.

J.Climate, No. 16 (22), 3650-3664.

National Research Council (NRC). 2000. Reconciling Observations of Global Temperature Change. Wallace, J.M. et. al.

(eds.). National Academy Press. Available online here.

Seidel, D.J., J. Angell, J. Christy, M. Free, S. Klein, J. Lanzante, C. Mears, D. Parker, M. Schabel, R. Spencer, A. Sterin,

P. Thorne, and F. Wentz. 2003. Intercomparison of Global Upper-Air Temperature Datasets from Radiosondes and Satellites.

AMS Annual Meeting, 2003.

Thanks for your post. I must apologise that I haven't read the whole thing yet (I promise I will later), but I just wanted to note a few quick things. First, I probably should have clarified my intent. I wasn't trying to show "that anything can be proven from the data". Simply that people need to be careful about their use of statistics (and incomplete theories) because it's easy to interpret the same information in several different ways.

I have never denied the trend in the global warming data. Indeed, I specifically noted the global warming trend for ground temperatures in my original parody.

And if I may be allowed to think out loud... I wonder what the satellite trend would be if they used 1980 instead of 1979 as a starting point? I'm not saying it would be negative, but it would be interesting to see what difference that made?

As to who was "more right", it needs to be remembered what was being debated. I was never seriously attempting to show a "cooling trend" and this is well known. So the fact that there is a warming trend does not make me wrong. The issue of debate was whether I was "wilfully ignorant" in using statistics that are just plain wrong. I don't believe I was, and I don't think Tim's attack on me in the above post is appropriate or fair.

Thanks John. You are right to assert that we all need to be careful about our use of statistics. In my experience, most of the really good climate change research being done today (what I've seen of it anyway) is pretty careful about this. There are exceptions though. About 1980 vs. 1979, we can get an idea of the impact of starting our trend then by looking at the graph we've been discussing. As I mentioned, the trend will come from an integration of the anomaly curve it shows - that is, from

(∫ t f(t) dt)/(∫ dt)

(I wish it were easier to show math in blogs!). So the impact of removing 1979 will vary as the area under the curve for that year ratioed over the total integrated area. An examination of the curve reveals that this will be noticeable, but slight. And it could equally go the other way at the far end (2005). As a matter of fact, Prabhakara et al (2000) did their trend estimate for 1980 - 1999 and obtained a tropospheric trend of 0.13 deg K/decade - higher than either UAH or RSS. There were other differences in their merge and target factor characterizations, so it's not apples and apples to the curve we've been using. But it illustrates the point that, properly done, a year or two here or there doesn't impact things as much. That's the beauty of using an integrated anomaly curve for trend estimates - it acts as a kind of low-pass filter on short-term noise.

Also, bear in mind that the curve we've been using represents (I believe) UAH Version 5.0, which though not without its strongpoints, has issues. These are being debated by many. For comparison, see the latest RSS global 1979 - 2003 trend plot (a global color map rather than a trend graph) at http://www.ssmi.com/msu/msu_data_description.html#figures. The figures to note are those for Channels 2 and 4 (troposphere and lower stratosphere - Channel 3 is "polluted" by the tropopause, and not as useful). In my personal opinion, which is shared by many climate scientists today, RSS's target factor characterization and merge methodology are on more solid ground. Time will tell.

I appeciate you taking the time to discuss these issues. Unlike you -- I'm quite happy that maths isn't easier to show on blogs. :)

One of my favourite thought experiments is to consider how current life will be written up in the history books of the future. As you say -- time will tell. My money is on the continued decline in the prices of renewable energies, making this whole debate largely academic. Something like Malthus v the Green revolution in the 19th century maybe? But that is a different debate!

A side note -- I still think Tim's attack on me was inappropriate an unfair.

Scott, you can get an integral sign by typing ∫ - I put them in your post.

I already told John what the result was if you started in 1980, but he refused to correct his statement. I don't think anyone could reasonably interpret his statement as being about comparing two months instead of being about trends.

John is being disingenuous when he says that his intent was to show that people should be careful with their statistics. He implied that the scientists arguing for the existence of global warming were being selective with their statistics. That was certainly how his post was interpreted by readers. See here for example. John did say that his data was selective, but he also said that his statistics were true. And he wrote:

"For the record, I do not believe we are facing global cooling, and I accept that there has been an increase in surface temperatures."

implying that the satellite record showed cooling while the surface record showed warming.

Now this could have been accidental, but when I asked him to make a correction he refused and repeatedly accused me of being dishonest. All he has to do to prove his good faith is make a post on his blog correctly stating what the satellite record shows. If he does, I'll be happy to retract and apologize.

Oh, and John, if you think that the prices of renewable energies are going to keep decining, then Kyoto won't cost anything, so why are you so opposed to it?

I didn't say I thought that the price of renewables was going to be competitive tomorrow. Also, I don't endorse the idea of introducing a government policy because it will have no impact.

Of course I haven't corrected my statement Tim - because what I said wasn't wrong, as Scott and I have both pointed out. The latest temp is lower than the 1980 temp. While I didn't specifically say "the latest temp" in my original post, I thought it would be fairly obvious (I meant it this way, it seems Scott took it this way, as have other readers).

I think it fairly clear now that your attacks on me were inappropriate and unfair. You imply that I don't change my mind when the facts change. I do. You imply that I don't admit certain facts. I do. You seem to liken me with creationists (which I think was intended as an insult and I took it as one). You neglect to point out that my post was a parody. You accuse me of wilfull ignorance and have had a number of other pot shots at me. Your post clearly gives an inappropriate impression of my views. And when it turns out that I was correct you refuse to adjust the record or offer any apology.

All you have to do to prove that my assessment of your character was wrong is make a post on your blog correctly stating what the satellite record shows. But you won't.

The important thing to note here is that the trend, and our contributions to it, are the real issue - and a proper characterization of this is not obtained by comparing the global average temperature at one point (1980 for instance) with the present. So discussions about whether or not we were right about the temperature in 1979 vs. now is off the subject (though I don't mean to slight discussions about someone's motives). And yes, a correct presentation of the full satellite record and resulting trends by all 4 analyzing teams to date, would clear the air. BTW - I got to thinking about it yesterday, and I realized that the equation I gave in my last post was incorrect. Sorry 'bout that! Actually, I failed to take my own advice - it's a temperature anomaly curve, not a temperature curve. So the equation I gave you would actually work out to zero because it's based on the anomaly with respect to the average temperature over the period in question (in this case, the last 25 years). What is needed is some sort of running average that smooths the short-term fluctuations out of the curve, leaving behind the long-term trend. The period of smoothing, δt about any time t would have to be long enough to cover all extraneous phenomena that hide the long-term trend under (in the case of the troposphere, this would be things like volcanic eruptions, el nino's, etc.), but short enough to let the long-term one (if there was one) through. The larger a part of the overall period these shorter-term fluctuations are - or to put it another way, the less monotonic the curve is (that is, the more frequently it changes from upward trending to downward trending and vice versa), the less meaningful it is to speak of a real "long-term trend" for it. El nino's and the contributions of Mt. Pinatubo and El Chicon, for instance, dominate the tropospheric temperature signal for more than a quarter of the last 25 years, so it is not at all clear that the trends we're getting are even meaningful yet. One last thought. All of this so far is qualitative - just a theoretical description of why data "smoothing" is important and we can't use arbitrary "snapshots". With real data you will have discrete measurements that have some scatter to them. For this, of course, we wouldn't know the function ahead of time. So in practice, we would use a least squares method to get the underlying trend by finding the best linear fit to the data points (i.e. the one with the smallest sum of the squared "residuals", or deviations of each point from the line). This is what each of the teams analyzing the MSU record did. UAH based their trend on 60 to 120 day smoothing and RSS based theirs on "pentad", or 5-day smoothing. Both methods do let through El nino's and volcanic eruptions though, and this is part of the debate. This comment was edited on Oct 10.

Scott, the 4 teams got together last October and discussed what you posted above. You can see the results here.

As I said on Chris Mooney's page, good work sir.



Neither the formula

(∫ t f(t) dt)/(∫ dt)

nor the formula

[1/δt] * ∫ f(t)

is quite right. The correct formula is (if I've got the constant right)

12/(δt) * ∫ (t-m) * T(t) dt,

where t denotes time, T denotes temperature, m is the midpoint of the time interval (over which the integral is taken), and δt is the time interval's length. With this formula, it doesn't matter whether the input data T(t) is the temperature or the temperature anomaly.

By the way, "hectopascals"? What's wrong with calling them millibars? That wouldn't be pompous and obscure enough? Not easy enough to misspell?

By Norman Yarvin (not verified) on 30 Aug 2004 #permalink

Uh, Scott, it's you who are misspelling the metric prefix "hecto", meaning 10^2. I found this out when looking up the silly thing in the Unix program "units", which didn't recognize "hectapascal", and confirmed it by doing a web search (6000+ Google hits for "hectopascal" vs <50 hits for "hectapascal"). As for the legitimacy of the stupid unit, see the one of the first of those Google hits:


which confirms that (as you say) it's a popular unit, and also that (as I say) it's silly; it's just a pompous renaming of the old unit "millibar".

As regards the formulas, "temperature anomaly" means the difference between the temperature and some constant reference temperature; the latter is not always the average temperature over the period being considered -- if for no other reason than that one might sometimes compute the trends over subperiods of the tabulated data, as indeed was the context for which you offered up the formula. My formula implicitly subtracts off the average temperature over the interval of integration. It's also normalized well: my formula, if fed a linear increase of X degrees over Y years (T(t)=t*(X/Y)+T0; period of integration t0 to t0+Y), will always output the number X, whatever X, Y, T0, and t0 are. You can easily check this, if you know how to integrate. You can also easily check that your formula won't work for this simplest of possible trends, even if (as you say must be done) you subtract off the average temperature before feeding the temperature curve to your formula. Indeed, if you subtract off the average temperature first, then the integral of the resulting temperature curve is zero, which is the result your formula will return.

By Norman Yarvin (not verified) on 31 Aug 2004 #permalink

"Hecta-" is, according to your source, a "Greek prefix"; "hecto-" is, according to the same source, a prefix from the SI system of units. The "pascal" is part of the SI system of units; thus "hectopascal" is appropriate, and "hectapascal" isn't.

As regards the formulas, my formula wasn't derived to "appear" right, but to produce correct results. Evidently you doubt it, but can't disprove it; so let me show you how one disproves a formula. It's most easily done by finding a counterexample. Take your formula for instance; it's obvious to me that it always returns zero when used as you dictate, but let's pretend I didn't know that. I'll just take some random trend, feed it into your formula, and see what it spits out. How about the simplest possible trend: T(t)=t, between the times t=0 and t=1; a straight line from T(0)=0 to T(1)=1. The answer should obviously be one: that's the slope of the line, and it's also the total change; you haven't said which of those two your formula returns, but it ought to be one or the other of them, and in this case they're the same number: one. Now, according to you, the mean of this line should be zero before feeding it to your formula; as it is, the mean is 1/2. So first we'll subtract off the mean, getting T(t)=t-1/2. Then we feed it into your formula. First we compute

∫ T(t) dt = ∫ (t-1/2) dt = (1/2)*t - (1/2)*t.

Taken between t=0 and t=1, this is

1/2 - 1/2 - (0 - 0)

which is zero. Then we divide by the square of the length of the interval, which is one; this of course doesn't change anything, so the result is still zero. Your formula returns zero, when it should return one. That means it's wrong. If we hadn't subtracted off that mean (1/2), it'd return 1/2, which would still be wrong.

That's how one disproves a formula. Now just try to do the same thing to mine. After all, it's got that crazy constant 12 in front of it, so it must be wrong; plug in some numbers and prove it. Rotsa ruck...

By Norman Yarvin (not verified) on 01 Sep 2004 #permalink

Bit late but... hecto not hecta, of course, but slightly more subtly its "hPa" not "HPa".

Civility is nice, as is accepting a correction in good faith.

As the the equations-for-trend, I'm baffled: you are of course both wrong: the trend is found by least-squares fitting.

Tim Lambert writes, "Fifth is our old friend Iain Murray."

Heh, heh, heh!

And what about our old friend Tim Lambert? Is he still saying (regarding the temperature of the atmosphere and its internal energy):

"In fact, Temperature T and internal energy U are related by the formula
where m is the mass and c the specific heat. It is true that it is possible for internal energy to change without affecting the temperature if there is a phase change, but the atmosphere stays way above the temperature of liquid nitrogen, so this makes almost no difference to temperatures."


"I mentioned nitrogen because that is what the atmosphere mostly is. Water vapour is less than 0.5% of the atmosphere. And the equation is true for water vapour as well if there is no phase change. The equation I gave is actually a very (close) approximation. Do you also complain that Newtonain physics is the wrong way to describe the atmosphere because it doesn't account for relativistic effects?"

Or does good ol' Tim now admit that his magical little equation is not as accurate in describing the relationship between atmospheric temperature and atmospheric internal energy as is using Newtonian physics (i.e. neglecting relativistic effects)?



Your old friend,
Mark Bahner

P.S. Tim, ol' buddy, I've been thinking...I really am desperate to give you that $20 I offered on my blog. Why?

Because I LIKE YOU, that's why! :-)

(Despite your snide "Newtonian physics" comment, which implied that I don't know what I'm talking about.)

So I'm extending the time limit for my offer. It was 3 months, expiring August 28. But--again, because you're such a good friend, and because I really, really want to give away that money--I'm extending the offer to October 1, 2004.

P.P.S. But please don't consult with Bob Maginnis on this matter. (I don't want to make this ***too*** easy for you!) ;-)

Tim Lambert wrote, "Mark, I'm not interested."

Yes...G@d forbid that you might ever change your opinions based on new information or further thought, eh, Tim? ;-)

"Why don't you attempt to answer your own questions?"

I will. But because of my new offer, I have to wait until after October 1. ;-) (Even though I really, really like you, I wouldn't want you to pick up the $20 without *any* effort at all. :-))

In fact (again, just because we're such good buddies)I'll release my request that you don't consult with Bob Maginnis. And even if you don't do that, you might want to read a few of his posts on your blog. The answers are there. (<----Hint, hint, hint.)


Mark (your very bestest friend) ;-)

Tim Lambert writes, "Take as long as you like, Mark."

Hmmmm...? OK. I gotta admit, though, it's a little puzzling to me how little interest you appear to have in discovering the truth in this matter. You're at a university, right? I thought universities were "into" that sort of thing...

I'm surprised you wouldn't run your assertions about the relationship between atmospheric temperature and the internal energy of the atmosphere by someone at your university who just might (possibly, just possibly!) have more expertise in the matter than you. Though it's pretty obvious from your comments:

"Wow. I guess we'll just have to ditch the entire field of thermodynamics then. In fact, Temperature T and internal energy U are related by the formula..."

...that you're an expert on the subject. (BTW, aren't those thermog@damnics classes a bitch? Not only that, "When will we ever use this stuff?!" ;-))

Not to mention your snide remarks (er, clever comments):

"The equation I gave is actually a very (close) approximation. Do you also complain that Newtonain physics is the wrong way to describe the atmosphere because it doesn't account for relativistic effects?"

So I'm puzzled. In fact, I'm so puzzled I can't even remember what post of yours started me on this train of thought. Was the title of your post, "Pot calls kettle black?" Hmmm...no. Something about information changing...and...changing opinions...?

Mark (your better-than-bestest friend) ;-)

P.S. As you may have heard, Hurricane Ivan struck the Alabama/Louisiana/Florida coast a few days ago. It's currently dumping quite a bit of rain right here in North Carolina, as a matter of fact. Curiously, there were no reports of a heat wave along the Alabama/Louisiana/Florida coast when Hurricane Ivan hit. A bit puzzling, eh? ;-)

P.P.S. Memo to Self: Stop giving Tim Lambert hints!!! ;-)

Tim Lambert writes, "As for the dispute between Fu and Spencer about how much warming the satellites show, I don't know enough about the science to know who is right and neither does Murray. However, Fu et al's work went through peer review and was published in Nature, while Spencer's criticism was published in Tech Central Station..."

Yes, Tim, that was exactly Spencer's point. Why did Nature publish Fu's work without asking Spencer or Christy to review it?

Was it because Nature is a biased pseudoscience magazine?

Or can you think of some legitimate reason why Nature wouldn't ask two leading experts on the subject to review the paper prior to publication?


(Sorry for the delay in responding; I've been away.)

The formula I gave is the formula for a least-squares fit -- the least-squares fit to a polynomial of order one (a straight line). In particular, if the polynomial is denoted by A*t+B, the formula I gave yields the coefficient A of the least-squares fit. (Well, to be precise, it yields A multiplied by the length of the interval; the latter has to be divided off to get A itself -- A being, in this context, the temperature increase per year, whereas my formula gives the total temperature increase.) Thus the formula I gave is, I think, the one generally used in this situation, although of course it will not be written the same way by everyone; and indeed people who just tell their statistics package "give me a least-squares fit" will never see the formula at all, it being hidden inside the package. But it's easy to rederive if you know the basic theory of orthogonal polynomials, so that's what I did.

As regards civility, if anything I've been overly civil. Scott, this exchange has established beyond doubt that you are a mathematical ignoramus. You can't just make up this stuff; it's a rigorous discipline. You clearly didn't even bother to subject your latest formula to the test I illustrated how to perform, of plugging some numbers into it and testing the sanity of the result. As a result it's as pathetic as your earlier attempts. If I sound casual about this math, it's only from long and deep familiarity with it -- and even so, I double-check what I write. This isn't "hate" or "poison"; it's just the only way to write something correct. The business of temperature trends and satellite temperature sensing is complicated and delicate; if you wing it, you flub it -- which is why I haven't offered any comment on those larger issues: I don't know enough to make any interesting comments with anything like certainty, and I'm not willing to wing it, because I know that if I did I'd flub it. To act otherwise isn't "productive"; it just amounts to erecting an edifice of error, which has to be torn down before something useful can be built.

Of course even a mathematical ignoramus can tell the truth about a mathematical subject, if he's repeating the words of people who know what they are talking about; but one constantly has to wonder whether something important was lost in rephrasing.

By Norman Yarvin (not verified) on 27 Sep 2004 #permalink