Physicists don't believe in global warming anymore?

intention · July 25, 2008, 9:24pm

tomndebb:

I note, however, that the confusion between the various numbers has already been explained in the poll citation. When the question is asked of “scientists,” (including physicists, biologists, chemists, astronomers, and any number of other folks, then the results come out as one in four have not yet been persuaded of AGW. However, when a similar question is asked of “climate scientists,” then the number of persuaded individuals jump to over 80%, with some portion of the less than 20% who are not fully persuaded including those who lean toward that explanation but may be holding out for a bit more data points. Given a base line poll from a (presumably) neutral pollster, then the APS would seem to have already met the conditions that would allow them to avoid publishing “personal” beliefs in the name of their membership.

tomndebb, I’m not clear what you mean by this. Obviously, the editor of the APS journal doesn’t think that there is unanimity among the APS membership’s opinion on climate change, that’s his stated reason for bringing the debate to the fore by publishing.

Given that acknowledged lack of unanimity within APS, how would a poll of some other group meet “the conditions that would allow [the APS leadership] to avoid publishing “personal” beliefs in the name of their membership.”

All the best,

w.

PS - I didn’t understand your distinction above between “climate scientsts” and “scientists”, as the citation seems to use the two terms interchangeably. And the citation says:

jshore · July 25, 2008, 9:47pm

intention: Just to go you a few examples -

Table 9.1 looks at estimates of aerosol forcings obtained by data from Upper air, surface and deep ocean space-time temperature data, global mean and hemispheric difference insurface air temperature data, and global mean ocean heat uptake data. Fig 9.3 looks at satellite data measuring the shortwave flux radiation anomalies for the Mt. Pinatubo eruption. Section 9.4.1.2 looks at global temperatures during the 20th century as well as the geographic pattern of such changes (e.g., Fig. 9.6). Figures 9.7 and 9.8 look at the variability as a function of timescale of global mean temperatures and of continental scale temperatures. Fig. 9.14 shows re-analysis data of tropopause height. Section 9.5.1.1 looks at data for ocean heat content changes for the various ocean basins. Section 9.5.2 looks at data for sea level changes. Figure 9.17 looks at column-integrated water vapor data. Fig. 9.19 looks at data for rainfall in the Sahel. Section 9.6 discusses the use of observational constraints to determine climate sensitivity, including 20th century data of various sorts (such as global temperature, hemispheric difference in surface air temperature, and global ocean heat uptake), major volcanic eruptions, Northern Hemisphere temperatures from 1270-1850 from temperature reconstructions, and data from the last glacial maximum (LGM), i.e., the last ice age.

Anyways, I am bored of listing all of this so you can go find other stuff yourself. The point is that there is plenty of data out there that is used. As in all sciences, the observational data is used to build, test, and constrain models and, as in all sciences over the last half century, computers are used to do much of the modeling. Computer models are also used in interpretation of data, e.g., to convert between observational data and the equilibrium climate sensitivities implied by that data (since much of the data is over short enough time intervals that one is not very close to “equilibrium”…although the LGM data is an exception).

If you don’t want to deal with there being computer models in science, I suggest that the easiest thing for you to do is to time-travel back at least 50 years. The endless harping about computer models by the so-called “skeptics” like yourself is really getting rather boring and silly if you ask me.

jshore · July 26, 2008, 1:36am

By the way, this interview with one of the two editors of the newsletter, Al Saperstein, is interesting. Among the highlights:

“Maybe I am unduly naive, but it seems the issue is not as clear cut as whether the Earth is flat or not. We’ve been getting comments saying ‘Now that you’ve published this will you publish my paper showing the Earth is flat?’ In my view, not being an expert, they are not in the same class,” says Saperstein.

Yet, in hindsight, Saperstein admits that the newsletter might have been better off focusing on the impacts of climate change, not the causes. He also regrets the statement at the front of the newsletter which reads “There is a considerable presence within the scientific community of people who do not agree with the IPCC conclusion”.

“It is clear that we made one error in my colleague’s introduction. In retrospect, the word ‘considerable’ should have been dropped.”

sleestak · July 26, 2008, 1:39am

Well, the question regarding computer models is obvious: are the computer models accurate? The models apparently do not back test well which seems like it ought to be a giant flag that something isn’t right.

The only information I can find on back testing the models show that they suck . Do you have other studies showing that the models, when back tested, give accurate numbers? If not, why should we trust the models when they do not back test well?

Slee

jshore · July 26, 2008, 2:18am

That is hardly a credible source, to say the least. It contains the usual lies, deceptions, and outdated information that we have already debunked countless times before.

To some extent, the chapter that I referenced in the IPCC report talks of back-testing…at least in the sense of comparing the observations with that predicted by the models with the forcings as they are understood. (There are certain uncertainties in this, of course, one being the uncertainty in aerosol forcing.)

The more relevant chapter to read is the one before it, Chapter 8.

GIGObuster · July 26, 2008, 2:29am

I have to say it is your link that sucks, in the context of where we are with the OP, almost none of the cites are from serious scientific sites (many of the links are broken also) but are coming from skeptic blogs with points that are already debunked.

As I have seen, the consensus among scientists, physicists and educators is that computer models are very good tools to use:

Checking the history of computer modeling in relation to climate science it is clear to me that most of the criticism was somewhat valid 10 or more years ago, but climate models are using less of the “invalidating” tune ups that they were accused of using. And the worst thing for the skeptics is that most of that tune ups were found by the new models to had been good educated and evidence based “fudges”

http://www.aip.org/history/climate/co2.htm

By now there were a dozen teams around the world using computers to integrate every advance in observation or theory. As the 21st century arrived, the growing agreement among the rival teams, and the consistency of their models’ results with many different kinds of observations, became overwhelmingly convincing. Scarcely any reputable expert now doubted that CO2 and other greenhouse gases were at least partly responsible for the unprecedented warming all around the world since the 1980s. A final nail in the skeptics’ coffin came in 2005, when a team compared computer calculations with long-term measurements of temperatures in the world’s ocean basins (it was not in the air but the massive oceans, after all, that most of any heat added soon wound up). In each separate ocean basin, they showed a close match between observations of rising temperatures at particular depths, and calculations of where the greenhouse effect should appear. This was telling evidence that the computer models were on the right track. Nothing but greenhouse gases could produce the observed ocean warming — and other changes that were now showing up in many parts of the world, as predicted.

The computations pinned down an imbalance. The Earth was now taking in from sunlight nearly a watt per square meter more than it was radiating back into space, averaged over the planet’s entire surface. That was enough energy to cause truly serious effects if it continued. James Hansen, leader of one of the studies, called it “smoking gun” proof of greenhouse effect warming.(56)

Yet amid all the uncertainties about how carbon cycles operated, how much could we trust the computer models? Scientists are more likely to believe something if they can confirm it with an entirely independent line of evidence, preferably from somewhere nobody had looked before. Just such new evidence came up in the 1990s, thanks to an unexpected alliance of paleontology and plant physiology. Studies of plant species that had changed little since the rise of the dinosaurs (magnolia for one) showed that if you exposed them to a higher level of CO2, the structure of their leaves changed. Ancient fossil leaves showed just such changes. Several kinds of chemical studies confirmed that the level of the gas had swung widely over geological ages, and the temperature too.

Eventually geochemists and their allies managed to get numbers for the “climate sensitivity” in ancient eras, that is, the response of temperature to a rise in the CO2 level. Over hundreds of millions of years, a doubled level of the gas had always gone along with a temperature rise of three degrees, give or take a couple of degrees. That agreed almost exactly with the numbers coming from many computer studies.

http://www.aip.org/history/climate/GCM.htm

Another persistent problem was the poor quality of models that included the oceans. Modelers had long understood that reliable climate prediction would require coupling an atmospheric GCM to a full-scale ocean model. Such coupled models were beginning to dominate attention. They all tended to drift over time into unrealistic patterns. In particular, models seemed flatly unable to keep the thermohaline circulation going. The only solution was to tune the models to match real-world conditions by adjusting various parameters. The simplest method, used for instance by Suki Manabe in his influential global warming computations, was to fiddle with the flux of heat at the interface between ocean and atmosphere. As the model began to drift away from reality, it was telling him (as he explained), “Oh, Suki, I need this much heat here.” And he would put heat into the ocean or take it away as needed to keep the results stable. Modelers would likewise force transfers of water and so forth, formally violating the laws of physics to compensate for their models’ deficiencies.(95a)
The workers who used this technique argued that it was fair play for finding the effects of greenhouse gases, so long as they imposed the same numbers when they ran their model for different gas levels. Some other modelers roundly criticized flux adjustments as “fudge factors” that could bring whatever results the modelers sought. A few scientists who were skeptical about global warming brought the criticism into public view, arguing that GCMs were too faulty to prove that action must be taken on greenhouse gases. If a model was tuned to match the present climate, why believe it could tell us anything at all about a changed situation? The argument was too technical to attract much public attention. The modelers themselves, reluctant to give ammunition to critics of their enterprise, mostly carried on the debate privately with their colleagues.

Around 1998, different groups published crudely consistent simulations of the ice age climate based on the full armament of coupled ocean-atmosphere models. This was plainly a landmark, showing that the models were not so elaborately adjusted that they could work only for a climate resembling the present one. The work called for a variety of ingenious methods, along with brute force — one group ran its model on a supercomputer for more than a year.(96*) Later studies not only reproduced the ice age climate as well as it could be measured, but got an excellent match to another period when the weather patterns had been different from now, the warm “mid-Holocene” about 6000 years ago.

Better still, by 1999 a couple of computer groups simulating the present climate managed to do away altogether with flux adjustments while running their models through centuries. Their results had reasonable seasonal cycles and so forth, not severely different from the results of the earlier flux-adjusted models. Evidently the tuning had not been a fatal cheat. Models without flux adjustments soon became standard.(97*)

But for a climate more or less like the present, the factors that modelers failed to understand seemed to have gotten incorporated somehow into the parameters. For the models did produce reasonable weather patterns for such different conditions as day and night, summer and winter, the effects of volcanic eruptions and so forth. At worst, the models were somehow all getting right results for wrong reasons—flaws which would only show up after greenhouse gases pushed the climate beyond any conditions that the models were designed to reproduce. If there were such deep-set flaws, that did not mean, as some critics implied, that there was no need to worry about global warming. As other experts pointed out, if the models were faulty, the future climate changes could be worse than they predicted, not better.

Those who still denied there was a serious risk of climate change could not reasonably dismiss computer modeling in general. That would throw away much of the past few decades’ work in many fields of science and engineering, and even key business practices. The challenge, then, was to produce a simulation that did not show global warming. It should have been easy to try, for by the 1990s, millions of people owned personal computers more powerful than anything commanded by the climate modelers of the 1970s. But no matter how people fiddled with climate models, whether simple one- or two-dimensional models or full-scale GCMs, the answer was the same. If your model could reproduce something resembling the present climate, and then you added some greenhouse gases, the model always displayed a severe risk of future global warming.(105)

The modelers had reached a point where they could confidently declare what was reasonably likely to happen. They did not claim they would ever be able to say what would certainly happen. Different model runs continued to offer a range of possible future temperatures, thanks mainly to the different ways they tried to figure how clouds worked. Worse, the various GCMs stubbornly continued to give a wide range of predictions for particular regions. Some things looked quite certain, like higher temperatures in the Arctic (hardly a prediction now, for such warming was becoming blatantly visible in the weather data). But for many of the Earth’s populated places, the models could not reliably tell the local governments whether to brace themselves for more droughts, more floods, or neither or both.

By the dawn of the 21st century, climate models had become a crucial source of information for policy-makers and the public. Much of the modelers’ attention now focussed on working up results for the reports of the Intergovernmental Panel on Climate Change, rather than formal publication in obscure scientific journals. Struggling to provide a better picture of the changes that would come upon the planet as a whole and in each of its regions, the community of modelers grew larger and better organized. Projects to compare the models devised by different groups became a major ongoing activity. This cooperative framework forced the groups to agree on schemes for representing features of climate and formats for reporting their data, and once that was done, computer outputs could serve as raw material for groups that had nothing to do with the originators. That opened new paths for criticism and experimentation. A joint archive was established, which by 2007 contained more than 30 terabytes of data utilized by more than 1000 scientists.(105a)

There were about a dozen major teams now, and the pictures of overall global warming that their ocean-atmosphere GCMs computed were converging. The decades of work by teams of specialists, backed up by immense improvements in computers and data, had gradually built up confidence. It was largely thanks to their work that, as the editor of Science magazine announced in 2001, a “consensus as strong as the one that has developed around this topic is rare in the history of science.”(106*) Nearly every expert now agreed that the old predictions were solid — raising the CO2 level was all but certain to warm the globe.

The computer numbers got an independent check. Other scientists were studying past climates clear back to the ice ages, lining up temperature changes with changes in greenhouse gases and other influences such as aerosols blown into the atmosphere from volcanoes. By 2006 they had arrived at fairly good numbers, which agreed well with what the computers calaculated. When experts were pinned down to be specific, they might say there was scarcely one chance in twenty that the doubling of CO2 (what was expected to come before the end of the century) would warm the planet less than 1.5°C. The upper limit was harder to fix, since doubled CO2 would push the atmosphere into a state not seen for tens of millions of years, but a rise greater than 6°C seemed equally unlikely. The most likely number for climate sensitivity was 3°C.(107) However, experience showed that scientists tended to be too optimistic about their level of certainty. Some insisted that there was a very real chance of an outcome better than the IPCC expected—or far worse.

jshore · July 26, 2008, 2:29am

It is also worth noting and distinguishing the different ways that models are used. For example, the most famous way that they are used is simply to make a prediction of the climate sensitivity and of the future climate. However, that is not what Chapter 9 of the IPCC report is about. (It’s what Chapters 10 and 11 are about.) Chapter 9 is about using the models both to study what is responsible for the observed climate change…and also to study past climate changes in order to infer from it what range of climate sensitivities this change is compatible with. Note that in this latter case, the model isn’t predicting the climate sensitivity from first principles…It is merely serving as a conduit to infer what the climate sensitivity must have been to best reproduce the observed data. This approach is considerably less sensitive to deficiencies in the model than is the approach of just using the model to predict the climate sensitivity.

intention · July 26, 2008, 3:43am

jshore, thanks for the detailed references. I’ll take a look at them and get back to you. In the meantime, you say:

jshore:

As in all sciences, the observational data is used to build, test, and constrain models and, as in all sciences over the last half century, computers are used to do much of the modeling. Computer models are also used in interpretation of data, e.g., to convert between observational data and the equilibrium climate sensitivities implied by that data (since much of the data is over short enough time intervals that one is not very close to “equilibrium”…although the LGM data is an exception).

If you don’t want to deal with there being computer models in science, I suggest that the easiest thing for you to do is to time-travel back at least 50 years. The endless harping about computer models by the so-called “skeptics” like yourself is really getting rather boring and silly if you ask me.

You have totally mis-stated my position regarding computer models. It is a recurring frustration in dealing with you, but we persevere.

You are correct, all the sciences use models. Don’t you understand that I BOTH WRITE AND USE SCIENTIFIC COMPUTER MODELS MYSELF!!! How many times do I have to say that?

But climate science is unique among the sciences in my experience, by virtue of claiming that the model results constitute evidence of anything. They don’t. They represent a fast-calculating but bug-ridden version of our best scientific guesses, understandings, simplifications, and insights. IF (and it’s a big IF) those scientific understandings, simplifications, insights, and guesses are correct (and the bugs don’t crash the spaceship into Mars), then the model results will be borne out in reality … but there’s more bad models than good models in the history of science, so we can have no a priori confidence that the model results are correct.

This is particularly true given that we have no general theory of climate. We have a Theory of Relativity, and a Theory of Evolution, and a Newtonian Theory, but there is no Theory of Climate. This makes modeling much more difficult.

To highlight the difference between evidence and model results, consider the GISS Model E GCM, the computer model used by Hansen and Schmidt over at NASA. As reported in Present-Day Atmospheric Simulations Using GISS ModelE (Schmidt, Ruedy, Hansen, et al.), they ran a series of simulations of the historical climate. During those simulations, in their models the average global cloud cover was about 58% (p. 169). Does that finding constitute evidence about the average global cloud cover?

Now, the cloud cover is a critical part of any climate model because clouds reflect sunlight back into space. In this way, clouds control the heat entering the earth’s climate system. And it is a very powerful control system. In round numbers, if the cloud cover varies by 10%, the amount of sunlight reflected changes by about 17 W/m2. To get this is the same 17 W/m2 change, CO2 would have to rise from the current value (~ 397 parts per million) to the astronomical value of 9,800 ppmv.

A doubling of CO2 is the equivalent of a tiny, 2% change in global cloud cover. This means, of course, that until we can accurately model the clouds to within a couple of percent, any CO2 changes will be lost in the noise. So is the GISS model result of 58% cloud cover evidence of anything?

Well, yes. It is evidence of the complexity of the climate system compared to our still early, crude efforts at modeling it. Because measurements in the real world show that the cloud cover is actually 69%, an 11% difference from the model.

And lest you think that cloud cover is an isolated example, try this:

Climate Simulations 1880-2003, Hansen et al. including Schmidt

So when that GISS Model E computer model says that humans are heating up the earth … sorry, jshore, that don’t impress me much.

Because at the end of the day, all a computer model can do is calculate the results of the beliefs and understandings of the programmers.

It doesn’t, and never can, “prove” or “establish” or be “evidence” that those particular beliefs and understandings are right. I’m sure you can see the circularity of that argument.

A second, and totally independent, problem with computer modeling of complex systems is as follows:

How many computer runs have been done on lets say the GISS Model E and its predecessors?
Compared to the number published, how many of those were left on the cutting room floor, never saw the light of day?

I’m sure you see the problem in terms of sample selection.

A third, and totally independent, problem with computer modeling of complex systems is that the programs tend to be large, poorly documented, and the creation of a succession of programmers. As a man who has spent thousands of hours writing and debugging computer programs, I can assure you that this scenario guarantees bugs.

A fourth, and totally independent, problem with computer modeling of the climate is the complexity of the task. Climate involves important processes on temporal scales from milli-seconds to millions of years, and on temporal scales from the molecular to planet-wide. It involves six major subsystems (atmosphere, ocean, biosphere, lithosphere, cryosphere, electro-magnetosphere). The sheer complexity is overwhelming.

A fifth, and totally independent, problem with computer modeling is that the size and complexity climate forces the adoption of a whole host of parameters. Formally, each of these is a variable in the solution of the problem. As a result, it is easy to overdetermine the solution.

There’s more, including technical stuff like the need to have unphysical atmospheric viscosity to prevent model runaway, but the conclusion is clear.

Climate model results are not evidence of anything. They are interesting, they can be useful, they can give us insights, but they cannot provide evidence that humans are affecting the climate.

All the best,

w.

intention · July 26, 2008, 4:49am