It is time to stop quivering in our boots in pointless fear of the future and just roll up our sleeves and build it.
- Ray Pierrehumbert

Saturday, April 9, 2011

Engineer's Survey of Climate Sensitivity

Since the non-existent back channels are probably going to come down on me like a ton of bricks for the previous item, let me add something to soften the blow a bit. Steve McIntyre is right to demand an engineering-level survey of the climate sensitivity problem.

It's a tall order but it is a reasonable one. It is fair to say that IPCC WG I does not succeed. Archer's "Understanding the Forecast" text is a bit elementary for what an engineer would expect but is along the right lines.

Is there a deeper text along those lines? Any suggestions?

Is this a real gap, and if so, how should we go about filling it? Or has somebody already taken this on successfully?


Grypo Saurus said...

What would this look like? I'm not familiar with what engineering people expect and I don't have access to Archer's book. It is just bringing together all the elements of possible ways to look at climate sensitivity?

Like an updated version of what Hansen did in the eighties?

Michael Tobis said...

Nice find GS! Thanks!

An engineer expects an explicit list of assumptions and a rigorous derivation that starts with very few and very explicit assumptions of the reader's skills. This is rare in the earth sciences, and I share the frustration of others with an engineering education about it.

dhogaza said...

I'd sorta like a science-level examination of engineers' understanding of metal fatigue caused by pressurization cycles in 737s ...

And a re-evaluation of engineers' assessment of the likelihood of a lengthy power outage causing problems in boiling water reactors ...

I don't understand why we would think that an engineering-level evaluation of climate sensitivity would be better than what's done by scientists.

Their track record is highly exaggerated.

And McIntyre's not an engineer ...

And the captcha is 'bacchas' - disturbingly close to my real last name.

Andrew Dessler said...

This is really not a very easy request to satisfy. There is no single piece of evidence or first-order derivation that tells us the climate sensitivity is 3°C per doubled carbon dioxide. Rather, there are dozens of independent lines of evidence (from observations, theory, models), each one reasonably uncertain, but which all point in the same direction. Taken together, the evidence is really very very compelling, but the individual components are not. This is, unfortunately, not what you're looking for, and it gives skeptics lots of latitude to indict any single piece of evidence while ignoring the totality.

Michael Tobis said...

I think a clear Bayesian presentation across the various streams of evidence would suffice. Admittedly, it makes for a bigger project.

EliRabett said...

Engineering has a useful way of dealing with uncertain safety issues, it designs for what is known and then uses a multiplier for safety. The greater the uncertainty the greater the multiplier. To quote the Wikipedia
There is a near universal push towards conservatism in the calculation of safety factors, i.e. in the absence of highly accurate data, using the worst case configuration possible to make sure the system is adequate (to err on the side of caution).

SM does not appear to understand this.

Arthur said...

Ray Pierrehumbert's "Principles of Planetary Climate" does the job very well at explaining the fundamental physics that goes into modeling - which is all a physicist would need (paleoclimate or other evidence is just anecdotal :). Perhaps that's why McIntyre insists on the "engineer" and not "physicist" analysis...?

Paul said...

So how does James Annan's work fall short of your request for a Bayseian analysis?

I echo dhogaza's skepticism with regard to engineers. They get some stuff spectacularly right and some dangerously wrong. My major professor (low temperature solid state physics)was a man of great practical ability and amazing physical insight. He noted on numerous occasion that there wasn't much of anything an engineer could do that a competent experimental physicist couldn't do and also understand and expand on.

I am fully aware of the aura of omniscients that some physicists cloak themselves in. (Dyson?) I'm just sayin' that I have spent a lot of time with both engineers and physicists and if you are going to get an explanation down to the lever where your average retired engineer/libertarian can understand, you are really going to have to get down in the grass. Don't even think about computers and climate models.

Paul Middents

Michael Tobis said...

I am very surprised at Ray's book, wherein he explicitly eschews fluid dynamics altogether. Ray first appealed to me as a dynamicist and I consider all this stuff about radiative transfer to be a sort of slumming.

On the other hand, Ray has an engineering degree from MIT, so maybe that will help.

I think this all comes down to the same place - the system is complicated enough that some books will satisfy some people but no book will satisfy everybody, and no book will satisfy somebody who sets out to be dissatisfied.

In fact, that is true of many plain ordinary engineering books too. Somewhere in chapter two, a great deal tends to be asserted without proof (with a slew of references that reassure the undergrad but are not likely, as anyone who has ever been an undergraduate engineering student will attest, to ever be followed up). The rest of the book can be quite elegant, though.

Ray hints there is another book with lfuid dynamics possibly to follow. He's been working on this one for a very long time, so I won't be holding my breath. But I'd sure like to see it some day.

Steve Bloom said...

Andy, I'm very curious as to why you don't find the Pliocene-based Earth System Sensitivity work compelling all on its own. Is it just because it doesn't tell us how fast?

steven said...

Thanks Andrew. What SM was asking for was something "like" an engineering report.

Of course dehog and other misunderstand what this is, or why its important, and try to make it about SM not being an engineer.

Having talked to SM about this issue and shared with him my engineering experience writing reports for USAF and USN, what he basically was hoping to find was a single resource that covered the issue top to bottom with the completeness you find in a typical engineering report.

From a Baysian perspective have you see the stuff presented at the Newton Institute Symposium? Very cool. I'll link if you like

guthrie said...

If what Macintyre wants is a single source covering things from top to bottom, why doesn't he write it himself? Surely he is well placed to do so?

You could always split into chapter addressing specific areas such as radiative transfer by respective experts in their fields, with filler to tie them all together. I could probably do the job, if someone gave me enough money to live on for a couple of years and the emails of various experts.

EliRabett said...

Steven, that's just nonsense. If McIntyre had a clue we would "engineering reports" about physics, chemistry and biology, or, in Climate Audit land, perhaps there is one combining all three.

The closest thing are the IPCC summaries, and SM don't like those at all.

Roger said...

I am sceptical about the whole concept of "engineering" level as applied to science. I asked McIntyre in comments a few times to define exactly what he meant by "engineering" level, and got no real answer except that he wanted reports hundreds of pages long instead of your typical 5-20 page scientific article.

Until it is defined exactly what is meant by "engineering level", and why this is better than what currently exists, I don't see that the suggestion has any value. And why, for example, do advanced textbooks not suffice? What audience should the "engineering level" text be aimed at?

Michael Tobis said...

The nice thing about engineering texts is that they start with an explicit assertion of the knowledge and capacities of the reader, and that this is minimized for practical purposes. The nonstandard skills are developed in the first two chapters or an appendix, very quickly, but with reference to other materials.

I do share the feeling with other engineers that earth science textbooks are built on air rather than on a carefully built foundation.

It turns out that there is a foundation but it is informally embedded in the social context. This does not scale, and it reinforces the suspicions of the engineers.

guthrie said...

MIchael - I think others have pointed out the problems already with the difference between science and engineering.

Let me take another approach. In my library I have a book called "Understanding Earth" by Press and Siever. A 1st year undergraduate text from when I took geology modules during my chemistry degree.

Foundations required for comprehending what is in it:
Chemistry - since there are chapters on the formation of planets, (Which is a sub-topic by itself requiring cosmology), mineral composition etc. In fact several pages run through the periodic table of the elements and the elements commonly found in the earths crust.

Then onto igneous rocks, their minerals and how they are formed. What sort of foundation do you want to have there? If you accept chemistry then you can take a lot of it as read. The presentation method is a summary though, as required by a textbook. There simply isn't space to go through everything referring back to each individual paper which established something 80 years ago.

You really aught to think more about this issue, since at the moment it just confuses me. What precisely is wrong with the IPCC reports? Are they not big enough already?

Michael Tobis said...

IPCC WGI reports are admirable (I am not so sure about the other working groups). But while they serve well as a survey they do not serve well as a rigorous exposition, which is what I think outsiders are looking for.

There are reasons this is difficult. And in thinking about it, and in my experiences with the opposition of late, I can promise that they will reject every attempt. But I can well imagine a neutral, scientifically literate person moving from neutral to suspicious to hostile based on the limited visibility and limited quality of the sorts of intermediate level materials (above college freshman but below specialist) which are easy to obtain in engineering disciplines.

I can certainly attest to my own frustration with the didactic quality in the earth sciences in the 90s. I didn't progress to hostility because I was in the presence of many obviously competent and a few obviously brilliant people, including the likes of Suomi and Bretherton Sr. and less celebrated folks like Fred Best, the most effective engineer I've ever been exposed to.

I think it's improved since then. Of course, the problem is that people want an exposition on climate change before they understand climate, a process which has several important parts.

So I'm not saying it is anything but a tall order. But I am not dismissing the question.

Michael Tobis said...

In response to my:

"I am very surprised at Ray's book, wherein he explicitly eschews fluid dynamics altogether"

Ray grumbles in email:

Jeez tobis, how long did you want the thing to be? It's nearly 800 pages without even touching fluid dynamics. There are plenty of places to learn GFD (notably Vallis' book) and lots of people learn GFD. What's hard is for a person who knows GFD to also learn the other stuff needed to connect GFD to serious problems in climate. That's who the book is for. You can't do serious science in this area any more with just a knowledge of fluid dynamics. Most of the most serious issues in climate involve the fluids questions only peripherally. As I say in the last chapter, you need the fluids at some point, but it's not where you need to start.

For people interested in climate, I'd write a GFD book somewhat differently from Geoff's, and I would also make sure that the book came with software needed for students to build their own models, but at this point doing yet another GFD textbook variant is not high on my list of priorities, compared to the backlog of research I'm trying to get out."

guthrie said...

I have to thank Michael for thinking some more about this given that I demanded he do so despite it being me who was confused...

Paul said...

Science of Doom recently had a post on recommended text books in climate science.

Do none of these approach your standards for engineering exposition?

phil said...

Gerard Roe's review article is, I think about as clear as you can get in 10 pages:

-- Phil Austin

David B. Benson said...

Pierrehumbert's Principles of Planetary Climate certainly fills this engineer's needs; supurb in treating the hard stuff. As for the fluid dynamics there are many fine books which fill in that part at any required depth.

As for climate sensitivity, papers by Annan & Hargreaves seem to fit your perceived need...

Grypo Saurus said...

the phil austin article is available in full also

steven said...


Respectfully, I think I have a better idea what steve is talking about than other people. I've had a number of discussions with Steve on this matter. In fact, I pointed him at Science of Doom as the start of something could become more like what we refer to as an engineering type of report. It has little to do with being an "engineer" and more to do with a certain explicitness and exhaustiveness of explication and documentation.

So, for example, Lets take a sample article that reported a synthesis of the work I did long ago.

IN: AIAA Flight Simulation Technologies Conference and Exhibit, Boston, MA, Aug 14-16, 1989, Technical Papers (A89-48376 21-09). Washington, DC, American Institute of Aeronautics and Astronautics, 1989, p.383-389

Those 6 pages authored by the two managers I did the work for, basically boiled down literally thousands of pages of engineering reports into one short paper. A typical report on very simple model runs might run 600 pages or so. Of course I wasnt alone in writing thousands of pages, many people contributed to it.

Another example might be the surface record. A report on that would easily top 1000 pages.

steven said...

"If what Macintyre wants is a single source covering things from top to bottom, why doesn't he write it himself? Surely he is well placed to do so? "

Well, it's more than McIntyre that wants one. Should everyone that wants one be required to write one? makes no sense. The point is for someone like me, who believes in radiative physics, it has been ( until SOD) kinda difficult to answer the questions of those who are technically competant by pointing them at a resource where they can read for themselves. A resource that starts at the begining, lays bare all the things you know and dont know, a bible of sorts. So in engineering, you get a new guy in and he needs to get up to speed on a system or a process or anything. You hand him the one report that has it all. Here's the report on the engine dude, read that. And if that document is well written it has everything he needs to know.
Now obviously Sensitivity to doubling is not like an engine or a device. So in some respects the request for an "engineering" report is metaphorical. The gist of it is as MT points out. Like I said, I think SOD is a good start toward that.

However, some skeptics are not open to any kind of persuasion. The other thought I played with was getting engineers who actually worked with radiative physics to build things as "explainers" of things.

Finally, I think folks need to re evaluate the reaction that everything SM says or requests is automatically BS. I do remember times 4 years ago when people thought requests for code and data were out of line, but the community seems to have reached a consensus on transparency.

Michael Tobis said...

"Finally, I think folks need to re evaluate the reaction that everything SM says or requests is automatically BS. I do remember times 4 years ago when people thought requests for code and data were out of line, but the community seems to have reached a consensus on transparency."

Arguably so. But that silver lining comes with one hell of a cloud.

willard said...

Who said:

> Steve, your request for proof of 2.5C sensitivity doesn’t make sense in this context, which is why no one has responded

Answer here:

Jim Bouldin said...

McIntyre can "demand" away under his self-generated conception of the "right" way to do science. When he proves--repeatedly--that he's actually interested in what the truth is, rather than various forms of self-aggrandization, then others who've done so for years will begin to pay attention. And not until.

Alastair said...

As an engineer I find Eric Ghill's approach to sensitivity closest to mine. In other words the climate, like the economy, is not stable or linear, and so talking of a sensitivity is nonsense!

See his presentation to the EGU 55 minutes into this

James Annan said...

"Engineering-level" is just a sneer, as far as I can tell.

Michael Tobis said...

Via email from "Pete":

" survey of the climate sensitivity problem...." -

An imprecise and imho unnecessary new goalpost. It's the old wish for the perfect book (so I don't have to read ten books or work too hard). This book is only supposed to be perfect for a special group, engineers who still don't get it (but if they just had the right book....) and evidently must reduce science to engineering.

Meanwhile the focus on sensitivity is itself beginning to look like a distraction. Is there some reason to think we will get a break from radiation physics, or that climate response will be very weak compared to the last million years? And note, whatever sensitivity you calculate, sensitivity to CO2 forcing will become greater once CO2 feedback sets in. And it will last longer since the oceans will resupply what the rocks take out. The world will see a lot more sea before this is over.

Whatever the exact future temperature, weather of recent years suggests that climate disruption (drought, floods and heat waves) will get us before rising sea does. Very unfortunately there does not seem to be a model for this. But it's a more pressing matter than helping slow engineers.

This whole thread is part of a long running theme of yours: "The big problem with explaining the climate problem is that it is too complicated." I think nearly everyone has a comfort level with complications on nearly everything. Few want to get into physics at all, or feel that they need to in order to function. The big communication problem imho is the professional sowers of misguided doubt and worse. Absent that I don't think the complexity of nature would be the issue for very many folks.

adelady said...

"Absent that I don't think the complexity of nature would be the issue for very many folks."

Exactly. None of us can know everything about everything. I spend time on education and medical sites as well as climate science. I know a fair bit about many things. But there _always_ comes a point where I must decide. Do I put in time and effort on this particular technical detail? Or do I make a _judgment_ about relying on others' expertise?

Sometimes I find myself spending time on finding or reviving knowledge of a topic. Most of the time I have to accept that some person or group is reliable enough for me to accept conclusions or observations - at least for the time being.

The complainers, I find, are unwilling to both withhold judgment and to accept others' work, even if it's only for the time being. Insisting that we must have a definitive answer *now* and that someone, somewhere, has to front up with an explanation or explication that I understand *now*, instantly and unreservedly, is a step too far.

In fact, it's dog-whistle for lack of trust. If you say it's complex, or if I can't understand it at first glance without doing any work at all, then you must be hiding something. From me! How dare you!

I don't have to trust anyone with my life or my savings to trust them enough to accept their findings - and be willing to change the view I thus form if better information comes from elsewhere. My completely uninformed view of this group of people is that they are so afraid of changing their minds or their view that they're unwilling to accept anything at all. It's a kind of intellectual paralysis.

Grypo Saurus said...

Steve M just discussed this on his blog. This is what he is looking for...

Such an exposition would probably be 1200 or 2000 pages, not 10 pages. Some of it would be material available in textbooks e.g. description of the infrared bands that are affected by additional CO2 – information that is not in dispute, but which any engineer would include in a comprehensive exposition. The main area of scientific uncertainty is in cloud feedbacks. In an engineering quality report, there might be several hundred pages on this topic, describing precisely what is known and what is not known and how the scientific uncertainties might be reduced. In AR4, this important topic was covered in less than two pages.
I’ve raised this issue with climate scientists on a number of occasions. To date, I haven’t encountered a single climate scientist that remotely comprehended what was missing, while professionals from other fields often understand the sort of document right away. (DeWitt Payne, among others, has endorsed this on other blogs.)
Typically, if a climate scientist responds, they provide a link to some little article on climate sensitivity that is not remotely equivalent to an engineering quality exposition, with the citation merely showing that the climate scientist doesn’t have a clue about the form of communication employed in the professional world. (Computer professionals sometimes confuse this issue with properly QCed computer code but the points are different.)
Unfortunately, some climate scientists – Gerry North for example – have even sneered at the idea of such an exposition. BTW unlike some readers, I do not conclude that the absence of such an exposition means that it is impossible to prepare such an exposition.

Penguindreams said...

McIntyre is inaccurate in his claim of no climate scientist taking interest. I was at his site some years back and after saying that he wanted an engineering analysis, he refused to name or describe what that was. When I asked if he meant 10 pages, 200, 10,000 or what, he refused to answer (making other comments in response instead).

I'd identified myself as a climate scientist, and one interested in writing a book that would address his alleged concerns. Yet he refused to say what he really wanted. That's the safest way to ensure that he can always complain about not getting what he wants, so can always complain.

I still have in mind something like 'An Engineer's guide to climate', so ideas for how to go about it are still welcome. My undergraduate degree was engineering, so I'm not entirely foreign to the culture.