I mentioned this in the comment thread on my earlier post on model IV&V, but I’m elevating it to a full post here because it’s an interesting point of discussion.

I had a very interesting lunch with David Randall at CSU yesterday, in which we talked about many of the challenges facing climate modelling groups as they deal with increasing complexity in the models. One topic that came up was the question of whether it’s time for the climate modeling labs to establish separate divisions for the science models (experimental tools for trying out new ideas) and production models (which would be used for assessments and policy support). This separation hasn’t happened in climate modelling, but may well be inevitable, if the the anticipated market for climate services ever materializes.

There are many benefits of such a separation. It would clarify the goals and roles within the modeling labs, and allow for a more explicit decision process that decides which ideas from the bleeding edge science models are mature enough for inclusion in the operational models. The latter would presumably only contain the more well-established science, would change less rapidly, and could be better engineered for robustness and usability. And better documented.

But there’s a huge downside: the separation would effectively mean two separate models need to be developed and maintained (thus potentially doubling the effort), and the separation would make it harder to get the latest science transferred into the production model. Which in turn would mean a risk that assessments such as the IPCC’s become even more dated than they are now: there’s already a several year delay because of the time it takes to complete model runs, share the data, analyze it, peer-review and publish results, and then compile the assessment reports. Divorcing science models from production models would make this delay worse.

But there’s an even bigger problem: the community is too small. There aren’t enough people who understand how to put together a climate model as it is; bifurcating the effort will make this shortfall even worse. David points out that part of the problem is that climate models are now so complex that nobody really understands the entire model; the other problem is that our grad schools aren’t producing many people who have both the aptitude and enthusiasm for climate modeling. There’s a risk that the best modellers will choose to stay in the science shops (because getting leading edge science into the models is much more motivating), leaving insufficient expertise to maintain quality in the production models.

So really, it comes down to some difficult questions about priorities: given the serious shortage of good modellers, do we push ahead with the current approach in which progress at the leading edge of the science is prioritized, or do we split the effort to create these production shops? It seems to me that what matters for the IPCC at the moment is a good assessment of the current science, not some separate climate forecasting service. If a commercial market develops for the latter (which is possible, once people really start to get serious about climate change), then someone will have to figure out how to channel the revenues into training a new generation of modellers.

When I mentioned this discussion on the earlier thread, Josh was surprised at the point that universities aren’t producing enough people with the aptitude and motivation:

“seems like this is a hot / growth area (maybe that impression is just due to the press coverage).”

Michael replied:

“Funding is weak and sporadic; the political visibility of these issues often causes revenge-taking at the top of the funding hierarchy. Recent news, for instance, seems to be of drastic cuts in Canadian funding for climate science. …

The limited budgets lead to attachment to awkward legacy codes, which drives away the most ambitious programmers. The nature of the problem stymies the most mathematically adept who are inclined to look for more purity. Software engineers take a back seat to physical scientists with little regard for software design as a profession. All in all, the work is drastically ill-rewarded in proportion to its importance, and it’s fair to say that while it attracts good people, it’s not hard to imagine a larger group of much higher productivity and greater computational science sophistication working on this problem.

And that is the nub of the problem. There’s plenty of scope for improving the quality of the models and the quality of the software in them. But if we can’t grow the pool of engaged talent, it won’t happen.


  1. Note that Integrated Assessment Models are another approach to making more policy-relevant tools: after all, GCMs aren’t really well suited to studying specific policies, so there are simpler models tuned to the big models. See, e.g., MAGICC.

    (Of course, there are also the EMICs, which are a different beast: simpler in many ways, but at least historically at the forefront of coupling different physical components together – eg, including ecosystem coupling and atmospheric chemistry long before the big GCMs)

    (well, and then there’s the REALLY simple IAMs like DICE where the climate model is just a simple energy balance model…)


  2. Hi Steve. I think ‘M’ has a good approach.

    Also possible: I have read on another blog that ensemble studies are starting to reach the point of diminishing returns. Assuming this is true, could the talent pools associated with some of the less-leading-edge GCMs be retasked to the creation of production models? Wouldn’t need to grow the overall talent pool then.

    In your opinion, any way GCM managers could make such an idea practical (e.g., get the funding)? I worry that, in light of Michael’s comment above, this approach might actually drive talent away from the overall pool.

  3. I’m surprised Michael didn’t mention one of the major issues that comes from weak and sporadic funding. Namely, the career path is unreliable and not very well paid. Typical is multiple postdoc positions, with corresponding multiple chances for funding to disappear and you to be out of the field. While in the postdoc positions, salaries are … unimpressive (unless they’ve changed a lot!). When I started my postdoc in 1990 — with doctorate in hand, and a relatively well-paid postdoc for the time, my salary was the same as I’d been offered for doing computer things — with my BS, in 1985. That ratio seems still to hold. So one question is why spend (or expect others to spend) 6 years (typical time to PhD) earning a PhD in order to make the same salary? (and probably no improvement in job security)

    Look, also, farther down the road. People in many fields accept relatively low starting salaries in hopes of big salaries down the road. Business/accounting/stocks are replete with examples. The high end there is millions per year and up (far up). Climate modeling … an SES position (Senior Executive Service, highest ranking civil servants) tops out $0.15-0.2 M. Tenured full professors raking in grants very successfully top out … ? $0.3 M. Not so sure about the latter, but probably not off by more than a factor of 2.

    One could also ask whether the skill set for being a climate modeler would transfer to other areas. Hmm, dealing with large data sets, complex systems, complex models, … somehow I don’t think climate is unique in that. Stocks/finance having taken a shine to such things, whether meaningful or not, and paying quite a bit more. But, more practically, several people have left my group in order to work in ordinary ‘make a computer do something useful’ — not even major finance — and increased their salaries by 50-100%. (I remind myself of this when things look shaky at work.)

    Those doing climate modeling do it for other reasons than good starting pay, prospects of high pay later, stable jobs, and a lack of ability to do anything else. Which is probably not the way to go about attracting and retaining the best possible people. On the other hand, those who do engage, are very passionate about the subject. And some are extraordinarily talented.

  4. Perhaps a good model for this would be the ab initio chemical programs, which evolved from a set of expert tools to something every organic chemist abuses. In the latter case it provides strong hints which can be combined with lab information such as NMR spectra to understand structure and reactions.

    The big jumps came with availability of reasonable computational power to everyone and when these programs went commercial with Gaussian first and now a bunch of competitors. There is a very well known, nasty story there, but the long and short of it is that the programs only made the jump when there was a commercial source which could provide help to customers.

    Regional air pollution models are probably going to be the first step because there are many customers. Right now they are roll your own, or freeware (RAMS which is Pielke Srs. thing for example).

  5. Robert said: “…the career path is unreliable and not very well paid.”

    Aren’t there tenured positions for climate modelers?

  6. Tim: Not really, particularly once you consider the funding uncertainty Michael mentioned. Sure there are people currently occupying tenured positions who do climate modeling. But if you back track, most (all?) got their tenure quite a few years back.

    Career path is the key term. In between freshly minted PhD and tenured professor doing climate modeling (2) are, now, 5-10 years of post-docing, with the positions being funded only a year or two at a time. Unstable funding drops the probability of the person making it through that period. Now suppose this is one of the lucky ones (1) who is hired in to a tenurable position. That’s another 5-10 years of needing continual funding, quite a bit more of it (if you can’t be counted on to bring in grants, year after year, you’re not ‘worthy of tenure’). So 10-20 years where an instability in funding means you’re out of the field. Not, I submit, a reliable career path.

    1) Universities have increasingly been moving away from tenurable positions. More and more, it is adjuncts and lectureships. And these are typically teaching-only positions.

    2) Having tenure doesn’t mean that the person is doing climate modeling. Universities typically only pay 2-3 months of salary for the professors (tenured or tenurable). The rest comes from grants. A gap in grants means either you’re gone or that you start doing more teaching (they’ll pay more of your salary if you teach more). I rather like teaching, myself, but to teach a class well means time you’re not working on doing that climate modeling. Teaching more classes means even less time on the climate modeling. So hiatus in grants, the unreliability, means hiatus in doing the science.

    None of this is unique to climate, except perhaps the degree to which the funding is hostage to political whims. There are a few other fields where that’s as bad or worse, but not so many.

  7. For students considering a career in climate modeling, the death threats and hate campaigns levied against current climate modelers doesn’t help. I am hoping that things will have calmed down by the time I can reach my career aspirations!

    Additionally, large scale climate modeling centers aren’t very common compared to other areas of research. There is a biomedical research lab at most every university, but (correct me if I’m wrong) only one GCM group in Canada. As others noted above, approaching cutbacks in funding won’t help this situation.

    For an area of research so important to current events and future welfare, students aspiring to join it are quite rare.

  8. Eli rarely disagrees with Bob Grunbine, but the percentage of salary paid by a university to someone on the tenure track varies like crazy. It is true that some places like MIT might only pay a few months, but other places, including many R1s pay the full nine months. Then you get into all kinds of crazy stuff like how a full teaching load is defined, what departments really assign as a full load and many other weeds including how much teaching relief is built into start up funds and much, much more.

    Here is a fairly typical (others may disagree) policy at an R1 (you may disagree), Texas A&M


  9. Eli does occasionally mis-spell people’s names. Apologies to Bob.

  10. Pingback: Why coupling should help with climate model verification but may not in reality « Rocky Dunlap's Blog

Leave a Reply

Your email address will not be published. Required fields are marked *