RealClimate logo

New On-line Classes and Models

Filed under: — david @ 6 February 2016

My free online class on entitled Global Warming I: The Science and Modeling of Climate Change has already served 45,000 people (started, not finished) in the four times that it’s run. Now it’s set up in a new format, called “on demand mode”, which allows people to start, progress, and finish on their own calendars. This would be an advantage if a teacher wanted to use the material to supplement a class; a new cohort of learners is launched every month, so the next class start date is never more than a month away.

A new, supplemental class to the first one has been added and will come online on Monday, called Global Warming II: Create Your Own Models in Python or Fortran. This takes advantage of new code-grading machinery at Coursera to automatically run your code through its paces. There is also a peer code-review step, where you will get feedback on your commenting and variable-naming skills, and provide feedback to others. The class gives detailed instructions to create simple models of: time evolution of global temperature, the ice albedo feedback drop into snowball Earth, an ice sheet, and a shallow-water circulation model. The class is intended for people who are new to programming, or new to Python, or wish to enhance their understanding and appreciation of some cool science of Earth’s climate system.

The classes are supported by the same interactive on-line interactive climate system models as before, at, with some new additions, both of which generate animations of their time-dependent solutions.

One is a Hurricane simulator using a model from Kerry Emanuel, which can demonstrate the sensitivity of ocean temperature, ocean mixing, and atmospheric structure on hurricane evolution.

The other, the Permafrost model, is a simulation of a soil or sediment column in which ice and methane hydrate can form. The model shows how the brine salinity thermodynamically excludes methane hydrate from forming until you get to the base of the permafrost zone, and also how long it takes to warm a soil column by warming the surface. This model shows why I do not believe in an imminent methane climate catastrophe from Arctic Ocean methane hydrates.

26 Responses to “New On-line Classes and Models”

  1. 1

    Darn, I wish you were using some other language than Python. Oh, well.

    [Response: Honestly I don’t use Python for work either. For serious industrial number-crunching like climate models, Fortran is infinitely faster than the interpreted Python. I used Python for this class because it’s freely available and extremely powerful, with all the external libraries you can import. I could have set it up to compile and run Fortran or C. Is that what you would have preferred? David]

    [Response: Stop the Press! I just realized I can tweak the auto-grader so that it compiles and tests Fortran submissions as well as Python. That’s in case you were lamenting that you couldn’t use Fortran. If you long for java or objective C some other weird language, I won’t help you. But there is definitely still a role for Fortran in climate science. I’ll get the new grader up and announced in a few days. Thanks for the feedback! David]

    [Response: Just finished the changes to the class grader and instructions; it now accepts Fortran as well as Python submissions. David]

  2. 2
    BojanD says:

    Interesting! Might give it a go, it really depends on the time available. But so far … cheers for the effort.

  3. 3

    Re Arctic Ocean methane hydrates
    There is a new study out on the potential for methane pockets under ice sheets, any thoughts on this? Press release

    Could this eventually mean nonlinear contributions from localized spots?

    Made a video edition

    Something similar was mentioned by Bill McGuire

    My concern, however, is that there may be a threat of submarine landslides around the margins of Greenland, which are less well explored. Greenland is already uplifting, reducing the pressure on the crust beneath and also on submarine methane hydrates in the sediment around its margins, and increased seismic activity may be apparent within decades as active faults beneath the ice sheet are unloaded. This could provide the potential for the earthquake or methane hydrate destabilisation of submarine sediment, leading to the formation of submarine slides and, perhaps, tsunamis in the North Atlantic.

    [Response: The lifetime of methane in the atmosphere is long compared to the atmospheric mixing time, so there are no local climate effects to local methane emissions. As long as tropical and human emissions exceed those from the Arctic by such a huge margin (factor of 10 or 20), I’m not going worry about the Arctic emissions. Actually, the largest Arctic methane source is the fossil fuel industry. Tsunamis are worrisome, but the methane they could release is generally not. David]

  4. 4
    Jeremy Fyke says:

    I totally recommend this course. In theory I’m a climate scientist myself ( and last year I audited this (in the online sense) with my father, a former high school geology/chem teacher, who took the actual paid course. Well worth it for both of us! Would definitely be great for, e.g, classrooms. Very interested in Python models too.

  5. 5
    Digby Scorgie says:

    Um, it’s an “imminent catastrophe”, not “immanent”. Please fix.

  6. 6
    Kevin Cowtan says:

    I did “Global Warming 1” while my first climate paper was in review, and it was a great experience. I’d recommend it to anyone who wants to actually get hands-on with the science, rather than just being a spectator.

    The new course looks great too. Python is an excellent choice – it’s a clean, modern and well designed language – it’s our default choice for teaching students, and we’re increasingly seeing students arriving having learned python in school. There are powerful libraries for just about everything imaginable from building web applications to user interfaces, graphics and animation. Netcdf libraries wrappers to other packages allow access to most climate data, and the numpy, scipy and statsmodels packages allow high performance calculations on very large datasets. I’ve written python programs which manipulate multi-gigabyte data at respectable speeds, as well as implementing more complex numerical calculations such as kriging.

  7. 7
    Edward Greisch says:

    Tried the hurricane and permafrost models a little. I think I need to play with them a lot more before I will understand them.

    Python course needs the on demand mode because last time I started learning it, I had to rush off to give information to the Illinois state legislature.

  8. 8
    Dee says:

    I’m nearly done with this round of Dr Archer’s Coursera course. The models are worth trying out. If you decide to work with the models, I highly recommend watching the videos explaining how to use them.

    I also recommend the course. The quizzes can be a bit frustrating at times (some of the questions aren’t all that clear), but the course gets into some depth and really helps pull all the different parts of the subject together. At least it does for me.

    [Response: There’s a way to lodge suggestions on question wording or grading. People do that, and I can go through and edit the questions, and gradually the rough edges get smooothed out. David]

  9. 9

    Thanks David; your effort is really valuable for both, teaching and reserach
    Cheers, Antonio

  10. 10
    Mal Adapted says:

    Kevin Cowtan:

    Python is an excellent choice – it’s a clean, modern and well designed language – it’s our default choice for teaching students

    Well, I feel old. When I was taking programming classes, Pascal was the clean, modern and well designed language. I write a lot of Python myself these days, mostly as a replacement for Perl. My Python scripts are still pretty much procedurally-oriented, but I profoundly appreciate all the work other public-spirited programmers have done, that I can just import. In time, Python scripts may need only one line:

    from __future__ import completeSolution

    O brave new world, That has such technology in’t 8^D!

    [Response: I learned Python for this class, never used it before. Yes, old. David]

  11. 11

    Thanks David. I always recommend your Global Warming: Understanding the Forecast book to my (environmental law) students. I’ll add a recommendation for your online lectures. Thanks so much for your ongoing and deep commitment to public education.

  12. 12
    Jacques Talbot says:

    You state that “there are no longer any time constraints built into the Coursera system” for your course. However, for me, as of to-day, Coursera says that “next session begins feb 22” and only the 6′ preview is available.
    Did I miss something?

    [Response: Yeah, I see what you’re saying. I’m not sure; I put in a query to my contact at Coursera and will advise. It could be that they start a new “batch” of people every month or something. The genius of the Coursera system is to get traffic going in discussion forums, so they do want to keep some sort of critical mass going, a large enough group of people at the same point in the course. But once you’re enrolled in the class, you can go as fast or slow as you like, although there are “suggested” deadlines to “keep you on track” with email reminders. David]

  13. 13
    pjkar says:

    I took this course 2014 and found it to be an excellent comprehensive overview of the sciece of global warming issues. The number cruncher exercises were challenging but provided insight into the modeling of climate science phenomena. I continue to use the climate models the course provided to test out scenarios of interest in a back of the envelope kind of way. Very insightful.These kind of things definitely bring you closer to the science. The best thing about this course is that the professor is a dedicated climate scientist and teacher who excells at both. I think anyone not a professional in the field but with a hight degree of interest in it (as most of us here) would find this course to be a very rewarding learning experience. I know I did.

  14. 14
    Jacques Talbot says:

    Thanks for a fast answer. Batching makes sense to me.

    [Response: I got this from my contact at Coursera

    The default now is a cohort model (new every 4 weeks) which groups learners together for a more structured experience to learn with others. Although there are deadlines, no one is penalized for lateness and they can transfer from one cohort into the next if for any reason they do not complete.

    This is an article in more detail for learners

    Also, please encourage the learners to reach out to the Learner Help Center, as that is the best place for them to get questions answered quickly. :)

  15. 15
    Chris Colose says:


    Brian Rose at University of Albany has a Python-based climlab package that is appropriate for studying “N-layer” greenhouse models, among other things like insolation distribution (e.g, plotting latitudinal/seasonal changes on orbital timescales), basic heat transport concepts, etc. You may find it useful in the future.

  16. 16
    dominik says:

    Isn’t there also at least one python-c crosscompiler, so you get the best of both worlds?

  17. 17

    Thanks for your excellent paper on why you don’t believe in an immanent methane catastrophe. I scanned it briefly, but it leaves open the question of rapid methane releases that could strongly influence Arctic regional climates. If “Spikes of methane emission during late deglaciation become somewhat more intense.” then should rapid increased heating by methane be possible because of too many unknown factors?

    Seeking info on land inundation – Sea level rise in the Arctic Ocean is not modeled at – does anyone know of the sea level rise mapping for the Arctic Ocean shorelines? They have a nifty map of the arctic ocean sea floor.

    Using Google Earth and just eyeing Arctic shorelines of parts of Russia, Canada and Alaska one sees about 10%-20% of shoreline within 3 feet of sea level. And extending in a ways… obviously serious map work is needed.

    Where are the tempestology studies needed to know how major storms in the Arctic will affect shorelines? What are the storm and tidal surges? If Sandy gave NYC a 14 foot surge, could that happen in the Arctic? How deep might turbulent storm churn extend into the Arctic depths? What about thermal churn?

    What happens when 2 or 3 feet of sea level rise inundates Arctic permafrosted lands? After a storm surge sends seawater inland from the Arctic shoreline, what happens to all that previously unexposed permafrost after a single storm surge of water? How deep is the refreezing? Does a subsequent surge from the next storm more impactful?

    What of the seismological destabilization of methane deposits in land and undersea? When a melting-ice-enlightened Greenland begins Isotatic Rebounding will there be tsunamis in the Arctic Ocean? Any modeling of this inevitability?

    Is permafrost melting based on linear heat increases? or non-linear/exponential rise in heating?

  18. 18
    Marcus says:

    Hello David,

    I enjoyed (and even finished!) your first version of that course very much, especially your colourful shirts in the videos.

    [Response: Congrats! Not many people finish (~7%). ]

    I am very interested in that Python modelling course, is that available in that “on demand mode” too?

    [Response: Yes]

    I am suffering a heavy workload till and of April, in the “normal mode” I could not start with that now.


  19. 19
    Hank Roberts says:

    > richard pauli …

    If the methane apocalypse happens, it will happen after many far worse things are already happening or in the pipeline. The argument focusing on that amounts to
    It’s All Gonna’ Blow so why not drill wells, sell and burn it…..

  20. 20
    Joseph O'Sullivan says:

    Does anyone know any good classes for understanding the statistics used in climate science? I generally understand the broad scientific concepts, but when I try to go deeper into numerical parts of the work I get lost.

    I have some basic math (college calculus and a intro to bio-statistics class) but that was 25 years ago. I am looking to learn more.

    Any suggestions?

  21. 21
    Marcus says:

    regarding the “what language is best” discussion: it works surprisingly well to combine C++ with Fortran i.e. calling Fortran spaghetti code from an object oriented driving framework as external C functions.
    This is justified because there exists so much well tested and highly performing Fortran code (in numerical mathematics)
    which should not be thrown away and does not have to be rewritten.


  22. 22

    Modern Fortran is not your grandfather’s Fortran. It has not needed “spaghetti code” since structured Fortran was introduced in 1977. The 1990 standard added modules, the 2003 standard added classes. I think the last version I heard about was 2008.

    Salford Fortran costs about $600, but you can download the Personal Edition for free. It uses the 1995 standard, and even has a GUI interface if you want to use it. The only reason I don’t write everything in Fortran is that I have Visual Basic available.

  23. 23
    sidd says:

    i use gnu fortran all the time, freely downloadable. Speed wise, the intel compilers make the fastest running code on intel hardware, with portland fortran almost as good. (these were the results i got some years ago, may have changed by now. ) Intel fortran used to be free for academic use, i dont know the situation now. Gfort is not slower by enough to make much of a difference to me, ymmv.


  24. 24

    Methane Spikes to Over 3,000 parts per Billion

    But we should wait until it hits 10k or 20k … before we start discussing the issue?

  25. 25

    Dr. Archer,

    I started taking the first course today. Lots of good stuff, well and clearly presented. I have a few minor nitpicks, of course:

    1. The astronomical constants seem to be about 40 years out of date. Some modern figures: S = ~1362 W/m^2 (Kopp and Lean 2011 give 1360.8 at Solar minimum); Bond albedos for Venus, Earth, and Mars respectively = 0.90 (!), 0.30, 0.25; surface temperatures respectively 735.3 K, 288.15 K, 214 K.

    2. In the discussion of band saturation, it might not be clear to the viewer that Doppler and Lorentz broadening apply to individual absorption lines and not to a whole band made up of hundreds or thousands of lines. Granted, the A-R-T properties of a band depend on those of the lines, but that way lies band models (i.e., madness).


  26. 26

    One of the quizzes is graded wrong. The question asks for the difference in outgoing radiation if T = 288.2 K as opposed to 287.2 K, given ε = 1 and σ = 5.67e-8. The flux densities produced are 385.76 and 391.16 W/m^2, respectively, for a difference of 5.4 K. I’ve checked this over and over again. But your quiz says this is wrong, and since subsequent questions depend on this result, I can’t pass the quiz.

    Here it is in Excel:

    σ T T^4 F
    0.0000000567 287.2 6803583861 385.7632049
    0.0000000567 288.2 6898837349 391.1640777
    Delta: 5.400872783