This reference from Dr. Melissa Bukovsky, our friend at NCAR should be useful. I’d like us to understand the models inside professional grade weather models CM1 and WRF.
Brandon, can you and Julian please estimate the work / benefit of either plugging those engines into our media system, or implementing some toy version? Do chat with Connor because he may have peeked at the CM1 refs last summer. (Kudos Connor)
Rather than reinvent the wheel we should take seriously Melissa’s recommendation re CM1: “This model is designed for idealized studies, unlike others I work with, so its code is much simpler and it will run much more efficiently. It
can also be used with different equation sets.”
What I’ve learned from doing scientific simulations for multiple audiences is that the smartest approach is
not to make toy models (like NetLogo) motivated by “education” or “public communications” bc that burns up infinite amounts of engineering labor for CS lab “demos” rather than effective pedagogy or credible science.
A more effective and rewarding way to invest our engineering expertise and labor is to
take a professional grade system and mask its content or features appropriately to fit the user. The weather modellers already do something like this — they call it
idealization.
Brandon, Josh:
There’s a very strong tendency among computer scientists to make toy models of math and science that satisfy us coders but don’t engage real science or science education. Instead, I want us to do more professional software engineering, engineering
with more anthropological tact and sociological acuity.
At Stanford, with a worldwide net of simulations researchers and teachers, we did this “masking" with a large range of Mathematica modules, from pharma physiology, geophysics, ODE’s, to elementary differential geometry and cosmology to good effect.
It is no more labor, and often actually less software engineering. More brain (reading other people’s science in papers and textbooks) and less brawn (coding from scratch based only on images output from other people). What we did was to code or acquire
code that did computations used by professionals in the field, be it physics, maths, medicine, or econometrics, and carefully make student versions by encapsulation, masking, thinning, scaffolding with multimedia narrative etc. This way instead of infantlizing
we give them a headroom. This way, the software environment can support their learning and growing to whatever extent they can, all the way to the top of the field’s practice.
The analogy would be giving an apprentice cabinet maker real hammers and nails to work with real wood, rather than plastic toy hammers and sponge.
Also, coding everything yourself from scratch is a typical novice programmer’s approach to the world, but software programmers with real world experience learn to reuse other people’s work. (With proper permission and attribution. :)
Live long and prosper,
Xin Wei
Begin forwarded message:
Subject:
Intel/NSF grant and modeling resources
Date:
February 5, 2015 at 4:23:30 PM MST
Hi Xin Wei and Chris,
…
Second, I had a brief discussion with a colleague here (George Bryan) about which numerical weather prediction/climate model it would be best to start with, and I've come up with a good option. It's his model (I knew it existed, I just needed
to know more about it before suggesting it). This model is designed for idealized studies, unlike others I work with, so its code is much simpler and it will run much more efficiently. It can also be used with different equation sets. I think it would be
perfect for expanding your current system to 3-dimensions while adding atmospheric variables (temperature, moisture, etc.). It has mostly been used for idealized thunderstorm studies too, which is perfect. It's webpage is here in case you'd like to see more
about it (the code is also there):
http://www2.mmm.ucar.edu/people/bryan/cm1/
Third, I found a few good educational modules for learning about weather/climate modeling, if you are still interested in them...
An introduction to climate models:
https://www.meted.ucar.edu/nwp/climate_models/index.htm
Modeling fundamentals:
https://www.meted.ucar.edu/nwp/model_fundamentals/index.htm
This one is much more complex, but has some good (if old) visualizations in it. It's about a specific type of thunderstorm system called a Mesoscale Convective System (MCS). These are an important part of the monsoon system.
https://www.meted.ucar.edu/convectn/mcs/mcsweb/mcsframe.htm
I look forward to talking tomorrow afternoon.
Cheers,
Melissa
