Inside a Computer Model – Hurricane Katrina

What you see here is a computed visualization of the computed flow of the hurricane over about a day. The track of the virtual camera was chosen by a human, and there was some tweaking of the color scheme to represent day and night. But the maximum wind tracks and the cloud formations were all computed. This may give you some idea of the complexity and skill nature of contemporary atmospheric simulations.

Comments:

  1. "This may give you some idea of the complexity and skill of contemporary atmospheric simulations."

    Very nice eye candy, but it's not at all clear to me how to learn anything about the skill of the simulation from the movie when it doesn't (a) compare the simulation to the actual wind fields and (b) compare that comparison to what you'd get from a naive model of the wind fields.

    All I can see is that it produces a nice cyclonic spiral with dramatically blooming cumulus towers, but I can't see anything to tell me whether this cyclonic spiral and its distribution of gusts and cloud towers is better than any other.

    I'm not saying this to trash the model---for all I know, it's incredible---but to caution against the tendency so many of us have (I'm very susceptible when I don't watch myself) of accepting something uncritically when it's presented with cool visuals.

    • The point here is simply to assert the complexity of the model and give some insight into its nature; specifically, how many features are kept in the system and how they interact on a physical basis.

      In actual fact, the models of Katrina were iextremely successful in predicting its trajectory and intensity.

      But many people fail to understand the difference between the sort of gross statistical models possible in other fields and the actual physical simulation possible in atmospheric science. That is the point here.

      That all said, welcome to P3, Jonathan!

  2. I'd be interested (from a layperson viewpoint, not as an expert) in the nature of the simulation in terms of what interactions were modeled, the scale of the finite elements, the time to generate both the data and the visualization, the nature of the hardware, how many people and person-hours were spent in the generation of the data and the visualization, etc.

  3. I read that for a time climate modelers had delighted in mischievously shown simulations to weather experts. They later told them the sequences were simulations and not satellite pics. Most hadn't noticed. That tells me their models capture many details of global circulation.

    The NCSA at UIUC specialize in scientific visualization. They helped create a sequence for the movie Tree of Life using a UT astronomer's data. That one draw of my going to see the movie but the sequence ended up meaning little as no explanation was included. As noted, that seems to be the case in the Katrina video.

    http://web5.cns.utexas.edu/news/2011/10/simulating-the-universe/

    Terrence Malick movies are so slow moving, I preferred Days of Summer.

  4. Yabbut what do the colored lines mean? At least, I guess the lines represent winds, but the colors?

    I further guess this is a question from the department of "one person is always more ignorant than the most basic explanation". I just wish it wasn't always me

    • Honestly, I don't know, but my guess is they passed through a specified surface at different tags and were given different colors of "virtual dye". Visualizing flow is tricky and this is one trick that is used sometimes.

      Again the distinction between what the model is doing and what the movie is showing is at issue. The movie is a way of looking at the output of the model, which originally appears as millions of numbers on a disk drive. And the coloring is a part of that visualization process, not of the modeling itself.


Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>