Unidata

The 2006 Unidata User's Workshop "Expanding the Use of Models as Educational Tools in the Atmospheric & Related Sciences" highlighted ways to integrate models into classroom. While it was generally agreed that models are useful tools, pedagogical approaches were heavily debated.

Should models be used as 'black-boxes'? At what level of education is it appropriate to expect students to understand the nuts-and-bolts of a model? Is it critical for a sophomore to understand the stability of a particular finite difference scheme, or should the focus be on the assumptions behind certain parameterization schemes?

Another interesting topic focused on how to address the questions above based on the institution size - how should an undergraduate only institution approach things when only one course provides the opportunity to discuss modeling?

What are your thoughts?

Sean, Excellent questions, and hopefully this issue will soon be further dissected by Leigh Orf and John Snow in a BAMS paper. Those 2 have somewhat differing views on the matter, if I recall. My personal views are middle-of-the-road. I agree that models can be used to teach about physical processes even without a complete understanding of what is "under the hood", but I also agree that there can be danger in that. If a model is "good", meaning that it is a legitimate representation of the true physical earth system, or if it at least has a realistic representation of a true physical process, then I am sure that it can be used to teach about that process. However, this is not a substitute for a complete understanding about how the real system works.

Also, there is a tendency for students to learn a great deal about how a *model* atmosphere (or other physical system) works, and how to manipulate that in the model world. There still needs to be some cross-verification to ensure that the model is accurately representing the system that it is supposed to represent. Those kinds of studies seem fewer in number than studies using the models themselves. And I admit that I am guilty of that as well- it is much easier to work with a model than it is to verify that it is realistic, especially for complex models.

Finally, I feel that for rigorous hypothesis testing and isolation of specific processes, I feel that models are a great way to go, provided that the model is "good". There are so many powerful things that can be done. And with simple models, allowing students to omit or alter a process allows them to deepen their understanding and elucidate behaviors and interactions in ways that would not be possible with observational data alone.