25 May 2012
Stages of numerical modeling
Posted by Jessica Ball
I’m currently working on some modeling for my thesis. For unrelated reasons, I happened to read a description of the Kübler-Ross model for stages of grief, and I realized that the cycle actually describes pretty accurately what the past couple of weeks have been like for me. Not only that, but it’s gotten to the point where even if I get my model to run, I’m immediately suspicious of the results. However, I guess since the model is running, I’ve made progress. That doesn’t mean I don’t still have issues…
1.Denial:
Not again! Is the model screwing up somewhere? It can’t be my inputs, I worked really hard on them.
2. Anger:
Stupid model. Why won’t you run? What’s wrong with you? I just wanted to do something simple and you’re crashing!
3. Bargaining:
Please, model, work just this once. I swear I’ll never shout at you again. I will lovingly craft each and every line of input and gently type ‘start’ on the keyboard, instead of swearing and slamming my hands down on the keys. I will speak encouragingly when you stall. Just please, please run all the way through this time.
4. Depression:
I suck at modeling. I’ll never get this thing to run. Everything I do is wrong. This is way too complicated to ever work properly, and everyone is going to look at my results and tell me they’re crap. I should be out in the field with a rock hammer, not in front of a computer screen trying to simulate things.
5. Acceptance:
I guess I can redo everything and try again…
6. Chagrin (this is where we deviate from the Kübler-Ross model):
That’s it? That’s all that was wrong? One measly negative sign? And now it runs? …….Wow, I feel really stupid.
7. Exuberance
Holy crap, it worked! My project won’t be a complete failure after all! I will have results! Hooray!
[Make changes to model. Attempt to run. Model crashes and/or outputs make no sense. Return immediately to stage #1. Repeat indefinitely.]
Geologists Orrin and Linda Pilkey wrote a bood a few years ago called “Useless Arithmetic: Why Environmental Scientists Can’t Predict the Future.” In it, they are particularly hard on groundwater modelers. coastal engineers, and fishery managers.
The jist is, modelers lack enough understanding of processes to identify enough variables, to make enough measurements, to collect enough data, to do a good job with numerical modeling. But what we can do to understand process is still hugely valuable.