Light-Bulbs in the Shower

I’ve heard the stories of many a grad student slaving away mentally, stumped day and night about their particular research problem. Stumped, that is, until inspiration strikes them. Oftentimes these “light-bulb” moments seem to happen in the shower, of all places.
As of my shower this morning, you can count me amongst them.
I’m not quite sure what the right word is, but it’d be fair to say that I’ve been feeling kinda down lately because of the status of my PhD research project (among other things). I’ve had a general idea for awhile of the main part of what I want to accomplish — to find the best configuration of a numerical weather prediction ensemble for the purposes of atmospheric transport & dispersion prediction. The primary challenge has been how to determine what that “best” configuration is (which combinations of model physics options to use, how to perturb the initial/boundary conditions, etc.). I’ve been told that to make my project PhD-worthy I need to come up with some sort of quantitative foundation, a mathematical or statistical way of determining what configuration is best. Fair enough. I’ve been under the impression for awhile though that that meant that I’d have to try to devise some way of predicting or determining a priori what configuration would be best. That’s what’s been getting me down and apprehensive for awhile, because I haven’t been able to see a way even to start that. Even though I know that all the other parts of my project are going to take work and that problems will undoubtedly crop up along the way, I really haven’t been worried about the rest of my project. If only I could figure out how to lay a mathematical foundation for my project, I felt like I’d be set.
I was becoming more and more convinced that it would be impossible to determine the best configuration beforehand, though. In a fully three-dimensional NWP model, there are just too many feedbacks and physical processes interacting with each other to try to isolate or predict the effects of changing any one model option (which is itself pretty complicated) across an entire spatial and temporal domain. So if I can’t make a quantitative determination of that beforehand, what about after the fact? Hmmmm… Then it struck me this morning. The core of my idea is to create my ensemble, let it run for a few months to a year (so that we have a big enough sample to allow for statistically significant conclusions), and then calculate ensemble means, spreads, errors, etc. for the ensemble as a whole as well as several mini-ensembles (all being various subsets of the whole). Having that information then for many “mini-ensembles” should hopefully allow me to make a determination or assessment after the fact about which ensemble configuration is “best.” It’s not exactly an earth-shattering idea, but hey, it’s an idea, when I was previously bereft of them! I told Sue about my idea this afternoon, and she reacted positively to it too. 🙂 The idea admittedly needs to be fleshed out quite a bit more, but hey, I only came up with it this morning in the shower!
So today I’ve had a much more positive outlook about my research project. And considering how much the ups and downs of research tend to dominate the life of a grad student, that also means that my outlook on life (or at least my life for the coming months) is much more positive now too, haha. Now I think I might actually be able to make some solid progress on this in time for the GMU conference in July and for my comprehensive exams in early August! It’s amazing what power a single idea can have. And it’s an answer to prayer, really.
Now I’ll actually have some serious time to spend on my PhD project too. First, I finished Advanced/Mesoscale Forecasting, my last-ever class (for credit, anyway), last week since it didn’t have a final. It’s amazing how much less stressful finals week (this week here at PSU) is when you don’t have any finals. 🙂 And second, I finally was able to submit the revised version of my first-ever publication to the Journal of Applied Meteorology and Climatology on Thursday last week. It was accepted with minor revisions back in February, so it shouldn’t be too much longer before it gets formally accepted in full now that we have the revisions in! The revised manuscript would’ve been submitted a few days earlier, except that I stumbled upon a bug in one of my programs. And hooray for bug-fixing that makes results look even better! Anyway, with that out of my hair, I can finally stop working on stuff from my Master’s project and actually start focusing on my PhD project. As much as I like it here at Penn State, it still would be nice to graduate eventually!