Day 73 - Output

Today was a great day. I met some new meteorology and CU people, got to know some people better, made progress on a GREAT new graphic design gig that I got a couple days ago (I'll tell you more later), and my nighttime cooling model is finally starting to do something (more on that in a minute).

The only downer today was lunch. It's so great to eat lunch at home while watching TV (usually Seinfeld or King of the Hill) or even while sitting at the computer working on something. But I can't do that at Windsor. And eating in a crowded cafeteria is totally not my kind of thing. It's impossible to have much of a meaningful conversation about anything in a situation like that (crowded, time-limited, etc.). I prefer sitting by myself and just thinking about stuff - but when I do that, everyone else probably thinks I'm some kind of weirdo. Still, it beats sitting there trying to think of idle things to say just to talk. Blah. Dinner tonight was a lot better because it thinned out and I actually got to talk to a couple people about more than just "so how were your lectures today?".

Alright, back to the model. It does something now! Here is a plot of my forecast temperature vs. the actual temperature at a Micronet site for the night of September 10, 2004 (the day I chose as the starting point for my calibrations).



As you can see, it performs beautifully for the first two hours. It nails the initial cooling rate and general look of the temperature trend. But something happens at t = 2hrs that the model doesn't pick up. Condensation! See the left-turn made by the actual temperature at 2hrs? The latent heat of condensation reduces the cooling rate once condensation begins (i.e., 2 hrs after sunset; and if that doesn't make sense, email me and I can explain it). The model doesn't take that into account yet. But the good news is, I think I can represent condensation with a fairly simple parameterization. We'll see. There may also be some other factors that I haven't considered yet. And this is just an easy ideal case. The real test will be predicting the wild temperature swings... well, you'll just have to wait and see.

posted by Matt @ 10:07 PM