Sunday, March 29, 2009
How models work
Ugh ... sorry folks, this post ain't about sexy Hollywood models with million dollars bods. But if I have your attention, I was really talking about computer models. There are a bazillion kinds of those things so let's narrow it down to Climate Change and Global Warming.
This is big news lately, with our President sounding like he wants carbon dioxide mandates and major folks like Joseph Romm (Climate Progress), Freeman Dyson (a major physics denier dude), and Thomas Friedman (NY Times Pulitzer winner) getting published all the same week, duking it out. And there are thousand of more major peeps banging the pots and pans as well.
What it all comes down to are several ensembles of computer models that predict sea level rise, ice coverage, differential heating, and technical stuff I won't bore you about ... other than all these models suck!
Say what, Sammy?
It's a rather existential question because any models that try to predict things into the future are based on humans, and most of the time Mother Nature has a different idea in mind. So global models for sea ice totally missed their mark, a horrible thing for the "modelers." What happened was that sea ice was supposed to break up in a decade or two, like in the Arctic and now even the western Antarctic. They blew it. The stuff is melting at an incredible rate never forecast by ANY computer model. What happened was a classic case of under-shooting because all models done by humans are fallible.
Then there are there is over-shooting, where some models show global heating right now, even in a mild cooling stage, although in the long term they are perhaps correct because the thinking is "Mother Nature took off a year or two from warming." Excuses, excuses, the models that predicted rampant global warming also failed. Poor Al Gore has been severely abused by these notions.
Then there are the ... what can I say, weirdos who predict "crazy weather" from global warming. These folks don't know the difference between climatology and today's weather and unfortunately must be relegated to the loony bin, as no model can support those conclusions except when modeled results conflict with each other, or somebody tries to bloviate about "super hurricanes." The only folks to get away with that trick was the original Farmers Almanac that predicted sun, rain, hail, sleet, and snow all in the same day in Boston sometime in the 1800s.
A model only works for about 5 years of predictive value and I'm going to leave it at that. We have absolutely no idea what will happen in 50 or 100 years. I'm serious! But before you dismiss me as having sentiments one way or the other, I can tell you that climate change is happening right now. I can also say that if the climate was not changing, we'd be in for some very serious entropy problems.
And so far, as good as our bazillion-dollar satellites and billion-dollar super-computers are, we really can't predict very much other than some obvious things that are already happening. This does not mean that they are bad. It means that climate modeling needs a new generation of thinkers who can plug in the right kinds of data without being so political about it.