justin said:
Well, they seem to be able to at least guess on the correct general direction, speed and likely strike zone of things like hurricanes. We know the approximate (within a day or so) time when cold fronts are coming through (= storms and lower temps). It's not 100% but it's a lot better than flipping a coin.
I check the "5 day forecast" from the local news channel and figure it gives me a pretty good indicator of (1) is it going to rain within 5 days and (2) is it going to be hot or cold the next few days. I don't see any equivalent "5 day forecast" for the market.
Sam said:
I disagree. Weather forecasting is a sound science. Not perfect yet, but it should get better with time.
I hear what you guys are saying, but I have to live with this crap every day. Spouse used to be the Ops officer at the Joint Typhoon Warning Center, her relief has his doctorate in tropical meteorology and now works at the National Hurricane Center, and she regularly went to conferences with guys like Stacey Stewart or had Reservists in similar jobs come out to JTWC for duty.
The most accurate weather forecaster in the world is a holy grail called "persistence". Persistence says that tomorrow's weather will be the same as today's weather. That's it. Persistence is usually right just over 50% of the time and the next-closest forecaster is no better than 45%.
Everything else is a forecaster looking at charts and talking to the guy who's an hour or two upwind. We've known about that concept since Ben Franklin, who first wrote up the idea, and that's based more on dead reckoning than on science. If some meteorological phenomenon deviates from persistence then the vast majority of forecasters are about to bust.
Hurricane forecasts are largely based on military-funded software called SAFA. It was originally developed by crunching all of the science's established forecasting models through various supercomputers and assigning predictive accuracy factors to them. Then those various models were collapsed into a historical database. SAFA is the world's best hurricane track prediction system, but it does it by looking at all the popular propagation models and essentially judging which model was the most accurate before in this particular location under these particular conditions at this particular time of the year. Ain't no science there-- just a slightly more educated guess that past predicts prologue.
Since it was installed it's been tweaked by various other models and its parameters have steadily improved, but that five-day forecast is only about half as inaccurate as it used to be a decade ago. The JTWC watch officers learned very quickly not to forecast outside the time/distance parameters of SAFA. It's given everyone the confidence to work on seven-day hurricane forecasts, but every step forward can be blown away by a butterfly wingflap.
The major forecasting improvements are based on better satellite sensors from better orbits, better on-site data (by windsondes dropped from aircraft like the hurricane hunters), and on better predictive software like SAFA. But it's not based on an improved understanding of the thermodynamics, fluid dynamics, or even chaos theory. Some of the predictive science is making small theoretical improvements but that modeling is very dependent on the grid-spacing teraflops of analysis of various supercomputers. The U.S. used to lead this field about 20 years ago but the Japanese were ahead five years ago. I don't think any of the world's top 25 supercomputers are even at weather centers anymore, even though they used to be more than half of that list.
I think, however, that SAFA would be one helluva predictive tool at multi-deck blackjack, as long as every card in the deck had been seen before. If something changed the deck (like a strong El Nino or global warming reaching a tipping point) then all bets are off...