|
Post by nonentropic on Jun 16, 2017 21:01:49 GMT
the probabalistic approach to weather forecasting is legitimate.
The issue I have relates to the usefulness of the representation. In oil and gas exploration it is the norm to state reserves as P10, P50 and P90 where each band is the probability of the produced reserves exceeding a defined value, believe me this has been very successful for all sorts of reasons in that industry. Somehow weather is different and that is the lack of test-ability unlike an oil field where you have constant reminders of how clever you were.
So a legitimate but poor method for weather, they in my mind don't mind this lack of apparent comparability.
|
|
|
Post by nautonnier on Jun 17, 2017 1:57:52 GMT
Well actually not. I remember doing a lecture to a set of meteorologists (at Boulder NCEP believe it or not) on probabilities and perception. Probability is perceived in different ways by different groups. A meteorologist giving a 50% probability is saying 'very likely'. The analogy I used was imagine playing Russian roulette with one shell in a 6 shot revolver.... now load another 2 shells - you are now at 50% but your perception of the risk has now risen to the forecaster's perception of a 50% risk. What is the probability that events will occur according to the odds, Mr. Naut? The problem is that all we have is conflicting forecasts. The wiggle watchers say this is what happened before when it wiggled like this - therefore we would expect it to wiggle that way again. In the short term this is a very accurate way of forecasting. "It will be what it was before". However, when things start to change, especially in chaotic systems, that model breaks down. I think, if Theo is right, that the weather will relatively suddenly alter not a 'tipping point' which the climate 'scientists' are so apt to claim. Just a change toward a different chaotic attractor. It depends on the states of multiple parameters that influence what the atmosphere and seas will do many of which we are probably unaware of. It is worth reading about Lorentz and his original discovery of chaos in weather forecasting. Or for that matter of Galileo watching the swinging chandeliers in church. James Gleick wrote a very readable book on Chaos that I found interesting many years ago - it is worth a read. Probability does not deal very well with 'chaos' as chaos despite its colloquial semantics, is actually deterministic, it is just that anyone less omniscient that La Place's Daemon will not be able to see all the various effects of all the minor changes in all the variables.
|
|
|
Post by nautonnier on Jun 17, 2017 2:24:38 GMT
Well actually not. I remember doing a lecture to a set of meteorologists (at Boulder NCEP believe it or not) on probabilities and perception. Probability is perceived in different ways by different groups. A meteorologist giving a 50% probability is saying 'very likely'. The analogy I used was imagine playing Russian roulette with one shell in a 6 shot revolver.... now load another 2 shells - you are now at 50% but your perception of the risk has now risen to the forecaster's perception of a 50% risk. I normally do not blow my own trumpet At the time I did that lecture I was Director of a Research Center of a university in Florida and was working with a group of meteorologists supporting the Collaborative Convective Weather Forecast Program (CCFP). www.weather.gov/zme/ccfp. I had worked with meteorologists and forecasters, such as the Aviation Weather Service and the European equivalent, for many years as an 'operational user' of their products. Since 1986 I have worked in research, development and testing of future air traffic management and aircraft systems. Many of you who have flown in Europe will have been in aircraft controlled by air traffic software that I helped design, and some by software that I actually wrote. However, typically, I am working on system architectures 15+ years ahead of current systems and spend my time selling, validating and implementing future ideas; now in a more commercial setting. It is only 90 years since the first air mail service between New York and San Francisco was started; and around the same time the first airline, KLM started operations. Some of the concepts they used are still enforced today - they are about to change significantly. Well you asked
|
|
|
Post by Ratty on Jun 17, 2017 5:19:26 GMT
Naut, I'm sure I saw you on Air Crash Investigations ... now you are an advisor to the Trump Administration?
|
|
|
Post by AstroMet on Jun 27, 2017 17:08:19 GMT
Well actually not. I remember doing a lecture to a set of meteorologists (at Boulder NCEP believe it or not) on probabilities and perception. Probability is perceived in different ways by different groups. A meteorologist giving a 50% probability is saying 'very likely'. The analogy I used was imagine playing Russian roulette with one shell in a 6 shot revolver.... now load another 2 shells - you are now at 50% but your perception of the risk has now risen to the forecaster's perception of a 50% risk. Well Nautonnier, forecasting is indeed a science, but weather and climate forecasting is even more of an applied science because it deals with the most difficult of mathematics, that is variable math. The problem with conventional forecasters is that they continue to depend on radar, and computer model probabilities. Of course, the great majority cannot forecast further out than 10 days, while monthly and seasonal forecasting is done by very few like me. Longer-range is easier for me because as a astronomic forecaster, I can extend out years (even decades) because I know that all of our climate and weather conditions are forced from outer space, and so I know where the Earth will be relative to the Sun, it's condition, as well as the Moon and planets relative to the Earth's position at any time in the future.
|
|
|
Post by AstroMet on Jul 6, 2017 21:30:02 GMT
|
|