Are Suggestions Encyclopedic?
Should predictions and suggestions be in this article? --SEWilco 03:15, 10 Sep 2003 (UTC)
- With model output approaching oberservational data (e.g. from satellite soundings) in resolution, the sheer size of the datasets means that data mining and data management will become equally important considerations in meteorological computing. In light of the decrease in density of surface and rawinsonde observations, new algorithms have to be developed to extract similarly accurate information from satellite data, for example about cloud type and distribution. Data management will become more global in nature, with some central archives storing a large number of numerical experiments from various institutions. This data needs to have a sufficient amount of metadata attached and can then be conveniently retrieved by a WWW interface from anywhere. These new archives will alleviate the important task of comparing experiments conducted with different models, which is instrumental for their further improvement. Also, grid computing may be a interesting way to harness the power of meteorological supercomputers more effectively. Of course, international cooperation is nothing unusual in modelling, but grid computing might automate the process of running a model where the right amount of computing resources are currently available and leave scientists more time for analyzing the results.
- Meteorological instrumentation that is used at the surface or in airplanes also has room for improvement. Radar and Lidar show precipitation and clouds by their effects on emitted monospectral electromagnetic waves. If radar measurements can be used to accurately determine the amount of precipitation (which as of now is only possible with rain gauges), this would be beneficial for numerical weather prediction. Lidar can be used to study clouds that are so thin that they cannot be seen by the naked eye such as certain types of cirrus filaments. The high-altitude clouds that can form from contrails may well be the key to understanding the effect of the increasing amount of air travel on global warming.
- Finally, meteorologists must educate the public more about weather and climate in general. Scientifically accurate and understandable information about topics like the ozone hole, global warming, the effects of rainforest deforestation, or sea level rise must be disseminated and misinformation by industry lobbyists be countered. Particularly in Europe, which may see an increase in extreme weather events as it already has in the 1990s, the population must be educated to pay closer attention to severe weather warnings or information about other detrimental health factors such as high tropospheric ozone concentration or high levels of UV radiation. Similarly, a better infrastructure to deal with natural desasters must be developed akin to similar services in the US. It is clear that political decisionmakers in Europe will rely on scientific assessment to validate the necessity for such spending.
-I agree, this is new research, speculation and opinion. geodynamo 5 July 2005 18:24 (UTC)
Shouldn't this article contain information on how meteologists do what they do?
Aside from the text on wind mapping, there is nothing in here.
- No, this article defines meteorology. See Weather forecasting. Hard Raspy Sci 04:01, 31 December 2006 (UTC)
This article needs a top to bottom overhaul, but I am not able to undertake this endeavor at the moment. For those just waiting for a good source to start off with, the University of Illinois has an excellent online guide to Meteorology: ww2010.amtos.uiuc.edu. It has everything you need to get started on rewriting the sections on current and future forecasting methods.
(also... to the previous poster: be sure to sign your comments... otherwise we can't tell who you are or when you wrote that.)--demonburrito 16:21, 10 August 2005 (UTC)
Without objection: subsection "Possibilities for future improvements" will be removed.--demonburrito 06:31, 11 August 2005 (UTC)
more major editing
I am completely rewriting the 'history' section. This is not just because I didn't like it very much, but because it appears to be entirely plagiarized.--demonburrito 09:26, 11 August 2005 (UTC)
- oops...nevermind about the plagiarism. It realized it was just a site not using GNU license properly. I still stand by liking the timeline better (it is now more complete than timeline of meteorology).--demonburrito 10:30, 11 August 2005 (UTC)
talking to myself
I think this paragraph is controversial:
In the 1960s, the chaotic nature of the atmosphere was first understood by Edward Lorenz, founding the field of chaos theory. The mathematical advances achieved here later filtered back to meteorology and made it possible to describe the limits of predictability inherent in atmospheric modelling. This is known as the butterfly effect, because the growth of disturbances over time means that even one as minute as the flapping of a butterfly's wings could much later cause a large disturbance to form somewhere else.
There are other articles on this wiki that state the belief of some that a complete model of the atmosphere can be described with diff eqs, but the complexity is beyond our means for the time being. Different than "inherent unpredictability".--demonburrito 11:13, 11 August 2005 (UTC)
- Yes, differential equations are solved numerically in weather forecasting models. However, no matter how complete and accurate the model is, uncertainty in the representation of the initial state of the atmosphere will always grow and lead to forecast errors. This is true of all chaotic systems, of which the atmosphere has been proven to be a good example. Therefore I don't think that the fact that the "atmosphere can be described by diff eqs" and stating that it is "inherent unpredictability" are contradicting each other. (see Chaos theory) PeterLean 10:40, 12 August 2005 (UTC)
- Demonburrito; I've reinserted the bit about the chaotic nature of the atmosphere because I believe that it is one of the most important aspects of atmospheric bahaviour for weather forecasting. If you feel that this theory is controversial then have a look at:
- -most of the major operational forecasting centers are beginning to use ensemble forecasts to take account of the uncertainty in the forecasts arising due to the chaotic nature of the atmosphere.
- But, thanks for all the work you've put in on the page; I think that it is a big improvement on what was there before.--PeterLean 11:32, 12 August 2005 (UTC)
- Thanks for the positive reinforcement. I was beginning to think no one was watching this page.
- Believe it or not, I intuitively agree with Lorenz, and have argued for the validity of his theories on many occasions. The popular books on chaos theory (like Gleick's "Chaos") in the late 80s had a big impact on my thinking, and I personally feel that time may demonstrate Lorenz to be correct as theoretical math progresses. However, having spent some time chatting on , I was unable to find a consensus on this issue. I worry that the paragraph may not be encyclopedic. Perhaps we can split the difference and mention chaos theory's role in the advent of ensemble forecasting, and provide a link to chaos theory?
- Chaos theory gives an unpredictable mathematical solution to a system, where the mathematical solution is unpredictable with respect to some exactness, but it does not mean that every solution is entirely unpredictable, just inherently so. (See Henri Poincaré or Phase space). But even this is not a great answer for all cases, however I am not sure if it is an oversimplification.
- The major problem with Chaos theory is that it is a mathematical theory. Most people have a problem with comprehending that not all mathematical (real, correct, or imaginary) solutions give a solution to real physical things. While the methods may be valid and give valid solutions, they are only correct in the realm of mathematics. But, and importantly, it should be understood that Physics (mathematically speaking) is a subset of Mathematics.
- Simply that means that there are solutions in math that don't work in physics, but all solutions in physics work in math! Hard Raspy Sci 02:03, 2 January 2006 (UTC)
Specific: Large scale wind mapping
The direction, strength, and variability of winds over various parts of the earth have been of great interest for a long time. The first large-scale wind mapping efforts were done to aid sailors who depended on winds to power their ships. In the last four decades a considerable amount of work has been done to map winds which might be used to generate power.
It is difficult to map the wind because it is a chaotic system. Weather forecasting is extremely difficult to do accurately for more than a few days. However, statistical methods can be used to predict general expectations of wind. Resolution can be low, e.g. maps of an entire continent; or high, where a specific site is assesed. High resolution maps are used for example to asses the viability of a site for wind turbines.
The inputs to the mapping process are: measured wind data and topography, and possibly other meteorological data which might relate to seasonal variation, pressure, and humidity. Humidity and temperature affect the density of the air, and thus the power produced from a given wind speed. Proximity of the sea has diurnal and seasonal effects on wind patterns. The 'WAsP' model uses a simple algorithm to predict wind over simple topography. It does not solve the Navier-Stokes equations which, in principle, give the correct answers to how the speed and direction of wind changes as it flows over 'complex' terrain.
A typical process for a large-scale wind map would take the wind data, and separate it into groups based on direction and speed. For each group, the wind is simulated using computer-based models, and the result is recorded. The results for each group are then averaged according to the frequency of that speed and direction 'bucket', giving the average wind over an area. This is only indicative, and before a wind farm is constructed, detailed local measurements are essential.
For small areas, the Navier-Stokes equations can be solved with some success, atechnique employed more often as computer power grows to match the task. With this method, the volume of air around the site of interest is cut into discrete chunks. Each is considered independently, like a pixel of a computer display. Each chunk interacts with its neighbours according to their pressure (although this can be neglected), and wind speeds. The smaller the chunk sizes, the closer the result is to reality, although large divergence is not uncommon. These models are extremely computationally expensive to run.
'Simple' terrain can be modelled using simple mass balance: if there is a hill in the way, the wind is accelerated. This is the Bernoulli Effect, and the core of the WAsP model.
Wind is powered by a temperature differential. It is slowed by obstructions and is generally stronger at high altitudes. Plains have high winds because they have few obstructions. Mountain passes have high winds mostly because they funnel high-altitude winds. Some passes have winds powered by a temperature differential between the sides of the ridges. Coastal areas have high winds because water has few obstructions and because of the temperature difference between the land and the sea. Off-shore also generally has high winds for the same reasons.
I think this is all fine work, but I also think that this is too specific to be included in the general meteorology article. I'm quoting it in entirety to preserve it on the talk page. --demonburrito 03:56, 12 August 2005 (UTC)
Relevancy of chaos theory for weather science
I agree that the term 'inherent unpredictability' is inappropriate for any form of dynamics that is 'chaotic' in character. Somewhat closer would be: 'inherent decay of predictability'. In the case of a chaotic system (chaotic in the specific technical sense of chaos theory), the reliability of even the best attempt at forecast drops exponentially with the time span of the forecast. A weather forecast 3 days into the future is pretty likely to be right on the mark, these days. The likelyhood of a forecast 6 days into the future being right is much smaller than half the odds of the 3-day ahead forecast being correct.
In the article it is stated: "the growth of disturbances over time means that even one as minute as the flapping of a butterfly's wings could much later cause a large disturbance to form somewhere else." That statement is simply wrong, the butterfly image is a metaphor, it is simply wrong to take it literally. --Cleon Teunissen | Talk 21:06, 13 August 2005 (UTC)
- Errm, well, hold on. The butterflys wing stuff is probably widely misunderstood, but isn't wrong. Any disturbance, even one far smaller than a butterflys wing, would eventually grown to the extent that different realisations (were that possible: it certainly is in a model) would be completely different. The error is in assuming that butterflys wings cause tornadoes in some direct sense, which is wrong. William M. Connolley 14:21:57, 2005-08-20 (UTC).
- In a sense one can say that the weather 14 days ahead being unpredictable is a consequence of a gazillion disturbances, sometimes reinforcing each other, sometimes annihilating each other. A truly exhaustive equation of motion would have to take each drop of a leaf into account, for a leaf still being attached to the tree or the leaf being torn off makes (theoretically) a difference, etc. etc.
- I see a twofold wrongness in taking the metaphor of the 'butterfly flapping its wings' literally. The one you mention, and the following: that an exhaustive equation of motion would have to take a gazillion disturbances into account simultaneously, each disturbance contributing on average one gazillionth to the decay of predictability. Each individual disturbance is rapidly diluted into oblivion in the process. The principle characteristic of systems that are categorized as 'chaotic systems' is that tracing cause-effect connections becomes exponentially harder to achieve with increasing time span. The path of any large development, such as a hurricane, is determined by innumerable tiny individual contributions. --Cleon Teunissen | Talk 10:55, 22 August 2005 (UTC)
- I agree entirely with William Connolley's comments. It's true that a perfect forecast would require a perfect model along with perfect initial conditions which would need to include all of the many small disturbance occuring. However, even just one small disturbance not included in the initial conditions can (if the conditions are unstable and allow perturbations to grow) eventually lead to a total loss of predictability. It seems probable that often a single butterfly flapping its wings would make no difference to a forecast at all in the long run if the conditions are such that the disturbance simply decays, but in unstable conditions it can grow and lead to a total loss of predictability. --PeterLean 16:32, 22 August 2005 (UTC)
I have linked the previous red-link thermoscope to Galileo thermometer. Am am pretty sure that is what is what is meant by thermoscope. While the name does sound reasonable for an early thermometer, I have not heard the word before. Gaius Cornelius 07:31, 1 October 2005 (UTC)