Weather prediction

Today weather forecasting is usually pretty reliable, at least over the short term of a few days or so. The ability to do this is especially important for warning people about upcoming storms, especially life-threatening ones like hurricanes. This prediction is done by tracking the movement of high and low pressure areas and of fronts and subsequently predicting where they will be in the future. This requires large amounts of data, much of which is gathered by Doppler radar. Like in police radar used to catch speeding motorists, radar (or microwave) waves are sent out which subsequently get reflected by clouds, etc. The time it takes for the waves to get reflected gives a measure of the distance to the object, and the type of reflected wave gives an indication of the density of the object. As well, if the source is moving, a Doppler shift in the frequency results for the reflected wave - recall that a moving source of waves emits a higher or lower frequency of the wave, compared to a stationary source, depending on whether the source is moving towards the observer or away. Thus, by comparing the frequency of the emitted and reflected wave, one can tell the speed and direction that the object is moving.

  This and other data is subsequently fed into large computers which predict, from the conditions reflected in this data, how these systems will change in time. Such predictions, over the short term, are fairly reliable, but we are not yet at the stage to be able to give reliable long-term predictions. Part of the reason for this is the lack of computer power to hold and process the huge amounts of data that would be needed, but also due to the fact that, over the long term, the equations used to predict movement of weather systems exhibit chaos: a small change in the initial conditions can greatly influence the subsequent evolution over a long time scale. This is also referred to as the butterfly effect: a butterfly flapping its wings in South America will produce an extremely small change in the air pressure, but over the long term this could affect the weather in North America. Thus, as well as the practical limitations of data gathering, there seems to be some barriers in principle to being able to predict the weather over a long time scale.