[--MAIN HOME--] [--ALL HABYHINTS--] [--FACEBOOK PAGE--]

 WILL FORECASTS ALWAYS BE PROBABILITY FORECASTS?

METEOROLOGIST JEFF HABY

The prediction of certainty is a prediction that has 100% chance of coming true. 100% is the maximum probability that an event will happen. When it is stated the chance of an event is 110%, it is an exaggeration meant to imply that the chance of the event is very likely. This phrase is often heard in the context of “I will give 110%” effort. This means giving an extremely strong effort, but the maximum measureable effort is 100%. A prediction of probability ranges from 0% to 100%. When given as a decimal, this range is from 0 to 1.

A 100% chance can verify if there is only 1 possible outcome for the event (100% chance the numbered dice will land on a number) or the statement is given that any of the possible outcomes happening is the prediction (tomorrow the chance for weather is 100%). A 100% chance can also verify if the event has already happened (there is 100% chance of rain for yesterday and it rained yesterday). Yogi Berra has a phrase that goes something like, “It is tough to make predictions, especially about the future”. This is true since options lead to probability. Nothing is for certain when there are options involved.

Generally, a 95% or above chance is used to express a very strong certainty. It should be very surprising (or disappointing) when anything with a 95% or higher chance does not verify. A probability of over 50% is more likely to happen than not to happen. Chances of rain of 70% and 80% are more likely to happen than to not happen. A probability of under 50% is less likely to happen than to happen. Chances of rain of 20% and 30% are more likely not to happen than to happen. A probability of 50% has an equal likelihood of happening and not happening. Flipping a fair 2 sided coin is an example. The lowest chance for rain typically mentioned is 20%. Once the chance for rain drops below 20%, then typically precipitation is not placed into the forecast since it is so unlikely to happen. You may time to time run into a forecast though that mentions the chance for precipitation is less than 20% or mentions the chance for precipitation is 10%. When it looks like precipitation is extremely likely, you may hear something like, “the chance for precipitation is near 100%”. If the rain is already happening and verified for a specific location, you may also hear the chance of precipitation today being 100%.

Future predictions in cases where there are multiple outcomes possible can not be guaranteed to an accuracy of 100% due to quantum fluctuations in matter, time and space. In a purely mechanical universe, predictions can have 100% accuracy but due to quantum mechanics it has been determined that there has to be an underlying uncertainty in a future event. Mechanical predictions in our universe can be very accurate but can not be perfect. Accuracy can be increased due to more data, better computational methods and more accurate equations. This is true in the field of meteorology. More balloon launches, better radar data systems, more ocean data, a higher density and accuracy of field reporting stations, more realistic forecast model equations, a better synthesis of data, better satellite weather data, etc., will all contribute to more accurate weather forecasts but this accuracy will have a limit that can not reach 100%. Very small size phenomena can not be accurately represented by the forecast models since they don’t have the resolution to sense data with this sort of density. Very small changes and errors turn into significant changes and errors when given enough time. Thus, weather forecasts will always retain the quality of having probability predictions. This is not really bad news since nothing else can be perfectly predicted either (i.e. NFL football games, future news about politics).