Weather 2000 - FAQ / Scientific Integrity in Weather Forecasting

Weather 2000 - FAQ Title Graphic

	I saw someone forecast that a Hurricane was going to hit Florida in a week to ten days. Can they know for sure?
	No. Hurricanes and Tropical Cyclones are relatively small and complex systems. In fact, the average forecast track error five days out is currently around 250 miles. Further into the future, track errors grow even larger; for comparison, the distance from Corpus Christi to Port Arthur is 250 miles. This is exemplified by the miscalculated tracks of Hurricanes Ivan (2004) and Ike (2008) and especially Tropical Storm Debby (2012) - see images below. A big contributor to this forecast error is Chaos Theory, which is further examined on our FAQ Page.

	Hurricane Ivan (2004)	Hurricane Ike (2008)	Tropical Storm Debby (2012)
	Five Day Forecasts by the National Hurricane Center


	I've heard that the weather can be forecasted for a single day several weeks from now, or a single week months from now. It sounds impressive, but is it really scientifically legitimate?
	Such claims are nonsense. Any academic institution would defend this point vehemently. Because the atmosphere is chaotic, an accurate daily weather forecast will lose validity after about 10 days or so. Even within the 5 - 10 day range, daily errors can be large (see graphic). You can calculate the historical average for a given day, or examine the record extremes for that day, but no greater precision is possible. Other weather parameters, such as storm tracks or precipitation amounts will display even greater forecast errors. "no verifiable skill exists or is likely to exist for forecasting day-to-day weather changes beyond two weeks. Claims to the contrary should be viewed with skepticism." - American Meteorological Society (AMS) policy statement on weather analysis and forecasting. "The theoretical limit on daily (and weekly) weather prediction is about two weeks. Weather forecasts for individual days and weeks beyond this limit cannot be skillful except through luck. Therefore, daily / weekly long-range forecasts are formulated in terms of time averages of the predictands (historical averages or normals). Because these long-range predictions pertain to times beyond even the theoretical limits on deterministic forecasts, they are now, and must essentially forever, remain uncertain." - Dr. Daniel S. Wilks, Cornell University Professor of Statistical & Agricultural Meteorology, Chair of the Probability & Statistics Committee of the AMS

	But with new research discoveries and computer models, isn't it possible that daily forecasting beyond two weeks will become possible in the future?
	No. That is the interesting thing about chaotic systems with infinite variables, such as our atmosphere. Regardless of computer power, or observational resolution, predictions will always become significantly divergent beyond about two weeks, due to chaos theory. However, computer modeling and new research is a valuable part of atmospheric science. Short range forecasting has and will continue to become more precise, as will long-range predictions *at monthly resolutions. "the sequence of [ daily ] weather events cannot be predicted precisely beyond 1-2 weeks, [ but ] the atmospheric circulation and precipitation, averaged for an entire season, are predictable."* - Center for Ocean-Land-Atmosphere Studies (COLA).


	*Speaking of chaos theory, is it possible that research in this* field might make long-range daily predictions possible?**
	No. Ensemble forecasting techniques, which developed around chaos theory, have improved and/or yielded information about the confidence surrounding different daily forecast scenarios in the 1-2 week range. But, mathematical chaos itself causes the inherent impossibility of long-range daily predictions, which will always be present.


	Can sunspots and other solar activity be used to predict long range weather?
	Correlations made between solar activity and daily, monthly or seasonal weather are unsubstantiated. No testable physical mechanism exists to explain correlations between sunspots and climate. "Despite much research, no connection between solar variations and weather has ever been unequivocally established. Apparent correlations have almost always faltered when put to critical statistical examination or have failed when tested with different data sets. As a result the subject has been one of continual controversy and debate. - National Research Council, National Academy of Sciences

	So how can I determine if all of these extraordinary forecasting claims, new secret and "proprietary" techniques and discoveries are really all just fiction?
	If you have suspicions about a new forecasting correlation or technique, ask to see publication references. Verifying all forecast claims and forecast products with the American Meteorological Society (AMS) (and specifically their probability and statistics committee), is strongly recommended. Unless a peer-reviewed science journal article has thoroughly analyzed the specific claim, then it is highly unlikely it has any validity. Unfortunately, the public is often manipulated, and this only damages the integrity of the science. Because we live in a scientific age, with enormous technological innovation, we are more apt to believe that a new discovery could be made that will unlock all the mysteries about weather forecasting, especially when fancy science jargon is utilized. No one objective entity will ever have a perfect solution, and even if a theoretical one existed, research has shown that a consensus of objective and subjective methodology will always yield superior accuracy.

Products & Services

Accuracy Statistics

Community & Affiliations