Almanac Forecasts: Mostly Hot Air
If you want to see steam come from the ears of someone who cares about the science of meteorology, ask him or her something like: â€śWhat does the Farmerâ€™s Almanac say December is going to be like?â€ť
One good answer from the meteorologist could be: â€śIâ€™ll toss a coin; heads December will be cold, tails it will be warm.â€ť The meteorologistâ€™s coin-toss forecast has about the same chance, maybe more, of turning out to be correct as those in the Almanac.
So, why donâ€™t you regularly read about someone whoâ€™s convinced coin-toss forecasts are 80% correct? Thatâ€™s what I read in newspapers that the Old Farmers Almanac claims for its forecasts.
Unlike the Old Farmers Almanac, coin-toss forecasters (if any exist) donâ€™t have a public relations machine to feed â€śstoriesâ€ť to the Associated Press and other news outlets. And, the coin-toss forecasters donâ€™t have gullible editors willing to repeat these claims.
These same editors would never let a politician get away with saying something like, â€śeighty percent of my campaign funds came from people who gave $25 or less.â€ť It would be as easy for a reporter who didnâ€™t know anything about meteorology to check out the Almanacâ€™s claims as the politicianâ€™s.
In fact, Jan Null, a retired National Weather Service meteorologist and American Meteorological Society Certified Consulting Meteorologist who runs the Golden Gate Weather Service in California, has done the job.
Most recently, Null reviewed the 2009 Old Farmerâ€™s Almanac winter forecast in his San Francisco Weather Examiner column, giving the forecast a cumulative grade of D based on his analysis of what the Almanac predicted and what actually happened.
Heâ€™s analyzed Almanac forecasts in the past with similar results.
Another kind of long-range forecast (for more than about 10 days in the future) you might hear about are the â€śseasonalâ€ť forecasts from the National Weather Serviceâ€™s Climate Prediction Center (CPC).
Unlike Almanac forecasts, which claim things such as a certain region will be cold for the first 10 days of a month and then turn warm and wet for the next few days, the CPC outlooks give only the odds that certain areas will be warmer than average, colder than average, or close to average for an entire three-month period.
The CPC forecasts are available on the Web.
When you look at the maps you’ll almost always see that large areas of the forecast maps are white. This means there are equal chances for above average, average, or below average temperatures or precipitation. The CPC forecasters are honest: When they don’t have good, scientific reasons for predicting what’s likely to happen, they say, in effect: “We don’t know.”
Null also examines the CPC forecasts. His latest such examination is on his Web site.
If you take a look at the CPC forecasts, youâ€™ll see a link to â€śSkill,â€ť which is how meteorologists measure how their forecasts turn out. This is pretty complicated. Nullâ€™s examination is easier to understand.
How can long-range forecasts be used?
As for the almanac forecasts, I wouldnâ€™t make any decision based on them that would cost me more than twenty-five cents if the forecast didnâ€™t pan out.
The CPC outlook can be of some use to those making certain kinds of decisions, such as whether to contract for fuel oil in advance at a fixed price, or to invest in improvements in a ski area. In both cases an outlook for a colder than average winter could make such investments seem less risky. But, the forecasts arenâ€™t guarantees. They are only one of the considerations that sophisticated investors could use.
Chapter 7 of The AMS Weather Book, which is the chapter on forecasting, explains all of this, including why the “butterfly effect” makes detailed, day-by-day forecasts for more a couple of weeks ahead so difficult. By the way, this chapter also explains (on Page 155) the difference between skill and accuracy as a measure of forecasts and why meteorologists use skill rather than accuracy to check on how they’re doing.