28 February 2014

Did The Farmers Almanac Winter Forecast Verify?? Nope!

Posted by Dan Satterfield

I suspect a lot of folks think the Farmers Almanac winter forecast was right, and the reason they think so is because of CONFIRMATION BIAS. Science in many ways is a vaccine against confirmation bias. Read the link on confirmation bias, and my friend Jim Gandy’s guest post. Then you will understand something that those who read horoscopes, wear magic bracelets, or buy pills for their prostate (hawked by retired football announcers) do not.

For Entertainment Purposes Only

Guest Post by Jim Gandy
Chief Meteorologist
Columbia, South Carolina
Jim Gandy

Jim Gandy

For almost a week there has been a persistent rumor that another snowstorm is on the way.  I traced this back to radio spreading the rumor that Farmers’ Almanac had predicted the past two snowstorms and was predicting a third.  I was frequently asked if there was another snowstorm on the way, for which I answered no.  The response was almost always “Well, Farmers’ Almanac is predicting another snowstorm and it predicted the last two.” What?

So I investigated this claim.  First let me point out that there is a Farmers’ Almanac (dating back to 1818) and an Old Farmer’s Almanac (dating back to 1792).  It was Old Farmer’s Almanac that was predicting snow.  Second the forecasts are so vague as to be of little value.  Neither ever forecasted sleet or freezing rain and thus the ice that was so devastating.
Yet there were a few that were adamant that Old Farmer’s Almanac got it right.  If that is the case, then why ever watch the media.  Let’s explore the claim.

I do not have the forecast breakdown for January, but I did get it for February.  For the period of February 7-14, the forecast is for “periods rain and snow, then sunny, cold”.  It was during this period that the snowstorm occurred (11-13).  But, wait!  The forecast is for the entire region from south Georgia to southeast Virginia.  It also includes the Piedmont as well as coastal areas.

Does this forecast apply equally to all locations?  Are we to apply this with rain in the south and snow in the north?  If so, where is the rain/snow line?  Where is there any mention of ice?  Where does it even say snowstorm?  Merriam-Webster Dictionary defines snowstorm as a disturbance of the atmosphere marked by a heavy amount of precipitation in the form of small white ice crystals.  In their forecast how much snow will fall?

Yes, but Jim, there was a snowstorm moving through the region at that time.  Ok, let’s look at the climatology of snow in Columbia, SC.  On average we get a snow once every two years.  The highest probability of snow comes in February.  I get this all the time; it’s January and we haven’t seen any snow.  People are disappointed until I tell them that February is our snowiest month.

Furthermore there is a higher probability of snow in North Carolina and southeast Virginia during this time. Plus, you have given yourself a week for this to happen. The odds are greatly in your favor for something to happen somewhere in the region. The odds favor periods of rain and snow in this region at this time. However, the forecast is of little value because it does not say where in the region, it just says somewhere in the region. This is not “calling for a snowstorm”.

Let’s look at the forecast for Feb. 21-26 that got everyone excited. It calls for “rain to snow, then sunny, cold”. It is this that people interpreted as saying that a third snowstorm was going to occur. For the record, it did not happen. Furthermore, the forecast for Feb. 27-28 is “Rainy, cool”. It will be cool, but it will also be bone dry for South Carolina.

To be fair Farmers’ Almanac was not calling for any snow during the Feb. 21 to 28 time frame, just some rain.

Both publications claim to have an accuracy of 80-85%. However, they provide no evidence to support this claim. How is the claim derived? Is it for the U.S. as a whole or does it apply to a certain region? Do all regions have the same accuracy? Why do the forecasts for each almanac differ? I find it very hard to believe that the accuracy in December, the farthest month out, is as accurate as the preceding January, a near month from when the forecast is issued.

Years ago an independent verification was done on the almanacs. The best that could be determined was that their forecast of departure from normal was right about 50% of the time. Otherwise, you were better off flipping a coin. It is amazing to see the almanacs make their claim of accurate forecasts, even pointing out specific cases. However, they never note their misses.

Look at the chart at the bottom of the page for Old Farmer’s Almanac. This is the forecast for departure from normal for each month. If you look at Columbia which is in the middle of the region for Old Farmer’s Almanac these are the results for the past four months:

Look at the chart at the bottom of the page for Old Farmer’s Almanac.  This is the forecast for departure from normal for each month.  If you look at Columbia which is in the middle of the region for Old Farmer’s Almanac these are the results for the past four months:
Old Farmer’s Almanac
Cold, Wet
Cold, Dry
Cold, Dry
Warm, Wet
Warm, Dry
Cold, Dry
Cold, Average
Warm, Dry
·       The information for February is through the 26th, but will not change even with just two days left.

Neither almanac forecast the unseasonably warm weather last week leading to a record high temperature on Thursday of 84° F. The average temperature last week was 6.4° F above normal.

The Bottom Line

he forecast for both almanacs are so vague that the forecasts are of little value. Their claims of accuracy have never been substantiated independently.

That is not to say that there isn’t useful information in the almanacs. There is, it’s just that the weather forecasts are not one of them.

The forecasts should be used for entertainment purposes only.