By WUWT regular “justthefacts”
In researching the use of tidal forces in long range weather forecasting, I came across an interesting August 30th, 2010 Associated Press/ MSNBC article based on interviews with Farmer’s Almanac Editors Sandi Duncan and Peter Geiger, and Ed O’Lenic from NOAA’s Climate Prediction Center:
“Good news, winter haters: After record snowfall in the mid-Atlantic and unusually cold weather down South, the Farmers’ Almanac is predicting a “kinder and gentler” winter.
After eyeing the skies, tidal action and sunspots, the folks at the 194-year-old publication say in their 2011 edition going on sale Monday that it’ll be cold but nothing like last winter, when 49 states saw snow and it got so cold in Florida that iguanas fell out of trees.
“Overall, it looks like it’s going to be a kinder and gentler winter, especially in the areas that had a rough winter last year,” said managing editor Sandi Duncan.”
“The Farmers’ Almanac, which claims 80 to 85 percent accuracy and says it correctly forecast heavy snow in Middle Atlantic states last winter, bases its predictions on a secret mathematical formula using the position of the planets, tidal action of the moon and sunspots.
Ed O’Lenic from NOAA’s Climate Prediction Center said the scientific community doesn’t accept tides, planetary alignment and sunspots as effective predictors of temperature or precipitation, but he stopped short of calling the almanac’s meteorological methods a bunch of hooey.
“In science you have to have an open mind. Someday, someone could conceivably find some scintilla of evidence that it’s useful,” O’Lenic, chief of the operations branch, said of the almanac’s methodology. “For the time being, we have to stick with what produces results for us.”
“For the record, NOAA’s Climate Prediction Center anticipates a warmer-than-normal winter for the mid-Atlantic and Southeast and colder-than-normal weather in the Northwest. That puts it at odds with the almanac, which calls for mild temperatures in the Northwest and cold in the Southeast.”
Here’s the October 21, 2010 Winter Outlook from NOAA’s Climate Prediction Center:
“The Pacific Northwest should brace for a colder and wetter than average winter, while most of the South and Southeast will be warmer and drier than average through February 2011, according to the annual Winter Outlook released today by NOAA’s Climate Prediction Center.”
“Northeast and Mid-Atlantic: equal chances for above-, near-, or below-normal temperatures and precipitation. Winter weather for these regions is often driven not by La Niña but by weather patterns over the northern Atlantic Ocean and Arctic. These are often more short-term, and are generally predictable only a week or so in advance. If enough cold air and moisture are in place, areas north of the Ohio Valley and into the Northeast could see above-average snow”
“Florida: drier than average, with an equal chance for above-, near-, or below-normal temperatures.”
“Central U.S.: equal chances of above-near-or below normal temperatures and precipitation”
Here’s Accuweather’s September 8, 2010 forecast:
“Wintry Battle Zone But No Snowmageddon
In general, the East Coast will be granted a reprieve from the tremendous snowfall that caused 2009-2010’s winter to be dubbed “snowmageddon.”
This does not mean a free pass for the Northeast. Bastardi predicts late November and December could get winter off to a fast start in the East, with a major thaw coming for much of the country in January.
Bastardi makes the early cold connection between this year’s active hurricane season and his winter forecast.
He said that years that see significant landfall, such as 1995, 2008 and 2005, usually also have cold for much of the eastern and central portions of the nation in December.
He said this year from the central Rockies to the Northeast a higher variance of temperatures will be present – “greater-than-normal swings between winter’s coldest and warmest days.” The conflicting warm and cold air masses contributing to these temperature fluctuations have placed this area into what Bastardi calls the “Wintry Battle Zone.”
Despite the wild swings in temperatures, cities like New York, Philadelphia and Washington, D.C., will still have near-normal snowfall. To put this in perspective, New York City receives an average of 28.4 inches of snowfall during winter.”
Here is the Old Farmer’s Almanac Atlantic Corridor Annual Weather Summary :
“Winter will be colder and drier than normal, on average, with below-normal snowfall in New England and above-normal snowfall elsewhere. The coldest periods will be in mid-December, January, and mid-February. The snowiest periods will be in early January and mid- and late February.”
Hmmm, “with below-normal snowfall in New England”. According to this January 28, 2011 Boston Globe article January 28, 2011 Boston Globe article “In Somerville, New England’s most densely populated city, some snowbanks are so tall that they deflect the plume of snow cleared by plow trucks and send it sliding back down to the street, said Michael Meehan, a city spokesman. Between storms, crews have been trying to clear snow piles and dump them on basketball courts, while the real estate trust planning a 50-acre redevelopment at Assembly Square has offered the city private land for use as a snow farm.”
In terms of The Old Farmer’s Almanac forecast that “The coldest periods will be in mid-December, January, and mid-February.” here are Weekly Mean Temperatures for the Northeast:
Week-Ending | Mean Temperature | Anomaly
20101204 | 33.76 | 0.90
20101211 | 26.15 | -4.02
20101218 | 23.80 | -4.01
20101225 | 22.68 | -3.12
20110101 | 24.45 | 0.35
20110108 | 24.73 | 2.00
20110115 | 22.28 | 0.37
20110122 | 20.86 | -0.54
20110129 | 19.89 | -1.64
20110205 | 19.93 | -2.15
20110212 | 20.84 | -2.22
20110219 | 25.70 | 1.20
20110226 | 26.21 | -0.20
20110305 | 28.28 | -0.41
Source: National Oceanic and Atmospheric Administration (NOAA) – National Climatic Data Center (NCDC):
I guess that you could call December 5th – 25th “mid-December”, but the first half of January had the warmest anomaly of an otherwise freezing winter and “mid-February” i.e. Feb 13th – 19th, was actually the only positive anomaly in the month of February.
Here are all of the Old Farmer’s Almanac Regional Annual Weather Summaries:
Note that you can verify the veracity, or lack thereof, of many of the weather predictions on the new WUWT US Weather History Reference Page:
And let us not forget about the UK MET Office who are apparently still trying to figure out what their forecast was, or at least what they renamed it and where they buried it on their website;
http://wattsupwiththat.com/2011/01/04/the-met-office-bullhockey/
but, according to this October 28th, 2010 article in the Telegraph;
“Although the Met Office no longer issues long-term forecasts, their latest data suggest a high probability of a warmer winter for London, the East of England, Scotland and Northern Ireland.
The South West, Wales and most of the North of England are less likely to enjoy such relatively pleasant temperatures but still have a 40 to 60 percent chance of being mild.
The statistics were generated by the Met Office’s new £33million supercomputer built by IBM.
Forecasters used it to analyse how likely temperatures and rainfall were to be above normal for winter but not how far above.
Related Articles
The average temperature for winter from 1971 to 2000 is 3.7C (39F). However, last year was 1.5C (35F), meaning anything above the 30-year normal this winter would be a marked improvement with far less chance of snow and ice.
As well as the milder winter, the computer concluded that almost all of Britain had a 40 to 60% of being drier than normal, with only the south coast more likely to see normal amounts of rain.”
How bad was the MET Office’s forecast? Per this March 4th, 2011 Article in Farmers Weekly Interactive;
“Britain’s worst winter weather for 100 years will cost farmers more than £100m, a Farmers Weekly investigation has revealed.
Farmers across the country have been left counting the cost of lost crops, collapsed livestock buildings and burst pipes.
Rural insurers said 2010 would go down in history as the worst year on record for cold weather claims.
Among the worst-hit are England’s 4000 sugar beet growers. Temperatures plunging to -13C in December followed by a mild and wet January left £52m of sugar beet rotting in the fields, according to Farmers Weekly calculations based on British Sugar figures.”
In summary, the long range forecasts of the Farmer’s Almanac, NOAA’s Climate Prediction Center, the Old Farmer’s Almanac, Accuweather and the UK MET Office all appear to be suspect. Why?
Why ? Because they’re not using wooly worms .
Why is predicting weather 3..6 months in advance so hard?
Weather is notoriously chaotic, from a mathematical perspective. Non-linear relationships with feedbacks on many timescales — *of course* it has sensitive dependence on the initial conditions and ongoing inputs.
I’m reading a textbook on Thermodynamics that was old fashioned when it was written 50 years ago. They could hand-calculate the limiting cases, but had to throw up their hands on anything realistic. I actually find that refreshing, and am in awe that they could design power plants with just the accuracy they could get on their slide rules. Nowadays, people just throw the whole thing into a computer and believe the results. It’s partial differential equations, after all, and they have a solver. What could go wrong?
It’s hard to predict exactly what a chaotic system with non-linear feedbacks will do.
If you build faulty assumptions about CO2 into your system, it’s even harder.
Not only have they been way wrong, but they seem to have been consistently wrong the last several years. That suggests to me that the methods they are using might not be as well founded as they think they are.
What is the predictor’s “skill”?
Why not accumulate statistics and see.
I’ll grab the easy one. I predict that today’s weather is the same as yesterday’s. I’ll miss all the changes, but, counting successful predictions, my statistics may be higher than a predictor using actual intelligence.
Same thing for annual predictions. Compare the forecaster’s results to the null hypothesis.
Could it be that there are some truly random processes involved, that make long-term weather forecasting impossible?
More details concerning the chaos of weather:
All chaotic systems have a strange attractor, and I have never seen anyone make a strange attractor of temperature, precipitation, or the number of fronts passing a location in a year. Everyone acts as if the weather acts according to some secret equation that only they know, when in fact, it is merely tracing paths in its strange attractor; and no one knows exactly what path it will take. However, it is only a matter of detail. I can accurately predict that January in the northern hemisphere will be cold and July will be hot, but I can’t predict exactly which days in January will be below zero in Colorado or which days will be above 100 and July in Colorado. All the weather forcasters are trying to push the boundaries of the amount of detail they can predict. They just need to stop trying to put more and more variables in their equations and concentrate on finding strange attractors of whatever they are trying to predict. That is the signature of chaotic systems.
Not that the soothsayers are unwilling to open their mouths to change feet! If you cover enough bases, your prediction has a 0-100% chance of being right!
Isn’t that what is being done by constantly changing the root “concept” to fit the data?
Engineer Bob says: April 9, 2011 at 10:09 am
What is the predictor’s “skill”?
On NOAA’s Climate Prediction Center they report on their “Skill” or lack thereof:
http://www.cpc.ncep.noaa.gov/products/predictions/long_range/
Unfortunately it seems that they stopped tracking their Skill, or at least providing it to the public, in mid-2007:
http://www.cpc.ncep.noaa.gov/products/predictions/long_range/lead04/off_skl.html
Here is how NOAA’s Climate Prediction Center defines Skill:
http://www.cpc.ncep.noaa.gov/products/predictions/90day/skill_exp.html
At 5.5 Months it is really bad:
http://www.cpc.ncep.noaa.gov/products/predictions/long_range/lead06/off_skl.html
At .5 Months, there seems to be a bit of skill on Temperature, but Precipitation looks very bad:
http://www.cpc.ncep.noaa.gov/products/predictions/long_range/lead01/off_skl.html
Because we don’t know what we don’t know. Our level of understanding our climate, or long term weather, if you will, is very poor.
My definition of arrogance: the inability to acknowledge how ignorant you really are. No statement is truer than, “the more you know, the more you realize you don’t know,” unless arrogance gets in the way.
They’re bound to get it wrong sometimes. The question should be what is each one’s level of accuracy in percentage terms?
The Met Office gave up on seasonal forecasts for the public probably due to their computer’s warming bias. The people of Britain are now grateful that they don’t have to endure anymore misleading forecasts. The government and other organisations who subscribe to there nonsense were mislead this past winter too.
http://www.bbc.co.uk/blogs/paulhudson/2010/01/a-frozen-britain-turns-the-hea.shtml
It’s said that prostitution is the oldest profession. I think prediction is. Or maybe that’s a distinction without a difference.
Old Farmer’s Almanac Atlantic Corridor Annual Weather Summary :
“Winter will be colder and drier than normal, on average, with below-normal snowfall in New England”
NOAA’s Climate Prediction Center: “Northeast and Mid-Atlantic: equal chances for above-, near-, or below-normal temperatures and precipitation. … Northeast could see above-average snow”
Here’s Accuweather’s September 8, 2010 forecast:
“Wintry Battle Zone But No Snowmageddon
In general, the East Coast will be granted a reprieve from the tremendous snowfall that caused 2009-2010′s winter to be dubbed “snowmageddon.”
====
Burlington, VT winter snowfall total as of last Sunday = 128.0 inches – third snowiest Winter on record. May still creep into second place as there is sometimes appreciable snowfall up here as late as mid-May. Cold? 6594 heating degree days vs normal 6718. So, a bit warmer, with much more snowfall than usual.
Old Farmers Almanac and Accuweather — Prediction wrong on all counts. NOAA — unspecific, but less awful. I reckon I could do about as well with a table of random numbers, a ouija board, or a magic eight-ball.
@justthefactswuwt: Thanks for the link to the NOAA skill page.
It looks like they have a bug in their script. They are obviously updating the *headings* periodically to reflect the current date, but the *data* stops at 2007 as you said.
It’s difficult to grasp data for a number of sources (e.g., the Old Farmer’s Almanac, NOAA, Accuweather, etc.), especially when each is associated with a number of variables (in this case, predictions for various portions of the winter). It would be helpful to organize the data as a table and include “Reality” as a source, showing what actually happened. This would make comparisons between the sources easier, and would show trends that are lost in the verbiage.
It is relatively trivial to make correct predictions about the future climate as long as you restrict such predictions to very large areas (nothing smaller than a hemisphere) and long timescales – nothing shorter than a month.
As another poster pointed out, there is no problem with predicting that July will be warmer than January in the Northern hemisphere, its when you try and predict the conditions in an area of less than 20% of the Earth’s surface for less than 5% of the year that major errors arise.
As others have suggested this is because local weather is a chaotic process – lots of no-linear effects interacting. But lower resolution predictions CAN be accurate about such chaotic systems because they deal in boundary conditions rather than specific or initial conditions.
An example invoking one of the simplest chaotic systems may help clarify.
The dripping tap may seem to be a simple periodic system, the drips occurring regularly with the same size according to the input flow and the size of the spout. But when measured accurately the timing of the drips and their size is NOT regular, it is chaotic with the values for drip size and the period between drips fluctuating by several percent in a deterministic but chaotically unpredictable manner. While there is some autocorrelation between drip size and time for a series of drips, it is inherently impossible to predict the exact time and size of the next drip given the size and time of the last seven. (any similarity with ENSO cycles is purely educational)
But given the flow rate and spout size it IS possible to predict the average size and timing for 1000 drips. Despite the chaotic nature of individual drops, the long term consistency of the system makes it possible to use the system as a clock as the development of water-clocks based on this principle in several civilizations shows.
It is also possible to predict the average size and timing if the flow rate or spout size changes. That will alter the average size and timing of drips, but will not make it any easier to predict the size and timing of individual drops.
Weather and climate – and the prediction of each have similar features to the dripping tap. Individual drops are like the weather, they will vary within a range and be chaotic, but climate is like the average size and timing of drips over MANY individual events, it shows consistent results for a given level of input and process. In one case water pressure and spout size, in the other energy input and thermal emission pathways.
“I can accurately predict that January in the northern hemisphere will be cold and July will be hot, but I can’t predict exactly which days in January will be below zero in Colorado or which days will be above 100 and July in Colorado.”
This is the Uncertainty Principle in particle physics, as applied to large scale object. You can know “what” OR “when”, but you cannot know “what” AND “when”.
Nathanael Herreshoff, the father of modern yacht design applied this to boats. You can have speed, comfort and low-cost. But never more than two at the same time.
“The Pacific Northwest should brace for a colder and wetter than average winter”
Was a warm winter here.
(A lot of people thought it would be cold because of linear misconceptions about the nature of relations with other indices such as SOI, N34, MEI, & AAM.)
izen says:
April 9, 2011 at 11:21 am
Yes but they are useless predictions. The whole purpose of predictions is avoidance. If you cannot be specific in where, when, how much and how and WHY, it’s useless. It might be approximately right in one aspect and completely wrong in another. If the climate of southern spain is warm in this epoque I assume you think that is climate prediction but in spain, costa del quoi, it snows sometimes. If a climate prediction cannot tell me whether that snow will become more persistent then it is useless and climate models cannot provable predict in any way shape or form so they are useless and should be shut down right now.
Why? The GCMs, upon which the aforementioned mis-forecasts are based, come pre-polluted with a warming bias (or trend). The result is one of two foregone conclusions:
a.) The warming trend continues in nature and the GCM’s are along for the ride or
b.) The cooling trend replaces the warming trend and the GCMs are lost in space.
The GCMs are computer programs incapable of dealing with climate that is not warming.
Contrast the GCM based results with a real metorologist who cracks the books.
P Walker nailed it with the first comment, and wooly worms are much cheaper than supercomputers.
Forget the long term forecasts. Why are we so bad at the short term? My father was an avocado grower, and for a while subscribed to a weather service who advertised an 85% accuracy of prediction. He quickly realized that they ensured their accuracy by constantly making revisions and issuing updates. He finally threw his hands up in the air (there was no CO2 then to harm him) and said, “Hell, I don’t need to pay somebody to look out the GD window. I can do that for free myself.
Never make predictions, especially about the future.
– Casey Stengel
@-stephen richards says:
April 9, 2011 at 11:38 am
“Yes but they are useless predictions. The whole purpose of predictions is avoidance. If you cannot be specific in where, when, how much and how and WHY, it’s useless. It might be approximately right in one aspect and completely wrong in another. ”
Unfortunately it is inherent in the nature of chaotic systems that the level of detail you demand is impossible. It is like asking for the ability to predict the size and timing of an individual drop.
The only possible predictions for such a system put constraints on the RANGE of values for drop size and time, and can give the probability distribution for that range in some circumstances. The boundary conditions – but it cannot give you predictions that are useful for avoidance of specific events.
-“If a climate prediction cannot tell me whether that snow will become more persistent then it is useless and climate models cannot provable predict in any way shape or form so they are useless and should be shut down right now.”-
Again it depends on the climate model and the resolution of the prediction.
It will never be possible to tell you how much snow you will get at a certain location during a certain month. But climate models using pen and paper (and maybe a slide rule) are capable of predicting the cooling from major volcanic eruptions.
Direct observation of the cooling after Agung El Chichon and Pinatubo combined with measurements of the solar dimming from the stratospheric emissions has provided an objective measure of the magnitude and duration of the effect of such volcanic events.
This will not satisfy the level of detail you are demanding, but the knowledge that a volcanic event of a certain size will reduce surface temperature by a certain amount for a certain time (with probability ranges) is still useful information for those that need to plan for future heating fuel use or agricultural productivity changes at the global or continental level.
The difficulty in predicting the timing and magnitude of the ENSO changes indicates the problem with specific predictions. But the 1LoT constrains the balance of positive/negative events so that the prediction that the overall effect on temperature for many cycles will be neutral.
The equivalent of the water flow rate for the dripping tap constraining the amount of water that drips over many events.