UPDATE: note to readers, Gray and Klotzbach are only discontinuing December forecasts for the season ahead due to limited predictive skill — for the time being. A main reason is the well-known “Spring barrier” in El Nino Southern Oscillation forecasts for the next year…
When is the last time you can recall any scientist suspending a highly visible public work because they decided it just didn’t have any predictive skill?
This is refreshing.
From: http://hurricane.atmos.colostate.edu/forecasts/2011/dec2011/dec2011.pdf
QUALITATIVE DISCUSSION OF ATLANTIC BASIN SEASONAL HURRICANE ACTIVITY FOR 2012
We are discontinuing our early December quantitative hurricane forecast for the next year and giving a more qualitative discussion of the factors which will determine next year’s Atlantic basin hurricane activity. Our early December Atlantic basin seasonal hurricane forecasts of the last 20 years have not shown real-time forecast skill even though the hindcast studies on which they were based had considerable skill. Reasons for this unexpected lack of skill are discussed.
…
Relationships between predictors and predictands which once seemed quite strong may fail to work in future years due to a phenomenon known as the ‘siege of time’. It is the failure of these once-promising relationships which requires the forecaster to demand as much understanding of linkages between predictors and predictands as possible.
…
We have developed a new way of assessing next year’s activity in terms of two primary physical parameters:
1. the strength of the Atlantic thermohaline circulation (THC)
2. the phase of ENSO
…
We strongly believe that the increases in atmospheric CO2 since the start of the 20th century have had little or no significant effect on Atlantic basin or global TC activity as extensively discussed in our many previous forecast write-ups and recently in Gray (2011). Global tropical cyclone activity has shown no significant trend over the past thirty years.
h/t to WUWT reader JohnD
Wow, truth from the horses mouth, I love them.
It’s the right thing to do. Now if only some other scientists would figure out their methods have no predictive value before they publish papers on a theory.
Those “increases in atmospheric CO2” has been actually an addition of one (1) CO2 molecule to existing three (3) CO2 molecules per ten thousand (10,000) other molecules during the last 200 years. Anyone who believes that this single molecule has an effect on hurricanes or extremes or whatever is just.. believer, of fool.
Is more money needed?
It is nice to see scientists admitting that models with terrific hindcast skill have absolutely no predictive skill. I’ve been saying this for years. Orrin Pilkey and his daughter wrote a book about it titled “Useless Arithmetic.”
“We strongly believe that the increases in atmospheric CO2 since the start of the 20th century have had little or no significant effect on Atlantic basin …”
I quote: “Yeah Baby, Yeah!”
Well, seems that reality still packs a punch.
Good on these guys for being honest enough to explicitly admit that their forecasts were worthless.
Their comments about the absence of any CO2 impact may wind up being the most significant part of this story.
That’s the difference between a scientist and a politician.
News Flash…..
Just because you can model the past…
…does not mean you can predict the future
Film at 11…………climate models pay attention
“We strongly believe that the increases in atmospheric CO2 since the start of the 20th century have had little or no significant effect on Atlantic basin or global TC activity as extensively discussed in our many previous forecast write-ups and recently in Gray (2011). Global tropical cyclone activity has shown no significant trend over the past thirty years.”
It is refreshing to see such honesty. Lets hope that more of those involved in using models to predict things will follow suit.
When it comes to Hurricane knowledge, William Gray seems to be the level/cooler head. A few years ago (Katrina time frame) he took a lot of flack from saying Hurricane’s have a natural high and low cycle. And we’re currently in a higher cycle which should end about 2030’ish. Not that I remember the dates, I remember he didn’t think highly of the cause or the band wagon it was riding.
wow . . . a moment of clarity and honesty from the Climate Scientology world.
Most refreshing.
Next up, true confessions from the Climate Modelers.
Reasons for this unexpected lack of skill are discussed…a phenomenon known as the ‘siege of time’
I would rather have an explanation as to why the discovery of their total lack of predictive power was unexpected. It should have been assumed from the beginning.
When is the last time you can recall any scientist suspending a highly visible public work because they decided it just didn’t have any predictive skill?
The UK met. office suspended their long range forcasts not long ago for the same reason Anthony.
“One of the primary impediments to successful forecasts at this lead time is likely due to the inability to predict ENSO through the springtime predictability barrier.”
http://www.seas.harvard.edu/climate/eli/reprints/Samelson-Tziperman-2001.pdf
“However, the rough correspondence between our results with those of Moore and
Kleeman (1996) who analyzed a regular, nonchaotic oscillation indicates that we need to keep looking for such a differentiating signal in order to understand whether the observed irregularity of El Nin˜o arises primarily from stochastic forcing or from low-order chaos.”
Hmmmm.
Looks like a reasonable approach. Now if climate scientists would stop their hysterical forecasts of impending doom, based on models and non-accelerating trends, we could put our efforts and resources into solving real problems.
OT, [snip . . indeed. Tips and Notes beckons you]
Despite global warming since the 1970s and the recent ‘hottest’ decade on the record we have just seen a 30 year record low ACE index.
A qualitative assessment will be even worse. Just give up and give the grant money back, or have they run out of grant money. Either way predicting chaotic behaviour is a mugs game.
In medicine it very commonly happens that high-profile trials of diagnostic methods are terminated for lack of clinical efficacy.
More broadly, it commonly happens in science an initially high-profile scientific trial becomes steadily lower-and-lower profile as the trial’s lack of efficacy becomes apparent, until the trial is quietly terminated, noticed by few and mourned by none.
Trials of earthquake prediction methods have historically followed this scientific arc, for example.
In summary, Klotzbach and Gray were following common-and-accepted scientific practice. For which, good on them.
An example of likely future importance for climate change research will be predictions of Antarctic ice flow rates, which now can be observed with high accuracy, and which in coming decades will provide a key test-bed for climate prediction science.
imagine that some use model that don’t have any hindcast skills …..
Our early December Atlantic basin seasonal hurricane forecasts of the last 20 years have not shown real-time forecast skill even though the hindcast studies on which they were based had considerable skill. Reasons for this unexpected lack of skill are discussed.
Newsflash: Our models are crap. Details at 11.
11: When we tune our models to the past, they can predict the past. But they do diddly-squat about predicting the future. It’s possible we don’t know enough about everything that influences weather and climate to make models with predicative skill about the future. But we still kicked butt on foretelling what the weather used to be.
“Our early December Atlantic basin seasonal hurricane forecasts of the last 20 years have not shown real-time forecast skill even though the hindcast studies on which they were based had considerable skill.”
This is an absolutely critical point. Hindcasting skill — even very impressive hindcasting skill — simply does not translate into forecasting skill. Are the temperature modelers listening?
Past performance is no guarantee of future results. That’s what financial services say, and they’re using computer models whose results they can test every business day.
“…have not shown real-time forecast skill even though the hindcast studies on which they were based had considerable skill”
Kudos to them! This is the way it has to be done!