CMCC FOUNDATION – EURO-MEDITERRANEAN CENTER ON CLIMATE CHANGE
A team of scientists, led by UK Met Office, has achieved a scientific breakthrough allowing the longer-term prediction of North Atlantic pressure patterns, the key driving force behind winter weather in Europe and eastern North America. CMCC scientists Panos Athanasiadis, Alessio Bellucci, Dario Nicolì and Paolo Ruggieri from CSP – Climate Simulation and Prediction Division were also involved in this study.
Published in Nature, the study analysed six decades of climate model data and suggests decadal variations in North Atlantic atmospheric pressure patterns (known as the North Atlantic Oscillation) are highly predictable, enabling advanced warning of whether winters in the coming decade are likely to be stormy, warm and wet or calm, cold and dry.
However, the study revealed that this predictable signal is much smaller than it should be in current climate models. Hence 100 times more ensemble members are required to extract it, and additional steps are needed to balance the effects of winds and greenhouse gases. The team showed that, by taking these deficiencies into account, skillful predictions of extreme European winter decades are possible.
Lead author Dr Doug Smith, who heads decadal climate prediction research and development at the Met Office Hadley Centre, said: “The message from this study is double-edged: climate is much more predictable than we previously thought, but there is a clear need to improve how models simulate regional changes.”
Advance warning of severe winter weather is imperative to those who make risk-based decisions over longer timescales.For example, better forecasts can help the Environment Agency plan water management and flood defences, insurance companies plan for the changing risks, the energy sector to mitigate against potential blackouts and surges, and airports plan for potential disruption.
Improving model simulations will enhance the countries’ response, resilience and security against the effects of extreme weather and climate change – influencing future policy decisions to protect people’s lives, property and infrastructure.
###
Read more:
The paper on Nature:
Smith, D.M., Scaife, A.A., Eade, R. et al. North Atlantic climate far more predictable than models imply. Nature 583, 796-800 (2020). https://doi.org/10.1038/s41586-020-2525-0
Source: based on Met Office press release
The current three month temperature outlook from the Climate Prediction Center shows that every square inch of the United States will be warmer than average. Somehow I doubt that.
This is touted as a climate prediction ‘breakthrough’, without any corroborating observation data from real locations on Planet Earth? Sounds more like unsupported speculation, to me.
If there is no steak, it’s fraud to sell the sizzle!
It appears someone ask where the money has gone after 6 years and they got this brain fart of predicting model data.
Wow – “highly predictable” but “much smaller than it should be”???
Well that’s Science for you.
I am convinced but less convinced than maybe, perhaps than I should be!!!
It sounds to me like if 100 times more ensemble members are required to calculate their “forecast” that it is extremely exposed to butterfly wings flapping in Brasil.
I think some of you may be missing the point of this UK Met study.
Yes, they are looking at “reanalyzed data”, which is generated by hind-casting historical data (using DA, data assimilation). Like any forecast, it is not always accurate, but it provides a convenient and plentiful platform for trying out testing out new algorithms and tools. But, the point you are missing, is that they noticed (about six years ago) that the models do not correctly predict the NAO patterns. (Surprise?) This current study announces that they have made a “fix” which enhances the accuracy of the model, compared to actual NAO trends. [This latest study paper is paywalled, so we are missing the complete details of the fix.] [I too am a skeptic of government “climate-scare” policies, but I also assume that there are some government researchers who are really dedicated to improving the accuracy of weather/climate forecasting tools.]
Here is what the authors said:
Now will they continue this study by making more predictions, into the future, and compare predictions to future weather results? That is a good question. I do not know answer.
I am also interested on the science behind the study. I posted a long comment above that went into moderation for unknown reasons.
Here is a “freed” copy of the article for anybody interested:
Smith et al., 2020. North Atlantic climate far more predictable than models imply.
The article is quite critical of models:
“Crucially, we find that the NAO signal is underestimated by an order of magnitude in the model ensemble. This adds to an increasing body of evidence that the signal-to-noise ratio is too small in climate models, as seen on seasonal20,35–37, interannual38 and decadal19,39,40 timescales. Consequently, the real world is more predictable than climate models suggest10,18 and uncertainties diagnosed from raw model simulations are too large. The cause of this error is not yet known, although there are several theories, including weak teleconnections to the quasi-biennial oscillation41, lack of persistence in the NAO42,43 and in weather regimes44, unresolved ocean–atmosphere interactions45 and weak transient eddy feedback46.
A key question is whether climate models also underestimate sig- nals on timescales beyond a decade. There is some evidence that the response of atmospheric circulation to Arctic sea-ice loss47 and to exter- nal factors10 such as volcanic eruptions, solar variations and ozone changes are too weak in models. Models also appear to underestimate the magnitude of multi-decadal temperature variability48,49, especially for the North Atlantic50,51. Furthermore, model-simulated winter climate change signals in the North Atlantic increase substantially as resolution increases52, consistent with the suggestion that eddy feedbacks are inadequately resolved46. If this is robust, treating current model simula- tions at face value leads to misleading conclusions about uncertainties and irreducible internal variability.”
What is disappointing is that they are missing totally the effect of solar activity on winter weather predictability. They won’t go far without it.
Javier,
Thanks for “liberating” the subject study.
Looking at it I find that the steps needed to repair the model are very simple and somewhat antiquated, at least for old-school practitioners of regression analysis like myself.
Seems like they performed the pre-processing analysis on this manually, which led to rules for selecting the optimal ensembles. There are tools to automate this, which would probably find additional parameters which can be optimized to futher increase predictability.
For example, you might recall the analysis I did to find predictability in the Labitzke data, in which I used Cubist/M5 (a 30-year old, open source tool) to creat a detailed regression model (a piece-wise linear regression tree, with human-readable rules) to predict 30 hPa temperatures using sunspot features. [The “holy grail”? I do not think so. 30 hPa is in the middle of the stratosphere and has very little effect on Earth’s daily weather. :-]
https://wattsupwiththat.com/2019/02/25/labitzke-meets-bonferroni/#comment-2641991
I didn’t recall it, but I have checked it again. Interesting, but as I said it is not the right analysis, so it is not surprising that did come negative. I showed the data in my comment:
https://wattsupwiththat.com/2019/02/25/labitzke-meets-bonferroni/#comment-2640975
The hypothesis that can be deduced from the data is that low solar activity results in very different 30 hPa conditions (geopotential height and temperature) over the winter North Pole depending on 30 hPa equatorial wind speed, and that the result is different when solar activity is high. If you don’t include the QBO you come out empty.
Except during the winter, which is what we are talking about. The polar vortex extends all the way from the surface to the stratosphere. When is strong and well formed stays over the pole and sucks all the heat. At the same time creates a very cold pole and keeps the mid-latitudes in a warm winter. When is weak and disorganized the pole is warmer and the mid-latitudes colder. The association between Arctic winter temperatures and solar activity is well known and studied.
Kobashi et al., 2015. <a href="https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1002/2015GL064764"Modern solar maximum forced late twentieth century Greenland cooling.
So what *IS* their prediction for Winter 2020-2021 for Northern Europe and the British Isles?
Their ability to predict next winter’s severity for Northern Europe will be correct, except when it’s not. Then it’s just random chaos that made them miss the prediction.
In reality what I see here is a massive Grift, or Rent-seeking if you will. Their statement:
“Hence 100 times more ensemble members are required to extract it,…. “ is the tell.
Bottom line: Send money. Lots of money.
Wow! Something is predictable for decades.
Well, we will know that in … a few decades.
Beware of predictions.
Especially about future!
I love the abstract! My caps…
“…Here we assess retrospective climate model predictions of the past six decades and show that decadal variations in North Atlantic winter climate are highly predictable, despite a lack of agreement between individual model simulations and the POOR PREDICTIVE ABILITY OF RAW MODEL OUTPUTS…”
Nice admission that the models were garbage despite how they were portrayed. Not to mention individual simulations couldn’t agree on decadal variations of “stormy, warm and wet or calm, cold and dry” winters.
This sounds pretty hokey, though…
“…To address these limitations, we implement a two-stage post-processing technique. We first adjust the variance of the ensemble-mean North Atlantic Oscillation forecast to match the observed variance of the predictable signal. We then select and use only the ensemble members with a North Atlantic Oscillation sufficiently close to the variance-adjusted ensemble-mean forecast North Atlantic Oscillation…”
Texas sharp-shooting?
JF
Great! Let’s put their predictions to the test and see how they compare to observations over the next four or five decades. Validating theories is the keystone of the scientific method. Let’s try science for a change.
I don’t understand some people here. They seem to be slagging off this paper purely because the UK Met office is fully signed up to the CAGW nonsense. This is a valuable piece of work that increases our understanding of how the atmosphere works.
Most people come for tribal reaffirmation. There used to be more people genuinely interested in the science of climate change.
Yes.
This used to be better.
I have abandoned this at times due to the US alt-right party politicals, which have no relevance to the science of climate.
the science is patently not leftist or liberal, but science.
They are not looking at the discrete solar forcing of NOA/AO anomalies, so this must be about probabilities according to things like the AMO phase. That doesn’t explain anything about how the atmosphere works.
How does looking at climate models and then throwing out the ones that don’t agree with observations increase our understanding of how the atmosphere works? There are too many PhD academics and not enough intelligent thought to go around.
NAO is fundamentally chaotic unpredictable.
https://en.wikipedia.org/wiki/Chaos_theory
I have had much success in predicting NAO/AO variability at roughly weekly scales since 2008 at very long range by means of the heliocentric planetary ordering of solar activity. Kepler used the same method to predict the very cold winter of 1595.
Day-to-day changes in the Arctic and North Atlantic Oscillations correlate with solar wind speed and relativistic electron precipitation.
https://www.sciencedirect.com/science/article/pii/S0273117713005802
They realise two things, for which they can be commended.
First the “ The chaotic nature of the climate system”.
Second that “the North Atlantic Oscillation [is] the leading mode of variability in North Atlantic atmospheric circulation”.
Who knows – maybe they’ll even rediscover the concept of the null hypothesis?
IMO the press release doesn’t find the core of the paper. The headline: “North Atlantic climate far more predictable than models imply” nails it. They find, that the NAO has a low frequency signal part which models don’t replicate, see fig.2a. They deduce, that CMIP5’s and 6’s do not show the climate of the real world when it comes to the dynamics of atmoshere. They find:” The fact that the NAO signal is much too weak in models implies that the effects of the NAO will be underestimated relative to other factors such as greenhouse gases. ”
Furthermore: ” If this is robust, treating current model simulations at face value leads to misleading conclusions about uncertainties and irreducible internal variability.”
An annoying comment in “Science” https://www.sciencemag.org/news/2020/07/missed-wind-patterns-are-throwing-climate-forecasts-rain-and-storms goes one step further:
“The missing predictability also undermines so-called event attribution, which attempts to link extreme weather to climate change by using models to predict how sea surface warming is altering wind patterns. The changes in winds, in turn, affect the odds of extreme weather events, like hurricanes or floods. But the new work suggests “the probabilities they derive will probably not be correct,” Smith says.”
This is an important outcome: Models do not show the real worlds climate and many “attribution” papers are not correct when the refer to model simulations.
When I lived in the UK, I had no problem predicting the weather. I tossed a coin.
Heads: dismal
Tails: rotten
I was nearly always right.
As far as the accuracy of their forecasts are concerned, a stopped clock is accurate twice a day.
Maybe they should have a look at Old Moore’s Almanac for information about future weather, it might be more (pun intended) accurate.
How can they possibly improve on the Cox-Attenborough settled science?
I do hope that they are not looking at multiple model runs and then engaging in post-hoc screening to find the best fit to the actual observations.
Really don’t see how the quote makes sense at all:
1. They claim “highly predictable” whilst asserting “lack of agreement between individual model simulations” and “poor predictive ability of raw model outputs”
2. They admit “…current models underestimate the predictable signal…of the NAO…by an order of magnitude
AN ORDER OF MAGNITUDE!
Maybe that explains why MetOffice/IPCC (a) don’t talk about the NAO upswing cycle from 1910-1940s (because its mostly natural) and (b) assign all of the warming on the upswing cycle from the 1970s onwards to AGW.
I don’t get the point of this study, surely if they improve their predictive skills they will no longer be able to claim:
“Nobody saw this flood coming coz it’s CO2 wot dunnit”
I used to think that the UK met office was incapable of even predicting weather 24 hours ahead. However after 6 decades of observation I realise they are incapable of accurately describing the current state of the atmosphere or even reading the data from their own radar. Decadal predictions? do me favour.
When I woke up yesterday my phone told me the rain would stop in 22 minutes. My eyes told me it sunny and dry. I don’t know which to believe.
When I last looked 80-90% of their yearly temperature forecasts were outside the maximum error bars. They never admitted that. Instead they stopped publishing these forecasts so we couldn’t laugh at them.
Without admitting their yearly forecasts were a failure, they now claim to be able to forecast on a decadal scale – which just means we’ll all be a decade older when we can laugh at them.
Regardless of the study, the press on it just sounds like something Squealer from Animal Farm wrote.
“Rejoice, comrades! Our science has determined what the winter will be for the coming decade. It was hard. The scientists found a signal – very faint- but it was there, and using difficult brain work they – after much labor and toil – were able to find it! That is why they need more resources than you. Their brainwork is very hard. They work themselves to the bone, so to speak! But it is all for YOU. All for the glory of Animal Farm.”
The question really is whether this “signal” can detect the the North Atlantic Oscillation (or possibly an extraterrestrial alien invasion) ?
You can’t predict the climate. It is a chaotic fractal which exhibits self-similarity.
See https://climatescienceinvestigations.blogspot.com/2020/05/9-fooled-by-randomness.html
and https://climatescienceinvestigations.blogspot.com/2020/07/17-noise-fractals-and-scaling-revisited.html
It doesn’t repeat but the average size of future fluctuations in temperature are predictable.
????
They analyzed sixty years of data and with extensive estimate adjustments managed to hindcast sixty years?
This is worth a press release?
Utter fools.