Latest Supercomputers Enable High-Resolution Climate Models, Truer Simulation of Extreme Weather
Berkeley Lab researcher says climate science is entering a new golden age.
Not long ago, it would have taken several years to run a high-resolution simulation on a global climate model. But using some of the most powerful supercomputers now available, Lawrence Berkeley National Laboratory (Berkeley Lab) climate scientist Michael Wehner was able to complete a run in just three months.
What he found was that not only were the simulations much closer to actual observations, but the high-resolution models were far better at reproducing intense storms, such as hurricanes and cyclones. The study, “The effect of horizontal resolution on simulation quality in the Community Atmospheric Model, CAM5.1,” has been published online in the Journal of Advances in Modeling Earth Systems.
“I’ve been calling this a golden age for high-resolution climate modeling because these supercomputers are enabling us to do gee-whiz science in a way we haven’t been able to do before,” said Wehner, who was also a lead author for the recent Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). “These kinds of calculations have gone from basically intractable to heroic to now doable.”
Using version 5.1 of the Community Atmospheric Model, developed by the Department of Energy (DOE) and the National Science Foundation (NSF) for use by the scientific community, Wehner and his co-authors conducted an analysis for the period 1979 to 2005 at three spatial resolutions: 25 km, 100 km, and 200 km. They then compared those results to each other and to observations.
One simulation generated 100 terabytes of data, or 100,000 gigabytes. The computing was performed at Berkeley Lab’s National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science User Facility. “I’ve literally waited my entire career to be able to do these simulations,” Wehner said.
The higher resolution was particularly helpful in mountainous areas since the models take an average of the altitude in the grid (25 square km for high resolution, 200 square km for low resolution). With more accurate representation of mountainous terrain, the higher resolution model is better able to simulate snow and rain in those regions.
“High resolution gives us the ability to look at intense weather, like hurricanes,” said Kevin Reed, a researcher at the National Center for Atmospheric Research (NCAR) and a co-author on the paper. “It also gives us the ability to look at things locally at a lot higher fidelity. Simulations are much more realistic at any given place, especially if that place has a lot of topography.”
The high-resolution model produced stronger storms and more of them, which was closer to the actual observations for most seasons. “In the low-resolution models, hurricanes were far too infrequent,” Wehner said.
The IPCC chapter on long-term climate change projections that Wehner was a lead author on concluded that a warming world will cause some areas to be drier and others to see more rainfall, snow, and storms. Extremely heavy precipitation was projected to become even more extreme in a warmer world. “I have no doubt that is true,” Wehner said. “However, knowing it will increase is one thing, but having a confident statement about how much and where as a function of location requires the models do a better job of replicating observations than they have.”
Wehner says the high-resolution models will help scientists to better understand how climate change will affect extreme storms. His next project is to run the model for a future-case scenario. Further down the line, Wehner says scientists will be running climate models with 1 km resolution. To do that, they will have to have a better understanding of how clouds behave.
“A cloud system-resolved model can reduce one of the greatest uncertainties in climate models, by improving the way we treat clouds,” Wehner said. “That will be a paradigm shift in climate modeling. We’re at a shift now, but that is the next one coming.”
The paper’s other co-authors include Fuyu Li, Prabhat, and William Collins of Berkeley Lab; and Julio Bacmeister, Cheng-Ta Chen, Christopher Paciorek, Peter Gleckler, Kenneth Sperber, Andrew Gettelman, and Christiane Jablonowski from other institutions. The research was supported by the Biological and Environmental Division of the Department of Energy’s Office of Science.
# # #
– See more at: http://newscenter.lbl.gov/2014/11/12/latest-supercomputers-enable-high-resolution-climate-models-truer-simulation-of-extreme-weather/#sthash.HIQAHanC.dpuf
Whoa, there is no fundamental reason models must be wrong. The notion that warming will cause more extreme weather is wrong. The geological record is crystal clear that warmer periods are more stable.
You just have to get the models right. No simple task, particularly when you start out upside down with a horrible misconception that the climate has high sensitivity to CO2. The task before the modelers is to deconstruct the “chaos” apology for failure. I do not believe in chaos. Chaos is groupspeak for “we don’t know squat”.
Resolution will help as it will rub their noses in the dogsquat all the faster, but they are in a position to run multiple scenarios very quickly. I think of it as when I have an unknown map projection and I go to arcmap start cycling through possibilities until one snaps in. A bingo moment. I fully believe the golden age of modeling is coming. It just won’t be tomorrow.
“tuning parameter”? That’s a “fudge factor” isn’t it? In other words they have not used the fine mesh to better model the actual climate they have used it to introduce a new farrago of fudge factors.
Navier Stokes with some fudge factors? Nah, too hard to get agreement with reality and the “deniers are onto it. Way too easy to spot the obvious flaw. Numerical Synthesis with lots of Tuning Paramaters is where its at. Upsize the kool ade too?
The caterpillar undergoes its chrysalis to become… a moth drawn to the light.
I don’t think Mazatlan or Los Cabos would be great tourist spots on that computer planet with as many cyclone hits as they took.
Wasn’t 2005 also the ending point for Mann’s model? I guess that was the year the science was settled.
One way to look at this: A chessboard has 64 squares and 16 pieces. The movement of the pieces is constrained by the rules of the game. It is only recently that super computers have been able to defeat the top human players.
How many squares and how many pieces does the Earth have?
In North America last year (2013) both Environment Canada (EC) and the USA National Weather Service (NWS) predicted a warmish winter, and winter 2013-14 was very cold in the central and eastern two-thirds of Canada and the USA.
It should be noted that certain private meteorologists made accurate predictions of a very cold winter as early as July 2013.
This year is shaping up much like last year – both Environment Canada and the USA National Weather Service have again predicted a warmish winter, and again the same private meteorologists have predicted winter 2014-15 will be very cold in the central and eastern two-thirds of Canada and the USA. So far the private prediction is proving accurate – November has been very cold.
I understand the EC and NWS government weather forecasts rely primarily on computer models that in recent years have a very poor predictive track record.
The private forecasters used analogs from previous years of actual weather history as the basis for their predictions, and recently have a very good predictive track record (I have not attempted to go back further in time).
This limited evidence suggests that the government’s best seasonal weather computer models have little or no predictive skill, whereas older methodologies that rely on historical analogues have a stronger predictive track record.
I further suggest that in science one’s predictive track record is perhaps the only objective measure of one’s competence, and in this regard both Environment Canada and the USA National Weather Service failed in their forecasts for last winter and probably again for this winter. I am a pragmatist, and I am interested in what works – it does not matter if the shiny new computer model is the biggest and the best if it has no predictive skill, and it does not matter if an analog methodology is hundreds of years old, as long as it has good predictive skill.
The reason this all matters is the Excess Winter Mortality Rate –many more people in Europe and North America die in the four Winter months than in the eight non-Winter months.
Repeating an earlier example for Europe and all of Russia:
Assume a very low Excess Winter Mortality Rate of 10% (it varies from about 10% to 30% in Europe);
About 1% of the population dies per year in Europe and Russia, or about 8 million deaths out of about 800 million people;
The Excess Winter Mortality of this population is (4 months/8 months) * 10% * 8 million = at least 400,000 Excess Winter Deaths per year – the real number probably exceeds 500,000.
This is an average number of Excess Winter Deaths across Europe and Russia – it varies depending upon flu severity, cold etc.
Many people in Europe, especially older people on pensions, cannot afford to adequately heat their homes so are especially susceptible to illness and death in winter.
The population of North America subject to cold weather is less than half the above and we have much lower energy costs due to fracking of shales to produce cheap natural gas.
I hope I’ve slipped a decimal or two – these numbers seem daunting. However, if I am correct then here are a few suggested conclusions:
1. Winter weather forecasts matter because people can be forewarned and prepared for a cold winter, or they can be misinformed and unprepared.
2. Government organizations that frequently get their winter forecasts wrong (especially due to a warming bias) are doing a great disservice to their citizens.
3. Even a small percentage increase in Excess Winter Mortality means an increase of tens of thousands of winter deaths that may have been preventable if people were properly forewarned.
4. Excess Winter Mortality particularly strikes down the elderly and the poor.
5. Cheap abundant energy is the lifeblood of modern society, and green activists and politicians who have driven up the cost of energy have done a great disservice to their citizens, especially the elderly.
I expect that global temperatures will start to cool within a decade or less, and Excess Winter Mortality rates will increase. That will put an end to global warming mania, after trillions of dollars have been squandered and many lives lost. As usual, I hope to be wrong.
Best wishes to all, Allan
I have not read the comments, and I suspect that others have noted this, but there will be no golden age in the near future because the past data base has become so horribly corrupted by endless adjustments, that it will be impossible to properly tune the computer, and thereby properly initialise it.
Any computer that is tuned to such past data is bound to give up garbage simply because of GIGO.
Further to Reg Nelson’s comment, it is not simply a question of the number of squares and number of piecies, it is also the different potential move that each piece can make. Since we do not know the precise consituents nor the upper and lower bounds of each and every component that encompasses the catchall esxpression ‘natural variation’ there is no prosepect that in the near future something worthwhile will be outputted from these glorifed games machines.
Again a proof about an AGW straw-hanging.
Again a model simulation run and model projections claimed to prove that climate models’ in high resolution simulation do project close enough to reality with a very good proximity and therefor the climate models in principle and in general still should be considered as correct models about the AGW projections.
One problem there though, this high resolution model simulation is not a proper climate model simulation perse.
It simply is a GW model simulation, as all climate models are simply run as AGW models of climate.
Considering the period of 1979 to 2005 as a Gw period is no wonder or unexpected that such a high resolution model simulation could project close enough to reality.
The projections in this case will be the projection of the GW impact on the weather.
The main problem of the climate models is that actually the models are run as AGW models of climate, and these models are not able to project after 2004 point a close to reality climate, they run too hot for a comfort.
The other problem is that some, for not saying many “climatologists” still do try to impose the AGW as a certainty through the misconception that climate and GW are one and the same always, that there is only one possibility or only one climate outcome, a GW or a AGW.
So, a GW or a AGW model simulation is considered as a proper climate model simulation simply because it’s projections come close enough to the reality for a GW period, aka a GW or AGW simulation being good enough for a given GW period should be considered by default as a proper and unquestionable climate model and there for the AGW still must be considered as a certanty.
To me that is a high resolution cherry-picking to prove a “scientific” obsession with a false certainty of AGW.
Cheers
GIGO.
It is the Golden Age of funding for models.
The thing with golden ages is not that they end, but that the end usually comes about due to underlying corruption or paucity of the underlying philosophy that was always there. In other words the golden age was an illusion.
A Computer Model is a Computer Game. A Computer Model is a Computer Game. A Computer Model is a Computer game. There I’ve said it thrice. What I tell you three times is true.
Add it to the golden age of super dooper computer models-
http://www.theguardian.com/environment/2014/nov/13/lightning-strikes-will-increase-due-to-climate-change
Or Swindlers’ List-
http://whatreallyhappened.com/WRHARTICLES/globalwarming2.html
Super fast BS in and Super fast BS out. or simply GIGO. There simply aren’t enough data points and never will be to accurately predict the weather or climate.
The golden age of climate science models….
No, it’s the golden age of academic grants.
No kidding! – http://www.whitehouse.gov/sites/default/files/microsites/ostp/FY%202015%20Climate.pdf
(Quote) “A cloud system-resolved model can reduce one of the greatest uncertainties in climate models, by improving the way we treat clouds,”
That sounds simple enough, but as a layman observer of this whole thing, I get the impression that it could be a while before we have a good enough understanding of cloud formation and behavior to properly apply their effects in climate models. Even then, how does one apply the more random solar events that influence cosmic radiation’s cloud forming ability? There appear (from common sense & observation) to be many more factors that must be loaded into the “crystal ball” before the chanting is started.
I’m looking forward to Allen’s promised posting on cloud formation.
I’m sorry, Anthony, for calling you Allen. I’m lousy at name retention.
We need more basic research on things like clouds as you suggest however, the hope that these can then be incorporated into models is a waste of money and time at this point. There is so little we know and most of there formulas are not proven facts. So, it all comes down to guessing and hoping a result looks reasonable but someone has to take the big picture and realize any attempt to do modeling over hundred years when we don’t have precise data to start, are guessing at the interactions and the couplings, missing entire data sets in some cases. They talk about 25km square area. We have 3 or 4 weather stations in areas like antarctica that represent 10-20% of the globe. I can’t believe we’ve paid billions for these people to play with computers and guess and guess and guess. Uggh can we please have someone cut them off and get back to science.
Unless the underlying assumptions, algorithms, and inputs are fixed, the models will still not be a valid representation of the real world.
Faster supercomputers can just yield the wrong results in less time
An analog computer would produce results at the speed of light. The precision of the input data is very low to begin with and there is no need for high precision results anyway. Is there really any point in predicting global temperature to anything more precise than a ~tenth of a degree?
If you take imprecise inputs and do a trillion trillion calculations on them what chance is there that the result is at all “predictive?” What they do is constrain the calculations so they produce what “looks like weather.” They used to have the problems that temps would plummet to -200C in some places etc… No doubt they’ve put in all kinds of funny things in the formulas to prevent bizarre results but this is tinker toy play. It has nothing to do with reality.
97.2873576% Consensus, in a snap!
Just more bad science, REALLY, REALLY, REALLY fast!!
Current models: rubbish in, rubbish out.
Future models: even more sophysticated rubbish in, rubbish out.
Advice to the modellers: take time off modelling and coding and think.
I couldn’t agree more but thinking won’t help. We need scientists to do basic research in this subject not more models or even model thinking. We are spending billions on models that have no chance of being valid because they are all based on completely unproven assumptions. I can’t believe we pay for this.
Seriously, is this a joke? The term golden age is funny. Does anybody else see the humor of how ridiculous their statements are: “concluded that a warming world will cause some areas to be drier and others to see more rainfall, snow, and storms.” So would a colder world and I bet if they ran the models with no change they would show increasing storms someplace and decreasing storms someplaces. I can’t believe we are spending money on this. Judith just published a blog explaining how we don’t know enough about the ocean to model anything very well. I believe we need to redirect funding from pointless models to basic science. We need satellites and more buoys and more measurements and more experiments and fewer models based on guesses.
“Give me four parameters and I will draw an elephant for you, and with five,I will have him raise his trunk.”
John von Neumann.
I have a rather simplistic view of computer modeling; since I do it all day long; mostly these days of Optical systems, both imaging, and non imaging. I actually expect that once I build the real system, it will perform just like the simulations say, to within the range needed for the final product. Hasn’t failed to do that yet.
So I’m intrigued by the availability of computer systems with probably one trillion times the computing power that I have to use.
Now science, and physics in particular, is quite fussy about “discrepancies.” We fret over repeatable discrepancies between the output of models, and our observations of the real universe.
Take planetary systems for example; specifically our own system.
A hundred years ago, we had a perfectly good model and theory of how it works as far as basic orbit variables are concerned. It was “Newtonian Dynamics.” Well I’m sure different people called it different things.
One thing Newton’s gravitation and his dynamics told us, was that the perihelion of Mercury (and other planets) would precess about some direction in space.
Trouble was, even then, it did not do so in agreement with Newton’s theory. There were discrepancies.
Specifically, the experimentally observed precession of Mercury’s perihelion differed from the theory to the tune of 43 seconds of arc per century. Astronomers were in no doubt, that their measurements were accurate, and not in agreement, with Newton’s laws.
Now today, you can set all that orbital motion up on your cell phone, and plot exactly what Newtonian dynamics says is supposed to happen, and you can also plot the real astronomically observed motion.
So if you sit in front of your ipad/ped/pid/pod/pud/whatever, and run that Mercury simulation in fast motion for 100 years, the two positions will diverge by 43 seconds of arc over that 100 years of plotting, which will take five minutes to run on your iphone.
The problem is, that the human eye, under the best visual conditions has an angular resolution limit of about one arc minute, so after watching 100 years of history, you still can’t discern the “discrepancy” that everybody fretted about.
Eventually, Albert Einstein waved the whole problem away, and fixed the model that was broke.
So “discrepancies” are a big deal.
So I don’t swallow any claims, that so and so is close enough.
My upper limit on the magnitude of any discrepancy between theory and practice is ZERO.
Now. I still do accept that certain “simplifications” can be made in a theory or model, to conveniently get reasonably close to correct, for rough estimation purposes. But when push comes to shove, If reality and theory diverge, they are NOT the same, and that should be kept in mind.
Now when it comes to modeling the climate, we are in luck, because apparently there exists about 165 years of peer reviewed carefully measured real data about aspects of earth’s climate, and specifically what we tend to call the earth’s mean Temperature.
So with recorded guard banded data, we do know what the earth’s Temperature has been for 165 years, and we have four of five, or more well regarded sources of such data, overlapping each other, and evidently pretty much in quite good agreement with each other. Two of those, the UAH and RSS satellite data sets, are less long lived, but desirable in other respects.
So I think we know within guard bands, what earth Temperature has been for 165 years.
So enter the megateracomputer gizmo, that these “modelers” are so happy about.
Great, I’m all for it. So here is your task, should you choose to accept it.
Twiddle the knobs on YOUR model of earth’s climate, and run it on your whizzmachine.
Publish the result, when your computer model of earth’s mean Temperature, replicates the 165 year long data set(s) that we have observational evidence of. Well of course I don’t expect you to get any closer to the numbers in the data sets, than the known error guard bands of those data set numbers.
At that point, we can all agree that “The science is settled”, and you do have a credible model of earth climate.
I should add, I do not expect them to reproduce the daily readings, or for that matter, even the monthly readings that Lord Monckton uses in his monthly stoppage calculation.
A 13 month rolling average like Dr Roy does now and then is fine, even a five year rolling average.
Well looncraz, of course I mean that we have that record of data that the purveyors of such assert is in fact a measure of the earth temperature.
Good or no good; that is the record that purports to tell us we are slowly roasting.
So that IS the record that the modelers should be trying to replicate
“we do know what the earth’s Temperature has been for 165 years”
I’d have to say that we really don’t. Even current estimates cover a spectrum of over 3°C. The relative changes between the datasets have rather high agreement, however.
That enables a poor model to perform “well” even if it is a degree or two off. Typical use of statistical misrepresentation, frankly.
A model that draws a straight line bumbling around the mean of the “observed” temperature data reconstructions can be almost as accurate as the best model depending on how you determine its accuracy. And, since the AMO/PDO are such powerful influences, and we know when they happened in the past, models are fed with that information, assumptions geared upon that knowledge, and so forth such that they will recreate the general features of that period somewhat decently, even if they diverge wildly outside of that range.
It will be a long time before we have demonstrably accurate models.
Well you will never have an accurate model if you don’t even model the real system. You can’t assume a steady state continuous 342 W/m^2 all over the earth at all points for all times, and expect it to conform to a system that goes from around 1360 W/m^2 max at the center, and drops to near zero just 12 hours later.
You cannot simply average the instantaneous variable in a non linear system; you always get a wrong answer.
And in the case of a black body like radiator, that undergoes a regular periodic Temperature cycle with a variation in Temperature of perhaps +/- 5-6% or more, the total radiated energy is always greater than what you calculate from a constant average Temperature. And the excess is well in the range of these various “forcings” they keep yakking about.
The earth cools faster than Trenberth claims, and it isn’t because it has a higher average temperature.
I tried explaining that but even Dr. S dismissed it