Since there has been a lot of discussion about Monckton here and elsewhere, I’ve offered him the opportunity to present his views here. – Anthony
Guest post by Christopher Monckton of Brenchley
At www.scienceandpublicpolicy.org I publish a widely-circulated and vigorously-debated Monthly CO2 Report, including graphs showing changes in CO2 concentration and in global mean surface temperature since 1980, when the satellites went on weather watch and the NOAA first published its global CO2 concentration series. Since some commenters here at Wattsup have queried some of our findings, I have asked Anthony to allow me to contribute this short discussion.
We were among the first to show that CO2 concentration is not rising at the fast, exponential rate that current anthropogenic emissions would lead the IPCC to expect, and that global temperature has scarcely changed since the turn of the millennium on 1 January 2001.
CO2 concentration: On emissions reduction, the international community has talked the talk, but – not least because China, India, Indonesia, Russia, Brazil, and South Africa are growing so quickly – it has not walked the walk. Accordingly, carbon emissions are at the high end of the IPCC’s projections, close to the A2 (“business as usual”) emissions scenario, which projects that atmospheric CO2 will grow at an exponential rate between now and 2100 in the absence of global cuts in emissions:
Exponential increase in CO2 concentration from 2000-2100 is projected by the IPCC on its A2 emissions scenario, which comes closest to today’s CO2 emissions. On the SPPI CO2-concentration graph, this projection is implemented by way of an exponential function that generates the projection zone. This IPCC graph has been enlarged, its ordinate and abscissa labeled, and its aspect ratio altered to provide a comparison with the landscape format of the SPPI graph.
On the A2 emissions scenario, the IPCC foresees CO2 rising from a measured 368 ppmv in 2000 (NOAA global CO2 dataset) to a projected 836[730, 1020] ppmv by 2100. However, reality is not obliging. The rate of increase in CO2 concentration has been slowing in recent years: an exponential curve cannot behave thus. In fact, the the NOAA’s deseasonalized CO2 concentration curve is very close to linear:
CO2 concentration change from 2000-2010 (upper panel) and projected to 2100 (lower panel). The least-squares linear-regression trend on the data shows CO2 concentration rising to just 570 ppmv by 2100, well below the IPCC’s least estimate of 730 ppmv on the A2 emissions scenario.
The IPCC projection zone on the SPPI graphs has its origin at the left-hand end of the linear-regression trend on the NOAA data, and the exponential curves are calculated from that point so that they reach the IPCC’s projected concentrations in 2100.
We present the graph thus to show the crucial point: that the CO2 concentration trend is well below the least IPCC estimate. Some have criticized our approach on the ground that over a short enough distance a linear and an exponential trend may be near-coincident. This objection is more theoretical than real.
First, the fit of the dark-blue deseasonalized NOAA data to the underlying linear-regression trend line (light blue) is very much closer than it is even to the IPCC’s least projection on scenario A2. If CO2 were now in fact rising at a merely linear rate, and if that rate were to continue, concentration would reach only 570 ppmv by 2100.
Secondly, the exponential curve most closely fitting the NOAA data would be barely supra-linear, reaching just 614 ppmv by 2100, rather than the linear 570 ppmv. In practice, the substantial shortfall between prediction and outturn is important, as we now demonstrate. The equation for the IPCC’s central estimate of equilibrium warming from a given rise in CO2 concentration is:
∆T = 4.7 ln(C/C0),
where the bracketed term represents a proportionate increase in CO2 concentration. Thus, at CO2 doubling, the IPCC would expect 4.7 ln 2 = 3.26 K warming – or around 5.9 F° (IPCC, 2007, ch.10, p.798, box 10.2). On the A2 scenario, CO2 is projected to increase by more than double: equilibrium warming would be 3.86 K, and transient warming would be <0.5 K less, at 3.4 K.
But if we were to take the best-fit exponential trend on the CO2 data over the past decade, equilibrium warming from 2000-2100 would be 4.7 ln(614/368) = 2.41 K, comfortably below the IPCC’s least estimate and a hefty 26% below its central estimate. Combining the IPCC’s apparent overestimate of CO2 concentration growth with the fact that use of the IPCC’s methods for determining climate sensitivity to observed increases in the concentration of CO2 and five other climate-relevant greenhouse gases over the 55 years 1950-2005 would project a transient warming 2.3 times greater than the observed 0.65 K, anthropogenic warming over the 21st century could be as little as 1 K (less than 2 F°), which would be harmless and beneficial.
Temperature: How, then, has observed, real-world global temperature responded?
The UAH satellite temperature record shows warming at a rate equivalent to 1.4 K/century over the past 30 years. However, the least-squared linear-regression trend is well below the lower bound of the IPCC projection zone.
The SPPI’s graph of the University of Alabama at Huntsville’s monthly global-temperature anomalies over the 30 years since 1 January 1980 shows warming at a rate equivalent to 1.4 K/century – almost double the rate for the 20th-century as a whole. However, most of the warming was attributable to a naturally-occurring reduction in cloud cover that allowed some 2.6 Watts per square meter of additional solar radiance to reach the Earth’s surface between 1981 and 2003 (Pinker et al., 2005; Wild et al., 2006; Boston, 2010, personal communication).
Even with this natural warming, the least-squares linear-regression trend on the UAH monthly global mean surface temperature anomalies is below the lower bound of the IPCC projection zone.
Some have said that the IPCC projection zone on our graphs should show exactly the values that the IPCC actually projects for the A2 scenario. However, as will soon become apparent, the IPCC’s “global-warming” projections for the early part of the present century appear to have been, in effect, artificially detuned to conform more closely to observation. In compiling our graphs, we decided not merely to accept the IPCC’s projections as being a true representation of the warming that using the IPCC’s own methods for determining climate sensitivity would lead us to expect, but to establish just how much warming the use of the IPCC’s methods would predict, and to take that warming as the basis for the definition of the IPCC projection zone.
Let us illustrate the problem with a concrete example. On the A2 scenario, the IPCC projects a warming of 0.2 K/decade for 2000-2020. However, given the IPCC’s projection that CO2 concentration will grow exponentially from 368 ppmv in 2000 towards 836 ppmv by 2100, CO2 should have been 368e(10/100) ln(836/368) = 399.5 ppmv in 2010, and equilibrium warming should thus have been 4.7 ln(399.5/368) = 0.39 K, which we reduce by one-fifth to yield transient warming of 0.31 K, more than half as much again as the IPCC’s 0.2 K. Of course, CO2 concentration in 2010 was only 388 ppmv, and, as the SPPI’s temperature graph shows (this time using the RSS satellite dataset), warming occurred at only 0.3 K/century: about a tenth of the transient warming that use of the IPCC’s methods would lead us to expect.
Barely significant warming: The RSS satellite data for the first decade of the 21st century show only a tenth of the warming that use of the IPCC’s methods would lead us to expect.
We make no apology, therefore, for labelling as “IPCC” a projection zone that is calculated on the basis of the methods described by the IPCC itself. Our intention in publishing these graphs is to provide a visual illustration of the extent to which the methods relied upon by the IPCC itself in determining climate sensitivity are reliable.
Some have also criticized us for displaying temperature records for as short a period as a decade. However, every month we also display the full 30-year satellite record, so as to place the current millennium’s temperature record in its proper context. And our detractors were somehow strangely silent when, not long ago, a US agency issued a statement that the past 13 months had been the warmest in the instrumental record, and drew inappropriate conclusions from it about catastrophic “global warming”.
We have made one adjustment to please our critics: the IPCC projection zone in the SPPI temperature graphs now shows transient rather than equilibrium warming.
One should not ignore the elephant in the room. Our CO2 graph shows one elephant: the failure of CO2 concentration over the past decade to follow the high trajectory projected by the IPCC on the basis of global emissions similar to today’s. As far as we can discover, no one but SPPI has pointed out this phenomenon. Our temperature graph shows another elephant: the 30-year warming trend – long enough to matter – is again well below what the IPCC’s methods would project. If either situation changes, followers of our monthly graphs will be among the first to know. As they say at Fox News, “We report: you decide.”
Thank you Joel, for your more mathematical and thoroughly clear explaination of my point.
I would also point out that the IPCC did not project an exponential increase in emissions all the way out to 2100. They seem to be limiting the growth rate after about 2040 based on peak oil, and substitution of non-fossil energy sources. So a simple exponential extrapolation using their 2100 end-point will not exactly match their decade-by-decade calculation of CO2 sources and sinks.
http://www.ipcc-data.org/ancilliary/tar-isam.txt
http://www.ipcc.ch/ipccreports/sres/emission/data/allscen.xls
In the A2 scenario, the population in 2100 is projected to be 15 billion, and coal use about 10 times current levels. Wow.
RE:Brad Beeson: (August 14, 2010 at 8:54 am) “Similarly, I calculated the linear trend observed at the end of each decade based on Mauna Loa data (subtracting data points 10 years apart, multiplying by 10 to arrive at the century rate of change) :”
“1960-1970 = 85 ppmv/century
“1970-1080 = 128 ppmv/century
“1980-1990 = 157 ppmv/century
“1990-2000 = 153 ppmv/century
“2000-2010 = 193 ppmv/century”
I have found that it is possible to represent the seasonally corrected Mauna Loa CO2 concentration data with relatively high accuracy (0.535 ppm RMS error, 1.77 ppm max) with an optimized, three-segment poly-line approximation. This was optimized using the Microsoft Excel Solver utility and using the Match() and Offset to select and access the applicable segment parameters for every decimal date from the source data.
The two internal break points were automatically selected for least error. The accuracy of this curve is slightly better than I achieved earlier using:
X=decimal_date – 1941.106
CO2=126.146 + 2.721347*SQRT(4516+X^2)
Poly-line approximations work best in cases where the data seems to show a series of stepped slope changes. In this case it seems strange that the CO2 slope went up at the time of the OPEC Oil Embargo.
In contrast to the curves above, my simple single-line linear fit has an average error of 2.59 ppm RMS and 6.65 ppm max
CO2 is increasing at 1.97 ppm per year and that rate is accelerating at 0.00176 ppm per year. So, in 2011, it will increase at 1.97176 ppm. Each 6 years, we add another 0.01 ppm to the growth rate.
One has to take a little longer timeframe in mind because the growth rate does fluctuate around these values quite a bit. It increases slightly faster in warmer years and slightly slower in cooler years. One could look at the last 10 years (perhaps because of the cooling trend) and see the exponential trend slowing somewhat but the numbers are close enough to the historical values that it is too early to make that call.
With the June, 2010 value and the historical trends, we will reach 650 ppm by 2100.
Of course, lots of things can happen in the next 90 years to emissions, to the ocean and plant absorption rate (resulting in the 50% airborne fraction), and to the supply of oil and coal, to the population and to technology.
What could also happen is that the IPCC will recognize their errors soon and rewrite the temperature response per CO2 ppm which is off 50% to date. Perhaps people could comment on that far more important issue rather than the IPCC’s CO2 projections which have a range of 500 ppm by 2100 (in other words, its the side of a barn).
I have posted a response to Monckton’s article here:
http://bbickmore.wordpress.com/2010/08/17/the-monckton-files-a-bold-monckton-prediction/
Brad Beeson says:
August 17, 2010 at 5:58 pm
“How is this “leveling off of the rate of rise” different from the previous times it levelled off at solar minimums? (1965, 1976, 1987, 1997, 2008)”
Gosh, you’d think maybe there was a connection between solar activity or something that is great enough to overwhelm any anthropogenic input. Nah, couldn’t be!
How is it different? I don’t see your point at all about 1965 and 1976 – you seem to be grasping at phantoms of noise. Try filtering the series. From 1987 or so onward, the rate has continually decelerated, in much the same way it would at the top of a sinusoidal curve. This becomes even more starkly apparent if you remove the outlier year of 1992.
To Brad Beeson and Joel Shore:
To get his temperature response projections, Monckton was feeding in false numbers for CO2 concentration that were supposed to represent the A2 scenario. See here:
http://bbickmore.wordpress.com/2010/08/17/the-monckton-files-a-bold-monckton-prediction/
But the fact is that the IPCC publishes a table with the central estimate for CO2 concentration every decade. All he had to do was look up the numbers for 2000 (369 ppm) and 2010 (390 ppm) right here:
http://www.ipcc.ch/ipccreports/tar/wg1/531.htm
Thus, 4.7 * ln(390/369) = 0.26 °C, and multiplying that by 0.8 you get 0.21 °C. That’s very close to what you guys said, but nobody can dispute that the numbers I quoted for the CO2 concentration are exactly what the A2 scenario projects.
I guess the moral of the story is that Monckton’s guesstimates might be in the ballpark if he didn’t insist on feeding in fake numbers. Of course, he could have just used the IPCC’s own graphs that say what the transient temperature response under the A2 scenario is supposed to be. (But that would mean admitting that the IPCC’s projections are right in the ballpark, and hence, defeat the purpose.)
Barry,
Interesting! So, the take-home message is that Monckton’s calculations, when actually done using correct numbers, support rather than undercut the IPCC’s projections! Maybe this thread should be retitled, “Monckton: Why Current Trends are Very Much in Line with What the IPCC Projects”. I guess it doesn’t quite have the same ring to it though, does it?!?
Smokey says:August 17, 2010 at 6:08 pm
“That more CO2 is beneficial is beyond question. But there is no evidence of any harm. Is there?”
Depends – how do you feel about a high starch, low protein diet?
http://www.ars.usda.gov/research/publications/publications.htm?seq_no_115=155952, Responses of Wheat Varieties Released since 1903 to Increasing Atmospheric Carbon Dioxide
“…newer varieties did not show a stronger carbon dioxide response when growth and yield were compared at a common CO2 concentration of 290 and 370 ppm.”
“In addition, the newer varieties showed a strong decrease in protein content and baking quality…”
We could eat less bread(not enough gluten) and and drink more beer(plenty of fermentable starch). Sounds harmless to me!
Joel Shore says:
August 18, 2010 at 7:24 pm
‘Maybe this thread should be retitled, “Monckton: Why Current Trends are Very Much in Line with What the IPCC Projects”. ‘
I’m not really paying attention to this because, well, I really don’t care. It’s pretty clear that CO2 is decelerating markedly and it is only a matter of time before it starts decreasing.
But, you two guys sure do seem about to wet your pants over the idea that a fit to a trend tends to have predictive value. It really doesn’t say anything about the underlying hypothesis of what caused the trend, you know.
I guess if the IPCC is projecting “No Warming”, then their projections are meeting the current trends.
The problem here is we are dealing with Math which is a logical, calculation MO. Some of the warmist posters think in emotive terms and hence 0.0C per decade or 0.1C per decade “feels” like 0.28C per decade to them.
Calculate for yourself, how much temperatures have to increase each decade to reach +3.25C in 2100 under A1B. Do the same calculation for the +3.8C under A2. Hint no. 1 : it is 9.05 decades until the year 2100. Hint no. 2 : the temperature has increased 0.7C to date. Show your work.
Re: Bill Illis
It is quite easy to get 3.8 C by 2100 with 836 ppm (A2 number), and a feedback factor of 3.0 typical of IPCC estimates.
Take the increase from 2010
836/390=2.14
ln(2.14)/ln(2)=1.10
CO2-induced warming per doubling =0.94 C
1.1*0.94=1.03 C
Feedback factor 3.0 gives 3.1
Add the current 0.7 gives 3.8
(Myself I would put the feedback factor at 2.7 to match current rates, but that is one way to get the number you showed).
[EDIT: Except the “current” warming isn’t 0.7 and the feedback factor is significantly less than 2.7 (since current warming isn’t anywhere near 0.7, more like half that), so lets try that again… – mike]
Dear Bill Illis,
Here is something that Lord Monckton has never grasped, and you don’t seem to have, either.
Climate models are good at predicting multi-decadal trends, not trends over only a few years. However, the climate models the IPCC uses DO PREDICT that there will be periods of a few years or a decade where there will be no warming, or even some cooling. Look at the IPCC graph at the following URL (the one in the upper left-hand corner is for the A2 scenario, which is what Monckton is supposedly referencing. Look at the individual lines, which are mostly individual model runs.)
http://www.ipcc.ch/publications_and_data/ar4/wg1/en/figure-10-5.html
So apparently you (and Monckton) are quite willing to pontificate about what the IPCC does and doesn’t predict, even though you haven’t really bothered to read up on what the IPCC says about it.
Likewise, Monckton here has not successfully responded to the charge that he likes to blame the IPCC for predictions about CO2 and temperature that the IPCC didn’t actually make.
mike: I went with the 0.7 from the Bill Illis post. He can defend it. It might be 0.6 in my view. What is the baseline? Feedback factor: using Monckton’s own 1.4 C/century
for 199 ppm/century gives 2.5. This is based on current trends, as is my estimate of 2.7.
4.7 ln (2) is 3.25 C for a doubling as Monckton attributes to IPCC, so I am using numbers that are within the bounds discussed already.
[EDIT: Firstly, using GISS or HADCRUT records is a waste of time since we know they are badly flawed due to UHI siting and homogenization issues, as well as the temporal manipulations that the Hockey Team has imposed upon them. Current UAH Global Anomaly is 0.49 C and trend is 0.14C/decade (where Monckton gets 1.4C/century). You would need to subtract PDO/ENSO/NAO/etc noise from current anomaly AND trend before making long term calculations. – mike]
Oh, Jim D, now you are in my sandbox and won’t win with that kind of smoke and mirrors which doesn’t phase me.
First, the feedback is 2 not 3.
Second, the feedback is not happening to date.
Third, +3.0C warming per doubling of CO2 is = 3/ln(2)*ln(CO2 560/CO2 280)
Fourth, if you want to take the IPCC’s 30 year ocean lag into account, one can change the formula to: = 2.4/ln(2)*ln(CO2 end/Co2 orig) = 2.4/ln(2)*ln(856/280) = +3.8C [under the understanding that it will eventually reach 4.32*ln(856/280)= +4.8C
Fifth, so let’s calculate the 2000 to 2010 temperature increase under A2 = 0.19C; How about 2010 to 2020 = 0.23C; how about 2060 to 2070 = 0.31C; how about 2090 to 2100 0.36C.
Now one should also understand that the A2 CO2 concentration level starts out slow and accelerates more towards the end which is not consistent with actual expectations.
Now how much have temperatures increased from 2000 to 2010; ZERO once you take out the impact of the ENSO cycles.
Barry Bickmore says:
August 18, 2010 at 9:39 pm
Are we really supposed to accept a few outlying model runs that have no temperature increase over a decade as representative of the IPCC’s position?
Some random downturns in a few model runs are now the consensus predictions?
Sorry, that is not reality. 1 out of 100 model runs have a random downturn and that is supposed to validate the models (which on average increase by 0.24C over a decade).
Secondly, one would have to show that some temperature-driving-forcing is not active for a full decade and in the current 2000-2010 decade there is NO negative forcing – there is only positive forcings in the models. In fact, every climate model should be predicting an increase now (since the Aerosols negative forcing should have declined by close to 50% in the last decade).
If “No Warming’ over the last decade is the IPCC prediction, then they should have said so beforehand. Instead, they said something like 0.2C per decade (when it is really 0.24C per decade).
Bart says:
Really? Do you want to make some sort of bet on that?
Interesting. People here weren’t saying that when they thought Monckton was correct. Now that we know that Monckton’s post here is all wrong, all of a sudden it is not important anyway.
Bill Illis:
You seem to be very confused about how systems with underlying trends + noise behave. While it may be true that a minority of the models (although certainly more than 1 in 100) show a negative trend over any particular decade, most if not all of them show that there will occasionally be decades when the trend is negative even when the forcings are increasing. Of course, if one is allowed to cherrypick the starting point (and increasingly the ending point, now that current temperatures are quite high) then it is that much easier to find such periods!
This really isn’t that complicated. As I have noted before, it is not at all unusual here in Rochester to have week or longer periods in the spring when the temperature trend is negative despite the fact that the seasonal cycle (which is very strong here in Rochester) predicts the trend should be strongly positive. The fact that your basic argument could be used to disprove the existence of the seasonal cycle ought to give you pause.
Bill, you aren’t getting it. The point is that climate models are not yet good at predicting WHEN you’ll get downturns, but they do predict that you WILL get them, from time to time. Different models produce different results in terms of WHEN those downturns will be, so the IPCC usually just gives “ensemble averages” as its projections. So like it or not, a decade of temperature data doesn’t say much about the reliability of the models, one way or the other. You have to look at at least a couple decades for it to be significant. (Just try calculating the significance of a trend in the global temperature data over the last decade. It is not significant at the usual 95% confidence level.)
Barry Bickmore says:
¨Bill, you aren’t getting it. The point is that climate models are not yet good at predicting WHEN you’ll get downturns, but they do predict that you WILL get them, from time to time. Different models produce different results in terms of WHEN those downturns will be, so the IPCC usually just gives “ensemble averages” as its projections. So like it or not, a decade of temperature data doesn’t say much about the reliability of the models, one way or the other. ¨
That sounds comical to me. Different models may predict downturns at different times, of different or unspecified magnitudes and durations, and therefore the ¨ensemble average¨ consists in predicting that unspecified downturns will occur ¨from time to time¨ or anytime. What kind of a prediction is that, where every outcome is predicted? Reminds me of that weather bureau that predicted a 50% probability of a warmer or colder winter. Or wetter or drier.
It´s just like that old song: ¨Che sera, sera, what will be, will be.¨ I agree with how the song assesses the future. Doris Day had the ¨ensemble average¨ technique pat down.
RE: Bart: (August 18, 2010 at 8:55 pm) “It’s pretty clear that CO2 is decelerating markedly and it is only a matter of time before it starts decreasing.”
As far as I can tell, the Mauna Loa CO2 data may indicate a decreasing rate of acceleration, much like an automobile that is accelerating up to its maximum self-limiting velocity, but I see no evidence of a slow-down. Of course, an Iran-Israeli nuclear war could make all these predictions pointless.
Joel Shore says:
August 19, 2010 at 5:59 am
“Really? Do you want to make some sort of bet on that?”
Sure. I bet you our respective credibilities. It is a dear bet. I consider mine more valuable than gold.
“Interesting. People here weren’t saying that when they thought Monckton was correct. Now that we know that Monckton’s post here is all wrong, all of a sudden it is not important anyway.”
Don’t conflate ME with everyone here. I haven’t even looked at your argument, and am conceding nothing. Maybe you’re right, maybe you’re wrong. I just don’t care because the question is moot, and not as significant as you seem to think it is.
Francisco,
Every outcome (in terms of upward, downward, or flat) is NOT predicted… over sufficiently long times. So if, for instance you picked a 20 or 30-year time period, all the models would have predicted that the global mean temperature would have gone up, but the amounts would be different. That’s why they plot error bars.
If you insist on looking at a single year of data, or 3, or 5, or 10 (like Monckton consistently does) you can get almost any trend you like, whether we’re talking about real temperature data, or model output.
Am I revealing some big secret that the climate modelers don’t want anyone to know? I think not. You can find the same kinds of statements on the Wikipedia page for “Global Climate Model”.
http://en.wikipedia.org/wiki/AOGCM
People who do numerical modeling know that models are always oversimplified, and so can’t predict everything. But they can be good at predicting SOME things, so if you want to challenge a model, you compare it to what its creators CLAIM it’s good for.
When Monckton insists on constantly comparing temperature data from only a few years to model predictions, he’s being dishonest. It makes it worse that he doesn’t accurately reproduce what the model predictions are.
Bart,
The last few years (until this year) have had flat or even cooling temperature trends. The hotter it gets, the less able the ocean is to take up CO2 (that’s basic geochemistry). Put those two facts together, and it is clear why CO2 buildup might be slowing down a little, lately.
But for climate models, data over just a few years might as well be random noise. If you take longer time periods, say 10-year trends in the data, for instance, you find that the rate of CO2 buildup has been increasing! See this commentary on Monckton’s CO2 claims:
http://tamino.wordpress.com/2010/08/09/mo-better-monckey-business/
You people need to get it through your heads. Unless you are talking about setting a new record, or something, a few years of data don’t mean ANYTHING in terms of “climate”.
Bart says:
Okay…although your prediction is sort of vague, which is why I thought a monetary bet might pin it down. (Also, you’re semi-anonymous here, which sort of limits the impact to your credibility.)
What sort of behavior of CO2 levels over, say, the next 5 years would cause you to rethink your views on the cause of rising CO2 levels…Or, do you think that that is just too short a period to verify either way?
Well, it is not so much that I think it is significant. However, I do think it is significant for the set of people who take what Monckton says as having any correlation whatsoever with reality. If you are not one of those people, then this whole thread is pretty much moot.
Francisco:
Just to amplify what Barry Bickmore says, modeling of weather and climate has two components that one can distinguish, aspects that are chaotic (extremely sensitive to initial conditions) and aspects that are not. The chaotic components are very difficult to predict because of this sensitivity (because initial conditions are never known exactly): If you take a climate model and perturb the initial conditions a little bit, you will get a different pattern of wiggles in the global temperature. Similar behavior limits the predictability of the weather…and, in fact, for weather forecasting, particularly over the medium range of 6 – 14 days, meteorology now relies on running the forecast models with a whole ensemble of different initial conditions. The results can help to show how reliable or unreliable various aspects of the forecast are.
Climate models are also run with ensembles of different initial conditions and, although the different initial conditions show a different pattern of wiggles, the overall trend in response to a forcing (such as a steady increase in greenhouse gases) is about the same for all of them when one looks over a long enough period that the response is dominated by that trend and not just the wiggles up and down.
The distinction is often made between modeling weather and modeling climate, although as we tend to use those two words, it is not really a very good place to draw the line. For example, a forecast about whether this fall will be hotter or colder than average for some location is often termed a “climate forecast” but can in fact be quite sensitive to initial conditions.
However, a forecast for the change in the climate with a forcing such as increasing greenhouse gases…or a forecast of a change in the average temperature between winter and summer at a certain location with a reasonably strong seasonal cycle…is an example of something that (while not necessarily easy) does not seem to be very sensitive to the initial conditions.