Roger Pielke Jr. and David Stockwell both asked… can we use 2001-2010 data to falsify IPCC values.
Now, as it happens, I don’t like to use already collected before a prediction to test a prediction. I don’t like that the IPCC does that.
But, oddly enough, the IPCC seems to do this all the time, and includes already existing data in their predictions.
So….
- if you think calling the IPCC’s “2001-2010” predictions, made in 2007 ‘predictions’ even including the eariler data is fair, and
- if the IPCC would normally use this 2001-2007 data to “confirm” their “prediction”, made in 2007
- then, it is, in some sense, fair to falsify their “predictions”, based on the same data the IPCC would use to confirm their predictions.
As you can see, the current Hadley trend is flat, flat, falt.
So yes, if the flat trend in Hadley data persists to the end of 2010, that would be inconsistent with projections of the predicted 2.0C/century trend. Only lower rates of increase would be consistent with the actual weather to a 95% confidence level.
Of course, then the arguments over who is right, Hadley or GISS will certainly come forward.
lucia,
How about looking at the previous IPCC report, i.e., pre-2000? That would not overlap the 2000+ data. What would the test look like then?
Barry W–
I’d have to look at the report. But, if you use data before 2001, there is a positive slope for both the HadCrut and GISS Land/Ocean. So, even without looking if we start any set before 2000, we probably aren’t going to falsify without doing hunting for “sweet spots”. Doing that is called cherry picking!
No matter what reason for warming you want to believe in, there was warming from the 60s through 2000.
Also, I looked at this because both Roger and David pointed out that the IPCC says “Our predictions start Jan 1, 2001”. So, on that basis, I figure doing the a 10 year consistency test starting in 2001, isn’t quite classic cherry picking because 2001 is the first year the IPCC wants to use for validating their own predictions.
FWIW, I think CO2 has a warming effect, and most the 1960-2000 warming is CO2, but some is recovering from aerosols. But the question is: How much warming is due go GHG’s? Where are we going? These are important questions, and figuring out if it looks like more than 2C/century is sort of important.
Yes there has been warming. The question is whether the IPCC predictions are robust, correct?
What I was proposing was that you use the predictions made based only on 20th century data (the 2001 or 1995 reports) and use the 21st century data to test them. No cherry picking involved. You would have at least 8 years of data for the 2001 report and more than that for the 1995 report.
1. IPCC made predictions in their previous report (either 2001(TAR) or 1995).
2 Data exists that was not part of that report nor did it exists at the time of the report.
Can you falsify their prediction from that report using the later data? If not then if the present predictions do not disagree with and are based on that report one could say that report was not valid. Doesn’t make the present report invalid but would throw a pall over the predictive capabilities of the IPCC since the basic numbers haven’t changed that much (1.1 to 6.4 °C for 2007 and 1.5 to 4.5 °C for TAR). (Notice that the range is widening rather that shrinking) My point is if those reports are invalidated then what certainty can you put on the present report?
now, wouldn’t they just say “This falls within the range of our projections” and leave it at that, thus shutting down any line of questioning?
I see a flat trend too—but I can see the Climate Puritans saying “this does nothing to falsify global warming because it falls within the range.” And then insert something about pipelines and such.
What you can say is that the range is truncated.
“A ten year trend of 0.0C makes a 100 year trend of greater than 2C unlikely (p=0.05).”
So all the model predictions above 2C would be falsified.
They will “insert pipelines” and other ad hoc hypotheses to salvage the models form adverse results. I suspect that the La Nina explanation for the cool 2007 is an example of an ad hoc excuse. While true, it was not anticipated ahead of time, and so reflects adversly on the skill of the predictions.
They will keep doing that, but post hoc excuses become less convincing every time they are used, until there is a complete lack of confidence. That confidence was exaggerated is clear from the revision of IPCC sensitivities to 1.1 to 6.4 °C for 2007 from 1.5 to 4.5 °C for TAR.
La Nina and El Nino are weather. The statistical tests accounts for that– it comes into the σT in the explanation on the other post. That standard deviation even includes volcanos. Though, obviously, if there is an eruption, one could just suggest at year 10 that this make the 10 year run exceed the 95% interval.
Sustained strings, if the happen, are what is required to falsify a claim there is no trend. The 10 year string with no slope hasn’t happened yet, but that’s what would be required.
Stop the presses! The boffins at Nasa actually made a falsifiable prediction:
“The solar minimum forcing is thus about 0.15 W/m2 relative to the mean solar forcing. For comparison, the human-made GHG climate forcing is now increasing at a rate of about 0.3 W/m2 per decade (Hansen & Sato 2004). If the sun were to remain “stuck” in its present minimum for several decades, as has been suggested (e.g., Independent story) in analogy to the solar Maunder Minimum of the seventeenth century, that negative forcing would be balanced by a 5-year increase of GHGs. Thus, in the current era of rapidly increasing GHGs, such solar variations cannot have a substantial impact on long-term global warming trends. Furthermore, recent sighting of the first sunspot of reversed polarity (reported Jan. 4 by, e.g., SpaceWeather.com and NOAA) signifies that the ~ 4-year period of increasing solar irradiance is about to get underway.”
http://data.giss.nasa.gov/gistemp/2007/
Soo… if I understand this correctly, they are saying:
Soo…. this would imply that if the 10 year trend lands in “falsification” territory, attributing that to solar variations is inconsistent with the +2.0 C/decade warming hypothesis.
I’m also tempted to look something up quickly. 🙂
There was another sunspot three days ago, not with reversed polarity. We are still in Cycle #23.
==================================
Lucia,
It looks to me that the most recent 12 month temperature data provide by GISS, UAH, Hadley and RSS, and usefully compiled by Anthony Watts on his blog, not only shows the largest anomaly [up or down] on record but highlights the absence of any warming observed over the past few years, somthing that has been flagged by several observers, including Roger Pielke Sr.
NASA’s solar “falisfiable prediction” appears completely at oods with what can be observed: we are in all probability still in cycle 23 and the sun has been very, very quiet indeed for a long time. The longer it takes for cycle 24 to start the greater the likelihood that cycle will be weak. For NASA to continue the “classic” AGW argument about GHG forcings when a recent graph shows R2 values of 0.002 and 0.005 respectively between Hadley and MSU temperature data and CO2 concentrations strains credulity.
Well done, lucia. I always liked your posts over at CA.
Someone is now tackling the issue of falsifiability of climate science, GCM and IPCC claims. This is the essence of science. It’s a somewhat arcane issue for the general public, but crucial for those of us who have some grasp of how science works. Were Karl Popper still alive, I’m sure he’d approve.
lucia,
You commented above that the globe warmed from the 1960s to 2000. That is not exactly accurate. The warming began in 1975/76. The globe cooled from 1945 to 1975. It is not a major issue perhaps, but correct dates allow for more precise thinking. The PDO switched to its warm phase in 1975/76 and I do not think this is a coincidence.
Ron– Yes, But if you do an OLS including the 60s, you still get warming. I guess that’s what I meant. The recent “up” is big.
But yes, if you know some meteorology, you might be able to line up changes in trends to specific events, and that could matter. As for the future: we need to wait and see!
http://ams.confex.com/ams/pdfpapers/100694.pdf
1964 is the changepoint
Lucia,
Anthony has an interesting post: http://wattsupwiththat.wordpress.com/2008/02/29/interesting-plots-of-temperature-trends-the-4-global-temperature-metrics-according-to-basil/#comments
It appears that 3 out of the 4 datasets had a 14 year trend <0.6 degC/decade from 1979 to 1992.
At first blush these dates look like shameless cherry picking (Pinatubo erupted in 1991). However, the endpoints appear to be choosen based on inflection points rather than the point that would give the minimum trend.
But this does raise of couple of questions:
- Does your statistic test still have meaning if endpoints are cherry picked?
- Can we use the past data to say that the trends >2.5 degC/century are not consistent with the observations?
I would say yes to both questions. But if the the answer is no to either then I am confused about what your proposed test is supposed to mean.
Raven– the sets I looked at had positive trends for 10 years. But, if I looked at specific ones, I might say otherwise. I haven’t looked at all possible ones!
Lucia,
I probably should clarify. None of the datasets show 0 trend to date. However, it appears that the 14 year trend from 79-91 weighs in around +0.6 deg/century which is the falsification point for trends >2.6 degC/century according to your numbers in a previous post.
If my understanding of your analysis is correct then we should be able to say with 95% confidence that future temperature trends cannot exceed 2.5 degC/century. In my view that would be huge progress even if we never get to the point where the 2.0 degC/century trend is falsified (which may be impossible since it could actually be right).
Incidently, your approach is similar to what James Annan tried to do with his paper on CO2 sensitivity but you are applying it to actual temperature measurements instead of a theoretical construct.
Raven– I see what you are saying. I think someone asked me about 5, 8 and 10 years?
One thing you should be cautious about: It’s fun to look back and see if anything is falsified anytime. But it’s also best not to pick and choose start dates arbitrarily.
Fishing through and picking and choosing actually affects the statistics, because you are being permitted to pick the string you like.
You say “14 years fails!” Someone will ask: 15 years falsify? If 15 years didn’t fail, clearly someone can just point out that your choice isn’t robust. And, they can also point out that my “falsification” has a confidence interval. One in 20 runs will “falsify” even if 2.0 is true. Given this, if anyone just get to sift through a lot of runs, they will find one that falsifies.
It’s for this reason that I am being very nit-picky in saying that to falsify IPCC predictions, we need to start with some data when they made those predictions. And, it’s for this reason, I wanted to do the analysis before the full data set arrives– so even before we get the final bit of data for the 10 year run.
Only the psychic can “cherry pick” before the data are even in!
Other than that: you are interpreting correctly. I would say that, based on past data, if you look at a lot of it, the past trend is not consistent with 2.0 C/century or higher. It’s been around 1.8 C/century or so, and holding for a while. It’s slower more recently. (Whatever that means.)
But we also need to recognize that the GCMs (and even Lumpy) predict non-linear behavior. So, they are actually predicting the warming rate to be higher in the next 20 years compared to the previous 20 years. So, logically,… you can’t prove a prediction 2.0C/century for future is wrong by showing it didn’t apply far in the past. The GCMs themselves suggest the future trend is higher than the past trend.
But, yes. I’m pretty sure if you do 1965-2008, and various past strings of data, you’ll find it inconsistent the higher ends of warming predictions for the future. The climate warmed warmed , but not at the higher rates, and certainly not at the very highest rates predicted for the future.
The search for falsifiable predictions is very important and probably the best way to make progress. What we need are ones that would only hold given greenhouse warming, and not any other sort. Here are a few that I have come across:
1) The troposphere should warm faster than the ground. (this was disputed on tamino but is asserted by McKittrick)
2) The stratosphere should cool (whether this is in the presence of overall warming, not sure)
3) The retained heat of the oceans should rise (Pielke, this one)
4) There should be changes in the Walker circulation (not sure what this is or why, it was on Pielke)
There must be more. Are the poles predicted to warm faster than the rest of the planet only if the warming is GG? Is there any pattern of increases of CO2 which have to precede water vapor rises which in turn have to precede temps? Is there any pattern in day vs night temps that are forcast only by GG?
What is really needed is something which would discriminate between GG warming and whatever caused the MWP and LIA.
In my experience, arguing with global warmers, no data could falsify anthropogenic global warming, any more than it could falsify the proposition that the wine and wafer miraculously transubstantiate into the blood and body of Christ.
Anthropogenic global warming may well be true. We have neither compelling evidence against substantial anthropogenic global warming, nor compelling evidence for substantial anthropogenic global warming, but if in the next decade or two we obtain substantial evidence against it, as seems increasingly probable, it is not going to make much difference – recall the silicone controversy.