Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
The graph below shows the predictions of James Hansen's 1988 climate model overlaid (in blue) with observed temperatures. Hansen's scenarios B and C have turned out to be very good predictions of what actually happened.
Of course, it is an article of faith amongst the global warming skeptics that the models are wrong, so what do they do?
Well, there are only two things you can do to make Hansen look bad -- you can misrepresent the results of his model, or you can misrepresent the instrumental record.
The first approach is the one taken by Pat Michaels, who dishonestly erased scenarios B and C from Hansen's graph. The second approach is the one taken by Willis Eschenbach over at Climate Audit. If you move your mouse over the figure above, you can see Eschenbach's version. By doctoring the instrumental measurements so that they were all lower, he makes it look like Hansen's model predicted more warming that what was actually observed.
The trick Eschenbach used was to use a single year for the baseline instead of the thirty year average that is normally used. Yes, it's another version of the disingenous baseline game that produced all those bogus "global warming ended in 1998" claims. Given the year to year variability of climate, by choosing the right year to use as a baseline you can manufacture almost any result you want.
Comments
His justification was that since the y-axis was delta-t then everything ought to start out at 0. Well, that is wrong if you are using the 61-90 period as a standard (what 0 is) so it is entirely possible to have a delta-t which is not 0 at the 1958 starting period.
But hey, Willis is a nut anyway -- a firm believer that the entire climate change problem is a conspiracy. All the models are wrong, measurements are doctored, proxy recons are a result of malfeasence, yada, yada, yada. In fact most of the people over there seem to hold that view...
Posted by: John Sully | August 29, 2006 1:48 PM
Could you update the article to make clear that the HADCRUT3 blue line(s) are Eisenbach's ?
It was not immediately clear to me, who does not follow the GW debate closely.
Posted by: _Arthur | August 29, 2006 1:50 PM
Tim: That is a really nice graphic - although perhaps not in the way Willis intended it. By mousing over it you can see how accurate it is.
John
Posted by: John Cross | August 29, 2006 1:52 PM
Hey, if you think Willy's graphing is a joke, you should read the "biology" paper he links in one of the comments. Wow, that site gets loonier by the day.
Posted by: Bob | August 29, 2006 2:25 PM
I've viewed this page in Firefox and IE and I can't see the graphic or a spot where it should be showing up. Help?
Posted by: PT&S | August 29, 2006 2:25 PM
Regardless of the observed temperature data that you use (hadCRUT3 or GISS) or the starting point (1958 or some average), the observed line still falls below Scenario "C". It can easily be seen that in 2006 the "C" scenario makes a big jump, while the observations do not.
But "... scenario C assumes a rapid curtailment of trace gas emissions such that the net climate forcing ceases to increase after the year 2000."
This didn't happen, as far as I'm aware, so it appears that Hansen's projections are way off the mark.
Posted by: nanny_govt_sucks | August 29, 2006 3:13 PM
I assume Hansen's model was predicting against the the 30-year baseline. If that's the case, then Eisenbach is comparing apples to oranges, something that's outright dishonest, not merely purposefully deceptive (like the 1998 cherrypicking).
BTW, Crichton said he'd trust models if they predicted the future accurately for 10 years. He'll be one of us any day, now. Or not.
Posted by: Brian S. | August 29, 2006 3:16 PM
Where is the justification for using this "thirty year average"? What do you mean by "normally used"?
I'm not sure that I see how obvious it is that an average should be used over a starting point. Don't the models have starting input parameters that are made as close to reality as possible? Why would adjusting these after the model has run make any sense?
Posted by: nanny_govt_sucks | August 29, 2006 3:21 PM
Still totally lost on this one... Won't each of the model runs (A, B, C) have different 30-year averages? Yet it looks like they all have the same starting point. What's going on here?
Posted by: nanny_govt_sucks | August 29, 2006 3:54 PM
The baseline for delta-t == 0.
Posted by: John Sully | August 29, 2006 5:05 PM
Tim: You meant to say that Michaels erased scenarios "B and C", not "A and B". A was retained. [Oops. Fixed. Thanks. Tim]
I don't understand why moving the baseline back to zero isn't valid - if all the scenarios start at zero you should compare them to a temperature trend that does so as well. Otherwise you already have some warming "baked in" to the comparison. Are you claiming there's some other year at which they matched, prior to 1958?
Anyway, with or without that adjustment it's obvious Scenario C is the line that comes closest to describing the temperature trend, yet C is supposed to reflect the outcome of a world in which there is "rapid curtailment of emissions such that net climate forcing ceases to increase after the year 2000". Did that happen? No. So the model overstated warming.
Posted by: Glen Raphael | August 29, 2006 5:21 PM
To explain John Sully's point further, nanny, scenarios are predictions of the future. The starting point is based on observations of the past and is the same for all scenarios.
Posted by: John Quiggin | August 29, 2006 5:26 PM
I did a whole series about Hansen's scenerios and the actual path that was followed. The short answer is that B and C pretty much follow each other until 2000, both in the emission scenerios and in the temperature profiles. Since many of the greenhouse gases have long atmospheric lifetimes and the amount builds over time the effect of changes in emissions becomes apparent only after appreciable time. Hansen predicted that B and C would diverge after 2010 from drastic cuts in CO2 emissions assumed to occur in 2000 kicked in. I believe Tim published a version of the graph from Hansen showing the various forcings. (towards the bottom on my page
This is a detailed description of Hansen's three Scenerios
This shows the actual emission profiles.
And here and here I compared the actual emissions up to date with those in Hansen's Scenerios.
There is also a long discussion in Prometheus after I left the premises
Posted by: Eli Rabett | August 29, 2006 5:27 PM
nannygovtsucks said:"Regardless of the observed temperature data that you use (hadCRUT3 or GISS) or the starting point (1958 or some average), the observed line still falls below Scenario "C". It can easily be seen that in 2006 the "C" scenario makes a big jump, while the observations do not."
We've got the data on 2006? Somebody forgot to wake me up!
Also, Hansen made assumptions about greenhouse gases (including methane and CFCs), and volcanic eruptions.
nannygovtsucks said:"I'm not sure that I see how obvious it is that an average should be used over a starting point."
Choosing a particular year changes the result based on whether that year was abnormally hot or cold - you don't see a problem with that? GW is a change in climate, which is the average weather.
Posted by: Tracy P. Hamilton | August 29, 2006 5:37 PM
Glen, Hansen constructed the baselines for the observed and calculated data using the same period. To arbitrarily pick the zero for the observations and the calculations using the average over different periods (btw, he had to have used a different period that 61-90 as the calculations were made in 1988) is to arbitrarily shift one vs. the others in a dishonest way.
Nanny, don't be a goat. The zero of the scale is chosen by separately averaging the Delta T for the observations and each scenerio over a period, say 1958-1978. One then graphs the difference by year between the observation and the average observation, and the scenerio A and the average scenerio A. This puts everything on a consistent basis.
Further, models are rung up, that is you have to start them from way before you start looking at the outputs. This is one of the things that was wrong with von Storch's model of millenial temperatures. By the time the GISS model got to 1958 it would be natural that there was a difference (although a small one) between the different scenerios. If they were the same something would be wrong.
Posted by: Eli Rabett | August 29, 2006 5:42 PM
So John Q, how does the baseline of a "30-year average that is normally used" fit in? The 30-year averages for scenarios A, B, and C will be different, yes? So why do they all have the same baseline in 1958?
Posted by: nanny_govt_sucks | August 29, 2006 5:47 PM
Eli, in your second paragraph, it sounds like you are talking about a graph of the anomalies. I think I understand that, and it makes sense then why the observed temps might start somewhere other than zero, but it doesn't explain why scenarios A, B, and C all start at the same point.
Certainly if you take the difference between the first calculated value for scenario A and the average for scenario A, and do the same for B and C you're not going to come up with the same value, are you?
Posted by: nanny_govt_sucks | August 29, 2006 6:03 PM
"So why do they all have the same baseline in 1958?"
You're asking why scenarios drawn up in the 1990's to predict events in the 21st century start from a common base point in 1958?
Think about it.
Posted by: Ian Gould | August 29, 2006 6:06 PM
Ian, no, I'm asking why using a 30-year average to produce an anomaly chart would result in 3 different calculated datasets having exactly the same starting point.
The starting point for scenarios A, B, and C appears to be 0 or very near to it. So if this is an anomaly graph, then the first calculated value for scenarios A, B, and C turned out to be exactly (or very near) the average for each series, respectively!
What are the chances?
Posted by: nanny_govt_sucks | August 29, 2006 6:18 PM
Well, if all of the forcings are considered to be the same until some point after 1958 I would hope that a decent model (and GISS is one of the better ones) would come up with a similar climate on each of the runs.
So, you ask what are the chances? I would say they very high that the model results for the 3 runs would be very similar.
Posted by: John Sully | August 29, 2006 6:29 PM
The runs started in 1958 with the same initial conditions. Hansen refers you to another paper for more details of the model, but it would obviously be wrong to use the temperature just in 1958 to initialize them -- instead a thirty year average centred in 1958 makes sense.
Hansen's graph shows delta from the 1951 to 1980 average. The numbers in HADCRUT3 are deltas from the 61-90 average, so I had to add the differences in the average to them.
Posted by: Tim Lambert | August 29, 2006 10:42 PM
Still struggling to understand here, Tim. Why is it "obviously wrong to use the temperature just in 1958 to initialize them"?
Why does a 30-year average make sense? That doesn't yeild an initial condition. Why use it as a model starting point?
How is a 1951-1980 average a "thirty year average centred in 1958"? Shouldn't the average be calculated based on 1943-1973?
Posted by: nanny_govt_sucks | August 29, 2006 11:26 PM
Tim, could you make the legend change when the image changes? Maybe "as published" and "as misrepresented" or something?
Nan, poke around with Google for any long time series you want -- inflation, rainfall, percentage fat in market hogs -- and you'll always find that any predictions are related not to some particular year but to a 20- or 30-year average.
It's a way of trying to be honest by avoiding picking some extreme year that's far from the average, then claiming that the following years (which are closer to the average, most likely) are a trend instead of just a scatter around the average.
Posted by: Hank Roberts | August 30, 2006 12:12 AM
As to 'rapid curtailment' -- the USSR collapsed in 1991. A bit earlier than the 2000 scenario, and not a permanent change in fossil fuel use, but certainly a big drop in fuel for a while. I don't have a cite handy.
Posted by: Hank Roberts | August 30, 2006 2:27 AM
Hank, scenarios A, B, and C are model runs, not some kind of statistical analysis. It is my understanding that models attempt to get the inital parameters as close as possible to reality. Using a 30-year average for a starting temperature is not reality.
Posted by: nanny_govt_sucks | August 30, 2006 2:52 AM
Nanny, even if the models were "perfect" there still would be deviations between the model result and the reality each year since there is a random component to climate no model is going to catch. Using 1958 to normalize results makes sense only if you believe that for that particular year the model got it perfectly right, which there is no reason to assume. Using an average over many years you will reduce the influence of these random errors.
Posted by: Thomas Palm | August 30, 2006 3:09 AM
Either way, the overlaid temperatures in blue most closely match Scenario C -- a scenario that hasn't happened. It may be too early to tell but so far Hansen's model isn't holding up.
Posted by: C Monte Alban | August 30, 2006 4:32 AM
Maybe it's just my eyes, but it appears that the model runs start just below the 0 average line.
But this line is apparently the 1951-1980 average which isn't centered around 1958 anyway.
So what did Hansen use for a starting point for his model runs?
More questions than answers on this one...
Posted by: nanny_govt_sucks | August 30, 2006 4:42 AM
Hansen used the same conditions for all three of the scenarios starting in 1958. The 1958 model annual mean (the starting point) was slightly below the 1951-1980 mean that's used as the baseline. The observed value in 1958 from both GISS's global mean (red line) and the HADCRUT2 global mean(blue line) was above the 1951-1980 mean. Even though the 1951-1957 values aren't plotted, you can see the problem in the two versions of the blue line. In the Eschenbach version, there are only 3 years above zero by my eyeball (1973, 1979, 1980). It looks like the 1958-1980 mean in Eschenbach's is about -.15 C, so that, to get to the 1951-1980 mean=0, the first seven years of the period would have had to have a mean of ~0.5 C, which from getting the data from the Hadley Centre, it's easy to see isn't true. In the version Tim shows here, the 1958-1980 HADCRUT2 mean is ~0. What Eschenbach has done is subtract off 1958-(1951-1980 mean) from the temperatures. Since 1958 was warmer in the obs that 1951-1980 overall, it results in lowering the HADCRUT2 curve.
Posted by: Harold Brooks | August 30, 2006 6:33 AM
1958 was unusually warm because it was an El Nino year, just like 1998.
Posted by: Tim Lambert | August 30, 2006 6:49 AM
I tried posting a link to your website but got flagged for "spam". So I posted this:
Would you care to comment on this?
http://tinyurl.com/rrbdt
Someone seems to think you are doctoring the data.
Here: http://www.climateaudit.org/?p=796#comment-43543
Lets see how honest he is.
Posted by: wacki | August 30, 2006 7:06 AM
oh crap, you already commented in that thread. heh, it's 6 AM and I just woke up. :-p
Posted by: wacki | August 30, 2006 7:08 AM
I would have thought that if you wanted to compare model projections with observational data, you would need to use a common origin, in order to register the true departure of the projections from the observations.
Is this some sort of handicap race, like the Stawell Gift?
Posted by: James | August 30, 2006 9:05 AM
James, the common origin is the AVERAGE over the 1951-81 period, not the value for a particular year (which is an average anyhow, so don't try and start that).
Posted by: Eli Rabett | August 30, 2006 9:28 AM
Nanny
Please do some reading on climatology. By international convention 'normal' is established using a 30 year window of data. In order to provide some coherence to this term, normals are recalculated every ten years (not every year), for example, 1941-70, 1951-80, 1961-90, 1971-2000. So normal is actually a moving target. Hansen made these model runs in the late 1980s, the normals during that time were based on the interval 1951-80. That period forms the baseline for comparison with the model runs. It is just that simple.
Whenever one examines departures (anomalies) from a normal it should be clear over what period normal was calculated. Otherwise you run the risk of comparing apples to oranges. Indeed many plots of temperature anolmalies (or other variables) are still shown relative to the normal over the period 1951-80. If you read the materials on the GISS website you will find a file that addresses this very issue.
Alas no vast government conspiracy here.
Posted by: J Hamilton | August 30, 2006 9:36 AM
Nan, two mistakes earlier -- it's the B scenario, not C, that shows a jump around this time (but also, remember, those are scenarios, not predictions, and the runs were made long in the past -- no particular jig or jag in the line matters.
You read this back in 1999 when it came out? If not, do; it answers a lot of the questions you raise here. Read the linked articles and brief summaries from later years too. For example, he answered the question about B vs C vs real numbers that you raise. http://www.giss.nasa.gov/edu/gwdebate/
And you say above that there was no reduction in fossil fuel use --- you've forgotten the former USSR!
That economy collapsed in 1991, followed by a dramatic reduction in fossil fuel use. The trend's still below what it would have been but for that event http://climatechange.sea.ca/fossilfuelemissions.gif
You can see what happened. Follow the link below the image for the source, or Google will find the info for you.
Posted by: Hank Roberts | August 30, 2006 11:56 AM
Everyone look again at the chart. The model runs do not start at 0. So apparently the average for 1951-1980 is not used as a "common origin" for the model runs.
Posted by: nanny_govt_sucks | August 30, 2006 1:25 PM
The mean of the model in 1951-1980 is set to be the same as the mean of the observations in 1951-1980. The value for 1958 does not equal the mean of 1951-1980 in either the model or the observations. If it did, then the model runs or the observations would have a value of 0 in 1958. Since the 1958 value for the model does not equal the mean for 1951-1980, it does not. The modelled 1958 was cooler than the modelled mean for 1951-1980, so the model value at 1958 is less than zero. The observed 1958 was warmer than the observed mean for 1951-1980, so the observed value at 1958 i greater than zero.
Posted by: Harold Brooks | August 30, 2006 2:01 PM
Hank Roberts: That economy collapsed in 1991, followed by a dramatic reduction in fossil fuel use.
Hank: So are you saying that Scenario C is happening because of this 1991 USSR event?
Posted by: C Monte Alban | August 30, 2006 2:21 PM
"Given the year to year variability of climate, by choosing the right year to use as a baseline you can manufacture almost any result you want."
Sily us, who thought the rate of change of temperature was the measurement at issue.
Posted by: z | August 30, 2006 7:16 PM
Scientists, check me here. I think you all are being too terse, too technical, and not spelling things out. Imagine you are talking to someone with the US average education and reading level, 7th grade -- who is going to inherit and have to manage everything you love, and you're trying to explain how to understand it.
Tolerantly and very patiently. I'm trying to undertstand it myself by rephrasing and rewriting below, as though to my young relations. Give it your best help please.
CMAlban -- none of the "scenarios" are happening.
Reality is happening. Each scenario used assumptions, to run the model, and see where it would go. One of the scenarios assumed a large volcano at some point. Reality gave us Pinatubo. Scenario C assumed a significant drop in fossil fuel use (after 2000) due to intelligent design. Reality included a significant drop in fossil fuel use by collapse (1991) of the USSR economy.
All the scenarios do is show how -- back in 1988 -- a computer model ran through its paces given the same assumptions for a long stretch of time (starting in the known past) and then three different sets of assumptions, starting from 1988.
Ok? The model started a long, long time ago. The model took the baseline average for a standard 30 year period, the same period many other models and what we call "normal" conditions uses for comparison.
The model ran three identical sets of assumptions through some past years. Each model run is a bit different (modelers, can you explain this? It's because there are too many things varying, each with uncertainties, but it confuses people who think it's like a sewing machine that should always stitch exactly the same each time).
So the models ran from the known baseline through the known past years, and matched the actual temperature well enough to be credibly working.
Then at 1988, the assumptions were changed (that was "now" at the time -- so the models were about to run into the then future). That's why models are used.
The models ran on out into the future.
So did we, at our own slow pace of one day per day.
The models ran up to 2018, is that correct? Modelers, please comment.
Each model -- each scenario plugged into the model computer program -- gave different results after 1988.
Reality, plugged into reality, also gave its own results, up through today at least.
In one model ("C") it was assumed some intelligent folks would decide to drastically limit use of fossil fuels at a certain point in time.
In one model ("A") it was assumed business would go on as it had before 1988.
In one model ("B") an intermediate set of assumptions was used.
In (one? two? all?) of the models, a large volcano was assumed to happen.
In reality, a large volcano happened.
In reality, a large drop in fossil fuel use happened in 1991 when the USSR collapsed -- and that was temporary, and fossil fuel use started up again.
Now. Don't keep thinking the models or the various scenarios are anything but old history -- they're just a look at how, long ago, the climate scientists were trying to figure out what might change.
"Reality is what, when you quit believing in it, doesn't go away." -- P.K. Dick
Posted by: hank | August 30, 2006 7:55 PM
Perhaps simpler -- read the Google Scholar results page, for the 24 articles here, all using a 30-year baseline:
http://scholar.google.com/scholar?sourceid=Mozilla-search&q=%2230-year+baseline%22+%2Baverage
Posted by: hank | August 30, 2006 8:17 PM
Hank Roberts:
Not sure what you are saying in that last post, but I think you summed it up well. Reality doesn't go away. And the reality best matches Scenario C, the scenario where we ceased emitting CO2. So what's right about Hansen's model? Out of the 50 or so comments no one explains that....
Posted by: C Monte Alban | August 30, 2006 9:01 PM
Yes, the answer's there to be read, and I posted the link above:
Hansen has answered the question about B vs C vs real numbers that you raise. http://www.giss.nasa.gov/edu/gwdebate/
Let us know if you have trouble understanding it. I think it's very clear.
Posted by: hank | August 30, 2006 9:08 PM
Even simpler, read just this excerpt from one paragraph of that January 1999 article:
"... the period of comparison is too short and the climate change too small compared to natural variability for the comparison to provide a meaningful check on the model's sensitivity to climate forcings. With data from another decade we will be able to make a much clearer evaluation of the model."
The "other decade" of data will be reported sometime during the year 2010.
Okay? You are looking at a picture up there. You can't believe -- I hope -- that you can tell more from the picture than the underlying math tells the people who drew the picture, eh?
So you know that it will take a few more years to have enough numbers for the math to be statistically meaningful.
The picture will change then. But it will still be just a picture, one that represents a larger and better collection of numbers sufficient to tell what's different from what.
Posted by: hank | August 30, 2006 9:23 PM
Hank, the article you reference is from Jan 1999. At that point there wasn't enough data to match a scenario. But now even my sixth grader says 'C'.
So we are (as we should) judging the model in the year 2006 which is closer to the other decade of data (up to 2010). OK, so the games not over for Hansen yet, but he's going to need a 4th quarter comeback.
Posted by: C Monte Alban | August 30, 2006 9:46 PM
Scenarios B and C don't diverge until after 2006. Results so far are close to both B and C. In a few more years we'll see if temperatures now stabilise (scenario C) or continue to increase (scenario B).
Posted by: Tim Lambert | August 30, 2006 11:02 PM
Can we get that picture again with error bars? Clearly they're lacking -- a point made in the recent House hearings about other charts as well.
Posted by: Hank Roberts | August 30, 2006 11:17 PM
Hank: "none of the "scenarios" are happening. Reality is happening."
This is well-phrased and so broadly applicable that it bears repeating.
Posted by: Ian Gould | August 30, 2006 11:20 PM
C Monte alban "Scenario C, the scenario where we ceased emitting CO2."
Scenario C does not assume that we "cease emitting Carbon dioxide", it assumes that the net radiative forcing from all GHGs in toto will slow in the 1990's compared to the 1.5% per annum growth in the 1970's and 80's and then stabilize at 2000 levels.
You may have read recently about the faster than expected phase-out of CFCs under the Montreal Protocol.
Posted by: Ian Gould | August 30, 2006 11:36 PM
I'm looking for more recent data and dat for total global emissiosn but emissiosn of the Kyoto Annex I parties fell by almost 7% between 1990 and 2002.
http://ghg.unfccc.int/graphics/graph2.gif
Posted by: Ian Gould | August 31, 2006 12:01 AM
I think CMA expresses what's happening -- people think this is like a horserace or a sports event -- where the announcer can be claiming that so and so is "ahead" or "winning" before the results are in.
Posted by: Hank Roberts | August 31, 2006 12:10 AM
I said "Scenario C, the scenario where we ceased emitting CO2." which is wrong. Ian, you are right, I meant to say ceased increasing CO2 emissions. I hope that's accurate.
I found another link on the same subject with some info about what assumptions were made in each scenario.
Mr. Pielke Jr. states that the assumptions for Scenario B were incorrect therefore if B does fit reality it would be for the wrong reasons.
And the graph looks slightly different there. Perhaps different data for the observations?
Also, are there other model runs and their predictions available to review? Is there a web site that tracks the prognosticators? Yes, Hank, I do tend to think of this like sports betting.
Posted by: C Monte Alban | August 31, 2006 12:38 AM
Maybe it's just my eyes, but even in the un-shifted version above I can see that the blue line is right on the "C" scenario from 2001 to 2005 while the "B" scenario floats above.
Sorry Tim, but I think it's time to remove your rose-colored glasses.
Posted by: nanny_govt_sucks | August 31, 2006 1:20 AM
http://www.cmdl.noaa.gov/aggi/
Total radiative forcing including from GHG's other can C02 increased by 0.18 Watts per metre squared between 2000 and 2005.
That's higher than the 0.16 watts for the period 1995-2000. The increase in CO2 emissions overpowered the effect of stabilization or decline in CFC and methane emissions.
However there's a lag between GHGs entering the atmosphere and their full impact on climate.
Posted by: Ian Gould | August 31, 2006 2:59 AM
Frankly I would not invest a dime in Roger Pielke's view on this. Among other things he does not understand that for small x, exp(x) = 1 + x. Thus a scenerio can call for an exponential growth of emissions, which over a decade are pretty close to lines. Roger P needs a Cal I course
All three scenerios call for about the same CO2 emissions up until 2000, where scenerio C goes flat. Between 1990 and 2000 (2006 for A and B) a straight line is a good fit.
If anyone is foolish enough to be interested you can find the scenerios here , but the point is that the differences between B and C are mostly in the other green house gases.
Finally if you want real observations of mixing ratios, you can find them here and they clearly show the effect of the fall of the Soviet Union and the ramping up of the Montreal Protocols in the early 1990s.
Posted by: Eli Rabett | August 31, 2006 9:44 AM
If people go to the climate Audit thread and scroll down ot message 50 Essenbacj summarises Hansen's scenarios and the outcomes:
Scenario A
CH4 0.5% annual emissions increase N2O 0.25% annual emissions increase CO2 3% annual emissions increase in developing countries, 1% in developed
Scenario B
CH4 0.25% annual emissions increase N2O 0.25% annual emissions increase CO2 2% annual emissions increase in developing countries, 0% in developed
Scenario C
CH4 0.0% annual emissions increase N2O 0.25% annual emissions increase CO2 1.6 ppm annual emissions increase
and the results were
Actual
CH4 0.5% annual emissions increase N20 0.9% annual emissions increase Developed countries CO2 emissions increase 1988-1998 -0.2% Developing countries CO2 emissions increase 1988-1998 4.4% CO2 atmospheric increase PPM (Mauna Loa 1998-2003) 1.6 ppmv
In the event, Scenario A was closest for methane, C was closest for CO2 (up until 2000, when it leveled off), and B wasn't closest for anything ...
So according to Essenbach, the observed 1.6PPM annual increase in C02 concentration since 2000 is in line with hte assumptions of Scenario C while the observed concentrations of methane and nitrous oxide are higher than predicted by Scenario C.
Referring back to the tables in the source I quote in my last post, carbon dioxide accounts for around 60% of total warming.
Let's note too that - according to Essenbach - real-world carbon dioxide emission growth is in line with the lowest of the Hanssen scenarios. Somebody refresh my memory, didn't Scenario C represent (according to the "skeptics") the radical unrealistic option that could only be achieved at the cost of enormous economic pain? I must have missed the news reports about the billions of people starving in the developing world as the green conspiracy to depopulate the Earth reached its final gruesome objective.
Finally, volcanism tends to produce cooling and is innately unpredictable. does anyone know how actual volcanism since 2000 comapres with Hansen's assumptions?
Posted by: Ian Gould | August 31, 2006 10:05 AM
Hi Ian, if you look at the links I posted, you see that actual CO2 emissions and A and B CO2 emissions are so close together there is no difference. C rides with them until 2000.
The methane emissions are a bit more difficult to deal with for reasons made clear here, mostly because systematic measurements were not started in 1958 and there is a significant variation of methane concentrations with latitude due to the fact that OH radicals oxidize it to CO2 in a time comparable to that which it would take to get from where it is emitted to say Antarctica. That being said C is a good fit up to today, but it will overestimate CH4 emissions in the future.
The other major forcing that decreased were the CFCs. Hansen speculated about that a bit in the 1988 article as I recall.
Essenbach does not have a clue
Posted by: Eli Rabet | August 31, 2006 11:15 AM
B and C have big volcanoes in 1995 and 2015, so they've matched up with volcanoes so far.
Posted by: Tim Lambert | August 31, 2006 11:31 AM
Nan, CMAlban -- it's just your eyes.
Teach your children not to believe theirs without knowing the real odds and doing the math.
Let's assume you're not just trying to get the last of my goats, or to yank my chain because you're hoping I'll get flushed, eh? So:
Imagine taking a flash picture while a Roulette wheel is spinning -- can you tell anything from where the little bouncing ball was captured by the flash? Nope, you've got to wait til every single one of the bounces is done and the ball's come to rest to know that one outcome.
Was it a fair spin? Say you wanted to know if the ball and wheel were biased, say a steel ball and a magnet hidden behind one of the numbers -- you'd have to spin many hundreds of times.
(Any statistician able to say how many, for a simple one tail test to detect that any one of a roulette wheel's squares is biased toward winning? -- all I know is it would be a big number.)
What would you know if you stopped before you had enough information to do the math properly? Nothing factual. You'd know what you wanted to believe.
Statistics -- decides on the question, first, then -- takes data -- a huge number of events over a long period of time, and ---after sufficient numbers are collected, calculates whether what appears to be a trend is probably* happening.
Why use statistics? Because people can't do this without doing math.
Posted by: hank | August 31, 2006 2:03 PM
Hank, I reject your analogy and it's a distraction. We have a lot more information than a flash pix of a roulette wheel. So let's talk about Hansen's model.
What I'm saying is quite clear. The best match so far is scenario C. Scenario C has a leveling off of GHG in 2000. That has not happened. To my knowledge CO2 has grown around 2% annually since 2000.
Now the temp is going to have to go up pretty quick in four years to make Scenario B the best match. I'd put my money on that not happening.
Could Hansen's model be wrong? You have to consider the possibility.
I'm still interested in links to other models and their runs and results.
Posted by: C Monte Alban | August 31, 2006 2:38 PM
CMAlban
These model runs were undertaken in 1988. The actual observations since 1988 match up reasonably well with the 'predictions' from two of the scenarios presented. You ask could "Hansen's model be wrong". I ask by what measure do you make such a judgement. Have you compared the actual and the predicted temperature anolmalies?
Considering the sophistication of the climate models used the late 1980s, relative to more recent models. I would suggest the performance of Hansen's model has been remarkable.
Posted by: J Hamilton | August 31, 2006 4:41 PM
Could "Hansen's model be wrong?"
1, neither of the two scenarios that match reasonably well actually happened.....(yes I know reality happened)
2, if you gave a sixth grader the observed data in 1988 and asked them to predict the future global temperatures, they'd probably just do a linear extrapolation. And their model would be better than Hansens.
So what's right about Hansens model?
Considering the sophistication of the climate models used the late 1980s, relative to more recent models. 20 years from now we'll be saying the same thing about the 2006 models. We have to have some way of assessing the usefulness of these models.
Posted by: C Monte Alban | August 31, 2006 5:42 PM
Question:
Why can't Hansen run his model again, this time revising his assumptions with the GHG 'reality mixture?' Scenario D would be very interesting. Wouldn't this test the skillfullness of the model?
Posted by: C Monte Alban | August 31, 2006 6:00 PM
There is no "match so far" there.
Have you the patience to read the article of which this is an excerpt? I think it might help.
"... Even when statistics is correctly applied, the results can be difficult to interpret for a non-expert. For example, the statistical significance of a trend in the data -- which measures the extent to which the trend could be caused by random variation in the sample -- may not agree with one's intuitive sense of its significance. The set of basic statistical skills (and skepticism) needed by people to deal with information in their everyday lives is referred to as statistical literacy."
--- http://en.wikipedia.org/wiki/Statistics
http://julesandjames.blogspot.com/2005/06/betting-summary.html
Posted by: hank | August 31, 2006 6:25 PM
C Monte: The CO2 level in 2000 was 369.47. In 2004 it was 377.43. Would you please justify your rate of 2% annually.
Thanks John
Posted by: John Cross | August 31, 2006 6:53 PM
John:
I meant to say PPM, thanks for noticing. Here are numbers from noaa
year ppm/yr
1959 0.94 1970 1.00 1981 1.44 1992 0.45 1960 0.50 1971 0.78 1982 0.71 1993 1.31 1961 0.96 1972 1.79 1983 2.16 1994 1.89 1962 0.65 1973 1.18 1984 1.35 1995 2.01 1963 0.74 1974 0.76 1985 1.22 1996 1.19 1964 0.30 1975 1.09 1986 1.51 1997 1.98 1965 1.07 1976 0.90 1987 2.35 1998 2.95 1966 1.26 1977 2.07 1988 2.11 1999 0.90 1967 0.68 1978 1.34 1989 1.28 2000 1.78 1968 1.04 1979 1.64 1990 1.31 2001 1.60 1969 1.37 1980 1.84 1991 0.99 2002 2.55 2003 2.31 2004 1.54 2005 2.54Posted by: C Monte Alban | August 31, 2006 7:28 PM
Why do you somehow imagine he can't? Where do you get these odd notions?
Here: http://data.giss.nasa.gov/imbalance/
Posted by: hank | August 31, 2006 8:19 PM
Hank, it was a thought that just popped into my mind. Why wouldn't Hansen include this in post discussions? I don't consider it an odd notion. Anyways, the link you gave points to runs for Hansen's 2005 model? But what I'm interested in is the exact model used in 1988, and it's not clear if that is available. Maybe you can offer additional help with this? And thank you.