Another Assessment of Hansen’s Predictions

The Blackboard has done a bit more work to do a better assessment of Hansen’s forecast to Congress 20 years ago on global warming than I did in this quick and dirty post here.  To give Hansen every possible chance, the author has evaluated Hansen’s forecast against Hansen’s preferred data set, the surface temperature measurements of the Hadley Center and his own GISS  (left to other posts will be irony of a scientist at the Goddard Institute of Space Studies at NASA preferring klunky surface temperature measurements over satellite measurements, but the surface measurements are biased upwards and so give Hansen a better shot at being correct with his catastrophic warming forecasts).

Here is the result of their analysis:

Hansenlineartrend

All three forecasts are high. 

Don’t be too encouraged at Hansen’s prediction power when you observe the yellow line is not too far off. The yellow line represented a case where there was a radical effort to reduce CO2, something we have not seen.  Note that these are not different cases for different climate sensitivities to CO2 — my reading of Hansen’s notes is that these all use the same sensitivity, just with different CO2 production forecast inputs. In fact, based on our actual CO2 ouput in the last 20 years, we should use a case between the orange and the red to evaluate Hansen’s predictive ability.

  • Scientist

    my reading of Hansen’s notes is that these all use the same sensitivity – you display your ignorance of how climate models work here. Climate sensitivity is an output from models, not an input.

    When assessing how well a model predicted subsequent climate trends, it is quite useless to think you can do so without mentioning the errors. GISS quote an error of ±0.1°C considering only spatial incompleteness. The true error will be somewhat larger. The models themselves were just single runs, rather than the more usual thing these days of averaging multiple runs, to smooth out weather noise. Considering errors, it’s clear that temperatures today are consistent with what was predicted 20 years ago.

  • And the lying goes on…….

  • Flashman

    Scientist:
    “you display your ignorance of how climate models work here. Climate sensitivity is an output from models, not an input.!

    I beg to differ.

    “Climate sensitivity” refers to a property of the physical world, a.k.a. as “reality”.

    A property such as this can, at least in theory, be measured by empirical procedures. Whatever the precision of the empirically derived approximation, the value derived may be put to valuable use as _input_ for theoretical constructs, be they formal research, wild speculation at the pub or fully coupled GCM:s. Applying computer _output_ to the physical world in order to define reality is considered futile, unless I’ve missed recent breakthrought in climate science.

    Physical properties of the universe remains, sad to say, _physical_. Sad but true. They do not emerge as output from computers. Some of them may even be politically incorrect.

    You confuse “reality” with “desired reality”. A classic human fallibility from which none of us are immune…

    In the same vein you confuse the issues regarding multiple model runs.

    Running multiple GCM:s and averaging the results smooths out the _model_ noise, as opposed to any noise derived from “weather” or any other aspect of the true evolution of Earth’s climate as manifest in the physical universe.

    Multiple model runs may be useful, but their utility relates to _model_ uncertainty, rather than any need to average out errors in _reality_.

    YMMV. In fact, I’m quite sure you will disagree 🙂

    /Flashman

  • Scientist:

    You’re only half right. Sensitivty is an output, but it is always an output relative to an input. It is sensitivity to *something*. In this case, scenario C was the case where there were actual mitigation efforts ongoing and CO2 had actually gone down (relative to today’s readings).

    So ceteris paribus, if current CO2 levels fall between the red and orange, tempaerature should as well. The question in my mind is whrether other inputs changed between the different runs.

    Has anybody put the actual inputs (e.g. CO2 levels, etc) into the original model to see if it tracks. Reconciling model outputs with the physical reality is the only way to test a model. One would think it would be a pretty simple process (unless of course their feedback values are wrong).

  • Demesure

    “Climate sensitivity is an output from models, not an input.”

    Climate sensitivities given by models are 1.5°C to 4.5°C (IPCC WG1) and this huge range of 1 to 3 hasn’t changed much since the Charney Report 30 years ago which gave the same estimated range.

    This means that 1) climate models haven’t improved and 2) anything can be claimed as “consistent” with their results.

    BTW, saying that models haven’t improved is an understatement. If you compare each scenario’s temperature prediction range between IPCC’s TAR and 4AR, you’ll see the intervals have widened, that means models are getting worse, not better, despite all the IPCC’s claims to the contrary :
    http://skyfal.free.fr/images/TempTAR_4AR.jpg

    If the science is so “settled”, why the need to sweep inconvenient truths under the carpet ?

  • dearieme

    “that means models are getting worse, not better”: I can report that after a career-full of listening to seminars on mathematical modelling, it is an empirical truth that by “better” most mathematical modellers mean “more complicated”, not “in better accord with reality”.

  • it’s clear that temperatures today are consistent with what was predicted 20 years ago.

    Yup, nothing to see here, move along. These are not the ‘droids you are looking for. Who you gonna believe, Scientist or your lyin’ eyes?

  • An Inquirer

    One of the best examinations of Hansen’s 1998 projections is at http://www.climateaudit.org/?p=2621

    Of the numerous weaknesses in Hansen’s work, one of the most frustrating is his lack of proper documentation or clear explanation. For example, when he says that “Scenario B has emission rates approximately fixed at current rates,” he means that the growth in CO2 emissions will be linear. Gavin Schmidt has helped clarify what data sets Hansen actually used, but Steve McIntyre does an excellent job in the above link in examining the role / record of Hansen’s various inputs. Although McIntyre is frustrated with Hansen’s poor documentation record, he is generous to Hansen in the above link.

  • D. K. Wells

    Climatology is not truly a predictive science, it’s actuarial tables for weather.

  • Scientist

    Flashman – you seem to be confused. Clearly, there is a ‘real world’ climate sensitivity. That’s what we’re trying to determine. One way of doing this is by constructing models. You do not ‘input’ a climate sensitivity into a model. You build a physically realistic model, run it, and determine what climate sensitivity the model produces. The author of this blog clearly thinks that the climate sensitivity for a given model is decided in advance. It is not.

    Tony K – So ceteris paribus, if current CO2 levels fall between the red and orange, tempaerature should as well – no, not at all. You seem to be suggesting that no matter what the details of the model, it should have been accurate. The model gave a climate sensitivity of 4.2°C for a doubling of CO2 concentrations. Subsequent results span a wide range but the best estimate is generally taken to be about 3°C for a doubling of CO2 concentrations. So, it should be no surprise if the model slightly over-predicted subsequent temperatures. But when one considers errors, as one should, the observed temperatures and model predictions certainly have error ranges which overlap.

    Demesure – this uncertainty is an inherent property of the climate system. Read Roe and Baker. It’s a rather simple mathematical consequence of the predominance of positive feedback.

  • D. K. Wells
  • Flashman

    Scientist:
    I’m not in the least bit confused. I firmly stand by my conviction that reality trumps models.

    If you admit to the existence of a “real world” climate sensitivity, the issue is resolved.
    But you do seem to get slightly mixed up in the remainder of your post.

    Scientist: “The author of this blog clearly thinks that the climate sensitivity for a given model is decided in advance. It is not.”

    It’s not about the models. It’s about The Real World(TM).

    – The climate sensitivity of The Real World(TM) is most assuredly decided in advance. It is what it is. Full stop. Scientific study means finding out what it is. Full stop again.

    – The climate sensitivity for “a given model” is as irrelevant as the oracle for future climate evolution predicted by the intenstines of a given sacrificial aardvark until _proven_ otherwise.

    The aardvark liver is relevant only when proven to have provided skillful predictions. But it is even then subservient to the reality it purports to mirror.

    I am quite prepared to proudly go to the Hansen Camps unrepentant on this issue.
    If this makes me a “denier” I’ll proudly wear the “D” embroidered on my jacket when the day comes…

  • Scientist

    Oh, you’re confused all right. reality trumps models? So you’re not into this whole idea of trying to understand the real world, then? Models are are a tool for understanding the real world. The climate sensitivity predicted by a model is something which can be compared to observational estimates. It is a number produced by the model, not an input into the model, as the blogger here seems to think.

  • Flashman

    Scientist:
    “Oh, you’re confused all right. reality trumps models?”

    Yes. (Are we in agreement on this issue?)

    “So you’re not into this whole idea of trying to understand the real world, then?”

    Can you please be as kind as to elucidate how this would follow from the previous statement?

  • mccall

    Again, like Scientist’s butchery of Svensmark position and his understanding of cosmic ray hypothesis in: http://www.climate-skeptic.com/2008/06/we-are-so-confi.html
    his post here does not inspire confidence. Best to wait for another perspective, ABS!

  • braddles

    There is an important feature of Hansen projection that is perhaps under-commented. In Hansen et al 1988 his Figure 2 (shown at http://www.climateaudit.org/?p=2611) gives the forcings due to each greenhouse gas. Even under Scenario A, CO2 only warms the globe 0.6 degrees between 1988 and 2050. Most of the 2 degree overall projected warming is in fact due to huge increases in “trace gases”, particularly CFCs and CH4.

    In reality these trace gases have increased little, and have more or less plateaued in the last decade. Today’s models, however, still predict the same disaster scenarios. I don’t know how: perhaps they have ramped up the presumed CO2 forcing to make up for the lost Trace Gas forcing.

  • Christopher Hanley

    To a layman, ‘Scientist’s’ quibbles sound like obfuscation.

    As I understand it, Hansen’s forecasts were included in a presentation to Congress and were intended to convince an audience of non-scientists to follow a course of action , i.e. dramatically cut CO2 emissions.

    Hansen’s unwarranted alarmism then, is clear, and therefore his current prognostications ought to be treated with a ‘pinch of salt’.

  • But when one considers errors, as one should, the observed temperatures and model predictions certainly have error ranges which overlap.

    I’m not sure this is true for any of the scenarios. Hansen didn’t give explicit uncertainty bounds for the data. But using baselines that I think are consistent with those used by Hansen, the Land/Ocean data are for 2006 and 2007 are more than 0.2 C lower than the Hansen A& B projections. (Be is supposed to be the best match.)

    2008 is only half over, so I show lagged averages in pink. The most recent 12 month lagged average is more than 0.2C below Scenario C’s 2007 and 2008 prediction.

    That said, measurement noise doesn’t include weather noise. But, the land/ocean data are currently noticably lower.

    This doesn’t mean the earth might not catch up to C or B. The Hansen runs didn’t include solar variability, and we are near a solar minimum.

    Also, I show Land/Ocean data. But some people think I should show land only data, and that “B” should fall between the Land/Ocean and Land data. The Land only data is higher than the land/ocean. I’ll be adding that in future go arounds. But, I do think the Land/Ocean data makes more sense as a comparison.

    I am planning to figure out whether simple linear extrapolation did better than the models. I suspect it would have, but I’m not sure.

  • Scientist

    ±0.1°C is the quoted uncertainty in GISS data due to incomplete spatial coverage alone. The true error is certainly larger. ±0.2°C would be a reasonable guess. Then, obviously, the model has an uncertainty. Only one run was done to produce each line (computing power then being about 1/1000th of what it is now). These days, you’d do a bunch of runs, and their spread would give you an idea about the range of possibilities. It seems clear that the uncertainties of model and observations overlap.

    Basically, the point to take away is that a simple model gave results that are rather close to what has been observed since. Today’s models are far more sophisticated, but they tell basically the same story. The quantity of CO2 added to the atmosphere by humanity is large enough to affect the climate, and we are observing the effects which were predicted.

  • An Inquirer

    Lucia,
    Although you are undoubtedly the best of judge of how to use your time, I am surprised that you would spend time on this blog. The blog has the most uncivil postings on it, and the quality of thought does not keep up to the standards that I associate with you.
    So what am I doing on this blog? 🙂 Great question!
    (Actually, I check the entries occasionally because the blog sometimes catches issues that are missed on the other two blogs I visit.)

  • davidcobb

    That’s only true if the +.12C per decade adjustment (Wentz 2005?, Hansen 1999) is accurate. Satellite and radiosonde ajustments are not emperically measured. They use deviation from the land temperature record.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Scientist–
    I’m not sure I need to explain why I read any particular blog, but I will. 🙂

    I visit lots and lots of blogs. If a blog links me, I always read that post. Most bloggers do that.

    I’ll probably return to this specific post to see if anyone happens to have questions about my graph.

    I realize that this blog is skeptical of AGW, where as I believe the theory. But, all climate blogs have some overlap, and I think conversation is the best way to discover the areas of agreement, disagreement, and what’s in between.

    I try to read every major climate blog at least once a week. I’ve read posts here, and I will continue to!

    I don’t always comment because it takes time, and sometimes I don’t have anything in particular to add to a conversation. But, when I can add to the conversation, and the blog doesn’t moderate or put hurdles in the path of commenting, I comment.

  • Thank you Lucia for writing up the chart.

  • kuhnkat

    Scientist:

    ” But when one considers errors, as one should, the observed temperatures and model predictions certainly have error ranges which overlap.”

    I’m sorry, but, the expert modellers claim they only have SCENARIOS!!! They NEVER predict. 8>)

    Now, you may ask why that may be. I believe it is because they realise that they are unable to map starting conditions closely enough to do a true prediction. I would add that their physics are a little lacking also.

    If you can’t get your mind around their admissions that they can’t model clouds and precipitation well, you should really give it up!!

  • D. K. Wells

    Hmmm? No comments from Scientist since July 3. Is he on vacation? I doubt we have changed his mind. Could it be that after his rantings about “deniers don’t know how to post HTML links” and I posted some once I knew the correct format that he decided that he lost his last “intellectual superiority” that he could claim? Don’t leave, Scientist, don’t leave. You give us such good practice for preparing our arguments for the people that we need to convince of reality.