| Citizen Runner |
| ||
I stared at that chart for awhile trying to understand what you're seeing and all I got was a headache. I did find this paper which seems to use the same data except it applies a 40 year moving average to each of the data sets. The rational, as I understand it, is this makes sample sets compiled from geographically limited regions to become more representative of global means by virtue of averaging out local/regional effects. In any case, this representation (figure 3) doesn't show any clear trend in that time period. It does show step changes in temperature around 1100, 1350, and 1450 in the pre-instrument era. Without looking into it further, I don't know what specific drivers have been proposed for these events.
Any model is an approximation of the system being modeled. The objective is to model the system with sufficient fidelity that it's behavior captures the essential behaviors of the system being studied. Confidence that the objective has been met in the validation process by running the model against historic data and seeing if it behaves essentially as the system being modeled did. To be fair when you start tweaking parameters that have been constant in your historic data such as greenhouse gas concentrations, there is some chance you've got it wrong, though it helps if the basic physics are well understood. For your specific case of cloud modeling, if there is uncertainty as to how apply that feature to the system, it might be better just to leave it out. That said, per my citation above, this is apparently something they feel they need to get a handle on. |
| rose colored glasses? |
| ||
|
Thanks for the graph BRM. Interesting to see the significant variaility between the different models. If science can't agree very well on past temperature patterns, I'm not sure how they can be expected to do a good job of predicting future patterns. Earlier in the thread somebody made the comment about how bad we are at predicting the weather, and somebody responded that predicting weather and climate are completely different, and we're supposedly much better at predicting climate. That's the part I have great reservation about accepting. I've looked at the climate model backward predictions (i.e. model calibration tests) included in the IPCC report and see very poor performance. They've managed to sort of capture the centory long trend, but have not captured any of the shorter period detail. In order to capture the century long trend, they hae evidently included consideration of many factors other than CO2, including, evidently, volcanic eruptions at the least. Events which had no obvious effect (at least not to the scale modelled) on actual recorded temperaure trends. Yet these are not (at least not obviously) included in forward predictions. Certain physical systems are well behaved and predictable. As an example, we can predict the tides, rotations and revolutions of planets etc with great precision out over extended periods of time, even though thei patterns are governed by a large number of factors. By contrast, other physical systems are more chaotic and unpredictable. Weather is a good example. We can usually predict it with good confidence a day ahead, with some confidence out 3, 7 or maybe even 14 days, but beyond tha (and even beyond 3-4 days in most places really), it's a crap shoot. Whay ca we predict the tides but not the weather? The two systems are governed by a similar range of different factors. But the factors that control weather are mostly poorly behaved themselves, or relatively "random." (Not truly "random" - they are all deterministic processes, but sufficiently varible that they can be described as appearing to be random). Therefore weather is chaotic and unpredictable. When you work with complex systems with many variables, some or all of which are stochastic or apparently random, you are working with deterministic chaos. Given sufficiently accurate knowledge of the input parameters, you should be able to predict the outcome. But behaviour can be wildy different with slight, possibly unmeasurable, differences in the input parameters. So it may not be possible to know the input parameters with sufficient precision to know anything at all about the outcome, even though the process is completely deterministic. There are plenty of good examples of deterministic chaos all around us. Financial markets are a great one. They fluctuate seemingly randomly, even though they should be controlled by a relatively small number of variables that are relatively easy to isolate. And yet predicting the bahaviour of any given financial market is extremely difficult, particularly if you try to predict behaviour very far ahead. The human body itself is another good example of deterministic chaos. It's all biology, chemistry and basic physics, we should understand it completely, no? And yet, if we did, then surely by now we'd have cancer licked. And mental health too. Yet we live in the dark ages still as far as treaing mental illness. I write this all out simply to give more perspective about my position. I have very little doubt that human activity has had some impact on climate, as it has had obvious impact on every other aspect of the physical environment. But I don't believe that it's possible for us to really know whether such impacts are significant or not. Because climate is chaotic and unpredictable, not to mention unmeasured over much of the globe for most of history, and entirely unmeasured everywhere for most of the earth's history. |
| Bobby1 |
| ||
|
My refrigerator makes all the ice that I need. |
| Citizen Runner |
| ||
Just noticed that I failed to provide a link to the paper I was looking at showing a smoothed version and more context for BRM's figure. http://www.pnas.org/content/early/2008/09/02/0805721105.full.pdf |
| rose colored glasses? |
| ||
|
Thanks kindly for the paper CR. Here is a look at the models of past temperature: Image: http://i44.tinypic.com/15wj5i1.jpg I think people should study that for a bit and dwell on it. I've read a little bit of the background information attached to the paper. These estimates are based on a large number of "proxy indicators" including tree ring data and numerous other bits of evidence. The first thing I'd like to point out is that while each individual model represents its own inherent uncertainty (as a broad colour band adjacent to their best fit curve), these are CLEARLY all grossly underrepresentative of actual uncertainty, given that none of the other models, presumably each developed by competent scientists, agree with each other. Now let's look a little closer at some specific aspects of the curves: Image: http://i42.tinypic.com/344tixs.jpg First look at the right side of the bottom graph. They are showing an overall measured temperature increase of about 1.3 degrees C since 1900. Earlier work we've looked at from the IPCC report shows about 0.5 degree increase. Is this paper using a different data source, or is there lack of agreement in the literature about actual recorded temperatures? Or am I maybe missing something? This is a VERY significant difference (more than a factor of 2), leading to potentially severe errors in interpretation. The second thing I'd like to emphasize from this second image is in the upper graph. Look at the huge range in modelledtemperatures around 800 AD. We have either increased global temparatures by 0.5C or 2C since 800 AD (if we are meant to believe the instrumented record, which I've just pointed out is suspect). That's a difference of a factor of 4. A factor of FOUR!!!! These graphs represent the "data" against which our forward modelling are based. If our backward modelling, based on real scientific data, gives us uncertainty measured in ORDERS OF MAGNITUDE, then how culd we possibly trust models attempting to predict furture climate? We can't in my opinion. Coincidentally, I was reading Maclean's magazine today (Canada's most popular English language national news magazine), and there' a lengthy article on cimate, related to the general absence of normal winter in Canada (ad extremely severe winter in Europe and Asia). In the article, it is noted tha "before winter had begun, weather experts were warning Canadias to brace for an exceptionally cold and snowy winter," and yet we've had one of the warmest and least snow on record. The "experts" got it COMPLETELY wrong. |
| gis |
| ||
|
Great discussion. Keep it going! |
| Should be noted |
| ||
You were doing better. Did you get too tired to even bother trying to hide your agenda? Come on, back to the "Oh, I am just someone with an interest in developing my understanding" act. |
| Roses, from the phone |
| ||
|
Your lame schtick is getting pretty old, pal |
| Should be noted |
| ||
Sorry if I insult your professionalism on the job. Carry on. Just try a little more subtlety, that's all. |
| Roses, from the phone |
| ||
|
I suppose I should be flattered. I must be making some good points? Seriously, who would pay to have somebody comment on a let's run thread... You really think there could be money in that? I will admit your badgering adds another dimension to the interesting social side of the discussion. You've got me a little bit intrigued |
| photofinish |
| ||
|
we are on the 12th page of a troll thread about f***ing Greenland. You dumba$$es. |
| Blowing.Rock Master |
| ||
|
Don't know if you've seen this yet. It's the Lower Tropospheric temperature data since 1979, when the satellite was launched. As far as I'm aware it's the only truly global temperature dataset. http://www.drroyspencer.com/wp-content/uploads/UAH_LT_1979_thru_February_2012.png It's updated monthly by Dr. Roy Spencer at the University of Alabama Huntsville. He also has a blog where he discusses many issues related to GW. You can read comments after each blog post where some very lively debates develop. |
| Citizen Runner |
| ||
I believe you're using a different definition of uncertainty than they are. My assumption is that they are giving a measure of the statistics of the data and the way it is combined without any a priori knowledge of "truth", that is, it is a measurement of residuals from the curve fit. Data can be biased (reasons for this are discussed in the text) without being "noisy". You're looking for an assessment of how far off of "truth" is their estimate.
The red line is instrument Northern Hemisphere Land-Only average temperature as discussed in the text. They include that because the reconstructions in that figure are NH. The grey line is the Global Average and is, I believe, the same as used in the IPCC report.
Several points: a) The disconnect around 900 between the CPS and EIV results is discussed in the text, comparison of the two reconstruction approaches being the gist of the article. They note that the CPS approach seems particularly sensitive to a particular subset of the observation data. In any case they note that there are increasing uncertainties in the reconstruction that generally increase the further back one tries to look. b) We have either increased global temperatures by 0.5C or 2C since 800 AD - Maybe this isn't what you meant, but there is nothing in AGW theory that claims "we", as in the human race, had any significant impact on global climate until the last few hundred years. Nor is there anything that denies natural variations. c) While it's possible that the people doing climate modeling gravitate toward validation using reconstructed data points with significant uncertainty, if it were me, I'd lean toward using data from the instrumented history or more recent reconstructions where more comprehensive data is available if my interest is in estimating relative near term behaviors.
That's weather, not climate, but this is the first time I've seen unseasonably warm weather in some region spun as an argument against AGW. Well done. |
| rose colored glasses? |
| ||
Thanks CR. Some food for thought there. A few comments in reply" (1) I'll not go off into an essay on uncrtainty, but this is an area I will claim some expertise in. The take home point from the graph, to me, is tha the fact that several different presumably competent scientists working in the field of climate studies have come up with wildly different conclusions from the same data, showing that actual temperatures in the past are not known with any particular precision. (2) OK thanks for the clarification, I should try to find time to read the whole text. But is it really appropriate to show that red curve on this plot, which shows past everage global temperature? I think it comes off as maybe a bit alarmist to show the worst recent trends against the average long term trends. (3) a) Agreed, hence my point (1) above. (3) b) I shouldn't have written "we" I just meant temperatures have either increased by 0.5 or 2 C. Poor choice of words notwithstanding, the point remains. (4) I'll admit it seems odd, and I'm not using the point against AGW specifically, I'm using it to suggest we have absolutely no ability to predict XXXXXX. Here, XXXXXX means some intermediary entity between "weather" and "climate," as the point was about prior predictions of winter conditions (not "weather" per se, which in my mind connotes conditions a few das ahead). Can you make a good argument why we should expect climate preditions to be more accurate than weather predicitions? The claim is widely made, and I frankly find it preposterous, but perhaps you have a good reason for believing it you can share? |
| Citizen Runner |
| ||
|
The data reconstruction in that graph is for the Northern Hemisphere only. They did this, as they note in the article, because there is significantly more proxy data available for the NH and most of this data is from land not ocean proxies. The "alarmist" red line is, as I noted above, the instrument data for NH land average and is there as an apples to apples comparison with with reconstructed land-only lines (orange and green). It is a fact that proxy data reconstructions have non-trivial levels of uncertainty. That said, the conclusion reached by the article, that 1) the current NH warmth "is likely anomalous" going back at least 1300 years and perhaps longer with strong caveats is supported and 2) we need more and better data for the Southern Hemisphere seem reasonable to me. Keep in mind that I pointed to this article in response to BRM's statement that the current warming trend predates the rise in atmospheric CO2. This data doesn't seem to support that claim. The difference between weather modeling and climate modeling is that in climate modeling you get to claim success or not on large scale averages in climate modeling whereas in weather forecasting you're going to get called out on missing any point forecast. If Canada's winter is warmer than expected, but Europe is comparably cooler then globally it might be a wash. That's not to say that getting the regional effects right isn't significant in climate modeling aren't issues. It's acknowledged that the ocean currents deemed responsible for this stuff are lacking in climate modeling. This certainly is a factor in determining the magnitude of the AGW problem. At the most fundamental level, the cause effect relationship for AGW is well understood, rising greenhouse gas levels cause retention of energy at the surface/lower atmosphere. There is significant observational evidence that indicates this effect is taking place There is, in my opinion, uncertainty as to the magnitude due to the uncertainty in feedback mechanisms, but the climate models capture the best available understanding. As an analogy, even though complaints about our ability to forecase weather is criticised by almost everyone, in its current imperfect state it still has great utility. |
| rose colored glasses? |
| ||
|
Thanks for another thoughtful reply CR. Quick points (won't comment much today): (1) Yes, I think that's an important point. Anomolous at least 1300 years. Why? (2) The reconstructions clearly show temperatures increasing since about 800 AD. That certainly predates the inferred rise in atmospheric CO2. Unless we like the other reconstructions showing much higher temperatures in 800 AD. But if we accept those, then temperatures today aren't significantly higher than "observed" in 800 AD. |
| gis |
| ||
|
lol. amazing, isn't it? |
| Citizen Runner |
| ||
|
1) Read the paper. 2) Read BRM's post. |
| rose colored glasses? |
| ||
That's a pretty lazy answer. If you know why, do us a favour and elaborate. If you don't know, that's fine. |
| rose colored glasses? |
| ||
|
Just a point of clarification. I misread the point about anomalous temperatures. Their point is that THE LAST DECADE is anomalous within the context of the last 1000-1300 years. I read that the last 1000-1300 years were anomalous within a longer context. My mistake, sorry for the distraction. Still, when I look at these reconstructions, I see that either the current decade isn't particularly anomalous in consideration of the uncertainties in the reconstructions, OR temperatures have been increasing since long before human-induced atmospheric CO2, depending on whether I like their lower or upper reconstruction. I will admit, though, that if the data are RIGHT, the last 10 years in their curve does seem to be relatively anomalous. I'm just not really convinced, though, that we can make direct comparisons between recorded temprature data and "reconstructed" temperature trends and draw meaningful, defensible conclusions. |