4 ms·

> The models don't tell us why the planet was warm and then it was in an ice age, they don't tell us why it stopped being in an ice age in the first place, they

> The models don't tell us why the planet was warm and then it was in an ice age, they don't tell us why it stopped being in an ice age in the first place, they don't tell us anything how our actions (or inaction) could cause the climate to change in a way that we wish it to change. This criticism is somewhat akin to trying to challenge the theory of relativity because our orbital models can't tell us exactly how the planets got arranged how they did. The reason they can't isn't because the models are broken; it's because there are many plausible hypotheses that could give rise to the outcome we see, and representing various possible combinations of historical parameters. And we can't necessarily know which set of parameters without the aid of a time machine. Similarly for the climate models. They can't tell us exactly how every historical change in the climate happened because there are many climate forcings that simply aren't recorded very well (or at all) in the historical record. So we can use the models to come up with various plausible scenarios, but we'd need a time machine to figure out which one is the correct one. To take another analogy, this criticism's akin to when creationists complain of the relative fuzziness of our knowledge of the history of life in the precambrian to argue that evolution cannot account for the major taxonomic divisions in the tree of life. Again, the truth is that the model is fine, what's inadequate is our knowledge of the historical conditions we would need to feed into that model. The distinction between those two sorts of knowledge gap may seem pedantic and hair-splitty, but an understanding of the nature and implications of that difference is one of the boundaries that separates science from pseudoscience.

steven2012 12y ago
The problem is that the models have all been wrong. At what point do you give up and say "the models don't work no matter what we do, maybe the theory is flawed?" They can't explain the past behavior of the Earth's climate, and they can't predict the future behavior. They're not even directionally correct, given the fact that global temperatures, albeit amongst the highest they've been on record, haven't increased since the late 1990s[1]. Given the incredible increase in CO2 over the past 16 years, the models would have predicted much higher temperatures, but it's simply not agreeing with their own data. 1. http://www.politifact.com/rhode-island/statements/2013/aug/25/steve-goreham/global-warming-skeptic-says-global-surface-tempera/ http://www.politifact.com/rhode-island/statements/2013/aug/2...
- rbobby 12y ago
  Seems like you're pining a lot on a model either being 100% correct or 100% incorrect. That's not really how models, especially statistical ones, work.
  - steven2012 12y ago
    That's not what I said, I never said I expected the models to be 100% correct at all. I said that the models aren't even directionally correct, considering the fact that the global temperatures haven't increased given the surge of CO2.
- Intermernet 12y ago
  Have a look at http://www.skepticalscience.com/climate-models.htm http://www.skepticalscience.com/climate-models.htm (there are many other links available, but that one has a nice graph of the data from "The Copenhagen Diagnosis" [1]). The correlation between the recorded satellite data and the previous 9 years of IPCC predicted data shows that the models are, if anything, conservative. There's more up to date data in the IPCC 2014 report [2] as far as I know. EDIT: Please note that IPCC web resources are terrible for searching. They have linked powerpoint documents and PDFs, but it's very hard to just search the data from their front end. Does anyone know of a nice HTML version of all of the IPCC resources? [1] http://www.copenhagendiagnosis.com/ http://www.copenhagendiagnosis.com/ [2] https://www.ipcc.ch/report/ar5/wg2/ https://www.ipcc.ch/report/ar5/wg2/
- anon1385 12y ago
  I'm afraid you have been led astray by people who were deliberately trying to misinform you. You should probably be angry about that. >The problem is that the models have all been wrong. All models are wrong, but some are useful. Climate models have demonstrated skill at modelling important parts of the climate system. >They can't explain the past behavior of the Earth's climate, and they can't predict the future behavior. They're not even directionally correct, given the fact that global temperatures, albeit amongst the highest they've been on record, haven't increased since the late 1990s[1]. Given the incredible increase in CO2 over the past 16 years, the models would have predicted much higher temperatures, but it's simply not agreeing with their own data. That is just factually incorrect. There are individual model runs that do match the observed surface temperature since the 90s. The recorded temperatures are still within the expected range of most model projections. Even Hansen's crude 1988 model is surprisingly close to what has happened[1], although that is probably in part down to luck. The most important thing to know here is that climate models are projections, not predictions. That means they aim to model what happens given a certain set of circumstances. They don't aim to predict what the actual circumstances will be. Those circumstances include things like volcanic activity, which you can appreciate is not something that is easy to predict decades in advance. Also internal variability like ENSO, PDO, NOA , and NAO will play a large part in determining the short term (decadal timescale) surface temperature. As far as I understand most models just model ENSO stochastically, so for any given model run it's going to be down to random chance whether the ENSO state in the model run matches what happened in reality. Over longer periods of time and over many models runs the result is a range of likely climate (for various different emissions scenarios). The aim of climate models is to give an upper and lower bound to what the system will probably do for a given emissions scenario, not to provide a single forecast for the exact state of the climate in x years. Consider writing a model of a die roll. If you know the die is fair you will be able to write a simple model that gives close results over a large number of throws, but it will have basically zero ability to predict a small number of throws. You could also adjust your model considering real data if it turns out that the die is weighted based on your observations. This isn't a direct analogy to a climate model, but it's a demonstration that models can be useful without the ability to predict the short term state of the system. Having said all that, there are some areas where the models do pretty badly. Arctic sea ice is one, but in that case the problem has been a big underestimation of the actual rate of melting. Finally, I'm not sure you read that link you posted because it discuses why 'no warming in 16 years' is a cherry picked figure[2]. It's a bit of a statistics game as well. If your null hypothesis is that the trend has continued unchanged then the data don't give any reason to think that the trend has changed[3]. Of course it's important to remember that surface temperature is only showing a small part of the total energy in the climate system, and our surface temperature record has gaps (it misses out most of the poles). Short term variability in the trend is to be expected; you can also cherry pick time periods when the short term trend was much greater, for example 1992-2006 [4]. [1] http://tamino.wordpress.com/2014/03/21/hansens-1988-predictions/ http://tamino.wordpress.com/2014/03/21/hansens-1988-predicti... this also shows some of the techniques that have probably been used to trick you into thinking models have been wildly wrong, which is handy if you want to know what to look for in the future [2] http://tamino.wordpress.com/2014/02/09/cherry-p/ http://tamino.wordpress.com/2014/02/09/cherry-p/ some relevant stats on cherry picking [3] http://tamino.wordpress.com/2013/02/12/2012-updates-to-trend-observation-comparisons/ http://tamino.wordpress.com/2013/02/12/2012-updates-to-trend... for a look at how recent years fit into the longer term trend. It uses 2000 as the cutoff - somebody redid it using 1998 but I can't seem to find that now, the story wasn't much different though [4] http://tamino.wordpress.com/2013/09/21/double-standard/ http://tamino.wordpress.com/2013/09/21/double-standard/

steven2012 12y ago

The problem is that the models have all been wrong. At what point do you give up and say "the models don't work no matter what we do, maybe the theory is flawed?" They can't explain the past behavior of the Earth's climate, and they can't predict the future behavior. They're not even directionally correct, given the fact that global temperatures, albeit amongst the highest they've been on record, haven't increased since the late 1990s[1]. Given the incredible increase in CO2 over the past 16 years, the models would have predicted much higher temperatures, but it's simply not agreeing with their own data. 1. http://www.politifact.com/rhode-island/statements/2013/aug/25/steve-goreham/global-warming-skeptic-says-global-surface-tempera/ http://www.politifact.com/rhode-island/statements/2013/aug/2...

rbobby 12y ago
Seems like you're pining a lot on a model either being 100% correct or 100% incorrect. That's not really how models, especially statistical ones, work.
- steven2012 12y ago
  That's not what I said, I never said I expected the models to be 100% correct at all. I said that the models aren't even directionally correct, considering the fact that the global temperatures haven't increased given the surge of CO2.
Intermernet 12y ago
Have a look at http://www.skepticalscience.com/climate-models.htm http://www.skepticalscience.com/climate-models.htm (there are many other links available, but that one has a nice graph of the data from "The Copenhagen Diagnosis" [1]). The correlation between the recorded satellite data and the previous 9 years of IPCC predicted data shows that the models are, if anything, conservative. There's more up to date data in the IPCC 2014 report [2] as far as I know. EDIT: Please note that IPCC web resources are terrible for searching. They have linked powerpoint documents and PDFs, but it's very hard to just search the data from their front end. Does anyone know of a nice HTML version of all of the IPCC resources? [1] http://www.copenhagendiagnosis.com/ http://www.copenhagendiagnosis.com/ [2] https://www.ipcc.ch/report/ar5/wg2/ https://www.ipcc.ch/report/ar5/wg2/
anon1385 12y ago
I'm afraid you have been led astray by people who were deliberately trying to misinform you. You should probably be angry about that. >The problem is that the models have all been wrong. All models are wrong, but some are useful. Climate models have demonstrated skill at modelling important parts of the climate system. >They can't explain the past behavior of the Earth's climate, and they can't predict the future behavior. They're not even directionally correct, given the fact that global temperatures, albeit amongst the highest they've been on record, haven't increased since the late 1990s[1]. Given the incredible increase in CO2 over the past 16 years, the models would have predicted much higher temperatures, but it's simply not agreeing with their own data. That is just factually incorrect. There are individual model runs that do match the observed surface temperature since the 90s. The recorded temperatures are still within the expected range of most model projections. Even Hansen's crude 1988 model is surprisingly close to what has happened[1], although that is probably in part down to luck. The most important thing to know here is that climate models are projections, not predictions. That means they aim to model what happens given a certain set of circumstances. They don't aim to predict what the actual circumstances will be. Those circumstances include things like volcanic activity, which you can appreciate is not something that is easy to predict decades in advance. Also internal variability like ENSO, PDO, NOA , and NAO will play a large part in determining the short term (decadal timescale) surface temperature. As far as I understand most models just model ENSO stochastically, so for any given model run it's going to be down to random chance whether the ENSO state in the model run matches what happened in reality. Over longer periods of time and over many models runs the result is a range of likely climate (for various different emissions scenarios). The aim of climate models is to give an upper and lower bound to what the system will probably do for a given emissions scenario, not to provide a single forecast for the exact state of the climate in x years. Consider writing a model of a die roll. If you know the die is fair you will be able to write a simple model that gives close results over a large number of throws, but it will have basically zero ability to predict a small number of throws. You could also adjust your model considering real data if it turns out that the die is weighted based on your observations. This isn't a direct analogy to a climate model, but it's a demonstration that models can be useful without the ability to predict the short term state of the system. Having said all that, there are some areas where the models do pretty badly. Arctic sea ice is one, but in that case the problem has been a big underestimation of the actual rate of melting. Finally, I'm not sure you read that link you posted because it discuses why 'no warming in 16 years' is a cherry picked figure[2]. It's a bit of a statistics game as well. If your null hypothesis is that the trend has continued unchanged then the data don't give any reason to think that the trend has changed[3]. Of course it's important to remember that surface temperature is only showing a small part of the total energy in the climate system, and our surface temperature record has gaps (it misses out most of the poles). Short term variability in the trend is to be expected; you can also cherry pick time periods when the short term trend was much greater, for example 1992-2006 [4]. [1] http://tamino.wordpress.com/2014/03/21/hansens-1988-predictions/ http://tamino.wordpress.com/2014/03/21/hansens-1988-predicti... this also shows some of the techniques that have probably been used to trick you into thinking models have been wildly wrong, which is handy if you want to know what to look for in the future [2] http://tamino.wordpress.com/2014/02/09/cherry-p/ http://tamino.wordpress.com/2014/02/09/cherry-p/ some relevant stats on cherry picking [3] http://tamino.wordpress.com/2013/02/12/2012-updates-to-trend-observation-comparisons/ http://tamino.wordpress.com/2013/02/12/2012-updates-to-trend... for a look at how recent years fit into the longer term trend. It uses 2000 as the cutoff - somebody redid it using 1998 but I can't seem to find that now, the story wasn't much different though [4] http://tamino.wordpress.com/2013/09/21/double-standard/ http://tamino.wordpress.com/2013/09/21/double-standard/