|
Post by socold on Jun 24, 2010 20:56:42 GMT
What I am saying is that the model has an equillibrium point N, a "local attractor" if you will. The problem is what is N. No need to give up, if you don't know what N is you can start the model with initial conditions X, leave the model running and it will attract towards the equillibrium point N. Then you can just read what N is from the output.
Then that you know N you can start the model run again but this time set the initial state to N, therefore you start the model at equillibrium. Any change you make to a forcing then can be measured in relation to how far that forcing moves the modelled climate away from N.
|
|
|
Post by nautonnier on Jun 25, 2010 2:12:23 GMT
What I am saying is that the model has an equillibrium point N, a "local attractor" if you will. The problem is what is N. No need to give up, if you don't know what N is you can start the model with initial conditions X, leave the model running and it will attract towards the equillibrium point N. Then you can just read what N is from the output. Then that you know N you can start the model run again but this time set the initial state to N, therefore you start the model at equillibrium. Any change you make to a forcing then can be measured in relation to how far that forcing moves the modelled climate away from N. "A" local attractor? The chaotic output from the sun heats the atmosphere which is chaotic and made up of multiple chaotic weather flows and precipitation that affect the Oceans which are a set of chaotic fluid flows, salinity changes and heat transports that affect the atmosphere which is chaotic made up of multiple chaotic weather flows and precipitation.......a continual nested interreacting iteration. Each of these chaotic interreacting subsystems with their own differential rotations and angular momentum may have several attractors that they flip between - which we see in our extremely brief period of observation as cycles such as MJO, PDO, AMO, MOC etc., some of these flips we see daily - sea breezes and convection modified by week long chaotic fluid circulation of weather patterns modified by monthly and annual changes.... and so on till after several millenia we are at ice age level attractors to which the entire climate can veer within a decade and which may be caused or influenced by the chaotic orbital behavior of the solar system..... Yet you claim that you can create a deterministic model with a few of the known variables and the rest as simple parameters based on some decades of observations - and that we should trust you - the model will come out just like the climate after a few iterations? I think that while there is a nice linear trend in some of the major cycles then your deterministic model may appear close (after much fiddling with parameters) leading to the current premature claims of success. However, once these cycles change perhaps interreacting in a way that may not have been seen for a millennium, the models will get further away from the truth possibly quite abruptly - in a way that will make 'error bars' superfluous. This is just what has been seen with the models of the Sun's behavior which their authors were using to confidently predict that SC24 'was going to be a doozy' - the models are wrong and its been a dozy. As we are talking climate it may take a decade or so - but the climatologists will almost certainly be in the same position as the solar physicists trying to rework the models to see where they went wrong. Unfortunately, the politicians are still acting on the simple climate model forecasts of tipping points, 6C increases and sea level rises of tens of meters - as it gives them power, taxes and kickbacks by salami slicing carbon credits. If, as implied by the author of this thread, basic flaws are found in the models' formulae, algorithms and assumptions, the climatologists involved may find that they are somewhat less than popular with the world's population.
|
|
|
Post by northsphinx on Jun 25, 2010 13:59:46 GMT
Socold wrote; "leave the model running and it will attract towards the equillibrium point N. Then you can just read what N is from the output." But dont You think that our real world with many million years of interaction should reached this equillibrium point N a long time ago? Maybe because of there is no point N. In the real world.
|
|
|
Post by socold on Jun 25, 2010 18:28:28 GMT
Here's a graph of a control run. You can see the initial temperature state of the model was started higher than the equilibrium value, and as a result the climate shifted to equilibrium. In this control run, the model boundary (input) forcings are held constant at values for the year 1870 and the model is integrated forward in time for 500 years. During each year, the model is forced with the same year 1870 values. The natural variability of the simulated climate system will results in a non-constant, time-varying, surface temperature. After the initial adjustment period, the time-varying surface temperature of a balanced model will stay within a control climate envelope of the model denoted by the red and blue lines in Figure 1.www.gisclimatechange.org/runSetsHelp.htmlHere's another page explaining it: Control runs"Control runs establish the basic climate of the model. Control runs are long integrations where the model input forcings (solar irradiance, sulfates, ozone, greenhouse gases) are held constant and are not allowed to evolve with time. Usually the input forcings are held fixed either at present day values (i.e., for year 2000 or 2000 Control Run) or a pre-industrial values (i.e., for 1870 or 1870 Control Run). Note that in this context, "fixed" can have two different meanings. The solar forcing values are held fixed a constant, non varying number. The sulfate, ozone and greenhouse gases values, however, are fixed to continually cycle over the same 12-month input dataset every year. The CCSM is then run for an extended period of model time, 100s of years, up to about 1000 years, until the system is close to equilibrium (i.e., with only minor drifts in deep ocean temperature, surface temperature, top-of-the-atmosphere fluxes, etc)." With an animation of the equilibrium result obtained from one climate model: www.cccma.ec.gc.ca/diagnostics/cgcm1/animation2.shtml These results were obtained from a multi-century control experiment which was performed to simulate the natural variability of the coupled system. The model reproduces the observed annual the temperature cycle well. Note that the amplitude of the annual cycle is much larger over the poles and extra-tropical land areas than over the oceans. You might also notice dfference in the shape of the annual cycle which reflects the different thermal properties of the land, ice and water covered parts of the Earth's surface.
|
|
|
Post by sigurdur on Jun 25, 2010 20:32:37 GMT
Socold: The model is run in a slab type state and is suppose to prove something?
|
|
|
Post by socold on Jun 25, 2010 21:07:27 GMT
Not sure what a slab type state is.
|
|
|
Post by nautonnier on Jun 25, 2010 23:55:46 GMT
Here's a graph of a control run. You can see the initial temperature state of the model was started higher than the equilibrium value, and as a result the climate shifted to equilibrium. In this control run, the model boundary (input) forcings are held constant at values for the year 1870 and the model is integrated forward in time for 500 years. During each year, the model is forced with the same year 1870 values. The natural variability of the simulated climate system will results in a non-constant, time-varying, surface temperature. After the initial adjustment period, the time-varying surface temperature of a balanced model will stay within a control climate envelope of the model denoted by the red and blue lines in Figure 1.www.gisclimatechange.org/runSetsHelp.htmlHere's another page explaining it: Control runs"Control runs establish the basic climate of the model. Control runs are long integrations where the model input forcings (solar irradiance, sulfates, ozone, greenhouse gases) are held constant and are not allowed to evolve with time. Usually the input forcings are held fixed either at present day values (i.e., for year 2000 or 2000 Control Run) or a pre-industrial values (i.e., for 1870 or 1870 Control Run). Note that in this context, "fixed" can have two different meanings. The solar forcing values are held fixed a constant, non varying number. The sulfate, ozone and greenhouse gases values, however, are fixed to continually cycle over the same 12-month input dataset every year. The CCSM is then run for an extended period of model time, 100s of years, up to about 1000 years, until the system is close to equilibrium (i.e., with only minor drifts in deep ocean temperature, surface temperature, top-of-the-atmosphere fluxes, etc)." With an animation of the equilibrium result obtained from one climate model: www.cccma.ec.gc.ca/diagnostics/cgcm1/animation2.shtml These results were obtained from a multi-century control experiment which was performed to simulate the natural variability of the coupled system. The model reproduces the observed annual the temperature cycle well. Note that the amplitude of the annual cycle is much larger over the poles and extra-tropical land areas than over the oceans. You might also notice dfference in the shape of the annual cycle which reflects the different thermal properties of the land, ice and water covered parts of the Earth's surface.Excellent SoCold - a really precise model of the climate - I presume then that it clearly shows the Roman Optimum, the Medieval Warm Period and the the Little Ice Age extremely clearly - so it will be possible to see their effects? No? Whyever would that be?
|
|
|
Post by sigurdur on Jun 26, 2010 1:24:45 GMT
Not sure what a slab type state is. A slab type state is when the perimiters of h20 etc are held constant. In your example I didn't see any variability in clouds etc.
|
|
|
Post by socold on Jun 26, 2010 2:07:28 GMT
Both of you are making arguments that don't make any sense. I'll leave it at that.
|
|
|
Post by nautonnier on Jun 26, 2010 2:28:07 GMT
Both of you are making arguments that don't make any sense. I'll leave it at that. I shall take it then that your model produces a world theme that follows the progress of winter and summer - but does not appear to show the REAL variations in climate that are known like the LIA or the MWP. No ENSO or AO - just a nice steady as you go clockwork model. So - as Stranger has said - this is as realistic and useful as a model train set through a model town.
|
|
|
Post by icefisher on Jun 26, 2010 4:00:28 GMT
Here's another page explaining it: Control runs "Control runs establish the basic climate of the model. Control runs are long integrations where the model input forcings (solar irradiance, sulfates, ozone, greenhouse gases) are held constant and are not allowed to evolve with time. Usually the input forcings are held fixed either at present day values (i.e., for year 2000 or 2000 Control Run) or a pre-industrial values (i.e., for 1870 or 1870 Control Run). Note that in this context, "fixed" can have two different meanings. The solar forcing values are held fixed a constant, non varying number. The sulfate, ozone and greenhouse gases values, however, are fixed to continually cycle over the same 12-month input dataset every year. The CCSM is then run for an extended period of model time, 100s of years, up to about 1000 years, until the system is close to equilibrium (i.e., with only minor drifts in deep ocean temperature, surface temperature, top-of-the-atmosphere fluxes, etc)."
Another way of explaining a control run is it is a logic check of the model code. If parameters are set to a given value (for example the estimates of "forcing" in 1870) running the model should perform as designed and return the programmed equilibrium point with some allowable degree of variance to mimic unexplained natural variations). If it doesn't that means you have some coding errors that need correcting. Of course it is the developer who decides what the values are for the different forcings and adjusts his code so that the model returns the neutral value (say temperatures for 1870) when ever the model is run with those parameters. If it doesn't he goes in and fixes the code until it does. Of course always honoring whatever he believes the relative contribution to climate change is provided by each variable. Like an interest rate model. If you have an investment that is supposed to return 9% per annum, you adjust your code until the model consistently returns 9% per annum. One control run doesn't suffice for a model though because it tells you nothing about sensitivity. So you change the values to say the year 2000 and make another control run to ensure here it also returns the correct value. Here you can simply put in the necessary factor for sensitivity to produce the differing results. Now you are ready. Of course climate models are famous for not replicating history well (they run nicely through the 2 control points but fail to show consistency in replicating other years accurately). And of course it fails to predict the future because quite simply it is a travesty that they have no clue about what is really going on.
|
|
|
Post by socold on Jun 26, 2010 12:54:42 GMT
Both of you are making arguments that don't make any sense. I'll leave it at that. I shall take it then that your model produces a world theme that follows the progress of winter and summer - but does not appear to show the REAL variations in climate that are known like the LIA or the MWP. No ENSO or AO - just a nice steady as you go clockwork model. So - as Stranger has said - this is as realistic and useful as a model train set through a model town. I don't buy your reasoning. How can you conclude anything about LIA/MWP response from a single looped animation of an annual cycle?
|
|
|
Post by socold on Jun 26, 2010 12:57:53 GMT
Not sure what a slab type state is. A slab type state is when the perimiters of h20 etc are held constant. In your example I didn't see any variability in clouds etc. I don't think h20 is held constant in the model. Nor do I see any evidence in the animation for the presense or absense of clouds. I would think clouds are indeed included in the model, as they are in all GCMs.
|
|
|
Post by socold on Jun 26, 2010 13:07:24 GMT
Of course climate models are famous for not replicating history well (they run nicely through the 2 control points but fail to show consistency in replicating other years accurately). They do a lot better than 2 points over the 20th century. 14 models, 58 runs, red line is mean of model runs, black line is observed surface temperature: Just because there is difficulty measuring the energy budget precisely on short timescales doesn't mean the effects of observed forcings cannot be predicted.
|
|
|
Post by magellan on Jun 26, 2010 16:18:13 GMT
Of course climate models are famous for not replicating history well (they run nicely through the 2 control points but fail to show consistency in replicating other years accurately). They do a lot better than 2 points over the 20th century. 14 models, 58 runs, red line is mean of model runs, black line is observed surface temperature: Just because there is difficulty measuring the energy budget precisely on short timescales doesn't mean the effects of observed forcings cannot be predicted. Let's take this one step at a time. What TSI data is used in the models you cite? Waiting....... Edit: removed hint
|
|