Given that EVERY "climate model" used by the AGW people is really just weather software bastardized to make long term climate predictions, when the weather software was originally designed by its makers to, at best, provide a greater than 50% accuracy over a 5 day period, I would much rather trust a weatherman than someone who has spent their time on the Rainbow Warrior smoking bongs and chasing whaling boats.Maui wrote:IMHO anyone that claims Global Warming is "greatest hoax ever perpetrated on the American people" should not be accepted as a trusted source of dispassionate information. This year he's fished 237 more "scientists" out that are willing to call themselves skeptics. But this list builds off of last years 413 that included:
84 that are connected to fossile fuel industries
49 that are retired
44 that are television weathermen
Global Warming Concensus Broken
Re: Global Warming Concensus Broken
Whaa?!
Realclimate is complaining about someone cooking a graph.Here is a great but highly polemic post rom realclimate on the details of how Roy got his PDO data that matches so precisely. He selected to indices, and linearly combined them, and then added a big damping factor.
http://www.realclimate.org/index.php/ar ... y-lessons/
You know where the hockeystick came from, right?
Yours, TDP, ml, msl, & pfpp
molon labe
montani semper liberi
para fides paternae patria
montani semper liberi
para fides paternae patria
TDPerk,
It is always the detail that matters. So the answer to your question is no: what graph (in detail) was cooked, and how, and by whom, and when, with refs if possible?
I find the GW debate fascinating biut mainly because there are so many slurs (on both sides), and working out the real and complex situation is not easy.
But then that is usually the case, in any scientific field.
Tom
It is always the detail that matters. So the answer to your question is no: what graph (in detail) was cooked, and how, and by whom, and when, with refs if possible?
I find the GW debate fascinating biut mainly because there are so many slurs (on both sides), and working out the real and complex situation is not easy.
But then that is usually the case, in any scientific field.
Tom
In fact it is so complex and the details so ill known that any predictions made are suspect.tomclarke wrote:TDPerk,
It is always the detail that matters. So the answer to your question is no: what graph (in detail) was cooked, and how, and by whom, and when, with refs if possible?
I find the GW debate fascinating biut mainly because there are so many slurs (on both sides), and working out the real and complex situation is not easy.
But then that is usually the case, in any scientific field.
Tom
So the question is: if the predictions are so good where was the graph or predictions of a flattening of the slope for the last eight or ten years? Even in a 30 year smoothed average that should be visible.
Water vapor is the most important greenhouse gas and is responsible for the amplification of any CO2 warming. So what do the modelers say about the water vapor term: "we don't know the sign, we don't know the magnitude. But our predictions are excellent."
The PDO has been known for ten years. As I understand it it still has not been included in the models. The excuse? We don't know exactly how it works. Hasn't stopped them with the water vapor term.
No sunspots today. Are we headed for a Dalton minimum? If the solar guys are right about the 300 year cycle I wouldn't be surprised. Are we at the end of a 11,000 year solar heating cycle? Could be.
http://www.spaceweather.com
I'd have a lot more faith in the Climate Change guys if they said warming? Likely. Cooling? Definitely a possibility. We should be prepared to handle either. That would be real integrity given the shakiness of the data and coarseness of the models. Not to mention the known unknowns. It might also help if they copped that there might be significant unknown unknowns.
All we seem to get when there is a new factor found is another epicycle and the epicycles always say warming. Shouldn't they say cooling once in a while? Just by chance or even misunderstanding? Nope. The errors are always in one direction. i.e. not random. Biased.
Such bias is not unknown in science. It has happened before. The data on the precession of Mercury from 1919 by a lucky accident came out to what Einstein had predicted. We now know Einstein was right but that the measurement was in error. And then there is the Milikan oil drop experiment. Milikan was in error. However he was considered such a sound experimentalist that it took decades before the results had "drifted" to the correct value.
Normal interglacials don't show broad peak we have seen for the last 10,000 years. So has man's lad use been preventing a return to colder conditions? Maybe. Pielke has been saying that for a few years.
Oh, yeah. Galileo cherry picked the results of his experiments to get F=ma. Mendel did the same with his genetic experiments.
Heck even Lysenko has been proved right in a small subset of cases.
Humans are prone to biases.
And it very well could be I'm a victim of mine. We shall see.
Engineering is the art of making what you want from what you can get at a profit.
Anyone wishing to check the figures on their own is welcome to go on line to NOAA and access the federal climate data. The National Climate Data Center has a wealth of stuff available.
I once used it to debunk a well pump manufacturer's claim of lightning damage to a pump. The government records the time and location of every lightning bolt by the RF emissions, and has them all neatly catalogued.
Anyway, I just set the tool below ...
http://lwf.ncdc.noaa.gov/oa/climate/res ... /cag3.html
to show mean temperature for 2008, annual (the whole year). The results, state by state, show about an even mix of slightly warmer or colder than normal, or normal. Do the same thing for any given year over the last couple of decades and the bias is usually distinctly toward warmer.
There's been some discussion that the warming trend may have recently flattened out. I'm inclined to scrounge more data and see for myself.
I once used it to debunk a well pump manufacturer's claim of lightning damage to a pump. The government records the time and location of every lightning bolt by the RF emissions, and has them all neatly catalogued.
Anyway, I just set the tool below ...
http://lwf.ncdc.noaa.gov/oa/climate/res ... /cag3.html
to show mean temperature for 2008, annual (the whole year). The results, state by state, show about an even mix of slightly warmer or colder than normal, or normal. Do the same thing for any given year over the last couple of decades and the bias is usually distinctly toward warmer.
There's been some discussion that the warming trend may have recently flattened out. I'm inclined to scrounge more data and see for myself.
Tom -
This is nice data but alas no good for global warming - since it only tells you about US - not much of the globe! equally anecdotal evidence based on one place does not much help.
The problem is that the medium-term regional fluctuations caused by persistent weather patterns are quite large so you get more noise unless you consider average over globe - and of course data is not as good everywhere.
Best wishes, Tom
This is nice data but alas no good for global warming - since it only tells you about US - not much of the globe! equally anecdotal evidence based on one place does not much help.
The problem is that the medium-term regional fluctuations caused by persistent weather patterns are quite large so you get more noise unless you consider average over globe - and of course data is not as good everywhere.
Best wishes, Tom
tomclarke wrote:
The Earth's surface is vast and the atmosphere relatively deep. As it rotates daily and about the Sun the Earth's temperature fluctuates in wide range diurnally (+/- 15 K) and seasonally (+/- 20 K).
Are they really measuring the spatial and temporal fluctuations with enough local resolution to truly "average" the temperature enough to detect a 0.5 K variation over a century??
Sorry but my BS detector has just gone off the radar.
Climate scientists are a bunch of charlatans that should have stuck to making pretty pictures of clouds and tomorrow's weather predictions, instead of making grandiose claims about how much they know and how much we should be paying to them our taxes. In fact, it is a stretch to say that they are actually doing science.
So how exactly is it that the climate scientists "average" over the whole globe?The problem is that the medium-term regional fluctuations caused by persistent weather patterns are quite large so you get more noise unless you consider average over globe - and of course data is not as good everywhere.
The Earth's surface is vast and the atmosphere relatively deep. As it rotates daily and about the Sun the Earth's temperature fluctuates in wide range diurnally (+/- 15 K) and seasonally (+/- 20 K).
Are they really measuring the spatial and temporal fluctuations with enough local resolution to truly "average" the temperature enough to detect a 0.5 K variation over a century??
Sorry but my BS detector has just gone off the radar.
Climate scientists are a bunch of charlatans that should have stuck to making pretty pictures of clouds and tomorrow's weather predictions, instead of making grandiose claims about how much they know and how much we should be paying to them our taxes. In fact, it is a stretch to say that they are actually doing science.
Icarus
I would have some sympathy if you replaced climate scientists by economists in the above post. They try to predict the future of complex unstable systems with human agents and claim this is science.
Or weather scientists, on the grounds that it is inherently foolish to try to predict chaos.
Working out the change in average temperature for a given change in CO2 concentration assuming solar input stays the same is a valid physics problem. It gets even more complex when you include biology and geology - but these effects are longer term and ignoring them you get a valid medium-term problem which is complex but likely soluble. Not ignoring them you still have an interesting problem, but maybe one difficult to say much about.
The point is that average global temperature (averaged over time and space) is not chaotic - and the noise from the underlying weather chaos can be arbitrarily reduced by using a longer time constant.
So do you generally dislike scientists trying to predict emergent properties of complex physical systems or is it something specifically about this problem that you feel makes its disciples charlatans?
Personally I find this style of complex computational modelling rather unsatisfying - but it is not obviously stupid nor is it trivial.
Best wishes, Tom
I would have some sympathy if you replaced climate scientists by economists in the above post. They try to predict the future of complex unstable systems with human agents and claim this is science.
Or weather scientists, on the grounds that it is inherently foolish to try to predict chaos.
Working out the change in average temperature for a given change in CO2 concentration assuming solar input stays the same is a valid physics problem. It gets even more complex when you include biology and geology - but these effects are longer term and ignoring them you get a valid medium-term problem which is complex but likely soluble. Not ignoring them you still have an interesting problem, but maybe one difficult to say much about.
The point is that average global temperature (averaged over time and space) is not chaotic - and the noise from the underlying weather chaos can be arbitrarily reduced by using a longer time constant.
So do you generally dislike scientists trying to predict emergent properties of complex physical systems or is it something specifically about this problem that you feel makes its disciples charlatans?
Personally I find this style of complex computational modelling rather unsatisfying - but it is not obviously stupid nor is it trivial.
Best wishes, Tom
Jeff Kooistra is more than a little skeptical about global warming, and the measurements are a very large factor in his skepticism.
Jeff sent me some photographs of a few of the installations used to record the temperatures used in these records. Typically, there is a little outdoor "cupola", a louvered box standing a convenient height off the ground.
I think these are typically painted white, but some don't look as if they've been painted in a decade or two. This would tend to raise the readings.
One of his favorites is located right beside a building, and close to what appears to the a heat exchanger for a heat pump. We doubt the heat pump has been there all of the past century.
Many are located in urban heat islands, which have clearly grown over the last century.
I've taken enough temperature measurements to be very cautious about the amount of error this sort of thing can introduce. When half a degree is significant ....
I realize the US-only data may not be representative of the world trend, but they were just completed and dropped into my lap. They absolutely buck the recent trend, and have me interested in looking at the world trend.
The world trend, judged by the melting of glaciers and the ice caps, clearly shows warming even without the thermometer stations. The real question is, how much of that is natural and how much anthropogenic? Looking back a few centuries, there are some really cold spells that ended naturally. Some of the ice melting now is only a few centuries old, and losing it could easily be normal climate fluctuations. It could also start coming back on its own in a decade. Melting very old ice concerns me more.
I don't know the causes. I'm quite interested.
Jeff sent me some photographs of a few of the installations used to record the temperatures used in these records. Typically, there is a little outdoor "cupola", a louvered box standing a convenient height off the ground.
I think these are typically painted white, but some don't look as if they've been painted in a decade or two. This would tend to raise the readings.
One of his favorites is located right beside a building, and close to what appears to the a heat exchanger for a heat pump. We doubt the heat pump has been there all of the past century.
Many are located in urban heat islands, which have clearly grown over the last century.
I've taken enough temperature measurements to be very cautious about the amount of error this sort of thing can introduce. When half a degree is significant ....
I realize the US-only data may not be representative of the world trend, but they were just completed and dropped into my lap. They absolutely buck the recent trend, and have me interested in looking at the world trend.
The world trend, judged by the melting of glaciers and the ice caps, clearly shows warming even without the thermometer stations. The real question is, how much of that is natural and how much anthropogenic? Looking back a few centuries, there are some really cold spells that ended naturally. Some of the ice melting now is only a few centuries old, and losing it could easily be normal climate fluctuations. It could also start coming back on its own in a decade. Melting very old ice concerns me more.
I don't know the causes. I'm quite interested.
There are clearly a number of strong medium-term non-anthropogenic influences on climate: changes in solar radiation, persistent features in ocean circulation, volcanoes.
Then there are the anthropogenic changes:
land-use change (including change of albedo in built-up areas)
rain forest destruction (special case of above)
Changes in patterns of forest fires
CO2 & methane
Direct heating (I don't think this is significant)!
There are then the long-term feedbacks from change in temperature: how does ice, land & sea vegetation adapt, how does soil/atmosphere or sea/atmosphere methane balance change, how quickly is increasing CO2 extracted from the atmosphere?
(water vapour changes, very significant, are short-term feedback)
The long-term feedbacks, whether positive or negative, are not easy to quantify. We do not have much evidence and historical records are difficult to disambiguate since there are so many different mechanisms. We can try to put limits on them using fundamental physical calculations, but the error bars will be large.
As far as the AGW debate goes the existence of long-term feedbacks increases uncertainty and should make us more wary of doing things that change us from the past 10K years sweet-spot.
The existence of quite large natural variability is reassuring in that it means the feedbacks (assuming land-use changes have not disturbed them) are likely not to be locally overwhelmingly positive.
Against this uncertainty, the factor we can expect to determine easily is what is the overall (forcing + short-term feedback) effect of an increase in GH gasses. We can perhaps also try to estimate the effect of land-use changes - but it is not easy.
The climatologists think they have an increasingly good handle on the effect of increasing CO2: what temperature rise does doubling mean? It would be surprising if they could not determine this, and it does not depend primarily on flakey global temperature measurements, it depends on detailed physical models of heat transport, cloud formation, etc.
The argument against AGW focusses on the temperature record, and uncertainties in it. It assumes that the models used are primarily stochastic (i.e. they have parameters which are fitted, perhaps overfitted, to the temperature record). It sees current evidence of warming or cooling as somehow settling the AGW debate.
Unfortunately the popular presentation of the AGW case also focusses on recent temperature changes, perhaps because convincing a politician that they must make changes based on 100K lines of computer code is difficult.
In reality, we should look for better models heat transport in the atmosphere & oceans to answer the question. these do NOT depend primarily on global temperature measurements.
My reading of the rational pro-AGW arguments is that no-one expects any one model to be correct, and no-one justifies model purely (or even primarily) on the basis of global temperature predictions. But when multiple different models which have been developing over years converge as they get more accurate physics you can begin to trust them.
Suppose we are reasonably confident that doubling CO2 leads to 2 degree increase (on top of all the other variations). Should we be concerned? 4X CO2 = 4 degrees perturbation from the current sweet spot and that would worry me greatly because of the long-term feedback uncertainty. Even 2 degrees is a significant change.
Finally, this argument is not just about CO2, thoiugh that is easiest to quantify. it is also about change in land-use. Humans are now hitting real environmental limits where are actions collectively significantly disturb the global environment. we have been in this situation locally many times in the past, but never before globally.
Arguments about doomsayers always being wrong are historically naive. We know that civilisations expanding on new technology do often (in fact usually) collapse through consequent over-use of resources. Precisely when the collapse happens is not easy to quantify. Until the collapse, all the (too early) predictions will obviously be wrong!
We have the tools now, as never before, to look a little way into the future and determine our own path rather than march lemming-like over a Malthusian cliff. I have sympathy with those who look hard at the problem and believe that we are not there yet, so no need to change now. I have no sympathy with those who prefer to keep their eyes shut!
Best wishes, Tom
PS - the last paragaph sounds rather political. Not meant to be - my apologies.
Then there are the anthropogenic changes:
land-use change (including change of albedo in built-up areas)
rain forest destruction (special case of above)
Changes in patterns of forest fires
CO2 & methane
Direct heating (I don't think this is significant)!
There are then the long-term feedbacks from change in temperature: how does ice, land & sea vegetation adapt, how does soil/atmosphere or sea/atmosphere methane balance change, how quickly is increasing CO2 extracted from the atmosphere?
(water vapour changes, very significant, are short-term feedback)
The long-term feedbacks, whether positive or negative, are not easy to quantify. We do not have much evidence and historical records are difficult to disambiguate since there are so many different mechanisms. We can try to put limits on them using fundamental physical calculations, but the error bars will be large.
As far as the AGW debate goes the existence of long-term feedbacks increases uncertainty and should make us more wary of doing things that change us from the past 10K years sweet-spot.
The existence of quite large natural variability is reassuring in that it means the feedbacks (assuming land-use changes have not disturbed them) are likely not to be locally overwhelmingly positive.
Against this uncertainty, the factor we can expect to determine easily is what is the overall (forcing + short-term feedback) effect of an increase in GH gasses. We can perhaps also try to estimate the effect of land-use changes - but it is not easy.
The climatologists think they have an increasingly good handle on the effect of increasing CO2: what temperature rise does doubling mean? It would be surprising if they could not determine this, and it does not depend primarily on flakey global temperature measurements, it depends on detailed physical models of heat transport, cloud formation, etc.
The argument against AGW focusses on the temperature record, and uncertainties in it. It assumes that the models used are primarily stochastic (i.e. they have parameters which are fitted, perhaps overfitted, to the temperature record). It sees current evidence of warming or cooling as somehow settling the AGW debate.
Unfortunately the popular presentation of the AGW case also focusses on recent temperature changes, perhaps because convincing a politician that they must make changes based on 100K lines of computer code is difficult.
In reality, we should look for better models heat transport in the atmosphere & oceans to answer the question. these do NOT depend primarily on global temperature measurements.
My reading of the rational pro-AGW arguments is that no-one expects any one model to be correct, and no-one justifies model purely (or even primarily) on the basis of global temperature predictions. But when multiple different models which have been developing over years converge as they get more accurate physics you can begin to trust them.
Suppose we are reasonably confident that doubling CO2 leads to 2 degree increase (on top of all the other variations). Should we be concerned? 4X CO2 = 4 degrees perturbation from the current sweet spot and that would worry me greatly because of the long-term feedback uncertainty. Even 2 degrees is a significant change.
Finally, this argument is not just about CO2, thoiugh that is easiest to quantify. it is also about change in land-use. Humans are now hitting real environmental limits where are actions collectively significantly disturb the global environment. we have been in this situation locally many times in the past, but never before globally.
Arguments about doomsayers always being wrong are historically naive. We know that civilisations expanding on new technology do often (in fact usually) collapse through consequent over-use of resources. Precisely when the collapse happens is not easy to quantify. Until the collapse, all the (too early) predictions will obviously be wrong!
We have the tools now, as never before, to look a little way into the future and determine our own path rather than march lemming-like over a Malthusian cliff. I have sympathy with those who look hard at the problem and believe that we are not there yet, so no need to change now. I have no sympathy with those who prefer to keep their eyes shut!
Best wishes, Tom
PS - the last paragaph sounds rather political. Not meant to be - my apologies.
The boxes are called Stevenson Boxes after Robert Lewis Stevenson who invented them.Tom Ligon wrote:Jeff Kooistra is more than a little skeptical about global warming, and the measurements are a very large factor in his skepticism.
Jeff sent me some photographs of a few of the installations used to record the temperatures used in these records. Typically, there is a little outdoor "cupola", a louvered box standing a convenient height off the ground.
I think these are typically painted white, but some don't look as if they've been painted in a decade or two. This would tend to raise the readings.
One of his favorites is located right beside a building, and close to what appears to the a heat exchanger for a heat pump. We doubt the heat pump has been there all of the past century.
Many are located in urban heat islands, which have clearly grown over the last century.
I've taken enough temperature measurements to be very cautious about the amount of error this sort of thing can introduce. When half a degree is significant ....
I realize the US-only data may not be representative of the world trend, but they were just completed and dropped into my lap. They absolutely buck the recent trend, and have me interested in looking at the world trend.
The world trend, judged by the melting of glaciers and the ice caps, clearly shows warming even without the thermometer stations. The real question is, how much of that is natural and how much anthropogenic? Looking back a few centuries, there are some really cold spells that ended naturally. Some of the ice melting now is only a few centuries old, and losing it could easily be normal climate fluctuations. It could also start coming back on its own in a decade. Melting very old ice concerns me more.
I don't know the causes. I'm quite interested.
They used to be painted with whitewash. Now a days it is titanium based paint. I forget which way the delta T goes but the two different paints are different based on how they reflect sunlight.
http://surfacestations.org/ has done some work on that.
And the latest eqpt. Doesn't use Stevenson boxes.
Engineering is the art of making what you want from what you can get at a profit.
If no one model is correct and the results are due to overfitting then what you have is numerology not science. It has no predictive value.My reading of the rational pro-AGW arguments is that no-one expects any one model to be correct, and no-one justifies model purely (or even primarily) on the basis of global temperature predictions. But when multiple different models which have been developing over years converge as they get more accurate physics you can begin to trust them.
give me four adjustable parameters, and I can fit an elephant, give me five, and I can fit the tail
None of the models so far have subtracted out the PDO. Which even if it can not be forward predicted can be dealt with historically.
None of the models have subtracted out solar variability. Which even if it can not be forward predicted can be dealt with historically.
And based on this steaming pile we are expected to give the shamans and their government acolytes trillions?
Let me run it down again: the grid squares used are large. Thus lots of things have to be parameterized rather than based on physics, chemistry, biology and land use. Rising thermals are usually less than a couple of miles across. And yet such thermals are critical for heat transport from the ground/ocean to the upper atmosphere. They are the heat pipes of the system. How many of them are in a given model area? It is parameterized. How much heat is transported? It is parameterized. How does it affect cloud formation/albedo/heat reflection back to earth? It is parameterized. The kinds of clouds formed and their effect on albedo/heat reflection back to earth? It is parameterized.
Now such models are not useless. They can answer questions like "if it works this way that is what will happen" - trends. But exact numbers? Hah.
===
I use models in electronics. And what do I use them for? Trends. A very good model of a system containing transistors (which are parameterized) is good to around +/- 5%.
So let me see 5% of 300K = +/- 15K.
And the models used in electronics are very simple. Simple (but inaccurate) models are more trustworthy.
Engineering is the art of making what you want from what you can get at a profit.
Simon -
While I agree with much of your post - especially about overfitting - the generalisation from electronics to climate modelling is good only as a debating point. You know that it is disingenuous.
Error bars on electronic circuits depend on errors in constituent devices (typically never better than a few %) and how those constituent errors combine to produce the final result. this latter is highly variable, and depends which parameter is being measured. For example, the output voltage of a common emitter stage is high independent of hfe so that even though tis parameter may be specified -50+100% the characteristics of the resulting circuit are tightly specified. I could equally give examples of circuits very sensitive to parameter variation. So your figure of 5% is plucked from the air. (5% of what?). I agree - it is a good rule of thumb for what is normally acheived, for many parameters of interest - but no more than that! (And for example of much better - consider DC output of op-amp voltage follower which is typically specified to within 0.1% of the output signal).
The analogy with climate modelling is not precise in the following way. Electronic circuits are relatively simple and unrobust. Removal or change of any one component usually breaks the circuit. The circuits quoted above rely on negative feedback to reduce parameter sensitivity. You might argue that climate system has no such negative feedbacks but of course it does! However perhaps not with a gain margin of 100dB as in the op-amp example.
The (real) climate is determined by many different physical interactions - some more & some less important. A non-chaotic emergent property - such as a temporal average of global temperature - will be affected by many parameters. Some more, some less.
We know the actual (instantaneous) global average temperature is relatively stable. +/-4 degrees which is a+/-1.5% tolerance. We are concerned with the dependence of this parameter on CO2 forcing - knowing that this +/-1.5% change is "noise" caused by various other external inputs - changes in solar radiation, earth orbit, etc. This noise is however in principle calculable, so calling it noise is a misnomer.
Finally - and this is where we part company definitively - you believe that climate models are stochastic models over-fitted to the data.
If this is true I agree with you - they have no skill.
However I do not agree with your reason for concluding that over-fitting must exist. Because the models are physical, even though many elements of the physics may have adjustable paraemeters, overfitting depends on what data is used to tune said parameters. If all are adjusted against historic global temperature record this is obviously no good. The particular parameters you worry about relate to thermal transport in the atmosphere. The primary data for fitting these is not global average temperature, but detailed temperature measurements over time & space (vertical and horizontal) in the stmosphere. So whether these parameters can be reliably inferred from data is a complex matter, as always, but it is not a case of over-fitting the model to the final result. You prefer to think that all these climatologists are incompetent. I would perhaps agree, had not this area been studied and subjected to challenge and peer review by many people for many years now. Obvious problems, like overfitting of crucial parameters, would have emerged and been quantified.
BTW the maths to quantify overfitting mathematically is Bayesian statistics and very beautiful - though exceptionally difficult to apply precisely in realistic cases. Approximations to Bayesian analysis underlie a lot of the rules of thumb used to determine model reliability.
Best wishes, Tom
While I agree with much of your post - especially about overfitting - the generalisation from electronics to climate modelling is good only as a debating point. You know that it is disingenuous.
Error bars on electronic circuits depend on errors in constituent devices (typically never better than a few %) and how those constituent errors combine to produce the final result. this latter is highly variable, and depends which parameter is being measured. For example, the output voltage of a common emitter stage is high independent of hfe so that even though tis parameter may be specified -50+100% the characteristics of the resulting circuit are tightly specified. I could equally give examples of circuits very sensitive to parameter variation. So your figure of 5% is plucked from the air. (5% of what?). I agree - it is a good rule of thumb for what is normally acheived, for many parameters of interest - but no more than that! (And for example of much better - consider DC output of op-amp voltage follower which is typically specified to within 0.1% of the output signal).
The analogy with climate modelling is not precise in the following way. Electronic circuits are relatively simple and unrobust. Removal or change of any one component usually breaks the circuit. The circuits quoted above rely on negative feedback to reduce parameter sensitivity. You might argue that climate system has no such negative feedbacks but of course it does! However perhaps not with a gain margin of 100dB as in the op-amp example.
The (real) climate is determined by many different physical interactions - some more & some less important. A non-chaotic emergent property - such as a temporal average of global temperature - will be affected by many parameters. Some more, some less.
We know the actual (instantaneous) global average temperature is relatively stable. +/-4 degrees which is a+/-1.5% tolerance. We are concerned with the dependence of this parameter on CO2 forcing - knowing that this +/-1.5% change is "noise" caused by various other external inputs - changes in solar radiation, earth orbit, etc. This noise is however in principle calculable, so calling it noise is a misnomer.
Finally - and this is where we part company definitively - you believe that climate models are stochastic models over-fitted to the data.
If this is true I agree with you - they have no skill.
However I do not agree with your reason for concluding that over-fitting must exist. Because the models are physical, even though many elements of the physics may have adjustable paraemeters, overfitting depends on what data is used to tune said parameters. If all are adjusted against historic global temperature record this is obviously no good. The particular parameters you worry about relate to thermal transport in the atmosphere. The primary data for fitting these is not global average temperature, but detailed temperature measurements over time & space (vertical and horizontal) in the stmosphere. So whether these parameters can be reliably inferred from data is a complex matter, as always, but it is not a case of over-fitting the model to the final result. You prefer to think that all these climatologists are incompetent. I would perhaps agree, had not this area been studied and subjected to challenge and peer review by many people for many years now. Obvious problems, like overfitting of crucial parameters, would have emerged and been quantified.
BTW the maths to quantify overfitting mathematically is Bayesian statistics and very beautiful - though exceptionally difficult to apply precisely in realistic cases. Approximations to Bayesian analysis underlie a lot of the rules of thumb used to determine model reliability.
Best wishes, Tom
Icarus -
Is your view based on the fact that they have taken political action (IPCC etc) or is it based on the opinion that their science is bad?
If the latter - I think your argument would be more convincing with detailed evidence. We have been pursuing this debate for a while now and the main substantive issues have pretty well emerged.
But unless you object in principle to any scientist taking political action a critique of their work must be scientific - and that means making arguments and replying to the counter-arguments until the matter is resolved.
Best wishes, Tom
PS - in UK there is now a strong push to make scientists more socially responsible, political, etc. I am not sure I agree with it!
Is your view based on the fact that they have taken political action (IPCC etc) or is it based on the opinion that their science is bad?
If the latter - I think your argument would be more convincing with detailed evidence. We have been pursuing this debate for a while now and the main substantive issues have pretty well emerged.
But unless you object in principle to any scientist taking political action a critique of their work must be scientific - and that means making arguments and replying to the counter-arguments until the matter is resolved.
Best wishes, Tom
PS - in UK there is now a strong push to make scientists more socially responsible, political, etc. I am not sure I agree with it!