The LI Revisited (Climate Modelling Using a leaky integrator)

[jethro]

Congrats on the promotion Rob, does it come with a chic new hat?

[jethro]

Stumbled across this whilst looking for something else... Boar - thought you may find it interesting. Having lost the will to live after reading no more than the abstract, I can't comment on the content, never mind the conclusions.

http://www.earth-syst-dynam.net/3/173/2012/esd-3-173-2012.pdf

[VillagePlank]

Stumbled across this whilst looking for something else... Boar - thought you may find it interesting. Having lost the will to live after reading no more than the abstract, I can't comment on the content, never mind the conclusions.

http://www.earth-sys...-3-173-2012.pdf

I am always wary of those offering gifts dressed up as untractable mathematics. In this case, I think I am right to be wary: cointegration, whilst useful, has been shown to be quite dangerous, and unreliable in the past with respect to financial methods, amongst others. I wouldn't pay to much attention to it. I am not skilled enough to dismiss it out of hand, but I wouldn't trust it not because of the authors, but rather the technique.

[Methuselah]

It's great to see the LI thread up-and-running again!

Although I don't, for one minute, doubt the basic tenets of AGW theory, anything that can help us understand prehistoric climate fluctuations (some of which have been more extreme than what's going-on currently) has got to be a step forward...

It's simples: the more the know, the better equipped we'll be to mitigate present any future climate-concerns!

Edited January 13, 2013 by Rybris Ponce

[Gray-Wolf]

Agreed Pete! It seems a shame that those who dismiss the current run of global extremes by comparing them to past instances of such record events never go the full mile and bring into context what brought about the past record?

None of us dismiss natural climate extremes driven by ev ents like volcanic activity or an unusual collection of like natural forcings but when folk say " oh we had droughts/floods/heat/cold in the past they never explain what drove that extreme then and see if similar conditions are driving that extreme today?

The Greenland 'Summit melt' this year is not alone in the series of snow records at the summit. Some folk are content to say " look! it's happened before!" and leave it at that without looking into what drove that event. As it is it appears that event was amidst the climate chaos caused by the eruption of Krakatoa. Where is that forcing last year? if that forcing is absent then what drove the melt this time?

I would see it as the equator being impacted by the dimming the eruption brought and lessening the temp Gradient between pole and equator. I see the same mechanism driving the melt this time but instead of the equator cooling to lessen the grad. we have the pole warming.

We are then left with explaining what is driving the warming of the pole.

The same with past climates. We can find many mechanisms to bring similar end results and so it is important to isolate the individual mechanisms that brought about past periods of change so we can compare that forcing to what is occuring today and see if we can attribute todays changes to that forcing.

If anything L.I. shows us a mechanism that can ape early AGW impacts but , now climate inertia is increasingly overcome, will the L.I. theory still be up to the job?

Most of our current warming has occured since the 1980's and we can see the 'natural' warm drivers that helped in this process. Now we are in a suite of colder drivers we continue to see high temps but also major physical changes that are driven by exposure to a continuous warm driver (Arctic sea ice loss/permafrost melt to ever greater depths/ice sheet mass loss) even though logic begs that we should be cooling from the high's of the 80's?

Of course the loss of ice leads to 'new' energy forcings in the form of albedo flip and also the energy freed up by the loss of ice to melt.

Basically we have the continuous AGW driver (slowly growing year on year?) but now, suddenly, other energy sources within the climate model that did not exist prior to the late 90's (and also growing year on year).

I do not remember the L.I. mapping a period where it's forcing brought about the introduction of an energy budget that altered it's deployment and also had new sources of energy that could be introduced into the model when certain parameters where reached?

[Admiral_Bobski]

Congrats on the promotion Rob, does it come with a chic new hat?

Sadly, no - but I did get a new haircut

If anything L.I. shows us a mechanism that can ape early AGW impacts but , now climate inertia is increasingly overcome, will the L.I. theory still be up to the job?

Most of our current warming has occured since the 1980's and we can see the 'natural' warm drivers that helped in this process. Now we are in a suite of colder drivers we continue to see high temps but also major physical changes that are driven by exposure to a continuous warm driver (Arctic sea ice loss/permafrost melt to ever greater depths/ice sheet mass loss) even though logic begs that we should be cooling from the high's of the 80's?

Logic begs that we should be cooling? The LI doesn't show cooling, and the LI is a mathematical formula - it doesn't get much more Logical than mathematical formulae.

Of course the loss of ice leads to 'new' energy forcings in the form of albedo flip and also the energy freed up by the loss of ice to melt.

Basically we have the continuous AGW driver (slowly growing year on year?) but now, suddenly, other energy sources within the climate model that did not exist prior to the late 90's (and also growing year on year).

I do not remember the L.I. mapping a period where it's forcing brought about the introduction of an energy budget that altered it's deployment and also had new sources of energy that could be introduced into the model when certain parameters where reached?

We don't have other energy sources as such, merely a reallocation of the distribution of energy that came from the Sun, which is precisely what the LI is all about. Look at it this way - some energy from the Sun has gone into the melting of the ice and now, in your words, that energy is "free". It has taken some time for the ice to melt, so the incident energy has come into the system, performed a "task" (the task of melting sea ice) and then been "freed up". That means that energy has been trapped in the system, which means that there has been a lag between the energy coming in and its being freed up into the atmospheric system. Is this not just one of many ways in which there is a lag in the climate system?

This being the case, have we not just proven a central tenet of the LI: that the climate system exhibits hysteresis?

(Also, changes in albedo are factored into the LI, so that "forcing" is accommodated by the hypothesis.)

[BornFromTheVoid]

Just a few question, if it's the solar input adding energy to the Earth system and warming the oceans, thus driving things like ENSO, then isn't heat release by ENSO just a way of the solar energy moving through the system? So in this sense, El Nino isn't adding any energy to the system?

What different forcings have been added so far? I'm aware of ENSO, sea ice, volcanic eruptions and solar activity. Is there anything else? Is the sea ice just for the Arctic?

Cheers

[VillagePlank]

I do not remember the L.I. mapping a period where it's forcing brought about the introduction of an energy budget that altered it's deployment and also had new sources of energy that could be introduced into the model when certain parameters where reached?

'New' energy isn't required, and I'd certainly be unhappy about the notion of the creation of energy. That's the realms of science fiction ...

I'll try to explain it in a different way. Our normal measure of energy is temperature. Most climate frameworks assume a nil energy balance, that is to say just as much goes out as it goes in. AGW says that we are altering that balance by not allowing some of the energy to leave, thus we see more energy left, which means we have a higher temperature.

The LI concludes the story differently. It says that GhG have a neglible effect, and that very small changes in the incoming energy hang around - that there's no nil balance, that the climate has only ever been quasi-stationary, and that the balance sheet does not balance at whatever time period we are looking at. Furthermore the rate that we lose energy is proportional to the energy itself; the warmer it is the faster it cools. The cooler it is the faster we warm.

We know that extra energy is in the system; we are recording higher and higher temperatures all the time, the question is "why?" The LI sets out to answer that question, not by dismissing GhG forcing, but rather to see how far we could go without it. I do not doubt AGW, but I doubt the conclusions regarding the magnitude therein.

Just a few question, if it's the solar input adding energy to the Earth system and warming the oceans, thus driving things like ENSO, then isn't heat release by ENSO just a way of the solar energy moving through the system? So in this sense, El Nino isn't adding any energy to the system?

Yes, I alluded to this when I spoke of degrees of freedom. When checking whether ones work is statistically significant, the normal process is to assume each parameter is independent. As you've correctly point out, most of these parameters aren't. It's a big problem.

Edited January 13, 2013 by Boar Wrinklestorm

[Admiral_Bobski]

Just a few question, if it's the solar input adding energy to the Earth system and warming the oceans, thus driving things like ENSO, then isn't heat release by ENSO just a way of the solar energy moving through the system? So in this sense, El Nino isn't adding any energy to the system?

What different forcings have been added so far? I'm aware of ENSO, sea ice, volcanic eruptions and solar activity. Is there anything else? Is the sea ice just for the Arctic?

Cheers

That's correct: ENSO isn't a source of heat per se. However, the timings of the ENSO events are relevant in that they modulate the basic hysteresis effect (or, to put it another way, they refine the output to give a closer match to reality). As I said the other day, the timing of the heat in and heat out of ENSO can make subtle differences depending upon the timing of solar events.

The only factors we included were, as you say, Solar Input (as proxied by sunspot data), ENSO, Volcanic activity and Albedo (as proxied by sea ice). The sea ice data was total global data, as I recall, and not just Arctic sea ice.

[songster]

The sea ice data was total global data, as I recall, and not just Arctic sea ice.

How do you account for seasonality if you're looking at the total? Extra ice during a given hemisphere's winter has next to no albedo effect. If ice area is up/down in June/March, it matters whether it's in the North or the South.

[VillagePlank]

How do you account for seasonality if you're looking at the total? Extra ice during a given hemisphere's winter has next to no albedo effect. If ice area is up/down in June/March, it matters whether it's in the North or the South.

All data was averaged to annual.

[BornFromTheVoid]

Cheers for the responses. If the time comes to re-jig the model, might I suggest adding the AMO and snow cover too?

Also, hope ye don't mind, but I thought I'd post this back up, if there's a more recent one, then I'll remove this.

Time for a prediction (until 2015, that is!) - for a bit of fun ....

• For sunspots to 2015 I have used data from here
  
• For volcanoes I have simply copied volcanic forcing from 1975->1994 to 1995->2015
  
• For ice I have assumed no change - zero deviation from mean
  
• For enso I have assumed an index of zero
  

Since then, we've had annual sea ice a little lower than predicted. I'm unsure about volcanic activity. With ENSO, we've averaged a weak La NIna, and solar activity has been lower than expected.

Presuming sea ice isn't the dominant driver of climate and has been balanced somewhat by Antarctic growth, does this indicate that the drop should have been steeper?

[songster]

All data was averaged to annual.

That doesn't answer the question. You could have two years with identical annual extent, but in one year you have a million km^2 less ice in the Arctic during June (affecting albedo) but a million extra km^2 in the Antarctic (not affecting albedo). Total sea ice is not a good proxy for albedo.

[VillagePlank]

We used sea ice extent if I recall correctly. I don't remember, tbh. Check out the pages on the original LI thread, it might be in there. Also, of course, so does depth of the ice, glaciers, quality of the ice, permafrost and whole host of other things not accounted for.

Edited January 13, 2013 by Boar Wrinklestorm

[VillagePlank]

Since then, we've had annual sea ice a little lower than predicted. I'm unsure about volcanic activity. With ENSO, we've averaged a weak La NIna, and solar activity has been lower than expected.

Presuming sea ice isn't the dominant driver of climate and has been balanced somewhat by Antarctic growth, does this indicate that the drop should have been steeper?

And we'd need to check the sunspot predictions against actual, too, I guess. I would consider dropping ENSO from the model, too.

[VillagePlank]

Presuming sea ice isn't the dominant driver of climate and has been balanced somewhat by Antarctic growth, does this indicate that the drop should have been steeper?

I'd never say that sea-ice is the dominant driver, but it account for > 30% of the "loss" of energy

[Admiral_Bobski]

As I recall, adding ENSO really snapped the LI output to match observed trends to a greater precision - I'd be reluctant to drop it on the basis that, although it could be considered to be a generic lag-causing phenomenon and hence already included in the "size of the hole", it is a greater modulator of atmospheric heat than any one other thing. In a way it almost could be considered a source of heat at certain times, while a sink of heat at others.

As for the sea ice extent/albedo issue, we accepted at the time that it was not a perfect proxy, but decided that, in the absence of anything better, it should show a general trend of decreasing albedo. I am happy to disregard the albedo data until it can be replaced with a better dataset. Bear in mind, songster, that this is currently a fairly low-res chart, using only 1 datapoint per year - averaging the northern and southern hemisphere summer-only data would be a better way of getting a yearly average, I guess, but that was more work than we were able to take on at the time.

[BornFromTheVoid]

A slightly more effective (though far from perfect) way of using sea ice might be to discount the anomaly from November to January for the northern hemisphere, and May to July for the southern hemisphere, seen as the ice doesn't have much of an effect at those times. Then generate the annual anomaly based on the remaining data?

Also, June has been losing snow cover at a very high rate in the northern hemisphere. Given that it's the month we have the longest days of the year, I should think adding that snow cover data would be important.

[VillagePlank]

My thoughts about redoing the LI (apart from the being absolutely being petrified) is to create a model of the solar system - on this basis: the premise is that sunspots for some reason not known is the primary driver of climate, and that sunspots are magnetic phenomena .... just putting it out in the open

Edit: I should add I think we start from first principles: ie the sun is, by far, the primary source of our energy, so we need to account for sun-earth distance. And we don't need howls of protest, either. As far as I can tell, such things have already been dismissed in the literature. If we can dismiss it too, then that's a good thing. It shows we're doing 'good' science.

I am currently fighting HM Prison Service for endless encroachment into the local community, and onto the strategic gap where we live with no regard for the environment, for emmissions, or for the amenity of the local community. My wife wanted to help so she looked back to well into the 1400ADs with the records of the largest landlord in our area. She didn't find a 'smoking gun' (although lots of Prison and War office documents relating to abusing Crown Immunity) and I explained to her: not getting the conclusions you think might be right are more important than getting the ones you think are right.

If you want me on onboard, that's the only philosophy I would work under. I like being wrong, but I can't abide with spurious enquiries just for the sake of spurious enquiry - which is the sole reason the LI project got shelved, King Commander Bobski left the forum for three years etc etc etc. The problem with doing it openly commits any posters to be 'in the full picture' rather than reading one sentence and committing time to answering what should already be obvious. Clearly, we'd need an administrator to judge what is fair and what isn't, but in that case, it isn't open. And so on, and so forth ....

Therefore constructing the project from first principles is the only way to go, which, as far as I can see includes, mathematics, computer science, climatology, meteorology, astronomy, quantum mechanics, ...............

Edited January 13, 2013 by Boar Wrinklestorm

[PeteG]

Look at it this way - some energy from the Sun has gone into the melting of the ice and now, in your words, that energy is "free". It has taken some time for the ice to melt, so the incident energy has come into the system, performed a "task" (the task of melting sea ice) and then been "freed up". That means that energy has been trapped in the system, which means that there has been a lag between the energy coming in and its being freed up into the atmospheric system. Is this not just one of many ways in which there is a lag in the climate system?

It's been a long time since I did science at school but hasn't that energy been used to melt the ice to water and therefore it cannot be "freed up" into the atmospheric system unless the water refreezes.

[Gray-Wolf]

The 'new energy' is within the 'L.I.' model and not just created out of the blue!

With no extra input from the 'tap' the ongoing changes within the leaking receptacle leads to energy that was once spent on enabling a 'state change' ,from the cargo of ice into water, is no longer 'spent ' on that task and so is 'freed up' to be used elsewhere?

If we already have accounted for the ongoing (and increasing) albedo flip within the model then surely such 'new energy' needs accounting for to? Over time the warming of the deep ocean will also bring impacts to the atmospheric energy budget by lessening the energy needed to 'warm' the upwelling bottom waters as they re-emerge at the surface.

These impacts must also appear as mutually reinforcing as part of the new energy will do into the changes bringing this 'new energy' into the L.I. Model?

[VillagePlank]

The 'new energy' is within the 'L.I.' model and not just created out of the blue!

With no extra input from the 'tap' the ongoing changes within the leaking receptacle leads to energy that was once spent on enabling a 'state change' ,from the cargo of ice into water, is no longer 'spent ' on that task and so is 'freed up' to be used elsewhere?

If we already have accounted for the ongoing (and increasing) albedo flip within the model then surely such 'new energy' needs accounting for to? Over time the warming of the deep ocean will also bring impacts to the atmospheric energy budget by lessening the energy needed to 'warm' the upwelling bottom waters as they re-emerge at the surface.

These impacts must also appear as mutually reinforcing as part of the new energy will do into the changes bringing this 'new energy' into the L.I. Model?

We had this one before with Iceberg, I think (apologies, if I've got it wrong)

There is no new energy, and no energy is created. The energy is accounted for becuase whilst most models assume it's already disappeared out into the cosmos, the LI suggests it's still in the system. Rising temperatures are the evidence for this. Unless you think temperatures aren't rising

Edited January 13, 2013 by Boar Wrinklestorm

[songster]

If you rerun the LI at any point, can I ask you to try a simple experiment? Run it once using all your inputs, and once leaving out the sunspots. I'd be willing to bet that you can get a good fit (perhaps even equally good) by using just ice extent, ENSO and volcanoes. That's because each of these three inputs captures different aspects of the real-world temperature record.

a ) Sea ice extent has a long-term declining trend (with an accelerating decline) which will get fitted to the long-term trend in real-world temperature data

b ) ENSO, is a short-term quasi-periodic oscillation, which will get fitted to the (ENSO-caused) short-term oscillation in the real-world temperature data

c ) Volcanoes produce irregular pulsatile forcings, which will get fitted to the (volcano-derived) spikes in the real-world data.

If you were looking for further variables to add to improve the fit, then sure, the sunspot cycle might improve things a little - however I suspect that aerosol forcings would be even more powerful.

Unfortunately, there is a much more serious problem with the model in that ice/albedo is a feedback loop. Yes, ice loss causes increased temperature by increasing energy absorption. However and more problematically, increased temperature also leads to ice loss. In using ice extent as an input to the model, you are thus effectively using the temperature data to explain itself.

Let's assume for a moment that your model is entirely correct. Everything is explained by sunspots, ENSO, volcanoes and albedo. Now, I come along with a magic dragon and warm the planet with its fiery breath, melting the polar ice caps. Will your model detect the dragon? I suspect not. The model "sees" the ice cap melting (because you're using ice extent as an input) and thus expects that temperature will rise. Hey presto, this fits very nicely with the observed temperature changes. Therefore the warming is "explained" by the albedo change, and the dragon is just a myth.

Fundamentally, you cannot attribute causality this way.

Edited January 13, 2013 by songster

[VillagePlank]

Whilst I am sure you think you've stumbled upon something, Songster, I can assure you the feedback problem is well known. Please look back, even over this thread, the independence of variables is already known to be a problem.

As it is with CO2, incidentally, and, fundamentally, in the same way, you cannot attribute causality, this way, either. In fact, CO2 is subject to multiplicative factor of about 200 in order to 'make it fit' if my memory serves me well.

Edited January 13, 2013 by Boar Wrinklestorm

[songster]

It's nothing to do with independence of variables, it's the fact that you're using a proxy for long-term temperature change to try and predict long-term temperature change - explaining one variable in terms of itself. A good fit isn't surprising. It's like trying to model the changing height of the British population over time by using trouser length as an input. Sure it agrees well, but it's not going to tell you anything about what happens in 50 years' time.