## Can we make better graphs of global temperature history?

Chris Gittins sends along this post by Gavin Schmidt, who writes:

Some editors at Wikipedia have made an attempt to produce a complete record for the Phanerozoic:

But these collations are imperfect in many ways. On the last figure the time axis is a rather confusing mix of linear segments and logarithmic scaling, there is no calibration during overlap periods, and the scaling and baselining of the individual, differently sourced data is a little ad hoc. Wikipedia has figures for other time periods that have not been updated in years and treatment of uncertainties is haphazard (many originally from GlobalWarmingArt).

I think this could all be done better. However, creating good graphics takes time and some skill, especially when the sources of data are so disparate. So this might be usefully done using some crowd-sourcing . . .

In general, I’d give the advice that multiple graphs are a good idea, and that many graphics difficulties come from people trying to come up with some clever method for cramming too much information on to a single plot.

That said, when I think of the challenge of coming up with a single graph representing a time scale of many orders of magnitude, I think of this amazing poster, L’histoire de la Terre et de ses habitants, which cleverly uses nested helixes to display recent centuries on a single unified scale going back billions of years:

1. khms says:

many graphics difficulties come from people trying to come up with some clever method for cramming too much information on to a single plot.

And I’d argue that this amazing poster is actually a prime example – far too busy. That Wikipedia graph above is downright friendly compared with your poster.

• Andrew says:

Khms:

I think the poster is beautiful for the way it conveys so many orders of magnitude of time in one plot.

• Rahul says:

It’s all subjective but the graph pundits will never admit it. Like there’s no real objective reason to defend the nested helixes graphic against the Wikipedia one.

2. John Hall says:

Even ignoring the x-axis on that first graph, I think these climate charts don’t really give good a good sense of measurement error. Are we really to believe that we only measure the temperature with uncertainty when you go back 100mn years ago or more.

3. question says:

Does anyone know why they like using difference from the average of an arbitrary interval rather than just plotting the estimate of the temperature? Seems like a pointless extra step to me.

• Glenn Tamblyn says:

question

Basic reason for doing this is that we can estimate the difference between 2 values more accurately than estimating one value alone.

Every measurement or estimate we might make will have some error, some inaccuracy associated with it. Some errors and inaccuracies are random but some have a pattern. For example, if you have a thermometer on a wall how accurately it is read will depend on the height of the person reading it. A tall person will tend to read it differently form a short person. But if you take the difference between two readings from the short person and also two readings from the tall person, the bias due to their height will tend to cancel out so each will tend to report the temperature difference as being more similar than the absolute temperature.

This is why the use of ‘anomalies’ is so common in many areas; it helps to reduce the inaccuracies due to systematic biases because

• Chris G says:

> Does anyone know why they like using difference from the average of an arbitrary interval rather than just plotting the estimate of the temperature?

That’s always bugged me a bit too. Maybe the explanation is “human factors”? To wit, What’s 25.6-23.8? What’s 1.8-0? My bet is that most people can compute the latter more quickly than the former. (That doesn’t address how to choose as your zero reference though.)

• question says:

Interesting, this reminds me of https://en.wikipedia.org/wiki/Stochastic_resonance. Also, your answer appears to have gotten cut off and is different from that of “dab” below. Perhaps there are multiple reasons.

• dab says:

The reason to work with anomalies, rather than absolute temperature is that absolute temperature varies markedly in short distances, while monthly or annual temperature anomalies are representative of a much larger region. Indeed, we have shown (Hansen and Lebedeff, 1987) that temperature anomalies are strongly correlated out to distances of the order of 1000 km.

• question says:

Dab,

Thank you for that link. If the data shown in the chart is accurate they should stop using the term anomaly. The average temperature has only rarely been near that of 1960-1990.

The spreadsheet used to generate the chart can be found here:
http://gergs.net/?attachment_id=1014

It does not appear that the definition of “anomaly” used here is the same as that discussed in your link. The discussion on the GISS website appears to be about comparing “local anomalies”, which makes some sense. In the spreadsheet they simply use 14 C as the baseline.

• Nick Cox says:

Anomaly in climate science is equivalent to deviation in mainstream statistics, in essence value – mean. It doesn’t imply “weird” any more than deviation in statistics implies “perverse”.

• question says:

Nick,

I understand that. The difference between usual vs technical meanings of terms in statistics is a source of confusion for people, I wouldn’t hold up those practices as a desirable example.

I wonder how close we came to using “standard perversity” rather than “standard deviation”.

• K? O'Rourke says:

Errors I recall were once referred to as evils in an 1800’s paper on combining observations from astronomy (Simon Newcomb?) – so standard evils.

4. Q says:

Holocene global temperature:
Interesting that projections for 2050 (anomaly +4 degrees) and 2100 (+9 degrees) are plotted – given that, contrary to earlier projections, there has been no additional warming for at least 16 years.

• Chris G says:

> Interesting that projections for 2050 (anomaly +4 degrees) and 2100 (+9 degrees) are plotted – given that, contrary to earlier projections, there has been no additional warming for at least 16 years.

What’s interesting about it? The excess heat (due to anthropogenic CO2 emission over baseline) that the Earth has retained over that period has preferentially gone into the upper levels of the ocean as opposed to the lower level of the atmosphere, hence the reduced rate of increase in average land surface temperature. The trend is not expected to continue.

• Glenn Tamblyn says:

So if there hasn’t been warming, why are the seas rising? Sea level rises because of two possible factors:

1. Melting Land Ice adds water to the oceans. That requires extra heat – warming.
2. The oceans are expanding. Thermal expansion of the water in the oceans pushes sea level rise. And thermal expansion requires additional heat – warming.

• question says:

Q,

Do not bother with that. The only discussion should be why Venus temperature at a given pressure are 1.176 times those of earth at tropospheric pressures. You can calculate 1.176 by taking the square root of the ratio of orbital radii (eg relative temperature of the two planets appears to be determined by pressure and distance from the sun).

Everything else is a distraction while this blatant falsifier of the current narrative sits there unaddressed. Theories about climate will need to be modified to take this into account before they are worth considering further.

• As far as I can tell, the question isn’t how to explain that the earth’s global average temperature is about 290K or whatever it is, it’s how to explain the ways in which this temperature varies in time from it’s baseline average value. I don’t see how your commentary is particularly relevant to this question, but on the other hand, I don’t really understand what you’re trying to say.

• question says:

Daniel,

Venus and Earth have vastly different atmospheres, yet when comparing temperatures *at the same pressure* we can see that the difference is what we would expect due to distance from the sun. This data is not consistent with current theory that explains the temperature changes as due to high sensitivity to factors like albedo and concentration of gases in the atmosphere.

I agree that this is only tangentially related to the topic question, which is how to plot the data. I didn’t start it though!

• As we can see in the above graph, sort of, in the last 10M years the temperature has varied around +- 4 deg C. Since I’m pretty darn sure we don’t have Venutian data back 10M years, or even proxy reconstruction… and since the question of interest is about these wiggles of a few degrees. I don’t think your model addresses the question.

In essence, we have T(t) = K*sqrt(r) + epsilon(t), and you’re saying that K*sqrt(r) explains the vast majority of the difference between earth and venus, and others are essentially saying that epsilon(t) for earth is what we’re really interested in because no-one is going to be changing the orbit of the earth any time soon so r can be considered constant.

• question says:

Daniel,

I am not proposing a model. I am saying that given this observed relationship between venus and earth, we would not expect “epsilon(t)” to be due to changes in CO2 concentration. Perhaps this is because CO2 does not “trap heat” as is claimed, or the system acts to balance out this effect (eg increase albedo to compensate), or something else. I don’t know.

Any explanations for epsilon(t) need to be consistent with this data, and that does not appear to be true of current popular theories about climate.

• Chris G says:

Question:

I dismissed “The only discussion should be why Venus temperature at a given pressure are 1.176 times those of earth at tropospheric pressures…” as a second-rate troll but

“I am saying that given this observed relationship between venus and earth, we would not expect “epsilon(t)” to be due to changes in CO2 concentration. Perhaps this is because CO2 does not “trap heat” as is claimed, or the system acts to balance out this effect (eg increase albedo to compensate), or something else. I don’t know.”

That is sublime.

• question says:

Chris,

It is not trolling. I get upset when data is ignored by scientists. I’ve posted this before:

• dab says:

My understanding is the the vertical temperature profile of Venus is pretty well understood in terms of things like albedo and concentration of gases in the atmosphere. As you say, the the atmospheres of Earth and Venus are very different. The albedo of Venus is something like 70% compared to Earth’s 30% or so (I don’t recall the exact numbers). And there’s a lot more CO2 in Venus’s atmosphere than Earth’s as I am sure you are aware. These are competeing effects, which give rise to different vertical temperature profiles on the two planets. Choosing a 1 bar pressure for comparison between the two seems sort of bizarre and not well motivated by physical theory as far as I can tell. To me, a more natural comparison would be between the pressures (or altitudes) at which they each emit to space as effective blackbodies, which I think is something like 500 millibar for Earth and 100 millibar or so for Venus. If I recall correctly, the emission temperature for Venus is actually less than that for Earth despite Venus being closer to the sun and having a greater CO2 concentration, due largely to Venus’s much greater albedo.

• question says:

Dab,

Temperature is related to pressure (PV=nRT), comparing temperatures of different planets at different pressures does not make any sense. The non-bizarre thing to do is compare temperatures at the same pressure.

• Anonymous says:

Your physics is a little bizarre. Yes PV=nRT for gasses. But P = (n/V)RT, which means when you compare at constant pressure either you will find constant temperature, or you will find different density. But different density can occur because of different gravity, so there’s no real reason to prefer some kind of constant pressure comparison. What you really need, is a model for how density, temperature, and pressure vary in the atmosphere as a function of height, albedo, mixing, insolation intensity, re-radiation intensity, etc… and then use the observed patterns to infer via bayesian calculations what the effect of the various pieces of the puzzle are.

Your argument, as I understand it, is that Tvenus(Pearth)/Tearth(Pearth) = 1/sqrt(rvenus/rearth) and because of this and the fact that the composition of venus’ atmosphere is so different, we are left with the conclusion that CO2 concentration doesn’t matter on earth, only radius matters. I simply deny that the conclusion follows from the premises and say that I could rewrite this as

Tv(Pe)/Te(Pe) = 1/sqrt(rv/re) + epsilon(CO2e,…)

where “epsilon” is a small time varying function that is affected by CO2 concentrations and other things, and this is perfectly consistent with your observations (since epsilon is small and the data are noisy) so since everyone in climate science is talking about epsilon, your observations that the first order term is basically related to the radius is simply un-related to the question at hand, which is: how does epsilon behave ??

• question says:

Anon,

—-
“Yes PV=nRT for gasses. But P = (n/V)RT, which means when you compare at constant pressure either you will find constant temperature, or you will find different density. But different density can occur because of different gravity, so there’s no real reason to prefer some kind of constant pressure comparison”
—-
-One reason to prefer it is the observed relationship we are talking about here.

—-
“What you really need, is a model for how density, temperature, and pressure vary in the atmosphere as a function of height, albedo, mixing, insolation intensity, re-radiation intensity, etc… and then use the observed patterns to infer via bayesian calculations what the effect of the various pieces of the puzzle are.”
—-
-If a complicated model is needed, why does Tv=sqrt(Re/Rv)*Te relationship fit so well?

—-
“Your argument, as I understand it, is that Tvenus(Pearth)/Tearth(Pearth) = 1/sqrt(rvenus/rearth) and because of this and the fact that the composition of venus’ atmosphere is so different, we are left with the conclusion that CO2 concentration doesn’t matter on earth, only radius matters.”
—-
-That is not my conclusion. Once again, I have presented no model. This is data that needs to be addressed by the people who are coming up with these models. One could speculate that other factors adjust to compensate for any CO2 effects (eg Venus has high albedo) in order to maintain an equilibrium temperature. In that case, pumping CO2 into the atmosphere could change the climate, but this wouldn’t manifest as a change in global average temperature as is currently theorized. However, please do not take that to be any more than idle speculation. I do not claim to understand how the climate works.

—-
“epsilon is small”
—-
-Perhaps, but then what is all the fuss about global warming and runaway greenhouse effect? The preliminary model you have presented differs from the mainstream, they would need to modify their theories to account for the small epsilon.

• dab says:

Yes, the ideal gas law is one of the equations needed to get a basic understanding of the vertical temperature profile of a planet. And yes, Venus and Earth have vastly different vertical temperature profiles. They are very different planets, with different atmospheric compositions, different albedos, different distances to the sun, etc. Venus has a pressure of around 92 bar at ground level and emits to space at around 100 mbar with a scale height between 5 and 15 km; Earth has a pressure of around 1 bar at ground level and radiates to space at around 500 mbar with a scale height between 6 and 9 km. You seem to be saying that it is not enough for physics to explain the vertical temperature profiles of Venus and Earth, but that we must explain why the temperature at ground level on Earth happens to be different in some way from that of Venus at an altitude of $\ln 92$ scale heights (that’s between about 23 and 68 km above the surface for those who don’t see the point of logarithms). Gee, I wonder why that might be? Is there a climate scientist in the house?

• question says:

Dab,

“Venus and Earth have vastly different vertical temperature profiles.”

This is false in the case of altitudes corresponding to earth tropospheric pressures, one is sqrt(Re/Rv)~1.176 times the other.

“You seem to be saying that it is not enough for physics to explain the vertical temperature profiles of Venus and Earth, but that we must explain why the temperature at ground level on Earth happens to be different in some way from that of Venus at an altitude of \ln 92 scale heights (that’s between about 23 and 68 km above the surface for those who don’t see the point of logarithms). ”

I am not sure how you gathered that. There is data that the temperatures on Venus at 1 atm pressure are sqrt(Re/Rv) times those on earth at 1 atm pressure (actually the relationship extends beyond this, but lets keep it simple). This is true despite that “They are very different planets, with different atmospheric compositions, different albedos, etc”. Is that not interesting? How can this be explained by current theory which says that all the factors you mention must be known to determine temperature?

• dab says:

question, if you wish to use the Magellan data, you can download it from here:

http://atmos.pds.nasa.gov/Venus/venus.html

Then, you can get data from three orbits (numbers 3212, 3213, and 3214) rather than just 3213 as in the GIF graph on the site you linked. I would suggest “we” use that.

• question says:

Dab, thanks for finding that. The data available there is consistent with the 1995 plots, there is little difference between X- and S-channels or between orbits except for high in the atmosphere (> 60 km). So the 1994 paper appears to be in error and I only used one of the datasets for the comparison.

http://s23.postimg.org/k8terjtcr/venus2.png

You can see the relationship in this plot:
http://s13.postimg.org/bh2rwouk7/venus.png

The standard atmosphere was calculated as:
t=t0/(p/p0)^(1/C), where:
t= Temperature (K)
p= Pressure (mbar)
p0= 1013.25
t0= 288.15
C=-5.255876

• dab says:

So, question, you’re extrapolating the tropospheric layer (roughly 1000 mbar to 230 mbar) of the US Standard Atmosphere, 1976 both to pressures well above 1000 mbar (and hence below ground level on the earth) and well below 230 mbar (and hence governed by a different lapse rate in the exponent in the standard atmosphere), multiplying by the square root of the ratio of the mean distances of earth and venus to the sun, and claiming that the resulting fit to Magellan data which you’ve linked to is a falsification current climate models which must be addressed by the climate science community before any further discussion on climate change is had? Truly, you have a dizzying intellect.

• question says:

“So, question, you’re extrapolating the tropospheric layer (roughly 1000 mbar to 230 mbar) of the US Standard Atmosphere, 1976 both to pressures well above 1000 mbar (and hence below ground level on the earth) and well below 230 mbar (and hence governed by a different lapse rate in the exponent in the standard atmosphere), multiplying by the square root of the ratio of the mean distances of earth and venus to the sun, and claiming that the resulting fit to Magellan data which you’ve linked to is a falsification current climate models which must be addressed by the climate science community before any further discussion on climate change is had? Truly, you have a dizzying intellect.”

Dab,

1) Please distinguish between how the relationship was arrived at and the existence of the relationship. Imagine instead I said the red line was the output of some complex climate model taking into account greenhouse effect, albedo, etc. How would you then respond to that chart?

2) What ratio of Tv/Te would we expect if both planets had no atmosphere and zero (or the same) albedo? The answer can be arrived at from equation 4 put forward here: http://mathsci.ucd.ie/met/cess/FoundClim/archer_global_warming.pdf.

3) Yes, it is quite surprising that this relationship is so close without making any effort to modify “the exponent”. The further we get from earth troposphere pressures the less it works.
You can find out the motivation and how it is calculated here: https://docs.google.com/file/d/0B2UKsBO-ZMVgQV83S2loaGs4dnc/edit?pli=1 . You will find that the fit only gets better when we make the adjustments you imply should be made. By looking at the topleft panel of http://s23.postimg.org/k8terjtcr/venus2.png it is apparent that different things are going on above 60 km and below 45 km, which correspond to pressures outside those of the earth’s troposphere.

4) Do you know where I can get earth data to compare to the standard atmosphere?

• dab says:

>1) Please distinguish between how the relationship was arrived at and the existence of the >relationship. Imagine instead I said the red line was the output of some complex climate model >taking into account greenhouse effect, albedo, etc. How would you then respond to that chart?

Based on the clear patterns in the residual plot, I would suspect that something was very wrong with the model. And there is in this case. You need (at the very least) to adjust the exponent, which physically corresponds to accounting for different surface gravity and atmospheric composition. Strangely, you seem to be aware of that, which makes your overblown claims about data that must be addressed by climate scientists all the more bizarre.

>2) What ratio of Tv/Te would we expect if both planets had no atmosphere and zero (or the same) >albedo? The answer can be arrived at from equation 4 put forward here: >http://mathsci.ucd.ie/met/cess/FoundClim/archer_global_warming.pdf.

I’m not sure of the point of this question. Of course I am familiar with that calculation. I referred to a variant of it in my initial comment to you. And of course the different solar constants at the orbits of earth and venus are relevant to their temperature profiles. I never disputed that. I can only assume you are being patronizing as when you quoted the ideal gas law to me in a previous comment.

>3) Yes, it is quite surprising that this relationship is so close without making any effort to modify “the >exponent”. The further we get from earth troposphere pressures the less it works.
>You can find out the motivation and how it is calculated here: >https://docs.google.com/file/d/0B2UKsBO-ZMVgQV83S2loaGs4dnc/edit?pli=1 . You will find that >the fit only gets better when we make the adjustments you imply should be made. By looking at the

Yes, I am familiar with the calculations involved in the US Standard Atmosphere, 1976. I’m still not convinced that you are as familiar with them as you should be, though, based on the unphysical extrapolations you make. With the equation you quoted, you really should restrict your comparisons to pressures between ~230 and 1000 mbar. With that restriction in mind, the residual plot is even more telling.

>topleft panel of http://s23.postimg.org/k8terjtcr/venus2.png it is apparent that different things are >going on above 60 km and below 45 km, which correspond to pressures outside those of the earth’s >troposphere.

Yes. That would be above and below the cloud deck. See, e.g.,

http://curry.eas.gatech.edu/Courses/6140/ency/Chapter12/Ency_Atmos/Planetary_Atmos_Venus.pdf

>4) Do you know where I can get earth data to compare to the standard atmosphere?

Not off-hand. Googling suggests that this might be a good place to start:

https://earthdata.nasa.gov/data/standards-and-references/earth-data-science-disciplines/atmosphere/atmospheric-temperature

Now, I really must sign off from this discussion. I asked for the reference for your claims, you kindly provided it, and I must say I do not find it at all compelling. I would suggest that you stick to your strawman NHST crusade and let this one drop. However, we may have to agree to disagree on that point.

• question says:

“You need (at the very least) to adjust the exponent, which physically corresponds to accounting for different surface gravity and atmospheric composition. Strangely, you seem to be aware of that, which makes your overblown claims about data that must be addressed by climate scientists all the more bizarre. “

Dab, the whole point is that the two profiles are so similar without adjusting the exponent or accounting for the effects of the clouds, etc. But yes, lets end this conversation. The use of the standard atmosphere really just confuses matters. What is needed is the data for figure 3 here:

http://astrogeo.oxfordjournals.org/content/51/1/1.26.full.pdf+html

• Chris G says:

> Is there a climate scientist in the house?

David Archer, “Global Warming: Understanding the Forecast” covers the basics, including Earth vs Venus – see Ch.7.

• question says:

Is this what you are referring to?
http://mathsci.ucd.ie/met/cess/FoundClim/archer_global_warming.pdf

I looked at chapter seven and could see nothing mentioning nor explaining the relationship I brought up here. Could you point it out if this is the correct reference?

• dab says:

Chris G.: Thanks. I’ll look into that. FWIW, I like Ingersoll’s “Planetary Climates” as an introduction.

question: “Is that not interesting?”. No, one data point is not that interesting. “(actually the relationship extends beyond this,…”. Now that might be interesting, what’s your reference?

• question says:

Dab,

Glad you finally asked. First we need to discuss the venus data. For that we have data from Magellan found here:

http://nova.stanford.edu/projects/mgs/profile.html

If we follow the reference, there is a problem. Compare the temperature profile in the above chart with the one in this publication, and you will see that the publication data is shifted 10K higher at all altitudes. This is why I believe the relationship was originally missed.

Radio Occultation Studies of the Venus Atmosphere with the Magellan Spacecraft: 2. Results from the October 1991 Experiments
Volume 110, Issue 1, July 1994, Pages 79–94
http://dx.doi.org/10.1006/icar.1994.1108

A later paper with the same last author contains a plot of the supposedly same data, and is consistent with the chart on the stanford site:
Magellan radio occultation measurements of atmospheric waves on Venus.
April 1995. Icarus (ISSN 0019-1035), vol. 114, no. 2, p. 310-327

Do you agree that it is the 1995 reports of data that we should use?

• Tom M says:

From NOAA Global Analysis 2013:
Annually, 2013 tied 2003 as the 4th warmest year globally since 1880. Nine of the ten warmest years in the past 134 years occurred in the 21st Century. Only one year during the 20th Century–1998–was warmer than 2013.

Oh, right, 1998 was the year global warming “stopped” (unless you consider the sub-surface oceans and the cryosphere to be part of the globe, and unless you don’t cherry pick the biggest El Nino year on record as your baseline — try using “past 17 years” or “past 15 years” and you get a different result — or unless you understand the concept of variability on the time scale of climate change).

Oddly, it turns out that some domain knowledge is relevant in making claims about climate change. Realclimate.org is a good place to start. And lots of people on this site can explain the idea of variability along a trend line.

5. Chris G says:

For what it’s worth, I really like the idea of using a manifold representation of data to increase the information density in a graph. I’ve thought a fair amount about methods for learning manifold structures and ‘unfolding’ them. The thought of doing that in reverse – coming up with a low-D nonlinear structure which effectively presents information – is pretty neat.

6. John Mashey says:

(Statistics + graphics + good representations of data) are a really worthwhile topic, and hard, because doing this well is not mechanical, but takes experimentation and perhaps art to show data accurately and understandably.

Of course, this being related to climate, that had to be derailed into issues that would be covered in an introductory course, that have zero to do with the original topic, that have been done over and over elsewhere.

Sigh.

• K? O'Rourke says:

John:

> (Statistics + graphics + good representations of data) are a really worthwhile topic

Interesting point – to me statistics is about getting good representations of [the uncertainty in learning from] data.

And graphics is getting re-representations of those representations that are iconic (visual) and then good representations of data maximise the potential to see how these representations fail or fall short.

(OK now back to _real_ work.)

• John Mashey says:

Yes, as usual, one challenge is good visual representation of uncertainty for differing audiences.

Search this blog for Solomon Hsiang, who has done interesting work, and I’m pleased to say, got a well-deserved award from AGU last year.

7. Mike Liveright says:

I wonder how the Temperature of the Earth chart would look if the time were plotted on a log scale. It seems to me that breaking it apart would then be unnecessary, and it would be obvious? that the last few years is an anomaly.

• Nick Cox says:

A log scale for time going backwards (the opposite can’t be what you mean!) would imply equal steps for a year previous, a decade previous, a century previous, … In practice that’s far too strong a transformation. Square root of time before the present is closer to the mark. Any connection to the appearance of the square root of time in other contexts appears a coincidence, and in any case time going backwards and time going forwards are not equivalent here physically.

8. John Mashey says:

Back to the original topic, and the following is for CO2, not temperature, and it’s an animation, not a statistic graph, and I could wish it was interactive with sliders, but still, it’s a nice display by NOAA of CO2 history over a long time, with varying durations, runs 3 minutes.

This does however bear on the graph shown in this post, unless one rejects conservation of energy and known absorption/emission spectra of greenhouse gases..

• question says:

That’s a very nice chart. I think it would be better with “ghostlines” and vertical profiles. I’m not sure how they made it but it is possible to do it with R by exporting each plot (frame) as a png and importing into virtualdub (http://www.virtualdub.org/).

• Chris G says:

> This does however bear on the graph shown in this post, unless one rejects conservation of energy and known absorption/emission spectra of greenhouse gases..

+1

• John Mashey says:

yes :-) but perhaps my original comment may not have been obvious to all other readers.

It just meant that given real science, the large steep rise in CO2 implies a lagged, large steep rise in temperature, with the usual jiggles from ENSOs, volcanoes, solar activity, etc. Much of the warming over next 30-50 years is already “baked in” because of lags. Hence, the dots at the end of the graph can only be guesses based on different emissions scenarios,but are not out of line. I’d certainly prefer IPCC-style scenarios at the end, but that leads to another argument about time-scale for a static chart, again why I like interactive displays with sliders.

• dab says:

This one from HHMI is nice (with sliders) but requires that you own a tablet (it’s an app):

http://www.hhmi.org/biointeractive/earthviewer

On the plus side, the data sets used to make it are available for download on the same page.

9. John Mashey says:

Thanks, that’s the sort of thing I’d like to see more of.

I’d always prefer interactive systems that let people examine data, AND THEN perhaps select one or more static images, over starting with the goal of producing a static image.

10. Eli Rabett says:

A good reason for a log scale on the x axis is that the density of data gets a lot lower the further back you go in time. It’s not perfect, but judging by the graph it is a pretty useful method.

11. John Mashey says:

As a computer guy, I’ve often used semi-log scales, and that might be OK to start here.
On the other hand, for an interactive setup, I’d like to be able to select some period, and get linear, for isntance, in lookign at Milankovitch.

Of course, since eyes are not that good at comparing diagonals, I’d also like to see a parallel plot of a “first derivative”, perhaps a linear regression over some chosen interval, to see rates of change displayed as levels, not diagonals.

12. […] week Andrew Gelman picked up on a couple of graphs of extremely long time periods. Here they are again for your convenience (when one mentions a subject such as climate change, […]