Open Mind

CO2 Acceleration

January 12, 2009 · 50 Comments

The question arose recently, is it really true that atmospheric CO2 concentration is not just increasing but actually accelerating? Let’s take a look at some data.


Mauna Loa recently released the estimated CO2 concentration for December 2008, bring the year to a close. The initial value was a repeat of the value for November, which would be a very unusually low value (CO2 generally increases from November to December as part of its annual cycle). But this appears to be just a simple mistake because the value has been revised, the new figure fitting expectation well and conforming to the existing annual cycle pattern. Here’s the current data from Mauna Loa:

mloco2

CO2 concentration shows both a long-term increase (indicated by the smoothed curve) and an annual pattern. We can remove the annual pattern (and the average value) in order to define CO2 anomaly:

mloanom1

I’ve plotted a straight-line fit to the anomaly data, which indicates visually that the CO2 concentration follows a curved path rather than straight, one which curves upward. So let’s remove, from the original data, both the average annual cycle and a best-fit straight line, which will give us CO2 residuals:

mloquad

On this plot I’ve superimposed a parabola (a 2nd-order polynomial). The parabola indicates that the CO2 concentration does indeed curve upward, and the amount of curvature indicates that the growth rate of CO2 concentration is increasing at about 0.024 +/- 0.005 ppm/yr/yr. That increase is most definitely statistically significant (in fact it explains nearly 91% of the variance of these residuals). That does not mean that CO2 concentration is following a parabola, or that the growth rate is increasing with perfect regularity; it isn’t. But it does confirm that the growth rate is not constant, it’s been increasing.

We can also take the original CO2 anomaly data and determine the growth rate separately for each decade. We get this:

mlodecade

Clearly the decadal growth rates are on the rise, although just as clearly the increase in growth rate has not been steady. Nonetheless, even with only five decades we get a statistically significant increase in the growth rate of 0.029 +/- 0.009 ppm/yr/yr.

We can also compute annual average CO2 concentration, and from those we can compute the annual growth in CO2 concentration:

mloannrate

Once again we see that the growth rate has increased; linear regression indicates that it’s increasing by about 0.026 +/- 0.009 ppm/yr/yr, so this result too establishes the increase of growth rate with statistical significance.

In short, there’s no doubt whatever that not only is atmospheric CO2 concentration growing, the growth rate itself is also growing. The rate of increase of CO2 since 2000 is about 2.1 ppm/yr, and the long-term growth rate of the growth rate is about 0.025 ppm/yr/yr.

Categories: Global Warming
Tagged:

50 responses so far ↓

  • counters // January 13, 2009 at 12:16 am

    Thanks for working through the analysis for us, tamino. It’s a shame that even a thorough mathematical argument such as this isn’t enough to persuade certain people that the growth rate CO2 concentration is in fact increasing.

  • Colin Aldridge // January 13, 2009 at 12:30 am

    I don’t think even the most hardened sceptic would argue with your analysis. CO2 is obviosly going up and at an increasing rate and there is no sign of this increase slowing during the last 10 years or so when there has been a paause of a sort to the warming trend which undermines a popular scptic argument that warm globe drives CO2 not the other way round. Godd stuff.
    Any thoughts on the kink in the 85 95 period

  • Colin Aldridge // January 13, 2009 at 12:31 am

    I mean good stuff .. of course

  • David B. Benson // January 13, 2009 at 1:10 am

    Once again, very clear!

  • ScruffyDan // January 13, 2009 at 1:29 am

    You might want to take a look at a recent post over at the stoat blog.

    http://scienceblogs.com/stoat/2009/01/sea_absorbing_less_co2_scienti.php

    How does your post relate to that?

  • Hank Roberts // January 13, 2009 at 2:34 am

    William said a few years back it looked linear but if we kept increasing fossil fuel use that would change. Maybeso?

    Also
    http://www.pnas.org/content/104/47/18866

  • Hank Roberts // January 13, 2009 at 2:36 am

    > few years ago
    For a value of “few” equal to just slightly more than one, I mean:

    http://scienceblogs.com/stoat/2007/10/airbourne_fraction.php

  • Ricki (Australia) // January 13, 2009 at 3:01 am

    I would like to see how the emissions compare to the IPCC scenarios. It has been bandied around that we are tracking above the IPCC scenarios but I have not been able to set up the data to show this. Can you please do this?

  • Ricki (Australia) // January 13, 2009 at 3:04 am

    Also,

    The latest research on sea level rise by Grinsted et al. (Proudman Oceaneographic Laboratory and others) http://www.nbi.ku.dk/english/news/sea_level_rise_of_one_meter/

    The result: 0.9m to 1.2m by 2100 (11mm/yr by 2050, 20mm/yr by 2100) with long term rise projected to be much higher.

    Note that

    GHG concentrations used in the projections are still based on IPCC scenarios (eg. A1B) that are known to be below current annual increases in emissions.

    the long term rise (beyond 2100) is constrained to a min. of 0.5m per degree C based on historical and paleo data.

    Ice sheet dynamic melt contribution identified as the most likely contributor to the projections being 3 times IPCC.

    The equation is assumed to be linear while non-linear conditions have been known to occur in the past (see section 9).

    IPCC levels are stated as being much too low for both the fast and slow response times.

    No account is made of dynamic ice sheet collapse, change in albeido (planetary ice/vegetation cover), the effect of sea level rise on ice loss, acceleration in forcing due to increased methane from permafrost, collapse of ocean currents, etc. [although it could be argued that some of these are reflected in the paleo data].

    I have two problems with this paper:

    A) It assumes 3-4 deg rise by 2100 (a bit of an underestimate if we take no proper action). The result is that if we take concerted action, we will STILL see changes of this magnitude. If we take no action, as repeatedly looks likely, we are facing much worse sea level rises.

    B) The rates are assessed on related temperatures/sea level in historical and paleo data (as it is the only data we have). This is not the same as the current situation where forcing is accelerating as we put more GHGs up into the atmosphere with the temperature rising in a much faster way than ever before seen. Previous changes to forcing relate mostly to continental drift and Milankovitch cycles (as far as ice ages are concerned). Therefore, this paper can only approach the type of conditions we face, not define it – with consequently reduced certainty on the projections.

    My guess at long term rise from the rates given in the paper is around 5m to 15m by say 2200-2300 AD. Of course, if we melt all the ice on the planet, we get of the order of 70m sea level rise. This could still happen if we don’t get our act together.

    [Response: I'd say there's a lot of uncertainty about future sea level rise. We really don't know how the ice sheets will respond and how long they'll take to do so, and the thermal expansion of the oceans will take a very long time as it takes so long for heat to penetrate to the deep oceans.

    I'll look up some data on sea level and on CO2 emissions, and see what I can find.]

  • paulm // January 13, 2009 at 3:57 am

    Also look up data on sea level vs temp and the rate of rise.

    There are some scary precedents….

  • naught101 // January 13, 2009 at 4:43 am

    Tamino,
    I probably have a (very) slightly better grasp of climate science than the average person on the ‘web, but I have to say that I get massively confused following most if your science/stats-based posts. I can see that there are a huge number on here, and some are much simpler than others, but I don’t know where to start.

    I’d love it if you could create a new page to complement the “climate data links” page, with a list of links (either to your blog or elsewhere) ordered in a way you think appropriate, such that new comers have a good place to start. Such a page could serve as a massively helpful resource to all those reading about climate science on the ‘web.

  • Steve Bloom // January 13, 2009 at 5:10 am

    Colin (second comment), the annual plot shows that the kink lines up with the Pinatubo eruption (cooler SSTs => decreased absorption rate) and the Soviet industrial semi-collapse (which was accompanied by something of a global recession). I suspect that’s the answer, but I have no idea how the blame would be distributed among those factors.

  • William Connolley // January 13, 2009 at 8:33 am

    Nice post, thanks.

    Could you post a ref to the CO2 data as a service to humankind? - I can never find the most up to date stuff :-(

    Question (which SD has already raised): how far, from the atmospheric concentration, can you constrain plausible airbourne fraction values, and more topically how far can you constrain the fraction going into the ocean?

    Suppose you assume that 25% of emitted CO2 goes ino the atmos, and 25% into the ocean (or has, over the long term) can you demonstrate that “The results showed the amount of CO2 absorbed during 1999 to 2007 was half the level recorded from 1992 to 1999.” (reported by http://www.guardian.co.uk/environment/2009/jan/12/sea-co2-climate-japan-environment) can’t be true globally?

  • William Connolley // January 13, 2009 at 8:34 am

    …under a helpful banner such as “Climate Data Links”, perhaps. Ah well, I know now. Thanks.

  • koen // January 13, 2009 at 8:48 am

    @Ricki,

    You may find details & references here:
    http://www.globalcarbonproject.org/carbontrends/index.htm
    the .ppt has slide 7 comparing emissions to IPCC scenarios (and reality is worse than expected :)

  • michel // January 13, 2009 at 8:52 am

    there has been a pause of a sort to the warming trend which undermines a popular sceptic argument that warm globe drives CO2 not the other way round.

    These things operate, assuming they do, over very long periods. A divergence over 10 years can’t prove anything either way. Temps may start rising again. Or, less likely, CO2 levels may start to fall. But either way, there’s no reason they should move in lockstep over a period as short as 10 years.

    Back in Paleo, CO2 rises seem to have lagged by 700 years. The explanation given is that small initial temp rises caused CO2 rises that then led to large subsequent temp rises. Long time intevals though.

  • Chris // January 13, 2009 at 9:34 am

    http://www.esrl.noaa.gov/gmd/ccgg/trends/
    (Annual Mean Growth Rate)

    Back of the envelope says:
    growth rate for last 5 years (2004-8) = 1.91 ppm/yr
    For previous 5 (1999-2003) = 1.82 ppm/yr
    For previous 5 (1994-1998) = 1.99 ppm/yr

    Hence no (significant) acceleration in the last 15 yrs?

    [Response: This is getting ridiculous.
    Back of the envelope says:
    1994-1998: 1.99 plus or minus 0.6 ppm/yr
    1999-2003: 1.82 plus or minus 0.6 ppm/yr
    2004-2008: 1.91 plus or minus 0.4 ppm/yr
    Linear regression 1994-2008: +0.011 plus or minus 0.065 ppm/yr/yr.

    I shouldn't take it out on you, you just asked a question. But after all the effort I've put into increasing awareness of the statistics of trend analysis, having to deal (again) with the ridiculous comments from Lee Kington is pretty discouraging. There seems to be no end to claims about trends based on too little data data and no statistical analysis at all; are my efforts futile?]

  • Ray Ladbury // January 13, 2009 at 1:26 pm

    Tamino says: “There seems to be no end to claims about trends based on too little data data and no statistical analysis at all; are my efforts futile?”

    I will refer you to the wisdom of Mr. Twain:
    “Never try to teach a pig to sing. It doesn’t work and it annoys the pig.”

    For some folks, there is no evidence the learning curve has a positive slope. For some of us, however, you perform a valuable service and a reminder that any fool can lie with statistics, but properly used, statistics are one of the few tools we have to keep us from lying to ourselves.

  • Deech56 // January 13, 2009 at 1:56 pm

    Tamino says: “There seems to be no end to claims about trends based on too little data data and no statistical analysis at all; are my efforts futile?”

    What you have done is teach the rest of us critical evaluation skills - we’ve shaken off the cobwebs of our own stats courses (and regression courses - oh, yeah, residuals - for some of us) to make your arguments elsewhere. I am sure that a number of us saw the obvious mistakes in the Asher DailyTech post even before your thorough analysis.

  • Brad Beeson // January 13, 2009 at 3:34 pm

    I’ve enjoyed playing with the numbers published by Mauna Loa Observatory, too.

    If you overlap the data set with itself, with a 25.33 year offset, there seems to be a remarkable correspondence. (The peaks of 1973 and 1998 line up, as do many of the eralier and later features).

    25.33 is 3.619 * 7, which you’ve mentioned before as being a possible periodicity.

    Do you think this is significant, or just the human brain’s ability to find patterns in noise?

  • Dano // January 13, 2009 at 4:04 pm

    What you have done is teach the rest of us

    When you teach, you don’t teach for everybody. If you do, you’ll get frustrated real fast. You teach for those who want to learn. The rest you try to keep awake.

    Best,

    D

  • Jay // January 13, 2009 at 4:07 pm

    I used the rate data provided by the Mauna Loa site (available at their website) & got pretty much the same result & almost the exact same rate graph shown here. (Starting strictly from the annual, or monthly data as done here would have been easier by calculating the first derivitive from the data points & plotting that; the analysis here seems like a harder way to illustrate the point; but simply plotting the rate data from the Mauna Loa CO2 site was almost effortless….).

    Based on simple (emphasizing “simple”) curve fitting (using the “eyeball technique”) the overall trend is clearly up (and picking points from my graph matched the figures here within a tenth of the units of measure!), but some interesting subtrends are very apparent:

    The Mauna Loa data (for measures taken at Mauna Loa) shows a distinct deceleration (slowing) in the rate of CO2 acceleration starting in about 1992, give or take a little (not doing any kind of analysis to assess residuals, etc. this is plus/minus a year or two). After, about, 1992 CO2 levels are still going up, but clearly not as fast.

    THAT is a curious pattern. And totally unexpected given the unimpeded rate of industrialization. ….. but … that’s only for data measured at one site (Mauna Loa).

    IN CONTRAST, the world CO2 growth rate (using the data provided at the Mauna Loa website) shows a steady acceleration in CO2 without such a slowdown–just as one would expect.

    I’m not sure if this means anything more than the nuances associated with weather, etc. patterns at the Mauna Loa measurement site given the “noise” in the data.

  • george // January 13, 2009 at 6:01 pm

    Tamino:

    Rest assured. Your efforts are certainly not “futile”.

    Far from it.

    My bet is that someday, someone will probably write the book

    “All I really need to know about statistics I learned from Tamino’s blog”

  • TCOisbanned? // January 13, 2009 at 6:02 pm

    I agree that the added paramater gives 91% explanation of residuals, but think it may be more interesting to compare the rsq of the fits themselves. Obviously the second degree polynomial will be a better fit than the first degree, but I wonder how much difference.

    The acceleration seems pretty moderate. Between and 1 and 2% of the rate per year, which itself is a bit less than 1% of current total. Interesting to look at final extrapolated values at say 25 and 100 year marks, given a linear extrapolation and given a paraboloic one. Then we can consider how much difference there is functionally (in terms of temp) with the two fits.

    I assume that there are various factors of input for CO2 and maybe more than one factor of take-up. Would be interesting to look at these in a list and understand something about how each is changing, or will change. For instance some may be independant of concentration, some may be functions of it. While we can’t “predict the future”, playing with these functions should help us think about the mass balance situation over time, how different scenarious affect it, etc.

  • David B. Benson // January 13, 2009 at 6:06 pm

    The carbon dioxide information center has information about emissions. Here is a graph of yearly emissions:

    http://cdiac.ornl.gov/trends/emis/tre_glob.html

    Here is a news article about a recent attempt to estimate sea level rise (by a new method):

    “Sea Level Rise Of One Meter Within 100 Years”

    http://www.sciencedaily.com/releases/2009/01/090108101629.htm

  • Elery Fudge // January 13, 2009 at 6:25 pm

    Elery Fudge

    Tamino says: “There seems to be no end to claims about trends based on too little data data and no statistical analysis at all; are my efforts futile?”

    I know diddly about statistics but your efforts are appreciated.

    Elery

  • Zeke Hausfather // January 13, 2009 at 11:16 pm

    Ricki,

    Re: current CO2 emissions relative to IPCC scenarios, I dug up a few things in a reply over at Lucia’s: http://rankexploits.com/musings/2009/moncktons-artful-graph/#comment-8596

    You can see emissions relative to the IPCC scenarios here:
    http://67.220.225.10/~clim2165/cs/wp-content/uploads/2008/09/ipcc-scenarios.jpg

    And atmospheric concentrations relative to the (TAR) IPCC scenarios here: http://www.skepticalscience.com/images/ipcc_2001_co2.gif

    Wish I could find the actual numbers for the IPCC scenarios somewhere though… I figure I’d have to dig though the SRES datasets, though I’m not sure which are being used and how they have been modified for the AR4.

  • Jay // January 13, 2009 at 11:25 pm

    In reviewing the GROWTH in CO2 emissions its seems (intuition) that the greatest sources for the growth are in places like China–where there’s no significant, or negative, incentive to restrict this CO2 contribution.

    Question: what are the sources for the majority of the overall CO2 growth?

  • dko // January 13, 2009 at 11:32 pm

    A second-order fit through MLO trend data (to Dec 2008) shows an r^2 of 0.9988.

    Run the regression formula forward and it predicts 400 ppm in 2015 and 450 ppm by 2035.

  • blue // January 13, 2009 at 11:33 pm

    Tamino wrote: “There seems to be no end to claims about trends based on too little data data and no statistical analysis at all; are my efforts futile?”

    My guess is that people come here for the first time, read the most current post and comment. This way you’ll always start at zero again in the discussion. I kind of like naught101’s idea of having a “Tamino’s Essentials” page with links to selected posts of yours displayed prominently on the front page. It would be a very helpful resource and might reduce the number of comments rehashing well-known errors.

  • TCOisbanned? // January 14, 2009 at 12:23 am

    Right. And sometimes people don’t get my remarks either. What a bunch of newbies. NEW….BIES! Look at the NEW….BIES! (Said in a Mr. Garrison voice.)

  • David B. Benson // January 14, 2009 at 12:30 am

    Jay // January 13, 2009 at 11:25 pm — Look to the Carbon Dioxide Information Analysis Center for answers to such questions.

  • TCOisbanned? // January 14, 2009 at 12:38 am

    I’m just kinda noodling here. I’m trying….grasp towards inferences.

    Does the high explanatory power of a quadratic indicate some sort of quadratic process? Or just the better fit that you tend to get when adding terms? How would adding an exponential (instead of squared term) compare in terms of fit? How much of the residuals from the quadratic would be taken care of if we made it a cubic?

    Please don’t take any of the preceding babble as critical. If I have to make a prediction for 2020(and don’t have access to specific predictions from fundamental analysis), my “Bayesian bet” would be to use Tammy’s quadratic.

    [Response: The quadratic model doesn't imply an actual quadratic process. It really only establishes that the trend is definitely not linear, and the nonlinearity is such that the growth rate of CO2 is going up rather than down. I don't know whether an exponential fit would be better or worse, but it wouldn't establish an exponential pattern either; it would just show the same things, that the trend is nonlinear and the growth rate is going up rather than down.

    The noise level is high enough that i doubt it's possible to establish the form of the pattern, beyond saying that it's nonlinear and curves upward rather than downward.

    Any "smooth" function can be expressed as a Taylor's series, so in essence it can be expressed as an infinite-order polynomial. Fitting higher and higher order polynomials may be able to explore more and more terms in that series, but we'd need a lot more evidence to conclude that we've actually divined the form of the pattern.

    It's also very dicey to attempt to forecast the future using polynomial approximations. One can compute an "error function" (not the "error function" which refers to the normal probability distribution) which gives the probable error in such a forecast based on random noise alone (let alone due to incorrect specification of the form of the function!), and for polynomial models the error function explodes rapidly beyond the limits of the observed data. This is for a lot of reasons, including the fact that polynomials are unbounded. But the bottom line is: don't forecast with polynomial fits beyond a tiny future span, unless you have overwhelming support that the polynomial model is real.]

  • Mark // January 14, 2009 at 1:07 am

    Ray Ladbury wrote: “any fool can lie with statistics, but properly used, statistics are one of the few tools we have to keep us from lying to ourselves.”

    That line is almost worthy of Mark Twain, whom you quoted in the previous paragraph.

    It’s not a quote from Mark Twain, is it? Or someone else?

  • Ray Ladbury // January 14, 2009 at 1:48 am

    It is my own line. You are free to use it.

  • Kevin // January 14, 2009 at 7:25 am

    Please explain why temperature increase is not accelerating during the same period. CO2 causes the global temperature increase right? From what I see temperatures are decreasing over the same period that the CO2 increase is accelerating. How can this be?

    [Response: This "temperatures are decreasing" notion betrays a failure to understand even the basics, and it certainly betrays a lack of having read and understood what's on this blog.

    I really do need a "start here" link.]

  • EliRabett // January 14, 2009 at 1:16 pm

    Keeping with my betting policy, Eli bets a bunch of carrots and one beer that the rate of increase will fall sharply in 2009 mostly because the economy in the US, EU and China are tanking. You can see the same effect btw 1990 and ~1995 when the FSU and eastern europe collapsed economically.

    Crud is coupled.

  • Nick Zervos // January 14, 2009 at 1:45 pm

    Tamino, where did the values of annual CO2 growth in your last graph come from? They don’t seem to agree with the numbers shown in the table at the right side of the Mauna Loa web page
    http://www.esrl.noaa.gov/gmd/ccgg/trends/

    Just a couple of examples: your graph gives about 0.9 ppm/yr for 1960, 1.6 for 1970, 1.3 for 2000 vs. 0.51, 1.02, and 1.74, respectively, in the Mauna Loa table.

    [Response: I computed the annual average CO2 concentration, then took the difference between each year's value.]

  • Gavin's Pussycat // January 14, 2009 at 3:10 pm

    dko,

    there is a reason that Tamino goes from a linear to a quadratic fit… the interesting thing here is the non-linearity. That’s what his R^2 = 0.9091 is referring to.

    BTW you can only use Pearson’s correlation and significance value for time series that are not significantly autocorrelated. You can see already by looking at the residuals wrt the parabola that that just isn’t the case here: I would visually guess 5-10 years correlation length.

    From the fifth picture it looks like the derivative is close to “white”; if true, that would mean that CO2 itself is “red”.

    Tamino, 0.024 +/- 0.005, 0.026 +/- 0.009: those were derived properly with autocorrelation, right? (Your teachings have made me a paranoid wreck :-)

    [Response: Yes.]

  • John Mashey // January 14, 2009 at 4:52 pm

    If you do a “Start Here”, seriously consider doing a “Catalog of Cherry Picks” of which many have been well-illustrated here:

    1) Too short time series [purposeful or accidental]

    2) Series/graph without statistical significance [purposeful or accidental]

    3) Selection of carefully-chosen segment of a time-series, with graph to disappear important effects [one thinks of CO2 vs temperature, picked for any small sequence of years where temperature isn't rising].

    4) Selection by geography.

    5) Selection by subset of a time-series chosen to emphasize some effect or lack thereof. Example: ice-extent during Arctic winter, ignoring summer.

    6) Adding two sequences together, when one is just doing random jiggling: add Arctic and Antarctic ice together, and by picking right tiem period, you can prove anything.

  • george // January 14, 2009 at 4:56 pm

    Nick

    If i am not mistaken, the difference (between tamino’s values and those of NOAA) is caused by the different way that NOAA calculates annual mean rate of growth of CO2

    The table shows annual mean carbon dioxide growth rates for Mauna Loa. The annual mean rate of growth of CO2 in a given year is the difference in concentration between the end of December and the start of January of that year. If used as an average for the globe, it would represent the sum of all CO2 added to, and removed from, the atmosphere during the year by human activities and by natural processes. There is a small amount of month-to-month variability in the CO2 concentration that may be caused by anomalies of the winds or weather systems arriving at Mauna Loa. This variability would not be representative of the underlying trend for the northern hemisphere which Mauna Loa is intended to represent. Therefore, we finalize our estimate for the annual mean growth rate of the previous year in March, by using the average of the most recent November-February months, corrected for the average seasonal cycle, as the trend value for January 1. Our estimate for the annual mean growth rate (based on the Mauna Loa data) is obtained by subtracting the same four-month average centered on the previous January 1. Preliminary values for the previous year are calculated in January and in February.

    http://www.esrl.noaa.gov/gmd/ccgg/trends/

    compare that to tamino’s method:

    I computed the annual average CO2 concentration, then took the difference between each year’s value.]

    Though the method for calculating annual mean growth may be different, the slope of the trend (from 1959-present) obtained with the numbers from the NOAA table is basically the same as what tamino gets: 0.026ppm/yr/yr

    By the way, Tamino what noise model have you used for CO2?

  • TCOisbanned? // January 14, 2009 at 4:59 pm

    What is the rsq of the linear fit?

  • koen // January 14, 2009 at 5:20 pm

    Eli,

    I fear all bets are off. I agree with you that this crisis will lower a number of sources. But the articles from globalcarbon also indicate the ocean sinks lowering in capacity. I don’t know yet whether they do cliff-diving or not (hope not ).

    Also, Tamino reported on ABCs, which were said to cause lower temps (1-2° IIRC). With industrial output and shipping falling through the floor, we might see these ABCs diminish if not disappear.

    So in the end, we could have lower CO2 emissions and higher temperatures. Now that will be fun in some circles.

    (No I don’t take bets)

  • Richard Steckis // January 14, 2009 at 6:35 pm

    William Connolley:

    “Could you post a ref to the CO2 data as a service to humankind? - I can never find the most up to date stuff :-(”

    For those with R, the following script will give you a plot of the Mauna Loa data straight from their FTP server:

    url=”ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_mm_mlo.txt”
    mld<-read.table(url, skip=20, fill=T, na.strings=-99.99)
    names(mld)<-c("year", "month" ,"yymm", "co2", "co2interp", "co2trend")
    attach(mld)

    x<-yymm
    y<-co2

    plot(x,y, main="Mauna Loa co2 Data", xlab="Year", ylab="co2 (ppm)", col="red", pch=16, type="l")

  • Gerda // January 14, 2009 at 6:42 pm

    thanks, very useful.

    i’m still awake, and i understood nearly all of the statistics jargon!
    i agree though, some kind of ’stats 101′ list would be very handy.

    i sort of hope it is china rather than decreasing absorption by oceans that is the dominant factor in the recent acceleration. we will see - if it levels off again during this recession or continues to rise.

  • Richard Steckis // January 14, 2009 at 6:58 pm

    Just make sure you redo the double quotes around the url statement.

  • george // January 14, 2009 at 6:59 pm

    Eli bets a bunch of carrots and one beer that the rate of increase will fall sharply in 2009 mostly because the economy in the US, EU and China are tanking. You can see the same effect btw 1990 and ~1995 when the FSU and eastern europe collapsed economically.”

    Given the large variability from year to year, how could one know that even a fairly large decrease was due largely to a change in economic activity (ie, CO2 emissions)?

    eg, Mauna loa Annual Mean
    Growth Rate

    1990 1.31
    1991 1.02
    1992 0.43
    1993 1.35
    1994 1.90
    1995 1.98
    1996 1.19
    1997 1.96
    1998 2.93
    1999 0.94
    2000 1.74

    The dip from 1991 to 1992 does stand out, but is not clear that it can be attributed primarily to economic downturn.

    Pinatubo also may have had a significant effect.

    Large Volcanic Eruptions Help Plants Absorb More Carbon Dioxide From the Atmosphere (December 10, 2001)

    New NASA-funded research shows that when the atmosphere gets hazy, like it did after the eruption of Mt. Pinatubo in the Philippines in June 1991, plants photosynthesize more efficiently, thereby absorbing more carbon dioxide from the atmosphere.

    When Mount Pinatubo erupted, scientists noticed the rate at which carbon dioxide (CO2) filled the atmosphere slowed down for the next two years. Also during 1992 and 1993, ash and other particles from the volcano created a haze around the planet and slightly reduced the sunlight reaching Earth’s surface and made the sun’s radiation less direct and more diffuse.

  • Gareth // January 14, 2009 at 8:18 pm

    koen,

    IIRC, total GHG forcing at present is something like 430ppm CO2e, but aerosols reduce that to about 385ppm CO2e.

    A substantial reduction in aerosols due to a fall in industrial activity (or concerted efforts to clean up smokestacks) could provide a very nasty kick to overall warming.

  • george // January 14, 2009 at 8:34 pm

    From 1991 - 1992, the atmospheric CO2 growth dropped by 58%. (from 1.02ppm/yr to 0.43ppm/yr for Mauna loa)

    Actually, it’s fairly clear that was not due primarily to any downturn in economic activity (in eastern europe or anywhere else)

    In order for that to have been the case, the percent change in emissions (as a % of the previous year’s emissions), would have to have been similar to the percent change (drop) in the atmospheric CO2 growth from the previous year (ie, -58%). And it was not in that case (not even close).

    (The volcanic explanation I linked to above is a far more plausible explanation in that case)

    Even with a steep economic downturn, I doubt we would (will?) see a 50% drop in emissions (not unless the downturn lasted several years. 5 years at 10% yearly decline in emissions would be another story, of course.

    And it seems to me that even the effect (on atmospheric CO2 growth) of a 5-10% drop in total worldwide human CO2 emissions (more plausible if there is a rather severe worldwide economic contraction) might still get lost in the “noise” (assuming a 5-10% drop in total human emissions leads to about a 0.1- 0.2ppm drop in the atmospheric CO2 growth.)

    In fact, according to NOAA,
    “The estimated uncertainty in the Mauna Loa annual mean growth rate is 0.11 ppm/yr”.

  • Hank Roberts // January 14, 2009 at 9:57 pm

    George, you’re using logic without RTFM, mistakenly assuming the number reflects changes ” …from the previous year (ie, -58%). And it was not in that case (not even close).”

    Read the information at the Mauna Loa site on the time lags known between when changes in CO2 sources/sinks occur and when the change will appear in the mountain numbers.

    These can be tied back to seasonal changes, longterm large drought events, and such as well as economic changes in fuel use and methane leaks. No, we don’t know how leaky the natural gas system of the USSR was before they shut it off or how well they’ve fixed it. We have some idea of their fossil fuel use though.

Leave a Comment