Carbon dioxide (CO2), methane (CH4) and temperature anomalies time series in comparision

The above data was read in and displayed via javascript. The software is under GPL, please refer to the source for details. Screenshots of the image and the text are creative commons licence-by-sa. The script reads in local copies of CO2, CH4 concentration and temperature anomalies time series. The original CO2 file is from NOAA at: ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_mm_mlo.txt. The CO2 values in the original file are expressed as a mol fraction in dry air, micromol/mol, abbreviated as ppm. The CH4 concentrations are also from NOAA at ftp://aftp.cmdl.noaa.gov/data/trace_gases/ch4/in-situ/surface/mlo/ch4_mlo_surface-insitu_1_ccgg_month.txt. The original values are expressed as nanomol/mol, i.e. $10^{-9}$mol/mol as mol fraction in dry air often abbreviated as ppb. The temperature values are from Met Hadley Office at http://www.cru.uea.ac.uk/cru/data/temperature/HadCRUT4-gl.dat via the Met Hadley temperature page. Details about the derivation of the Med Hadley temperature data are described in the article: Quantifying uncertainties in global and regional temperature change using an ensemble of observational estimates: the HadCRUT4 data set.

The investigation of the values was motivated by the article by O. Humlum, K. Stordahl, and J. Solheim, "The phase relation between atmospheric carbon dioxide and global temperature", Global and Planetary Change, vol. 100, pp. 51-69, 2013. http://dx.doi.org/10.1016/j.gloplacha.2012.08.008 and the post: El Nino's effect on CO2 causes confusion about CO2's role for climate change on the blog real climate. In the Humlum et al. paper a time lag between annual changes (called diff12) of annual means of CO2 and temperature anomalies were observed. Following a discussion with Nathan Urban the annual changes of annual means means mathematically that if $(q_0,q_1,....q_n)$ is a list of monthly measurements then the annual mean of a value is defined as $$filter(q_i) := 1/12*\sum_{j=0}^{11} q_{i-5+j}$$ and the annual change (diff12) is defined as $$diff12(q_i) := q_{i+12}-q_i$$ and $$diff12(filter(q_i)) := filter(q_{i+12})-filter(q_i)$$ $diff12(filter(q_i))$ is sometimes for simplicity abbreviated as $diff12(q_i)$ or $diff12q_i$ and describes the annual changes of the annual means. The horizontal direction (red) denotes time. 0 denotes here month 3 of year 1958, where the NOAA CO2 measurements started. For the above diagram the values are sometimes scaled by constant positive factors and shifted in vertical direction by a constant in order to allow for a better comparision of the peaks and their corresponding time lags. The top three curves display the measured values of CO2, CH4 and temperature anomalies together with their means. The mean is drawn as a slightly darker curve around which the actual values fluctuate. The bottom three curves display the annual changes of the respective means. The computational part of the javascript part was done by me, Nadja Kutz without guarantee, if you spot errors then please send me an email (email adress in the source).

The temperature series, which were used in the Humlum et al article seem not anymore available in the internet, so Nathan Urban suggested to use HADCRUT4 data instead. The main result is approximately the same as in the Humlum et al article and as in the real climate post that is one can observe that a lot of the CO2 peaks of the diff12 curve seam to appear as lagging behind the corresponding ("corresponding" means here "approximately similar in size and form") temperature anomaly diff12 peaks, where it seems that this holds in particular for the peaks after time = 34, i.e. the year 1992. In general it seems that all the diff12 peaks correspond to a frequency of approximately one peak per every 2 years, which might be due to el Nino (?). So it looks as if there is a phase shift between the two more or less periodic curves. I suspect that the time lag between the diff12 peaks of CO2 and temperature might be connected to the decay of methane and therefore I wanted to see the corresponding diff12 curves of methane. Unfortunately methane measurements are available only since the eighties so one can't infer too much, but I find it looks indeed a bit as if most of the temperature diff12 peaks lag behind the methane peaks. It would be good to compare this also with other temperature data, in particular with raw local temperatures and other locations.
Update 6.12.2013: In the earlier version from 18.3.2013, the methane data was until 12, 2011. The current data is now 12, 2012. Moreover there was a mistake in the drawing routines offsets. This has been corrected. The correction doesn't change though the main observation that the temperature and CO2 diff12 curves "look" like as if lagging behind the methane curve (just not anymore as much as before). The reader should feel however strongly encouraged to check back for him or herself. The temperature and CO2 files are still the same. That is the CO2 values go until 2, 2013, the TEMP values until 1, 2013. The discussion can also be found in the Azimuth wiki. Further discussions are to be found at this Azimuth forum thread