Tuesday 3 June 2014

Carbon dioxide, temperature and natural variation.

Carbon dioxide can be both a cause of increasing temperature and an effect of increasing temperature.


The temperature over the last century has been increasing due to an increase in CO2 mainly from the consumption of fossil fuels. We can also see from paleo climate records that increasing temperatures can lead to carbon dioxide being released from the oceans and acting as a feedback. If the oceans warm then CO2 is released and this CO2 causes further warming.

Climate contrarians would like to suggest that the CO2 in our atmosphere has built up from the oceans warming and equally has no effect on warming. If true this would beg the following questions:- Where has the carbon dioxide from fossil fuel burning gone and what has caused the warming?  The contrarians don’t have coherent arguments for these questions though and instead try to find evidence to spread doubt on global warming being caused by increasing greenhouse gases.

Here I will describe some mathematical techniques that involve playing with data to try to convince you that CO2 lags temperature even in today’s changing atmosphere and temperatures due to the burning of fossil fuels. This in the hope that: - readers will falsely infer that CO2 in our atmosphere has built up from the oceans warming and equally has no effect on warming.

First here are graphs showing changes in temperature and CO2:-


                                               Figure 1 changes in CO2 over last 50 years.


                                      Figure 2. Changes in temperature over the last 160 years

In figure 1 the trend in increasing CO2 can be readily seen and also the seasonal variation that affects the CO2 in our atmosphere. This is numerically consistent with the amount of CO2 emitted from humans with about half being taken up by the biosphere and the oceans due to the increased partial pressure of CO2. This is in spite of oceans temperatures rising that would diminish this uptake.  Interestingly the trend seems to have small random variations due to natural variations of a secondary nature.

In figure 2 the trend in increasing temperature is also clearly seen along with variations due to natural variations. These natural variations include small changes in solar radiation, changes in ocean circulation such as the El Nino S.O. , and changes in aerosols from volcanic eruptions blocking sunlight. (There is also a period after 1940 of cooling.  This coincides with the anthropogenic release of aerosols, due to the increase of burning “dirty” coal.)

The CO2 trend is the main cause of the trend in the temperature, and the seasonal changes in plant growth in the Northern hemisphere is the cause the yearly cyclic variation seen in figure 1.
The cyclic 11 year cycle of solar activity with a slight decreasing trend cannot account for the warming trend being of the wrong magnitude and lately of the wrong sign but it CAN account for some of these variations.


If mathematically the trend is removed and the seasonal cycle is removed then the variations can be seen clearer. However if this is done we obscure any connections about the trend in CO2 and trend in temperature. There are good mathematical ways of removing the trend and the seasonal cycles but Ole Humlum crudely achieves this by “showing monthly values (in the graphs) of DIFF12. This is the difference between the average of the last 12 month and the average for the previous 12 months for each data series. This is shown here in figure 3



                                                 Figure 3 DIFF12 CO2 and temperature.

Ole Humlum produced this graph to falsely conclude that CO2 lags temperature in today’s atmosphere and Anthony Watts at WUWT falsely conclude that this is evidence that suggests that man made CO2 is not the driver ofglobal warming.

What does a “Diff12” do to data?

In short it smoothes out variation to some extent and then differentiates as in normal calculus.

1. As in normal calculus differentiation this will turn a linear trend to a horizontal line offset from the origin.
2. A sinusoidal cycle over twelve months will disappear.
3. Other variations will be differentiated as in normal calculus.

What can we observe from these graphs?
1. The blue graph will have the seasonal cycles removed and the average value which is just slightly positive will be the trend of the rising temperature over this time period expressed in degrees C per year. (This is the trend that is caused by the trend in CO2).
2. The blue graph also has variations (other than the seasonal cycle) which represent natural variation changes per year. (This is likely the rate of change of variations due to mainly to the ENSO cycles).
3. The green graph shows the rate of change of CO2 again with seasonal cycles removed. It shows if the last 12 month period had more or less CO2 than the previous 12 months.
4. The green graph shows (assuming no errors in data handling by Humlum) that these variations in CO2 lag the variations in the temperature.

(We can further see more clearly:- that the rate of CO2 build up is increasing  slightly and the rate of temperature build up has been roughly constant. The rate that temperature is increasing is fairly constant is clearly seen here because any peaks in temperature that are above the trend line in figure 1 are less likely to be deceptively used as a starting point.)

Speculation:-
There is absolutely nothing wrong with adjusting the data in this way as long as conclusions do not ignore the alterations that have been done but rather take into account what has been done. If this has been done accurately, one may speculate for example that some natural variation (ENSO)is causing the sea to warm at a different rate and the ENSO is also affecting  the short term variations of CO2.

(One may also speculate that the natural variation is causing the sea to warm at a different rate and this change in warming then affects the rate at which CO2 is absorbed into the oceans. This would imply some direct evidence of a positive CO2 feedback.... I think a speculation that Humlum or WUWT would resist).
.................................................................................................
However this later speculation is unlikely to be the case.
a). TheCarbon Cycle Response to ENSO has been discussed by Jones et al 2001 and this supports the former speculation.

"Climatic changes over land during El Nino events lead to decreased gross primary productivity and increased plant and soil respiration, and hence the terrestrial biosphere becomes a source of CO2 to the atmosphere. Conversely, during El Nino events, the ocean becomes a sink of CO2 because of reduction of equatorial Paciļ¬c outgassing as a result of decreased upwelling of carbon-rich deep water. During La Nina events the opposite occurs; the land becomes a sink and the ocean a source of CO2."

b). In addition to the land becoming a carbon sink during La Nina as described by Jones, Behrenfield et al suggest that oceanic phytoplankton would increase during La Nina also, due to the upwelling off nutrients. This would work in the same direction as suggested by Jones i.e. CO2 decreasing during La Nina years.

What can we conclude from Humlum's graph?
We can conclude the CO2 variations are not directly the driver of the natural variability in sea temperature changes. (If you want to see clearly the correlation between CO2 and the rise in temperature you must replace the trend that has been obscured.)

We cannot conclude however that the trend in CO2 is caused by the temperature as implied by Humlum.
.......................................................................................................
 The short term variations in SST are not the cause of the short term variations in CO2 but rather both of these variations have some other common causes including ENSO. Further these correlations can be explained and of course having nothing to do with the correlation in the trends.