ReinyDays
Gold Member
Well, living beings, man, dogs, other animals as well, do emit CO2 as they breathe. The major issue is causation. Research has shown now that Temperature rises causes more CO2 to rise.
Anyway, the tiny amount of CO2 at 30,000 feet is really not a problem.
Causality and climate
Guest post by Antonis Christofides, Demetris Koutsoyiannis, Christian Onof and Zbigniew W. Kundzewicz On the chicken-and-egg problem of CO2 and temperature.judithcurry.com
Investigating potential causes
So here is the question: given two processes, how can we determine if one is a potential cause of the other? We deal with this question in two papers we published last year in the Proceedings of the Royal Society A (PRSA): Revisiting causality using stochastics: 1. Theory (preprint); 2. Applications (preprint). We reviewed existing theories of causation, notably probabilistic theories, and found that all of them have considerable limitations.
For example, Granger’s theory and statistical test have already been known to be identifying correlation (for making predictions), not causation, despite the popular term “Granger causality”. What is more, they ignore the fact that processes exhibit dependence in time. Hence, formally testing hypotheses in geophysics by such tests can be inaccurate by orders of magnitude due to that dependence.
As another example, Pearl’s theories make use of causal graphs, in which the possible direction of causation is assumed to be known a priori. This implies that we already have a way of identifying causes. Moreover, insofar as those theories assume, in their use of the chain rule for conditional probabilities, that the causality links in the causal graphs are of Markovian type, their application to complex systems is problematic.
Another misconception in some of earlier studies is the aspiration that by using a statistical concept other than the correlation coefficient (e.g. a measure of information) we can detect genuine causality.
Having identified the weaknesses in existing theories and methodologies, we proceeded to develop a new method to study the question whether process X is a potential cause of process Y, or the other way round. This has several key characteristics which distinguish it from existing methods.
The only “hard” requirement kept from previous studies is the temporal precedence of the cause over the effect. Sometimes it can happen that causation goes both ways; for example, hens lay eggs and eggs hatch into hens (and it was Plutarch who first used the metaphor of hen and egg for this problem). Conveniently, we call such systems “potentially hen-or-egg causal”. Our method also identifies these, and also determines in these cases which of the two directions is dominant.
- Our framework is for open systems(in particular, geophysical systems), in which:
- External influences cannot be controlled or excluded.
- Only a single realization is possible—repeatability of a geophysical process is infeasible.
- Our framework is not formulated on the basis of events, but of stochastic processes. In these:
- Time runs continuously. It is not a sequence of discrete time instances.
- There is dependence in time.
- It is understood that only necessary conditions of causality can be investigated using stochastics (or other computational tools and theories)—not sufficient ones. The usefulness of this, less ambitious, objective of seeking necessary conditions lies in their ability:
- To falsify an assumed causality.
- To add statistical evidence, in an inductive context, for potential causality and its direction.
To deal with dependence in time, often manifested in high autocorrelation of the processes, we proposed the differencing of the time series, which substantially decreases the autocorrelation. In other words, instead of investigating the processes X and Y and find spurious results (as has been the case in several earlier studies), we study the changes thereof in time, ΔX and ΔY.
A final prominent characteristic of our method is its simplicity. It uses the data per se, rather than involved transformations thereof such as the cross- and auto-correlation functions or their Fourier transforms —the power spectra and cross-spectra. The results are thus more reliable and easier to interpret.
Atmospheric temperature and CO₂ concentration
In our PRSA papers we implemented our method in several case studies, such as rainfall-runoff and El Niño-temperature. One of the case studies was CO₂ concentration and temperature, and this one gave strong indications that temperature is potentially the cause and CO₂ the effect, while the opposite causality direction can be excluded as violating the necessary condition of time precedence.
However, the scope of these two papers was to formulate a general methodology for the detection of causality rather than to study a specific system in detail, and the case studies were brief. With regard to the relationship between temperature and CO₂ concentration, we hadn’t gone into details as to the effect of seasonality and time scale, or the exploration of many sources of data. So in our latest paper, published a week ago in Sci (“On hens, eggs, temperatures and CO2: Causal links in Earth’s atmosphere”), we studied the issue in detail. We used CO₂ data from Mauna Loa and from the South Pole, and temperature data from various sources (our published results are for the NCAR/NCEP reanalysis, but in the previous papers we used satellite data too). We used both historical data and the outputs of climatic models. We examined time scales ranging from months to decades.
The results are clear: changes in CO₂ concentration cannot be a cause of temperature changes. On the contrary, temperature change is a potential cause of CO₂ change on all time scales. As we conclude in the paper, “All evidence resulting from the analyses of the longest available modern time series of atmospheric concentration of [CO₂] at Mauna Loa, Hawaii, along with that of globally averaged T, suggests a unidirectional, potentially causal link with T as the cause and [CO₂] as the effect. This direction of causality holds for the entire period covered by the observations (more than 60 years).”
The math is a bit too complicated to present here. However all three papers have been reviewed extensively by referees and editors (notice in the last paper that four editors were involved as seen on the front page of the paper). The results in the earlier papers were criticized, formally by a commentary in the same journal and informally in blogs and social media. Some concerns expressed by critics, such as about lengths of time series, effect of seasonality, effect of timescale, are dealt with in this new paper. No-one has however developed any critique of the methodology.
How are they demonstrating this in the laboratory? ...
