Instead of focusing on probabilistic percentages and constructed correlations between two events in efforts to determine potential causality, 4 scientists have revisited the requisite conditions for causality in applying the scientific method. They find “the common perception that increasing CO2 causes increased [temperature] can be excluded because it violates the necessary condition for this causal direction.”
Image Source: Koutsoyiannis et al., 2022 (1) and Koutsoyiannis et al., 2022 (2)
In the modern, politically-charged version of climate science there is an emphasis on (anthropogenic) attribution and theoretical confirmation from the manufacture of visually-satisfying correlations.
The scientific method, in contrast, requires repeated and vigorous attempts to falsify the hypotheses used to construct the theory in question. Modern climate science often skips this process.
Instead, there is a presupposition that the determination of causality can be achieved via the construction of a curve-fitted correlation between two variables, then imagining any alternative causal variables can be “ruled out” if and when these alternative curves don’t align as well as the preferred variables do.
For example, if the theory is that polar bears cannot survive without thick sea ice in summer, the current emphasis is on seeking confirmation that polar bear populations are declining. A declining population trend can then be neatly paired with trends in declining Arctic sea ice, suggesting a causal link. The assumption that polar bears need thick sea ice to catch seals and survive (even though they may have evolved 4-5 million years ago, or when the Arctic was 22-25°C warmer-than-today) is not subjected to falsification. It’s simply accepted as established truth that sea ice is essential to polar bear survival. And if anyone disagrees with this presuppositional truth, she is smeared as a “denialist” deluded by misinformation-peddling internet bloggers (e.g., Harvey et al., 2017).
The Real Scientific Method: Observing Stochastic Processes
The first paper describes the flaws with claiming one can establish a causal link by associating trend shape fitment between two events. The authors assert the scientific method requires focusing on the stochastic processes or operational conditions necessary for causation. This allows an observational analysis to falsify assumed causalities that violate essential causal conditions.
For example, it is well known physics that causes cannot lag effects. If the paleoclimate temperatures began cooling 5,000 years before CO2 concentrations began falling, the decline in the CO2 could not be said to have caused the cooling due to the causality sequencing violation.
Chicken or Egg Causality? CO2 vs. Temperature
Over the years, there have been many papers published on the question of whether the CO2 concentration drives the modern temperature changes or the temperature changes drive the CO2 change.
Wang et al. (2013) indicate that only 10% of the variance in global CO2 growth rates can be explained by fossil fuel emissions. Instead, there is a “strong and persistent coupling (r² ≈ 0.50) between interannual variations of the CO2 growth rate and tropical land-surface temperature during 1959-2011.”
Image Source: Wang et al., 2013
Jones and Cox (2005) compare the “clear similarity” between the CO2 growth rate anomalies and their correspondence to El Niño events (warming) and the negative growth rate anomalies corresponding to La Niña events (cooling). They assert that it is “unlikely that these anomalies can be explained by an abrupt increase in anthropogenic emissions, as the anomalies are much larger than the annual increases in fossil fuel emissions.”
Image Source: Jones and Cox, 2005
Just how much growth is there in the year-to-year rate of fossil fuel emissions? As Wang et al. note, often it is only about 0.1 to 0.3 GtC/yr. Sometimes even less. (For example, in the chart below, there was no year-to-year change, and, in fact, a decline, in the annual fossil fuel emissions growth rates from 1980 to 1984.)
Dr. Jarl Ahlbeck (2009) assessed these growth rate anomalies were so “clearly insignificant” they could not explain the increase in the atmospheric CO2 composition. Thus, anthropogenic emissions were omitted from the correlation coefficient equation. In contrast, the temperature “is the dominating variable, and the efficient for the temperature is strongly significant” in explaining the CO2 growth rate anomalies from 1980 to 2008.
Image Source: Ahlbeck, 2009
Dr. Humlum and colleagues (2013) further clarify the sequencing of global atmospheric CO2 versus temperature and also conclude “changes in atmospheric CO2 are not tracking changes in human emissions.” They even specify the observed time lag for the T→CO2 sequencing. The CO2 growth rate changes occur 9 months after tropospheric temperature changes and 11-12 months after sea surface temperature changes.
Image Source: Humlum et al., 2013
Even researchers who clearly do not want the observations to show the temperature explaining the majority of CO2 growth rate changes are admitting this is what the data indicate. Coulombe and Göbel (2021) reluctantly report that global mean surface temperature anomalies “cause more CO2 than the reverse,” a finding that contradicts “common wisdom.”
Image Source: Coulombe and Göbel, 2021
The Koutsoyiannis Papers: Temperature As Cause, CO2 An Effect
Koutsoyiannis and Kundzewicz (2020) analyze the sequencing of the changes in CO2 growth rate versus temperature changes using freely-available data. They find that each and every result shows a temperature (T) leading CO2 (T→CO2) directionality. The probability values are 100,000 times more likely (“about 4 to 5 orders of magnitude”) to be in the T→CO2 direction and not in the CO2→T direction. Utilizing tropospheric temperatures available from UAH, the CO2 values are shown to lag behind T values by about 6 months.
Image Source: Koutsoyiannis and Kundzewicz, 2020
The companion paper (Koutsoyiannis et al., 2022 (2)) applying the scientific method to the chicken-or-egg, T-or-CO2 causation question does not support the “common conviction” that human CO2 emissions are driving modern global warming.
Noting that a causality process requires that causes cannot lag effects in sequencing potentialities, the authors suggest there is “a (mono-directional) potentially causal system with T as the cause and [CO2] as the effect.” The CO2 lag time is about 5-8 months.
“Hence, the common perception that increasing [CO2] causes increased T can be excluded as it violates the necessary condition for this causality direction.”
Perhaps the science is not settled after all.
Image Source: Koutsoyiannis et al., 2022 (2)
Dr. Koutsoyiannis provided this e-mail summary of the two papers in a comment.