Equation Soup

Most climate skepticism I encounter these days has transparently crappy technical content, if it has any at all. It’s become boring to read.

But every once in a while a paper comes along that is sufficiently complex and free of immediately obvious errors that it becomes difficult to evaluate. One recent example that came across my desk is,

Polynomial cointegration tests of anthropogenic impact on global warming

Abstract. We use statistical methods for nonstationary time series to test the anthropogenic interpretation of global warming (AGW), according to which an increase in atmospheric greenhouse gas concentrations raised global temperature in the 20th century. Specifically, the methodology of polynomial cointegration is used to test AGW since during the observation period (1880–2007) global temperature and solar irradiance are stationary in 1st differences whereas greenhouse gases and aerosol forcings are stationary in 2nd differences. We show that although these anthropogenic forcings share a common stochastic trend, this trend is empirically independent of the stochastic trend in temperature and solar irradiance. Therefore, greenhouse gas forcing, aerosols, solar irradiance and global temperature are not polynomially cointegrated. This implies that recent global warming is not statistically significantly related to anthropogenic forcing. On the other hand, we find that greenhouse gas forcing might have had a temporary effect on global temperature.

For me, this paper had a strong smell of a bull’s behind. The paper fails to find causal relationships that are known to exist from experimental physics, which is extremely unlikely to be right, and contradicts my personal experience with low-order energy balance models calibrated with Kalman filtering, which is roughly what the authors are doing.

However, the material is so dense that it wasn’t obvious where the problem might lie. They used a 1st-order energy balance model, which is likely to understate forcing-climate relationships by a factor of 2, but that shouldn’t entirely obviate relationships. It became clear to me that I’d have to fully replicate the paper to understand what was going on, and that would be extremely time consuming. So, it went to my “in another life” pile.

Fortunately, the intersection of the sets of people who know something about climate, know something about time series dynamics, and are willing to take on some hard analytical slogging is not zero.

Comment on “Polynomial cointegration tests of anthropogenic impact on global warming” by Beenstock et al. (2012) – Some fallacies in econometric modelling of climate change

Abstract. We demonstrate major flaws in the statistical analysis of Beenstock et al. (2012), discrediting their initial claims as to the different degrees of integrability of CO2 and temperature.

I think the critique is compelling. It would be even stronger with a demonstration of a proper analysis, though that is not required to refute the original, and would be time-prohibitive.

This interchange leaves me wondering what would motivate the original authors to risk their reputations by writing a paper about climate, without bothering to acquire (via reading or partnership) any domain expertise? And given that they evidently know enough about time series stats to be dangerous, why didn’t they bother using more of the tools of the stats trade to minimize vulnerability to stupid mistakes? Are they satisfied with creating the mere impression of doubt? Or are they merely too arrogant to have appreciated the danger up front?

I think this is suggestive of limits to model-driven policy. If unprincipled skeptics deliver increasingly sophisticated red herrings to journals, the time of the wise will be entirely consumed with refuting foolish propositions. This is a losing battle, as it can take pages to explain the problems with a few bad equations. I hope this is just an anomaly of peer review, and not a foreshadowing.

h/t Rabett Run

Update: While the critique appears to be right about the flaws in the original, I have to quibble with its final point and figure, “a simple bivariate plot of temperature and log(CO2ML) over the second period, matched by means and ranges, suggests the obvious: they are closely related.” While it’s true that they’re related, this is a violation of bathtub dynamics, because temperature integrates the CO2 forcing.

Leave a Reply

Your email address will not be published. Required fields are marked *

2 × 2 =

This site uses Akismet to reduce spam. Learn how your comment data is processed.