That peer-reviewed Colin Kelley et al. paper “Climate change in the Fertile Crescent and implications of the recent Syrian drought“, which the traddy media is portraying like Wired did—e.g. How Global Warming Helped Cause the Syrian War—is making the rounds. The Independent (surely not taking sides) said “Climate change key in Syrian conflict — and it will trigger more war in future“.
Axel Bojanowski, Science Editor at Der Spiegel, asked some questions about the paper. Here they are, with my responses (just as I gave them to him).
Hope these comments made it to you in time.
I find the paper poor, full of contradictory assumptions, misunderstandings and misuses of statistics, and with a conclusion which is not justified.
1) Is the time scale (115 years) really sufficient when it comes to “unprecedented changes”?
Especially considering the data is not the data. What I mean is that the data they present (like in Fig 1) are the result of a model, extrapolations from a computer using input from changing stations inside Syria. See the second paragraph of the supplementary material for words about this.
That means they are not justified in drawing single blue lines, as in Fig 1A. These lines should be accompanied by some measure of uncertainty, some kind of ± bounds. That Fig 1A blue line is just a guess, in other words, which has uncertainty attached to it.
The most important thing to get from this is that the uncertainty in the data *must* be carried forward in *all* analyses. That means that red line they drew (from a regression model) on top of Fig 1A is itself too certain. I mean, that “statistical significance” they claimed with a wee p-value might not be, and very probably isn’t, true.
Incidentally, did you notice they described “statistical significance” by p-values less than 0.05 and later by less than 0.1? Grasping at straws here. Anyway, p-values are awful measures of model performance. See this: http://wmbriggs.com/post/15465/
2) Do you expect the climate data of Syria beeing sufficient especially in the first half of the 20th century?
No, and for the same reasons. We barely have any real observational data. What we have model outputs as guesses of data.
3) Does the argument that models expected the detected change really satisfy when models often fail when it comes to regional climate change?
No, no satisfaction. We don’t even know if their models are any good. The only way we can is if they can predict new data, data never before seen. They give no indication they have paid any attention to actual model performance. They instead ask us to take their word for it that the models they use are fine and that the causes they ascribe are real causes. Why should I trust them?
They ask us to believe that their inefficient regressions (bottom of p. 3) “suggest anthropogenic influence” on the drought. It does nothing of the kind. Even giving them the “significance” (which we can’t because the data they use is not the data), it is not evidence that humans caused the drought. Statistical models do not prove causality, though to believe they do is a common mistake.
The burden is on them to say exactly how this drought could have been *caused* by humans by showing us the causal mechanisms. Their burden is a heavy one because observations show that world-wide temperatures have not done much these past two decades. I mean, the warming we were promised by climate models never materialized. That means—it absolutely means—those models are broken and cannot be trusted. Would you trust a weatherman who forecasted temperatures always too high?
4) Do you consider the climate connections named in the paper beeing plausible?
On p. 2 they say “1.2-1.5 million Iraqi refugees…arrived…at the beginning of the drought”, and that these refugees make up “roughly 20%” of Syria’s urban population. They also say Syria’s population rose. And some of this is from the Iraqi influx.
They also say on that same page “The rapidly growing urban peripheries of Syria, marked by illegal settlements, overcrowding, poor infrastructure, unemployment, and crime, were neglected by the Assad government and became the heart of the developing unrest”.
Can a drought exacerbate internal unrest? Of course. Who would doubt that? The authors even say (p. 5) “An abundance of history books on the subject tell us that civil unrest can never be said to have a simple or unique cause. The Syrian conflict, now civil war, is no exception.” But they start their next sentence, “Still…” and then go on to blame the civil war on the drought. And not only on the drought, but on a drought “made worse by human-induced climate change”.
But like I said, they haven’t even come close to proving, or even suggesting, that this drought had anything to do with global warming.
5) What else do you’d like to point out?
I won’t even get started on their section “Frequency of Multiyear Droughts”. More confusing probability for causality, and in an unnecessarily complicated way.
Let me know if you have more questions,