Following a brief reading of the World Weather Attribution report regarding the 2019 Australian bushfires, I felt the following extracts were worthy of comment:

“The study reveals the complexity of the 2019/20 bushfire event, with some, but not all drivers showing an imprint of anthropogenic climate change.”

That seems fair enough. But, of course, the press are only interested in those drivers that do show “an imprint of anthropogenic climate change”.

“In addition, ignition sources and type of vegetation as factors largely outside the meteorology play an important role. In this analysis we only consider the influence of weather and climate on the fire risk, excluding ignition sources.”

So they have conditioned their analysis with respect to “ignition sources and type of vegetation”. That’s one hell of a conditioning. It would have been nice to understand how the percentage risk changes when one conditions instead upon anthropogenic climate change and explores counterfactuals regarding ignition and combustion, e.g. forest management counterfactuals.

“Thus, while it is clear that climate change does play an important role in heat and fire weather risk overall, assessing the magnitude of this risk and the interplay with local factors has been difficult. Nevertheless it is crucial to prioritise adaptation and resilience measures to reduce the potential impacts of rising risks.”

Absolutely, it is crucial to prioritise. That is why it is so important to explore all factors, including the global and local, and the climatological and non-climatological.

“We therefore investigate the question how anthropogenic climate change influences the chances of an intense bushfire season, rather than focusing on a single episode of intense bushfires.”

That’s weird, because I got the distinct impression from the press that this report was all about a single episode of intense bushfires, i.e. Australia 2019. And for a report that didn’t focus on a single episode, it certainly had a lot to say about it! If the report wasn’t supposed to be about a single episode, perhaps the text ‘Australian Bushfire’ shouldn’t have appeared in its title.

“Attributing observed trends to anthropogenic climate change can only be done with physical climate models as they allow isolating different drivers.”

Further evidence here that the report is only interested in studying ‘observed trends to anthropogenic climate change’. Attributions for specific events can then made with little or no regard to local context, presumably.

“The attribution statements presented in this paper are for events defined as meeting or exceeding the threshold set by the 2019/20 fire season and thus assessing the overall effect of human-induced climate change on these kind of events. In individual years, however, large scale climate system drivers can have a higher influence on fire risk than the trend. Besides the influence of anthropogenic climate change, the particular 2019 event was made much more severe by a record excursion of the Indian Ocean Dipole and a very strong anomaly of the Southern Annular Mode, which together explain more than half of the amplitude of the meteorological drought (precipitation deficit). We did not find a connection of either mode to heat extremes.”

There you go! Buried deep in the report, an admission of the importance of the non-anthropogenic precipitation deficit, but no attempt to perform a counterfactual analysis that conditions upon the anthropogenic drivers so that the deficit’s impact in terms of percentage risk can be calculated. Could this be because no one was interested enough to do so, or was it because the factors were deemed too complicated? Either way, you will search in vain in the report for a statement along the lines of “A record excursion of the Indian Ocean Dipole and a very strong anomaly of the Southern Annular Mode increased the risk of the fire event by x percent.” All one can find is a strong hint that such a percentage would likely turn out to be a lot more impressive than the one attributed to anthropogenic climate change. As I have said before, when it comes to causal analysis, the storyline one creates depends largely upon the causal agents one conditions upon.

Caveat: I haven’t finished reading the report. I might yet change my position. There is also the possibility that someone like Jaime Jessop will upstage me by doing a lot more thorough job.


  1. I live on the edge of about 2 million acres of bush inn the blue mts.We had at one stage three fires threatening our home.The back burning over the years around here is nearly non existent .
    The fuel loads are huge and residents are allowed to build, in the bush with any design other than fire proof.The new legislation up here slows down the rate a house burns down to allow residents to flee.And nothing else.
    Yet “woke” people here are all on the grid/all own at least 1 car but seriously suggest we have to leave fossil fuels alone.(not them of course, everyone else)
    They do this by cool stickers on their cars and marching around with placards.
    Australia is a fire prone country.The aboriginal practises of fire stick burning around here are no longer practised..
    Fuel loads are huge and then when we have no rain for ages and arsonists running around (which the ABC and other Woke media tried to spin and hide) then there will be fires…
    But yup..its just Co2…

    Liked by 2 people

  2. John, I’m glad you’ve got the ball rolling on this. I was going to suggest a joint effort in analysing this paper. I’m sure there will be a great deal to question! I must admit, I haven’t really started reading it in detail yet, with other things going on.

    What did strike me as well is this. Just before your quote about ignition sources they say this:

    Wildfires in general are one of the most complex weather-related extreme events (Sanderson and Fisher, 2020) with their occurrence depending on many factors including the weather conditions conducive to fire at the time of the event and also on the availability of fuel, which in turn depends on rainfall, temperature and humidity in the weeks, months and sometimes even years preceding the actual fire event.

    The ‘availability of fuel’, according to the authors, is a function only of the rainfall, temperature and humidity preceding the fires. Which is odd, considering the furore generated by the demonstrable lack of prescribed burns and mechanical clearance which would have reduced the . . . . . availability of fuel!

    Also, the authors chose to do three attribution studies as follows:

    We perform the analysis of possible connections between the fire weather risk and anthropogenic climate change in three steps. First, we assess the trends in extreme temperature and conduct an attribution study using a seven-day moving average of annual maximum temperatures corresponding to the time scale chosen for the Fire Weather Index. Second, we undertake the same analysis but for drought defined as purely lack of rainfall in two time windows, the annual precipitation as well as the driest month in the fire season, with the latter again roughly corresponding to the input time scale of the FWI. Third, we conduct an attribution study on the Fire Weather Index (FWI) and the Monthly Severity Rating (MSR). These three attribution studies follow the same protocol used in previous assessments.

    So, they do a separate attribution analysis for extreme temperature trends, drought, and lastly Fire Weather Index. They don’t find any influence of anthropogenic climate change from drought. No surprise there then – drought is vastly more correlated to ENSO. But what’s interesting is that they choose the Canadian Fire Weather Index to do an attribution analysis, rather than the Australian McArthur Forest Fire Danger Index. Why I wonder? The FWI computes the fire risk from meteorological observations. The drought component consists of just a month’s worth of precipitation measurements immediately prior to the fires, whereas the FFDI is calculated using a drought index which goes back many months, even, I think, possibly years.

    The headline claim is that AGW increased the fire risk ‘at least 30%’, but it could be ‘much more’. That ‘fire risk’ is FWI, computed from meteorological observations, not observations plus a long term drought index. I wonder what result they would have got if they had used FFDI instead of FWI? My guess is that the authors chose FWI so they could separate out the influence of drought, which they knew could not be attributed to climate change and which might ‘muddy’ an analysis made using FFDI, which is, after all, the official Australian forest fire risk index. First impressions though. I might be wrong after reading further.


  3. Yes, it’s all a tad confusing for me at the moment. The FWI versus FFDI question is a very interesting one, as is the whole subject of conditioning upon sources of ignition and availability of combustibles. That’s why I spoke of ‘forest management counterfactuals’. I think the main takeaway at the moment is that their study, whilst purporting to be all about the Australian bushfires, is actually only answering the question ‘how much more likely are events like this in the light of anthropogenic climate change?’, rather than the more relevant question ‘to what extent was this event attributable to anthropogenic climate change?’ The first question can be answered by analysing climate models and exploring anthropogenic climate change counterfactuals. The latter question requires a much more sophisticated analysis involving a causal model that becomes very hairy very quickly. I guess they would say that they are determining the probability of necessity rather than sufficiency, but it’s as if they are choosing factors in order to minimise the importance of the absence of such a causal model. Maybe, as you suggest, the FFDI was not chosen because it carries too much causal baggage for their purposes.


  4. Having read the report more carefully, I am able to clear up a point of confusion on my part.

    Where the authors had written “We therefore investigate the question how anthropogenic climate change influences the chances of an intense bushfire season, rather than focussing upon a single episode of intense bushfires’ I had misinterpreted that as ‘bushfire seasons in general rather than a specific episode of intense bushfires’, e.g. Australia 2019. Instead, the point that they were trying to make is that they were looking at the bushfire season as a whole, rather than a specific/single outbreak within that season. That was an obvious point to make and so you could say that my misinterpretation had been perverse. The fact remains, however, that they are asking how much more likely bushfire seasons like 2019 are under anthropogenic global warming rather than asking to what extent the 2019 season could be attributed to it.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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