Bait and Switch

A new post at WUWT claims to “debunk” claims that “human caused warming doubled western U.S. area burned since 1984.” It was in response to a story in the LA Times of new research in PNAS (Publications of the National Academy of Science).

The LA Times article, like the PNAS research, is about area burned in the western continental U.S. But the WUWT post includes some bizarre statements, such as this:

The new study considered two time periods for comparing burned area acres and those were from 1984-1999 and from 2000-2015.

The number of U.S. wildfires for the period 1984 to 1999 is essentially unchanged from the number of U.S. wildfires for the period 2000 –2015.


Since the research is about area burned, why are they talking about number of wildfires?

It’s easy to figure out why: because if you look at number of fires there’s no real difference between the two time spans, but if you look at area burned the difference is obvious:



Attempting to “debunk” an increase in area burned by pointing to number of wildfires, is a classic example of changing the subject. It’s one of the most pervasive of all denier techniques.

The WUWT article goes on to say:

Additionally the latest year to date wildfire data ( for the last ten years shows absolutely no consistent upward trend whatsoever in either burned acres or number of U.S. wildfires which have occurred.


This is even more bizarre; the WUWT article points out that they compare the 1984-1999 time span to the 2000-2015 interval, but now they want to talk about the lack of trend in year-to-date values only, using the 2006-2016 span only? That’s only 11 years’ worth, which will inflate the uncertainty in a trend estimate immensely. Of course, there’s no actual analysis here so there’s no mention of how uncertain trend estimates are.

Using full-year figures for 2006-2015 the estimated trend is negative, at -165000 acres per year. But it’s not statistically significant (no surprise, since there’s only 11 years’ data); the uncertainty level (95% confidence) is +/- 629000 acres/yr.

In fact the whole idea of trend-analyzing a mere 11 years in isolation not only relies on a ridiculously small amount of data, it also ignores what came before their chosen starting point. It’s tantamount to using a “broken” trend, one in which what happened before 2006 isn’t allowed to inform what the trend really is. The model is therefore like the blue line shown here:


It makes no sense to call that the “trend,” unless you want to ignore what happened beforehand because it tells a different story.

We can even compute the uncertainty in the trend model for both time intervals and add them to the graph as dashed lines, here:


When the uncertainty range (shown by the dashed lines) is included, we can see that a single, unbroken trend using the entire span 1984-2015 is perfectly consistent, not just with the whole interval, but with the analysis of the separate intervals:


Criticizing a comparison of 1984-1999 to 2000-2015 by discussing a tremendously uncertain trend estimate using only data since 2006, they’re once again indulging in the beloved denier practice of changing the subject. Because, after all, they can’t deal with the subject.

But wait, there’s more! They add this:

The new wildfire study also fails to address that U.S. drought data does not support claims of increased nationwide droughts as driving increased number of wildfires.


It’s about wildfire, not drought … the funny thing (hilarious, actually) is that neither the LA Times article nor the abstract of the PNAS research mentions the word “drought.” Which makes it yet another case of changing the subject by those who can’t deal with the subject.

And by the way, where are the claims of “nationwide” droughts, or even “nationwide” anything, in the articles they’re attempting to “debunk.” It’s all about the western continental U.S., so including the entire 48 states means that regions with decreasing drought (in the eastern U.S.) cancel out regions with increasing drought (western U.S.). And that’s why the WUWT article does so: to cancel out drought increase in western U.S. regions by including regions the article is not about.

It turns out that U.S. drought data does support claims of increased drought in the west and southwest:



The whole post is an immense bait-and-switch. The subject was supposed to be area burned by wildfire in the western U.S., comparing the 1984-1999 time span to 2000-2015. Instead, we’re “treated” to misdirection about number of fires, a “trend” ignoring everything before 2006, and drought (not the subject) which has to ignore the “western” part of “western U.S.” to avoid giving away the game.

I guess they just can’t deal with the subject.

This blog is made possible by readers like you; join others by donating at Peaseblossom’s Closet.


10 responses to “Bait and Switch

  1. “Proceedings” of the National Academy of Sciences. Unrelated- saw a similar awful graph on other topic ( and thought of you. Scroll down to the “Longevity Limited” figure and you see similar shenanigans with trends. Oy!

  2. On the other thing- this is someone’s response- comparing the crazy change/trend model with simple linear model:

  3. Looks to me like the WUWT guest missed the main point. Raised in the comments but without any critique of the post.

    I was thinking to waste some time over there referring them to this and ripping them a new one but….

    How does the actual work incorporate the changes in forest management, development of fuel loads and long term changes to fire suppression since the end of WWII when we heard “Only you can prevent forest fires”.

    That’s actually a real thing. Makes it hard to figure what the effect of the warming was/is independent of it. I think yes, the areas are increasing and the situation is getting worse, but I have to read the actual paper and see what they actually say to form an opinion. You are right about the original post on WUWT. It is rubbish.

    Several of the responses though, have pulled in the confounding variable of direct human interference with the forests and the fires though, and that definitely confuses the issue and needs sorting…

    My take on CA is that here is a longer term trend besides what we do with fire suppression. The droughts were longer and worse a few thousand years ago, closer to the climate optimum, and as we shoot past that we’re going to see those again, but we’re changing too fast. It is becoming a blur.

    Good post – the question is now – whether it is worth finishing the job.

    {{ Your sisyphean task is to forever sort the sheep from the dip :-) }}

  4. Yet another example of cherrypick-a-local-max-then-note-“hiatus”-in-succeeding-few-years. Do deniers really think they fool anyone?

  5. Excellent, Tamino. Don’t ever stop!

  6. The formal name of the fallacy is “straw man.”

  7. Oddly enough, it wouldn’t surprise me if, at least in some areas, increased area of wildfires was caused by one of their favourite things – CO2 being plant food. That is, the growth of vegetation in areas where there is sufficient water has been enhanced by the elevated CO2 levels, and hence the fires are fiercer and harder to stop.

  8. If everything was burning, there would only be 1 single fire burning.

    Maybe one should tell the fire-brigades: There would be fewer fires, if they just lit up all the space between the different fires.

  9. OT note here but relevant to denialland distortions. Arctic ice has taken a sharp turn downward, a sickening lurch:

    So much for the hyperclaim campaign of “recovery” not long ago. Shameless, as always.

    Back to fires, there has been some really awful news from Siberia and the Amazon basin. Here’s one link:

    These guys are a bit strong, but there’s a lot of factual material in there; heard it while traveling a few days ago:

  10. Even assuming no change in precipitation regimes, warmer = dryer, because the rate of evapotranspiration is a function of temperature. The PNAS paper abundantly documents warming across the Western US. Ceteris paribus, it would be astonishing if AGW had not increased forest fire activity in the region.