The case for regime change
Some thoughts on wildfire change detection
Take j, k, and l, all constants. Let k = change. Now let j and l vary and we must concede that the only constant is…
Life is full of change. There’s the whole ageing thing, if we’re lucky, from baby to toddler to kid to tween and so on and so forth. We change clothes, change our diet and our hairstyle, our jobs. There’s technological change, social change, political change (or at least pivots). We change our mind. In science we change our ideas about the way the world is, based on careful evaluation of the weight of evidence and argument.
Sorry, I almost dropped the laptop in a fit of laughter.
Change is so absolutely ubiquitous that even after going through it all we still have spare change leftover.
The temperature record over the last 100 years provides stark evidence of change, beautifully illustrated in plots like the one shown above by Ed Hawkins. Looking at the whole plot, the change is unmistakeable. But what if your data doesn’t go back that far? Would you say temperature has changed in the last three years? The last seven? What if we time travel back to around 1935? There was a huge spike in temperature that may have led some to believe (temporarily erroneously, perhaps) that temperatures had changed. How can we tell?
To tell if there is change, we need a baseline. We must define normal, regular, typical, so that we can detect a departure from it. By convention, meteorologists often use a 30 year period to define normal weather. This means that you pick a continuous group of 30 bars from the plot above - from 1961 to 1990 for example - and use that as your measuring stick for any of the other years.
The first thing you will notice is that there is change within your 30 year baseline! Any 30 year period will contain quite a bit of year to year variability in temperatures. But we can take the mean, and maybe some other statistics like median, minimum, maximum and so on and we now have some handy reference points when new data rolls in. When we say that last year’s temperature was below average, we mean it was below the average of our 30 year baseline period. More likely, when we say that last year was unusually extreme, we mean that the very hottest days last year tended to be hotter than the very hottest days over that 30 year baseline period.
As you may have guessed, having a baseline comes in handy when talking about climate change. When you hear people talking about 1.5 degrees and 2 degrees of warming occurring under certain emissions scenarios, they are referring to an increase compared to a baseline period. You might like to ask your friendly neighbourhood climate scientist what this period is the next time you hear such talk.
So what about when we talk about future fire? If we say that fire will change (as it surely will) then what is it changing from? To answer that question we’ll need a baseline! What would a suitable baseline for fire be - 30 years? Maybe a bit longer?
But that brings up an even thornier question. What the blazes do we even mean by fire? Luckily, we have a few great starting places. The first is the fire regime. Allow me to quote from a rookie scientist:
The [fire regime] concept has evolved since its introduction and now generally includes the prevailing timing (frequency and seasonality), size, severity and type (ground, surface, crown) of fires at a given location. Fire regimes vary greatly among ecosystems; it has been said that there are no fire adapted species per se, rather there are species adapted to specific fire regimes (Pausas and Keeley 2009).
So a fire regime is what counts for normal in a particular fire prone landscape. If we want to check whether fire has changed, or is changing, or will change, we can just establish a baseline for the different parameters of the fire regime:
fire frequency;
fire seasonality;
fire size;
fire severity and
fire type,
and check whether we are outside the envelope of this baseline. As it turns out, this is quite hard to do! We have a good idea about many of these regime facets, but surprisingly little hard data, particularly at the kinds of spatial and temporal resolution that climatologists are used to. I’m not sure we even have agreement about what a suitable baseline period is. It would depend on the fire regime.
The passage above goes on to note that “variation in contemporary fire regimes can be traced to a large extent to variation in four drivers of bushfire incidence”. These are the four switches, which we covered in a previous post. They’d also make very handy variables for which to set up a baseline and check whether there’s been any change. This has been done a fair bit for fire weather (even I’ve had a crack), but not so much for ignitions, fuel moisture or fuel load.
I have no intention of stopping my efforts to predict the future of fire and how we will thrive alongside it, but at the same time, there is plenty of important work to do in characterising fire as it stands today and building up that all important baseline.
Nice figure
Ok, the figure’s terrible, but the data is something else. I have been looking at the weather conditions during Black Summer and this plot from Inverell Research Station jumped out. Over the last 20 years or so there have been a handful of days with both an extremely high maximum fire danger rating and an extremely high minimum fire danger rating. The 2019-20 season blew the previous years out of the water.