This thread is interesting. What are the real prospects for real time attribution of extreme climate events?
First thing to realize is that it depends on the kind of event. Heat waves, intense rainfall, fires, storms and so on have different dynamics (spatial scales, necessary background conditions etc).
Second point is that to attribute extremes in the way we have been doing (i.e. is there a greater chance of it happening now than before we affected the climate?), we need big data in order to estimate the likelihood of these rare events - tails of the distributions.
This needs a lot of modeling - statistical emulators or physics-based models, incl. climate models etc. But if you want real-time attribution that modeling has to exist already (since you don’t have time to do a bespoke study).
Which means you are limited in terms of the class of event you can look at to those that are definable and extractable from the database you have already. If that data set didn’t resolve fires, or hurricanes, then tough - you aren’t going to be able to use it to say much.
We have lots of modeling w/ & w/o anthropogenic climate change in #CMIP (w/more to come), and extremes like regional heat waves, cold snaps & precipitation intensity are well resolved, credible & evaluatable against the historic record. Not so good for storms, tornadoes or fires.
But there are now many bespoke studies done after the fact for many specific extremes - flooding in the UK, European heat waves in 2003 & 2010, downpours during Harvey, Katrina etc and they show a clear pattern: the closer an event is to a thermodynamic process....
... the more likely you can find a anthropogenic signal. So heat waves are more likely, cold snaps less likely & precipitation more intense. These results are now so ubiquitous you can skillfully predict ahead of time what the general attribution will be (tho not the exact frac).
For events that more dynamic in nature - depending more fundamentally on the weather regime or synoptic pattern - ice storms, derechos, tornado swarms, etc, it’s more ambiguous & both model credibility & signal to noise will be less. So real time attribution is less likely.
Finally, fires. This is tricky - not because of the anthropogenic part - deliberate arson, accidental human disturbances and suppression and climate (drying, lightning changes) are all human related of course. But we want to parse out which are causal.
To conclude. Real time attribution of increasing heat waves to human climate change is solid. This has been predicted, and predictions are playing out. Same with reductions in cold snaps and increasing intensity of precipitation. Everything else (IMO) is a work in progress.
But maybe I’ve missed something? Feel free to let me know.
