Nuclear sea breeze

Eli the climate bunny has gotten up my goat by suggesting that this study using fancy-pants shiny new climate models is the bees knees in nuclear winter prediction. Although I am not a climatologist, lemmings are nothing if not reckless individualistic animals willing to stick their necks out, so here I go. I reckon the paper is incomplete at best, and badly underconstrained at worst.

What they do:
-Get a newfangled climate model.
-Test the aerosol portion of the model using volcanic aerosols.
-Extrapolate volcanic aerosol performance to smoke
-Plug in smoke from nuclear winter

They do not test the model’s handling of smoke from real megafires, whether temperate, such as Canadian and Australian fires, or tropical, like the Brazilian or Indonesian fires. The result is that their model suggests particles can stay in the stratosphere for over 4 years, while actual fire research suggests that the residence time is 15-20 days (Ferrare et al. 1990)

The 3 order of magnitude difference between these results seems to be a result of the model lofting smoke particles into the stratosphere, a result seemingly at odds with observations of pyrocumulus clouds, which often show particulate stagnation at thermoclines penetrated by convecting water vapor.

Of course, their references suggest that megafires such as the 2003 Australian fire can directly inject smoke into the stratosphere. If their residence times are correct, that smoke should still be aloft. Evidence of its existence is not provided.

The mean annual fuel load of Australian wildfires is equal to their smoke injection mass (assuming 1 Mt = 1 Tg). Yet chronic catastrophic global cooling is not observed. As for the applicability of forest fire studies to urban fires, those of us in Canberra have recently conducted that experiment, finding the western suburbs to be slightly less flammable that the surrounding bush.

In short, the paper extrapolates volcanic aerosol response performance to smoke without performing any of the obvious tests of this extrapolation, then builds a model result on top of that unconfirmed hypothesis. While this may be dinky-di science in the bunny labs, here in the lemmingopolis, it is the research equivalent of playing fantasy computer games instead of doing homework.