Nuclear War and Nuclear Winter

by Brian Dunning
February 8, 2011

Put on your jackets; today we're going to prepare for the nuclear winter, the theorized period of catastrophic global cooling following a nuclear war. Some say that, if war happens, a nuclear winter is a certainty that will devastate agriculture and kill billions; some say that it's greatly overstated or even outright made-up by anti-nuke activists. Does the prospect of a nuclear winter truly constitute one more reason why nuclear arsenals should be dismantled?

Our purpose today is not to explore the myriad other implications of nuclear war, or to otherwise prove that it's a bad thing. That's pretty clear. The obvious effects of a nuclear blast, the initial explosion itself and the radioactive fallout, are well established and not in dispute. As destructive as those are, their long-term environmental effects are negligible. So what, then, causes the nuclear winter?

The concept of a nuclear winter entered the mainstream in 1983, when the "TTAPS" team, named for its authors Richard Turco, Owen Toon, Thomas Ackerman, James Pollack, and Carl Sagan, reviewed existing work and ran computational climate simulations to see what would happen when huge amounts of smoke were added to the atmosphere. The source of this smoke is not the nuclear explosion itself, but all the building fires and wildfires that would follow each one. Take 150 nukes striking major population centers worldwide, and that's a lot of fires, with a smoke output greatly exceeding anything in human history. Members of the TTAPS team have also published many followup papers, revising and improving their estimates, but generally with similar results.

Almost all of the simulations run by scientists replicating these results agree, at least in broad strokes, when the input variables are the same. It should be stressed that there are a lot of these variables. The principal weakness that these studies all share hinges on one of these variables, and it's a very important one: Exactly how much smoke will the fire following a nuclear bomb produce? In 1986, Joyce Penner from the Lawrence Livermore National Laboratory published an article in the journal Nature in which she pointed out that this specific variable is responsible for determining whether the effects will be minor or massive. She also found that the published estimates of this varied widely.

Penner's paper was not the only one critical of the TTAPS predictions. However, this fact is often misinterpreted. The differences that have been found among the various simulations are of degree. Laypeople who hear of the criticism often think that the idea of a nuclear winter has been "debunked" or that it's some kind of discredited myth. This is not the case. Perhaps the most often cited and most critical paper was "Nuclear Winter Reappraised" by Starley Thompson and Stephen Schneider, published in 1986 in the journal Foreign Affairs. They did not dismiss the idea at all; rather they recharacterized it as a nuclear autumn.

So here we get into the meat of the question. We know with pretty good certainty how a given amount of smoke in the atmosphere, distributed a certain way, will affect the climate and for how long; but what we can only guess at is how much smoke is produced when a city burns after a nuke. Our guesses are educated, but they're all over the map. Cities also vary wildly in just about every relevant aspect. Let's look at what we know from history.

An obvious question to ask is whether these effects have been seen with any of the nuclear tests that many nations have conducted. Some 2,000 nuclear bombs have been detonated, somewhat less than half of which were in the atmosphere and are comparable to what would be used in a war. In none of them were any harmful smoke-induced environmental effects produced. However the reason for this is quite simple. Nuclear tests are not performed in cities filled with tens of thousands of combustible buildings; they happen way out in the desert or over the ocean, and no subsequent fires are created.

But what about the two cases when atomic weapons were used on real-life cities, Hiroshima and Nagasaki? Discussion of the subsequent fires in both cities are hard to come by, as they were not really what people were focusing on. Hiroshima developed a firestorm — where it builds into a single large fire with a central heat core that draws in oxygen with a powerful wind from all around — that peaked two to three hours after the explosion. Six hours after the explosion, nearly everything combustible within a one-and-a-half kilometer radius had been consumed, and the fire was almost completely out, leaving over 8 square kilometers destroyed. Descriptions of residual and secondary fires outside the radius of the firestorm are rare and hard to find, but it seems likely that several hundred or thousand small fires continued for the better part of 24 hours. Photographs taken of Hiroshima over the next few days do not show any significant evidence of vast amounts of smoke.

Nagasaki was hit with a larger bomb, but its geography spared it a firestorm. Whereas Hiroshima is centered in a large flat plain, Nagasaki is irregularly shaped among hills and valleys, and cleft by a large harbor. Secondary fires were widespread, and Nagasaki firefighters had to cope with a damaged water system. It took several days to get the many small structure fires controlled or burned out. But Nagasaki's geography meant that there were far fewer fires than in Hiroshima. Again, the post-nuke photographs don't show vast atmospheric plumes of smoke.

When the Iraqi army set 700 of Kuwait's oil wells on fire when they retreated in 1991, the wells burned for eight months, lofting about a million tons of smoke into the atmosphere. The TTAPS team predicted global climate change effects, that fortunately failed to materialize. Carl Sagan discussed this error in his book The Demon-Haunted World, and later research discovered the reason. The smaller individual smoke plumes, spread over a wide area, did not generate sufficient uplift to get the smoke into the upper atmosphere, even though theoretically enough smoke was produced. Temperatures did drop over the Persian Gulf, but the effect remained localized.

Other cataclysmic events have proven that the nuclear winter scenario is not at all far-fetched. The eruption of Mt. Pinatubo in the Philippines, also in 1991, threw some 17 million tons of particulates into the upper atmosphere that caused global temperatures to drop by about a degree for several months. Sunlight dropped by 10%. This temperature drop did not, however, have any long-term effect on agriculture.

Pinatubo was only a blip compared the the K-Pg extinction event of some 66 million years ago, when a theorized asteroid hit us with one hundred million megatons of destructive force, lighting virtually the entire world on fire. The evidence of this is called the K-Pg boundary, a layer of clay found all around the world. Sunlight was reduced by 10-20% for ten years, which caused a massive cascading extinction of species from plants to herbivores to carnivores.

But we shouldn't expect anything like this to happen from a nuclear war. Times continue to change, including the nature of warfare. Nations no longer stockpile the megaton class weapons popular in the 1950s and 1960s; typical yields now are a fraction of a megaton. The United States' conventional capability is now so good that it can effectively destroy an entire nation's ability to wage large-scale war overnight, using only conventional weapons. But that doesn't mean the nuclear forces are no longer needed. Should a superpower strike first against the United States with nuclear weapons, the response would more than likely be nuclear, bringing Mutually Assured Destruction into play. But what about a small nation striking first? What about nukes in the trunks of cars parked in major cities? In the modern era, it's much less clear that any superpower would necessarily have anyone to shoot back at.

Increasingly, non-superpower nations are building nuclear stockpiles. India and Pakistan might get into it with one another. Israel's foes might surprise it with nuclear weapons. Who knows what North Korea and Iran might do. Smaller regional nuclear wars remain a very real possibility. According to the worst-case estimates in the TTAPS papers, about one million tons of smoke would be expected from the fires resulting from each nuclear strike. And these smaller regional nuclear combats are expected to use about 50 nuclear weapons (compare this to 150 nuclear weapons for a broader global nuclear war). Thus, today's most likely nuclear scenario would be expected to produce climate effects similar to three Pinatubo events, according to the worst estimates, and still many orders of magnitude less than the K-Pg extinction.

And so, while the nuclear winter scenario is a good prediction of the effects of a worst-case scenario, when all the variables are at their least favorable, the strongest probabilities favor a much less catastrophic nuclear autumn; and even those effects depend strongly on variables like whether the war happens during the growing season. A bomb in Los Angeles might result in history's worst firestorm, while a bomb in the mountains of Pakistan might create no fires at all. The simple fact is that there are too many unpredictable variables to know what kind of climate effects the smoke following nuclear fires will produce, until it actually happens. Obviously we're all very mindful of the many terrible implications of nuclear combat, and if it ever happens, the prospect of a nuclear autumn will likely be among the least of our concerns. The physicist Freeman Dyson perhaps described it best when he said "(TTAPS is) an absolutely atrocious piece of science, but I quite despair of setting the public record straight... Who wants to be accused of being in favor of nuclear war?"

Correction: An earlier version of this referred to the K-T boundary of 65 million years ago. Since the International Commission of Stratigraphy split the Tertiary period into the Paleogene and Neogene periods, this formation is now called the K-Pg boundary, and the specific age of 66 million years is more accurate.

By Brian Dunning
Follow @BrianDunning