Wednesday, October 24, 2007

Wildfires and statistics

It is awful to watch the disaster unfolding in southern California, where wildfires are threatening major cities and nearly 1 million people have been forced from their homes. It is my sincere hope that the fires quickly abate, and that people are able to work together for a fast recovery. (And, at the same time, maybe we should re-double our efforts to speed the recovery of the Gulf Coast, which is still sputtering along after the devastation from hurricanes Katrina and Rita.)

Unfortunately, disasters such as wildfires, strong hurricanes, heat waves will continue to strike new areas as climate change due to global warming continues. Due to global warming, some areas of the world will feel improved weather, such as warmer winters in Canada, increased rain in currently try climates, and other such "nice" things. But, weather will worsen in other areas. Normally wet areas will dry out, and in the transition, wildfires will occur.

So, then, are the California wildfires a direct cause of global warming? I don't know, and anyone who claims to know is probably wrong. The reason is that extreme weather, such as the droughts in southern California over the last year, or strong hurricanes in the Atlantic, have occurred before. Even without global warming, such weather would eventually occur again. Climate change theory predicts that extreme weather will become more common, but it cannot tell us whether a specific event was due to global warming.

Let's look at another example. Suppose I were a casino owner with a craps table (a dice game). In a normal craps game, getting "snake eyes" (or a "one" on each of the two dice) is typically a very bad roll for the gambler (and therefore good for the casino owner). With a fair set of dice, snake eyes will appear on average once every 36 rolls of the dice. It could be that snake eyes will come up twice in a row, or even a hundred times in a row, but if I were to roll the dice thousands of times, on average snake eyes would appear once in every 36 rolls.

Now, suppose I am a bit of a cheat, and I replace the dice with loaded dice (dice with tiny weights in them) such that snake eyes appear once every 24 rolls. It's still rare enough that most people wouldn't notice the change, but, over time, the casino would win much more often than in a fair game. Now, suppose a gambler rolls snake eyes. Is this one set of snake eyes due to the loaded dice? Maybe, maybe not. It may be that, with fair dice, she would have rolled snake eyes anyway. Then she rolls snake eyes on a second roll. Again, is it due to the loaded dice?

Again, we could not be sure. Such is the funny nature of statistics. With completely fair dice, it is possible to roll snake eyes four times in a row --- it's a one-in-a-million chance, but remember that millions of people play craps each year! The next gambler may roll 100 times with loaded dice and not see snake eyes (a 1 in 100 chance with my loaded dice). It would only be after many, many games that we could figure out whether or not the dice were fair or loaded.

So, it is the same way with strange weather. It may be that, even in the absence of global warming, Katrina and Rita would have both hit the Gulf Coast in the same year, and that a drought would hit California this year. It is only over time that we can build up enough statistics to say that more storms and droughts are occurring. Likewise, a year without a strong hurricane doesn't mean that the danger has passed (ask the folks in Central America whose homes were destroyed by hurricane Felix last month). Even a string of lucky years doesn't mean the danger hasn't increased.

What's the take-home point of my ramblings? It is that you cannot say whether one specific weather event is due to global warming or not. What we can say is that, if we do not change the human impact on the environment, strong hurricanes and severe droughts will become more common in places that rarely had trouble before.

No comments:

Post a Comment