Ultra-Long Range Weather Forecasting
If our great-grandchildren are still around a century from now, they will almost certainly be living in a world with different climates from what we know today. Carbon dioxide, methane, and more complex chemicals are building up in our atmosphere, and given the economic realities of life on an overcrowded planet, they will continue to build up at least into the 21st century. While there is plenty of room for doubt that the heat and drought of the summer now coming to an end were caused by these changes, it is almost certain that the changing chemistry of the atmosphere will produce clime changes of this magnitude or greater, probably within the next few decades.
The big unanswered question is what these changes will be. It is fairly simple to work out the direct changes in radiation and say that the coldest clear-night winter temperatures will be a little warmer next century, or that late spring and early fall radiation frosts will be slightly less common. But most of the variability of our weather is due to the variable horizontal and vertical movement of air and water vapor. This movement is driven by the difference in the solar and thermal infrared radiation between various places on the earth, so if changes in the atmosphere change the radiation, the air movements will change as well. And if air movement changes, so will cloud cover, rain, and snow, which in turn will produce further changes in the radiation distribution and thus more changes in air movements. The whole thing is called a feedback process, and we don't understand it very well.
One tool for trying to understand what will happen is climate models. These models are made up of equations which try to reproduce the way the atmosphere behaves, and which are evaluated by computers -- generally by the fastest computers available. Even on these computers, the best and most complete models would take days of computer time (if not months) to run enough years to get any climate information, so the models used to simulate climate are much simpler than those used for daily weather forecasts.
In order to test the models, they are sometimes used to "hindcast" climates we know occurred in the past, such as the ice age climate. The ice age climate in Alaska, although not fully understood, is intriguing in that large parts of the state remained ice-free and apparently quite dry. Because of this, and because Alaska is often warm when the central part of North America is cold, I have felt for a long time that Alaska may be a key region for model testing. But the models cannot include enough detail that the precipitation and temperature changes they forecast can include the effects of the Alaska and Brooks ranges. We know these ranges are now, and almost certainly have been in the past, very important in the distribution of temperature and precipitation in our state.
I'm going to be spending the next year visiting climate modelers and trying to work out a method of using the large-scale air motions predicted by the models to estimate more realistic values for temperatures and precipitation in Alaska than the models can compute directly. While I may be able to send in an occasional article for this column, I cannot be the lead author from several thousand miles away. So this article is good-bye for a while. I've enjoyed writing the Science Forum, and I may be back a year from now, but this winter you'll be hearing from other scientists.
