There are several questions dealing with global warming that come up frequently...and it happened again this week when I gave a lecture at Google in Kirkland.
"How can you meteorologists possibly predict global warming fifty years from now when you have problems forecasting the weather three days ahead?"
(Some questioners use more earthy language to describe the current state of meteorological prediction skill, while others suggest that weather forecasters use dice for all the skill they have).
Well, we CAN predict what is going to happen fifty years from now and the forecasts are worth listening to. How can that be?
First, let me say that weather forecasts have gotten much more accurate during the past decade or so. And that is due to several reasons: far better computer forecast models (better understanding of physical processes, better resolution due to more powerful computers) and FAR better and much more observations that allow us to start our computer models with a better starting description of the current state of the atmosphere. We rarely get big storm forecasts wrong today and the day 4 forecasts has the skill of the day 2 forecast 25 years ago. Major progress. And these are basically the same computer models used to predict climate.
But forecast skill of even our new models degrades with time and by 1-2 weeks ago there is little forecast skill left. So knowing that, how can we forecast 50 years from now what will be happening with increasing greenhouse gases?
First, a fifty year climate forecast is completely different than a 2 day weather prediction. In a weather prediction we attempt to forecast the exact configuration of the atmosphere at specific times. Exactly what the temp will be at Olympia at noon, the strengths of all the highs and lows and their exact positions at 7PM Thursday, the wind conditions everywhere at exact times. Pretty challenging. And we know based on theoretical studies that our ability to do this degrades rapidly in time. But for climate predictions we don't do this. We forecast averages..like the average temp over 10 years or average precipitation for all the Marches for a decade. This is MUCH easier to do.
Second, the nature of a climate forecast is very different. For weather forecasts our predictions are dependent on the details of the initial state and small errors or differences grow during the forecast. For climate predictions averaged over decades, there is far less sensitivity to the initial state and far more dependence on the overall forcing by the sun's radiation and how the composition of the atmosphere changes its ability to absorb and transfer infrared radiation. For example, if the earth's atmosphere acts to hold in more radiation, than the planet will tend to warm. Technically, one is an initial value problem and other a boundary condition problem. We have run our climate models over the past hundred years and have been able to duplicate the climate variations over decadal time scales...providing confidence in climate forecasts.
The bottom line of all this is that useful climate predictions ARE possible even when weather prediction skill is lost in a few weeks. These climate prediction aren't and won't be perfect. But they provide great insights into what will happen over the next 50-100 years.
My next blog will deal with the other major argument used by those questioning the potential for human-induced global warming..the lack of warming over the past decade. And then I will talk about the hyping of global warming by some scientists and the media.