Sources of Uncertainty

Confidence in simulations of present climate by models

Weather models and climate models have much in common, including tendencies to err for the same reasons. What are the main reasons that a climate model may represent the current climate imperfectly?

1) A model represents the world with data at discrete points at discrete times. Whatever really exists between these points is simply ignored. Phenomena that happen between these times are overlooked. So climate patterns on small scales (say, in the Willamette valley) are not depicted by global models which have widely separated grid points.

2) Some processes must be omitted in models. Some of these are “unknown unknowns”: we do not yet know that such-and-such is important for climate, so it is not represented in the model.

3) Models differ among themselves on “climate sensitivity” – the amount of warming that eventually takes place because a certain “forcing” changes. If the concentration of CO2 in the atmosphere is doubled over 19th Century levels, how much will the planet warm (over the long run)? Different climate models give different results. This remains one of the greatest challenges in modeling of climate.

(However,  no climate model predicts that global temperatures would remain stable or would cool in this situation.)

4) Models differ in how they represent climate “feedbacks.” This is perhaps the most important reason why models have different values of climate sensitivity. Two feedbacks merit attention here:

  • Cloud feedback:

    Low clouds reflect most of the sun’s energy back to space; the reflected energy does not warm the planet and is “lost” to the climate system. Many things influence the formation of clouds, and any one of them may influence climate in important ways. Current models struggle to simulate the correct frequency of occurrence of low clouds.

  • Ice / Snow feedback:

    Fresh snow reflects about 90% of the incoming sunlight. When snow is replaced by bare land, or sea ice melts and is replaced by dark sea water, most of the incoming solar energy is then absorbed.  Energy that is absorbed, of course, warms the planet. Modeling of climate in polar regions is very sensitive to the distribution of ice and snow.

Confidence in modeling future climates

Our knowledge of the future is arguably imperfect. The future concentrations of greenhouse gases depend both on past releases of these gases (which we know fairly well) and on future releases (which we do not know). Future emissions of greenhouse gases depend on the very policies and incentives that Congress and the world leaders in Copenhagen have been debating. About one-half of the future uncertainty in temperature comes from our uncertainty in the future emission of CO2 .