What Improvements do Models Need to be More Accurate?
Computer and data storage requirements

Before we can make computer models more complex, we need faster computers to run the code on.

Global climate models have to run on supercomputers, which are made up of many thousands of individual computers. Even on very sophisticated computers, it can take months for a GCM to compute a century's worth of weather and climate for the world.

Typically, researchers will develop models or a modify a model on a personal computer. They may even be able to run short test runs with it on a small machine. But as soon as they run a "production" run on it, which could be from a model "month" to many years in length, a supercomputer is required. Grants are written to obtain research time at these supercomputing centers.

As the models run, they output a lot of information that requires large storage capacities. For example, at each time step at each point on the globe, the model outputs information about the temperature, humidity, wind speed and direction, amount of incoming solar radiation, and precipitation rate, just to name a few. Researchers can tell the model what data they want to output and how often to output it but even so, this can add up to an enormous amount of data.

The cutting edge of climate modeling includes models that run within a model like Cloud Resolving Models. So Atmospheric Scientists are pushing current supercomputers to extremes and are some of the first to utilize faster machines as they come out.

Next page   -> improvements, continued

Links and resources