Predicting how Earth's climate will change in the coming years is a deeply important task for science. It also seems fairly fundamental -- they're just clouds! But cloud dynamics are incredibly problematic, to the point of being unknowable in some instances.
The Verge has a very cool feature today about the problem. Essentially, it's impossible to predict future climate changes without knowing how cloud behaviour will change. And that's especially important in places like the tropics, where a chance in rainfall patterns could prove disastrous.
The trouble with making those predictions basically breaks down to an issue of scale. The models used by the general circulation model used by the UN is far too vast to be used for basic cloud dynamics. Those only simulate atmospheric phenomena larger than 100km, while most clouds are no bigger than 1km. Problem!
The solution is smaller models (obviously). Some track various types of clouds' movement from satellites, while others use supercomputers to crunch numbers on various, hellish conditions input by the scientists. It's all pretty complicated. Check out the rest of the details over at The Verge.