When it comes to weather models it's probably a good idea to have at least two of them for validation and performance comparisons. These things are (a) quite co

When it comes to weather models it's probably a good idea to have at least two of them for validation and performance comparisons. These things are (a) quite complex (in the order of 100k-1M LOC), (b) important for humanity for various reasons and (c) not as expensive as you might think. To give you an idea, for the price of one F-22 fighter jet one could go and build the world's best weather prediction, including the price of the supercomputer (as in the world's fastest one) as well as the initial software development.