Ah, I see. Makes me ponder the potential future of NWP. I wonder how much improvement is feasible before they start running up against the "limits of chaos".
Also, even if they can improve the equations and throw more supercomputer power at it, there is also the factor of the actual input data in terms of resolution and accuracy. Although I have heard that improving weather satellites can help a lot there, since apparently a lot of the assimilated data comes from satellites these days.