Nice! The terrain issue that you mention is the limitation of the grid resolution. For example, HRRR, the operational model used in the charts in your link, has

Nice! The terrain issue that you mention is the limitation of the grid resolution. For example, HRRR, the operational model used in the charts in your link, has 3 km horizontal spacing. Individual peaks and slopes are poorly resolved at that resolution. We are already (internally) experimenting with 1 km and sub-1-km grid resolutions using HRRR as initial and boundary conditions. Creating custom forecast domains at fine resolutions is something that we are currently developing and should be ready in the next 2-3 months. Thank you very much for your input, it's super helpful!