[chriswatt]

Hi,

Although I am experienced with CFD and numerical modelling of fluid flow, I have some difficulties to solve the following question :

My problem is the following : I wonder down to which time resolution, the solution provided by a numerical weather prediction (NWP) model can be considerred to be realistic and physical, providing the typical hypotheses involved within operational NWP models. For example, let say I can measure the wind at a given place. On a minute to minute basis, I can observe the wind to vary. These variations come from turbulence (mechanical and thermal) and from the large-scale weather motion. Of course, in a numerical solution, variations of the wind component also exist, but what is the limit down wich these variations are purely numerical, and have nothing to see with reality ?

What would be the limit in time below which timestepped solutions should not be kept, and only time averages should be considerred instead ?

The typical hypotheses used in NWP models are :

- Treatment of small scale turbulence by turbulence closure of zero-order (mixing-length type treatments)
- Treatment of larger scale turbulence id. LES models ; NWP models respect CFL conditions. For example, a 15 km resolution model will use a Dt of 45s.

I imagine the response is a function of the spatial and time resolution of the NWP model, and probably of local wind velocity... but I can not figure it out on my own, since most of the problems I have studied in the past are steady over time !!! any advice or thought on the subject is very welcome !

Chris