Wrong automatic definition of the laplacian coefficient
Context
The viscosity and diffusivity coefficients that are produced are incorrect when you activate:
-
ln_traldf_lap = .true.
and use 31 fornn_aht_ijk_t
-
ln_dynldf_lap = .true.
and use 31 fornn_ahm_ijk_t
The coefficients should be true to what is written in the commentary, namely it should be ahm = 1/2 Uv*Lv
instead of ahm = 1/12 Uv*Lv
which is what is computed for momentum. Similarly, for tracers, aht
should be 1/2 Ud*Ld
instead of 1/12 Ud*Ld
.
-
Branches impacted: current release and main -
No reference configurations or test cases are currently impacted by this bug -
Dependencies: Independent
Analysis
The laplacian coefficient will be 6 times smaller than expected. This will result in excessive diffusion of tracers and/or momentum in your configuration.
Fix
Replace 1/12 by 1/2 for the case "31" in ldftra.F90 and ldfdyn.F90