Hello everyone,
I would like to perform simulations using different vertical mixing schemes, but without activating any convection parameterisations (such as EVD or MFC). I turned off all convection parameterisations and tested two simulations using the TKE and GLS schemes, respectively. In both cases, the runs crashed after about two months due to unrealistic salinity values developing at around 150 m depth.
I noticed in the NEMO v5.0 manual that it is recommended to always activate convection when using these vertical mixing schemes:
The TKE and GLS turbulent closure schemes presented in subsection 11.1.4 and subsection 11.1.5 are, in theory, able to handle statically unstable density profiles. [….] However, in the vicinity of the sea surface (first ocean layer), the eddy coefficients computed by the turbulent closure scheme do not usually exceed 10−2 ms−1, because the mixing length scale is bounded by the distance to the sea surface. When using either the TKE ( ln_zdftke=.true. ) or GLS ( ln_zdfgls=.true. ) turbulent closure scheme, it is therefore recommended to also enable the enhanced vertical diffusion parameterisation ( ln_zdfevd=.true.).
In theory, both GLS and TKE should be able to handle convective processes, as the buoyancy term is included in their equations. Why, then, is it necessary to explicitly activate a convection parameterisation? Is this requirement due to numerical instabilities?
Thanks!
Best regards,
Lucia Gualtieri