Which dynamical core fields for overwriting Snr->Jnr?

The dynamical core fields are

  • u, v, w: velocity components
  • θvd: virtual dry potential temperature
  • Dη/Dt: Lagrangian derivative of vertical coordinate with time
  • ρd: dry density
  • π: Exner function (related to pressure)
  • mv, mcl and mcf and optionally mcf2, mr and mgr: mass of water vapour, mean liquid water content, ice water content, 2nd prognostic for cloud ice, rain water content and graupel.

Overwriting ρd

  • Expected to ruin mass conservation
  • Causes massive warming in Jnr
  • Jnr crashes after about 8 months

Running for 2 years

Overwriting u, v & θvd

   

Overwriting u, v, θvd, π and πs

Overwriting all dynamical core fields except ρd & π

Overwriting all dynamical core fields except ρd

Overwriting π and πs seems to increase drift.

Running for 4 years

Running job which overwrites u, v & θvd

for 4 years.

  • The overall drift continues to increase
  • There's a seasonal cycle
    • Largest peaks on in mid-July
    • The other peaks occur in mid-February
  • Are the surfaces diverging?
    • Colin has suggested looking at sensible and latent heat flux