CCAPS - use time centred rho in step 4's grav forcing

[JOThurgood]

In step 4 should probabally calc a time centred rho (from 0.5 * (rho^n + rho^n+1 ) to use in calc of U*

Currently uses rho^n+1 (as advect_dens is called prior, which updates the rho array to the new time level).

Should test and then (if good) implement in CCAPS, CCAPS_vardens and CCAPS_atm_const

[JOThurgood]

Now done in CCAPS_atm_const - carry over to others when get a chance.

[JOThurgood]

Now done in CCAPS_atm_const - carry over to others when get a chance.

Did you mess it up - looks like used rho0(iy) (i.e. hydrostatic rho at the start of the step) rather than just the rho + perturbations or perturbations (whatever is appropriate)

Check this!!!!

[JOThurgood]

This might create problems with corrections (which use the end rho?)

[JOThurgood]

Now done in CCAPS_atm_const - carry over to others when get a chance.

Did you mess it up - looks like used rho0(iy) (i.e. hydrostatic rho at the start of the step) rather than just the rho + perturbations or perturbations (whatever is appropriate)

Check this!!!!

Must have been tired - it looks fine.... At least, for the choice on the coefficient in-front of the gravitational source.

There is a question over what factor of beta/rho (in atmosphere case, or 1/rho otherwise) to use in front of the lagged pressure gradient. Since grad p is lagged, should you use a similarly lagged density? Or should we use the time-centred density? What currently happens is the density at n+1 is used.

[JOThurgood]

This is essentially issue #115 but for a specific part of the code. There are other places where it might be appropriate to use rho(n+1/2) etc. Closing, see that issue