Vertically-integrated frazil_3d diagnostic

[rmholmes]

This commit adds a diagnostic frazil_3d_int_z that vertically sums the frazil_3d diagnostic for space saving (note this is not the same as frazil_2d, which takes only the surface value). See discussion here.

Successfully tested in a 1-degree ACCESS-OM2 run.

[aidanheerdegen]

I'm fixing the travis CI compilation tests in this PR #292

In the meantime, @rmholmes have you tested this and confirmed that your diagnostic is the same as the 3d one when that is summed in post processing?

[russfiedler]

Need to change the description of the diagnostic.
This isn't the vertical integral of the heat flux. It's the sum of the heat fluxes or the integral of the heat changes due to the formation of frazil.

Otherwise looks good.

[rmholmes]

@russfiedler Right thanks. I've changed "z-integral" to "Vertical sum".

@aidanheerdegen yes I ran one year of 1deg_jra55_ryf, vertically-summed frazil_3d using ncap2 and it matched the frazil_3d_int_z diagnostic to within +-2e-6 Wm-2 (where the diag itself is ~1-30Wm-2).

One thing that I wasn't 100% sure of -> The part inside the if (pop_icediag). Does this go to the ice model diagnostics?

[russfiedler]

@rmholmes Yes it does, so you have to do the calculation in two places like you've done.

[aekiss]

Thanks @rmholmes.

For clarity, should the description for frazil_2d be changed from

'ocn frazil heat flux over time step'

to something like

'ocn frazil heat flux from the top level over time step'

?

[rmholmes]

Good point @aekiss, I've changed it.

[aekiss]

@russfiedler, @aidanheerdegen - if this PR looks OK, can one of you pls merge it in so I can compile new exes for Andy before the weekend? Ta!

[aidanheerdegen]

Will be in tomorrow, will check it then and merge if all ok. I borked the CI checks and can't get them back online.

[aidanheerdegen]

@rmholmes can you rebase this on to the current master branch so the travis tests will fire?

[aidanheerdegen]

Should just be a change to the .travis.yml file in the top level

[aidanheerdegen]

@rmholmes Don't worry. Seemed to have trigged them with the test PR I did. All good.