Implement energy and water checker for YOG

[MarionBWeinzierl]

This PR is an implementation of a water and energy conservation checker to investigate the conservation errors observed in #34 .

The implementation is based on lines 404ff in check_energy.F09.

The checker should be called at various places in nn_interface_cam.F90->n_convection_flux_CAM.

The output can be found in the atm.log.***.gz files.

[MarionBWeinzierl]

The code runs through and gives the expected outputs, but both the energy and water sum for SAM seems to be off by a factor 100:

Might be an error in the parameters I hand in?

[MarionBWeinzierl]

The output of the built-in energy checker (not the new implementation) is:

• actual value (energy/water)
• expected value
• change (in static energy/total water)
• tendency (from boundary fluxes)
• expected tendency
• relative error

[MarionBWeinzierl]

OK, this is where the factor 100 comes from:

pres_int_cam, in the "before the conversion" call is in Pa (see

CAM-ML/src/physics/cam/nn_interface_cam.F90

Line 135 in 6e1b04a

!! interface pressure [Pa] from the CAM model

f).

presi, as used for the "after conversion" part, is in hPa, see

CAM-ML/src/physics/cam/nn_interface_cam.F90

Line 42 in 6e1b04a

!= unit hPa :: pres, presi

. It is explicitly converted to hecto Pascal in SAM_sounding_init:

CAM-ML/src/physics/cam/nn_interface_cam.F90

Line 562 in 6e1b04a

presi(:) = presi(:) / 100.0

.

[MarionBWeinzierl]

Last time step with @jatkinson1000 's new changes:

We loose energy (increasing from 0.04% in the first iteration to 0.168% in the last), and we can a bit of energy. :-/