Saving fields from CM2 for driving OM2

[MartinDix]

@wghuneke wants to drive an OM2 0.1 model with atmospheric fields from a CM2 0.25 run.

Fields are https://github.com/COSIMA/01deg_jra55_iaf/blob/master/namcouple#L86-L178

• Field 01 : swflx down 1,235 TOTAL DOWNWARD SURFACE SW FLUX
• Field 02 : lwflx down 2,207 DOWNWARD LW RAD FLUX: SURFACE
• Field 03 : rainfall 5,214 TOTAL RAINFALL RATE: LS+CONV
• Field 04 : snowfall 5,215 TOTAL SNOWFALL RATE: LS+CONV
• Field 05 : surface pressure Use UM sea level pressure to avoid spurious low values for coastal points near Antarctica and the Andes. OM2 config actually uses JRA psl rather than surface pressure too.
• Field 06 : runoff. Runoff is passed on the destination grid. UM river outlow
• Field 07 : near surface (2m) air temp For coastal points calculate this as a difference between 3,236 TEMPERATURE AT 1.5M and 3,341 LAND MEAN TEMPERATURE AT 1.5M
• Field 08 : 2m air humidity As above with 3,237 SPECIFIC HUMIDITY AT 1.5M 3,342 LAND MEAN SPECIFIC HUMIDITY AT 1.5M
• Field 09 : 10m wind (u) UM 3,209 10 METRE WIND U-COMP. Note that this is on the UM C grid U points
• Field 10 : 10m wind (v) UM 3,210 10 METRE WIND V-COMP. Note that this is on the UM C grid V points
• Field 11 : land ice calving flux, like runoff it is passed on the destination grid In CM2 this is a pre-determined pattern with the actual amount set to match the model snow accumulation.

[wghuneke]

My understanding is that this task involves writing out the above fields from the coupling stage.

Runoff is probably not needed for my purpose, but might still be worth writing it out for the sake of completeness.

[MartinDix]

Suite cj877 modified to save 3 hourly data for the above fields. Ran 18 months from 0501 to allow creating a shifted set of 12 months following the JRA RYF. Take 501-01-01 to 501-04-30 as 1900-01-01 to 1900-04-30 and 500-05-01 to 500-12-31 as 1900-05-01 to 1900-12-31.

Difference between the grid average and land-only air temperature gave unreasonable estimates of the air temperature over the ocean fraction at some grid points (inconsistent time averaging?). Therefore just use the grid box average for driving the model.

[MartinDix]

ACCESS-OM2 1deg_jra55_ryf experiment modified to use the cj877 forcing is https://github.com/MartinDix/1deg_cm2_ryf. Coupling weights generated using suite u-cx958 with OCEAN_GRID_DIR='/g/data/ik11/inputs/access-om2/input_20201102'

River outflow and ice calving still from JRA55, everything else from cj877.

Some analysis of 5 year run at https://github.com/MartinDix/notebooks/blob/master/cj877_ryf_analysis.ipynb