Turn on additional output for SORRMv3 mesh

[darincomeau]

Turns on the ocean conservation analysis member and ocean eddy analysis member for the SOwISC12to30E3r3 mesh.

This allows plotting of mass flux anomaly due to land ice fluxes and EKE in MPAS-Analysis.

[BFB]

[darincomeau]

Tested

SMS.ne30pg2_r05_SOwISC12to30E3r3.CRYO1850.chrysalis_intel
confirmed mpaso_in settings turned on here, and

SMS.ne30pg2_r05_IcoswISC30E3r5.WCYCL1850.chrysalis_intel
confirmed mpaso_in settings changed here for SOwISC12to30E3r3 remained default off.

[proteanplanet]

Can you provide an estimate of the added computational overhead from this change?

[darincomeau]

@proteanplanet turning on the ocean eddy analysis member is in the noise/machine variability as far as I can tell.

Example PACE with ocean eddy analysis on:
https://pace.ornl.gov/exp-details/201662, 8.13 SYPD

Example PACE with ocean eddy analysis off:
https://pace.ornl.gov/exp-details/199419, 8.13 SYPD

I don't have an estimate for the computational overhead of the ocean conservation analysis member here, since we always run with it on (and need it on). When I turned it on for v2 Cryo meshes, I found minimal impact in one-year B-case tests.

[cbegeman]

Thanks @darincomeau! I agree that this is what we want. Let me know if you'd like me to do any additional testing.

[darincomeau]

@proteanplanet I ran a pair of small one month ne30pg2_r05_SOwISC12to30E3r3.CRYO1850 tests to estimate performance cost:

Baseline (additional analysis remains off): 1.85 SYPD
This PR (additional analysis on): 1.80 SYPD

So that's an estimated 2-3% overhead for the additional analysis members here.

[proteanplanet]

Approved. Additional computation costs are minimal and necessary.

[jonbob]

Passes:

• ERP_Ld3.ne30pg2_r05_IcoswISC30E3r5.WCYCL1850.chrysalis_intel.allactive-pioroot1

merged to next

[jonbob]

merged to master