...
RESULTS: All tests pass with roundoff level changes to c_k10h, and fail (detect statistically different results) with small changes in c_k10h. With the default thresholds, PGN is the most sensitive (detecting a statistical difference with a 1e-8 change, followed by TSC at 1e-53, and MVK at 2e-2). These results are correlated with the timescale of each test, with PGN looking at physics columns after 1 timstep, TSC time step convergence with 300 timesteps, and MVK examining 1 year climatologies.
...
12 member ensemble of 5 day simulations. Takes about 10min on 11 nodes.
c_10kh | Test Result (pending possible bug fix in scripts. Below using PASS = land and ocean are both PASS) | TestStatus.log Metrics region by region results Global, Land, Ocean |
---|---|---|
0.35 (default) | PASS | PASS, PASS, PASS |
0.350001 | PASS | PASS, PASS, PASS |
0.35001 | PASS | FAIL, PASS, PASS |
0.3501 | PASS | PASS, PASS, PASS |
0.351 | FAIL | PASS, FAIL, PASS |
0.36 | FAIL | FAIL, FAIL, FAIL |
Hack to reuse a ā-cā case to run multiple experiments used in MVK and PGN tests does not work. For TSC, during the RUN phase, the user_nl_eam_???? ensemble member namelists will be created anew by cime/scripts/lib/CIME/SystemTests/tsc.py. This script can be edited to append user_nl_eam to each user_nl_eam_???? file, and then one can set parameters in user_nl_eam.