General Info

Continuing Cybershake 100m runs around Wellington region.

Run Folder: /nesi/nobackup/nesi00213/RunFolder/Cybershake/v22p11

Environment: /nesi/project/nesi00213/Environments/cybershake_v22p11

Gmsim Version: 20.4.1.1 

Stat File: /nesi/project/nesi00213/StationInfo/non_uniform_whole_nz_with_real_stations-hh400_v20p3_land.ll

Used core hours: 750k

Initial Faults (list.txt)Run Faults (list2.txt)

AkaOtaki 33r
BooBooALL 35r
BooBooEAST 33r
DryHuang 31r
Fisherman 34r
Manaota 35r
Moonshine 29r
OhariuC 31r
OhariuS 32r
Okupe 32r
Otaraia 28r
PukeShep 32r
WellTeast 32r
WellWHV 35r
Wharekauhau 33r
Whitemans 28r

AkaOtaki 33r
BooBooALL 35r
Fisherman 34r
WellTeast 32r

513 Realizations134 Realizations

Run Details

Wrong parameters were set in the root_defaults Ds_multiplier was 1.2 and PGV_Threshold -1, but VM's were generated.

These gave the wrong nx,ny,nz and sim duration values and so the estimation was a lot higher than expected.

Estimated core hours

Then a single realization test was run with VM_PARAMS to get the correct sim duration for the events. We used these values to calculate the difference and expected new estimate based on if we changed the sim duration  but kept the old VM's.
The below file contains information calculated to determine the faults to run for to fit the 750k left on Maui to run by the end of November 2022.

mag_info_final.csv

The PGV value was calculated for each fault based on the current VM's generated that had the wrong settings. These were then looked at to see which ones were close enough to the expected value of 2 for cybershake runs.
From that 4 faults were very close to the value 2 for PGV and their estimates added to 715k core hours which was determined a good selection to run for 22p11.

The adjusted sim durations and old sim durations are in the file above.

There seemed to be an issue with EMOD3D where if too many runs were going at once at the same time some processes would just die and so EMOD3D would fail (Guess is IO problem) and so there were alot of reruns for EMOD3D.
AkaOtaki_REL01 however was run with the old sim duration when running EMOD3D and has issues with merge_ts with 6 rows of empty 0's at the end of the PGV entries and so therefore fails the test for completed merge_ts.

Results

Dropbox (.csv/.info for all rels and vm_params.yaml/nzvm.cfg for each fault) - This currently includes all 16 faults info
Dropbox (compressed BB.bins) - Only contains the 4 faults listed in list2.txt
Dropbox (IM Data) also on /isilon/CS_IMs/v22p11 - Only contains the 4 faults listed in list2.txt

  • No labels