When an end to end test was run with the correct timeshifts it was found that some of the IMs had larger values than with the incorrect timeshift.
The IMs with significantly different values are the pSAs with period of at least 3s, as the studied event was run with a 400m grid, and flo of 0.25, this is very near the switch over threshold.
The waveforms of each of these simulations were investigated, the LF and HF waveforms were the same, with the exception that the new LF waveform had an 8 second longer duration than the old one, as expected with the changes to the workflow.
The results were considered to be the result of the timeshift change, and no further work is needed at this stage.
Details
A table with a basic statistical breakdown of (new_ims/old_ims) is available below:
| PGA | PGV | CAV | AI | Ds575 | Ds595 | MMI | pSA_0.02 | pSA_0.05 | pSA_0.1 | pSA_0.2 | pSA_0.3 | pSA_0.4 | pSA_0.5 | pSA_0.75 | pSA_1.0 | pSA_2.0 | pSA_3.0 | pSA_4.0 | pSA_5.0 | pSA_7.5 | pSA_10.0 | |
| count | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 | 519 |
| mean | 1.000142 | 0.998486 | 1.000878 | 1.000006 | 0.9998 | 0.999842 | 0.99954 | 1.000117 | 1.000205 | 1.000035 | 0.999897 | 0.999983 | 0.999969 | 1.000011 | 1.000001 | 1.000137 | 1.000515 | 1.001262 | 1.02395 | 1.062839 | 1.111222 | 1.104871 |
| std | 0.002858 | 0.039352 | 0.003502 | 0.000174 | 0.0041 | 0.002437 | 0.009184 | 0.0028 | 0.002253 | 0.001539 | 0.001317 | 0.001432 | 0.001665 | 0.001917 | 0.002793 | 0.003492 | 0.012626 | 0.035171 | 0.119366 | 0.16971 | 0.167845 | 0.158326 |
| min | 0.991528 | 0.881659 | 0.989233 | 0.999268 | 0.962916 | 0.980311 | 0.970436 | 0.991623 | 0.992775 | 0.994799 | 0.996007 | 0.99475 | 0.994187 | 0.992435 | 0.989321 | 0.987698 | 0.956618 | 0.879601 | 0.690023 | 0.627881 | 0.716029 | 0.67907 |
| 25.00% | 0.998378 | 0.974984 | 0.999142 | 0.999909 | 0.999005 | 0.999461 | 0.994619 | 0.998372 | 0.998691 | 0.999048 | 0.999103 | 0.999127 | 0.998996 | 0.998993 | 0.998609 | 0.998239 | 0.994192 | 0.981783 | 0.948519 | 0.949866 | 0.99273 | 0.995576 |
| 50.00% | 0.99985 | 0.998775 | 1.000388 | 1.000004 | 1 | 1 | 0.99974 | 0.999884 | 1.000057 | 1.000047 | 0.99993 | 0.999984 | 0.999955 | 1.00003 | 0.999905 | 1.000041 | 1.000358 | 1.001328 | 1.012833 | 1.042543 | 1.087355 | 1.090826 |
| 75.00% | 1.001811 | 1.021862 | 1.002241 | 1.000097 | 1.000812 | 1.00042 | 1.004597 | 1.001833 | 1.001485 | 1.000934 | 1.000728 | 1.000861 | 1.0009 | 1.001074 | 1.001346 | 1.002208 | 1.006398 | 1.021219 | 1.093904 | 1.151701 | 1.210073 | 1.206002 |
| max | 1.008882 | 1.144559 | 1.021907 | 1.000825 | 1.027938 | 1.018225 | 1.035728 | 1.008675 | 1.00659 | 1.004869 | 1.004768 | 1.004191 | 1.008049 | 1.007692 | 1.011825 | 1.011286 | 1.050045 | 1.142534 | 1.490298 | 1.967473 | 2.002373 | 1.801219 |
The station with the greatest IM ratio was selected for further investigation (11006e3).
As the significant differences were found at periods greater than 3s the waveforms were filtered for frequencies below 0.2Hz (period of 5s), no site amplification was used.
| Old LF |
| Old HF | |
| Old unshifted merge (LF+HF) | |
| Old shifted merge (HF starts 200 ts behind LF) | |
| New LF (same as old LF) | |
| New HF (same as old HF) | |
| New unshifted merge (HF and LF start at ts 0) | |
| New shifted merge (HF starts 800 ts behind LF) |
