@kkappler The aurora tests are now all passing except for the Parkfield tests comparing the aurora computed transfer function with the emtf transfer function. Specifically, TestParkfieldSingleStation::test_single_station_comparison_with_emtf [line240] in test_parkfield_pytest.py. There is a new method in zfile_murphy.py that will compare another TF. It basically interpolates over the same period range for easier comparison.
When comparing the aurora computed TF and the emtf computed TF using this method there is a negative sign floating around somewhere that is causing the comparison to fail. Below is an example where tf1 is the aurora computed TF and tf2 is the emtf computed TF. You can see that they are off by a minus sign.
I couldn't track down where the negative sign is introduced. I would think that the aurora version is correct as the yx component should be negative in a right-handed coordinates system with z+ down. Which would give it a phase between [-90, -180].
The z-files looks like (left is aurora right is emtf):
There is the factor of -1, and the covariances are very different. The estimated error is about 2x larger in the aurora TF. Could you track down the -1. I think if we can do that we should add a similar comparison to the other tests in test_parkfield_pytest.py where a TF is estimated.
@kkappler The
auroratests are now all passing except for the Parkfield tests comparing theauroracomputed transfer function with theemtftransfer function. Specifically,TestParkfieldSingleStation::test_single_station_comparison_with_emtf[line240] in test_parkfield_pytest.py. There is a new method inzfile_murphy.pythat will compare another TF. It basically interpolates over the same period range for easier comparison.When comparing the
auroracomputed TF and theemtfcomputed TF using this method there is a negative sign floating around somewhere that is causing the comparison to fail. Below is an example wheretf1is theauroracomputed TF andtf2is theemtfcomputed TF. You can see that they are off by a minus sign.I couldn't track down where the negative sign is introduced. I would think that the
auroraversion is correct as the yx component should be negative in a right-handed coordinates system with z+ down. Which would give it a phase between [-90, -180].The z-files looks like (left is
auroraright isemtf):There is the factor of -1, and the covariances are very different. The estimated error is about 2x larger in the
auroraTF. Could you track down the -1. I think if we can do that we should add a similar comparison to the other tests intest_parkfield_pytest.pywhere a TF is estimated.