Issue with Radial component polarity for Instaseis/Syngine Force source. about mtuq HOT 3 CLOSED

I also have a functional test ready, that will download synthetic seismograms from Syngine and check that E = -R. If the test fails, it means that Instaseis has been patched and that we can revert this fix.

We also tweak the sign of the GFs in our code lsforce. And we have a test here as well: https://code.usgs.gov/ghsc/lhp/lsforce/-/blob/main/tests/test_lsforce.py#L123-L163

This compares [sign-modified] Syngine GFs with CPS GFs and fails if the two are different, so it should also fail once Instaseis is patched.

from mtuq.

Thanks @thurinj @liamtoney

It seems like the example above involves a horizontal force oriented in the direction of outward propagation from the source to receiver, so the radial displacement at the receiver should be positive.

In fact we see, however, that the displacement returned by instaseis is negative, suggesting the whole radial component is flipped.

I agree with the proposed workaround.

from mtuq.

As a side note, do we understand the character of the impulse response in the figure above, beyond just the radial component sign convention? e.g. what explains the shape of the pulse?

Perhaps we could continue the discussion offline at the Air Force meeting next week, if anyone here is attending.

from mtuq.