SW4 seems to only read first set of point sources from multi-segment SRF rupture

[sndybing]

Hello! I'm working on simulating the M7.1 Ridgecrest earthquake using SW4. I've built an SRF version 2.0 file with four fault segments from a published inversion model of the earthquake (Goldberg et al., 2020). Following the SRF 2.0 format, I have the header for the four planes, which is then followed by four groups of point sources separated by the POINTS labels for the four segments. The four segments have 2130, 380, 270, and 340 point sources each, so the entire rupture model should have 3120 point sources.

When I run SW4 with my SRF file organized such that the largest fault segment (2130 point sources) has its points first in the order in the SRF, SW4 reads the file and says the following:

Opened rupture file 'RC_M7_multiseg.srf'
Version = 2.0
Number of segments in header block: 4
Seg #1: elon=-117.533, elat=35.7161, nstk=21, ndip=20, len=63.5649, wid=23.5743
stk=318.522, dip=89.525, dtop=0.1496, shyp=8.6493, dhyp=7.9268
Seg #2: elon=-117.539, elat=35.7199, nstk=4, ndip=19, len=10.399, wid=23.5668
stk=309.45, dip=89.9, dtop=0.1797, shyp=4.6311, dhyp=6.6393
Seg #3: elon=-117.651, elat=35.8163, nstk=3, ndip=18, len=6.685, wid=23.5687
stk=309.84, dip=89.9, dtop=0.1226, shyp=3.3287, dhyp=7.4848
Seg #4: elon=-117.439, elat=35.6097, nstk=4, ndip=17, len=12.172, wid=23.0142
stk=319.76, dip=88.9213, dtop=0.0924, shyp=6.0854, dhyp=8.7137
Number of point sources in data block: 2130
Read npts=2130, made 2054 point moment tensor sources, nu1=2054, nu2=0, nu3=0

It seems that SW4 is only reading the first block of point sources corresponding to the first listed fault segment?

As another check, I changed the order of the segments in the SRF such that the 340 point source fault is listed first. In this case SW4 reads the file and similarly says the following:

Opened rupture file 'RC_M7_multiseg_newsegorder.srf'
Version = 2.0
Number of segments in header block: 4
Seg #1: elon=-117.439, elat=35.6097, nstk=4, ndip=17, len=12.172, wid=23.0142
stk=319.76, dip=88.9213, dtop=0.0924, shyp=6.0854, dhyp=8.7137
Seg #2: elon=-117.651, elat=35.8163, nstk=3, ndip=18, len=6.685, wid=23.5687
stk=309.84, dip=89.9, dtop=0.1226, shyp=3.3287, dhyp=7.4848
Seg #3: elon=-117.539, elat=35.7199, nstk=4, ndip=19, len=10.399, wid=23.5668
stk=309.45, dip=89.9, dtop=0.1797, shyp=4.6311, dhyp=6.6393
Seg #4: elon=-117.533, elat=35.7161, nstk=21, ndip=20, len=63.5649, wid=23.5743
stk=318.522, dip=89.525, dtop=0.1496, shyp=8.6493, dhyp=7.9268
Number of point sources in data block: 340
Read npts=340, made 329 point moment tensor sources, nu1=329, nu2=0, nu3=0

The Mw that SW4 calculates based on my SRFs (6.9 and 6.4 for the above examples respectively) is also lower than I expect given that the SRF is based off of an inversion for the M7.1 Ridgecrest mainshock.

I'm wondering if this is a bug in the software or if I have an issue with my SRF file. I don't think I'm able to attach the SRF, but here's a screenshot of the beginning of the file where the largest fault segment is first, as well as one of the transition points to the next segment deeper into the file.

SW4 runs fine and outputs waveforms as expected - I'm just not sure it's using all of the point sources.

Thank you!

[sndybing]

Just to add something else I tried after this, if I take out the differentiation between the segment point source blocks and just put everything in one POINTS 3120 block, SW4 does read in all of the point sources and it gets the Mw up to 7.04. Seems like a reasonable workaround, but I'll leave this open in case there's anything I'm missing about a way to execute it with the multi-segment point source blocks!

I'd also be curious to know why SW4 reduces the number of point sources from the input number each time - in this case, it reads npts=3120, but makes only 3012.

[houjun]

Hi @sndybing, I have only used SRF files with a single plane, so there might be a bug with multi-plane data.
Can you upload your SRF file to a Google Drive/Dropbox/Box/OneDrive location and paste a link here? Then I can download and use it to debug SW4.

[sndybing]

Hi @houjun, sure! Here's the file. Let me know if this doesn't work! https://uoregon-my.sharepoint.com/:i:/g/personal/sdybing_uoregon_edu/EVE9fnr2Ny5GtIV6mAiA598BheLd5WDcvz5OrSBGgli15w?e=TepQ1I

Thanks!

[andersp]

Hi @sndybing. If I remember correctly, we only implemented the single block SRF case. If you need the more general case, feel free to generalize the code and submit a pull request.

[sndybing]

Hi @andersp, thank you for the info! For my current use just putting everything in one block seems to be the simplest solution. If I need more in the future I'll consider looking into doing this!