SPECFEM / specfem3d

SPECFEM3D_Cartesian simulates acoustic (fluid), elastic (solid), coupled acoustic/elastic, poroelastic or seismic wave propagation in any type of conforming mesh of hexahedra (structured or not).
GNU General Public License v3.0
403 stars 225 forks source link

half duration and numerical noise #1481

Open SeismoFelix opened 2 years ago

SeismoFelix commented 2 years ago

Hi,

I am looking for an appropriate half duration in the CMTSOLUTION file for avoiding numerical noise in the simulation visualization (movie).

Quoting page 34 of the manual:

(1) if a CMTSOLUTION moment-tensor source is used, a (pseudo) Heaviside step function with a very short half duration is defined for elastic cases to represent the permanent slip on the fault.

Besides, for making the movie quoting page 74 of the manual:

(2) As usual, for a point source one can set half duration in the CMTSOLUTION file to be 0.0 and HDUR_MOVIE = 0.0 in the Par_file to get the highest frequencies resolved by the simulation.

However, if I set the half duration in the CMTSOLUTION to be 0.0 as is stated in (2) for getting the highest frequencies resolved by the simulation, I got the error message:

Error hdur too small for movie creation, movies do not make sense for Heaviside source

I understand I cannot set the half duration as 0.0 in the CMTSOLUTION file because I am going to use the Heaviside step function. But I have to use a Heaviside step function If my source is defined by a CMTSOLUTION file (1) . Therefore I do not understand how to follow the instruction in (2) because setting the half duration to be 0.0 will yield an error because the Heaviside source is implicit using the CMTSOLUTION file (1).

Thanks in advance for your aid,

Best regards,

Félix

Jiefangsong commented 1 year ago

Dear SeismoFelix, I have met same this problem with you, do you have solved it?

homnath commented 1 year ago

Hi @Jiefangsong and @SeismoFelix If you want a movie, I would suggest setting a positive nonzero value for the half duration in the CMTSOLUTION, for example, corresponding to the actual half duration for the source or the value larger than 1.1 x smallest period resolved.

Best, Hom Nath

Jiefangsong commented 1 year ago

Ok @homnath,thank you for your detailed answer. I have solved the problem with your suggestion.

danielpeter commented 1 year ago

thanks for bringing this up again, there was a small inconsistency between the documentation and the actual code routine, throwing this error message.

let me try to fix the code check such that you can have both, i.e., a zero half duration in the CMTSOLUTION file and a zero HDUR_MOVIE setting in the Par_file.

ps. this will assume that as a user, you know what you want - but, it would probably be an ugly movie output with lots of high-frequency jitters. for getting Green's function seismograms, just turn off the movie output.