[CIG-SEISMO] simulation of seismic wave propagation excited by a force

Tue Sep 11 09:53:40 PDT 2012

Hi Yingzi,

At the moment all the source time functions (Gaussian, Ricker,
Heaviside) are defined in comp_source_time_function.f90 which is located
in src/specfem3D.
Then, the proper function name needs to be specified in
compute_add_sources_*****.f90

I guess we could design a way to read an external file for the source
time function if USE_FORCE_POINT_SOURCE is on, but I don't know how much
of a priority this is for the main developers - Dimitri would be the one
to tell.
Or we could have a single source name [comp_source_time_function]
associated to the source for USE_FORCE_POINT_SOURCE, and users could
then implement their own source in comp_source_time_function.f90
(basically a line), without having to change names in
compute_add_sources_*****.f90 ? Any thoughts ?

Christina

On 09/11/2012 09:41 AM, Yingzi Ying wrote:
> Hi Christina,
>
> Can you also add a feature to use user-defined source form inputted from
> a file.
>
> Best,
> Yingzi
> On 09/11/2012 05:34 PM, Dimitri Komatitsch wrote:
>> Hi Christina and Yang, Hi all,
>>
>> OK, great. This is excellent news.
>>
>> So let us just move this "USE_FORCE_POINT_SOURCE" flag from
>> setup/constants.h.in to DATA/Par_file, and let us mention it in the manual.
>>
>> Jo, could you please do that and commit the changes?
>>
>> Thank you,
>> Dimitri.
>>
>> On 09/11/2012 06:17 PM, Christina Morency wrote:
>>> Hi Dimitri,
>>>
>>> Yes, I can confirm what Yang is saying, it is already in the package,
>>> with interpolation if the source is not located on a GLL point (as it is
>>> indeed done in the 2D btw).
>>>
>>> Christina
>>>
>>> On 09/11/2012 09:14 AM, Yang Luo wrote:
>>>> I think Daniel (or someone else) has already copied necessary lines from old package.
>>>>
>>>> There is an option "USE_FORCE_POINT_SOURCE" in constant.h, which allows one to use a point force
>>>> (which Yingzi already figured out, or maybe it's already in the manual)
>>>>     logical, parameter :: USE_FORCE_POINT_SOURCE = .false.
>>>>     double precision, parameter :: FACTOR_FORCE_SOURCE = 1.d15
>>>>     integer, parameter :: COMPONENT_FORCE_SOURCE = 3  ! takes direction in comp E/N/Z = 1/2/3
>>>>
>>>> Interpretation is included, if the point force is not located exactly on a GLL point.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ________________________________________
>>>> From: Dimitri Komatitsch [komatitsch at lma.cnrs-mrs.fr]
>>>> Sent: Tuesday, September 11, 2012 12:01 PM
>>>> To: Yingzi Ying
>>>> Cc: cig-seismo at geodynamics.org; Joseph Charles; Paul Cristini; 谢志南; Yang Luo; Christina Morency
>>>> Subject: Re: [CIG-SEISMO] simulation of seismic wave propagation excited by a   force
>>>>
>>>> Dear Yingzi,
>>>>
>>>> You are right, in the current code there is no option in the Par_file to
>>>> use a point source, but we added this to the todo list right before the
>>>> summer and thus we should/will definitely do it. It is easy to do: just
>>>> add the Ricker wavelet to the accel() vector (which is in fact a force
>>>> vector before we divide it by the mass matrix), to the component you
>>>> want (vertical or horizontal) and you are all set.
>>>>
>>>> Joseph, could you maybe do it? below are a few lines from one of my test
>>>> codes, in which I do just that; let us cut and paste these lines in
>>>> SPECFEM3D (it is just a matter of selecting the right grid point; which
>>>> we already do for a CMT source, thus we could add a force instead of a
>>>> CMT at that location; the only difference being that the CMT source is
>>>> added to all the points of a given spectral element, while a force
>>>> vector should be added to a single point).
>>>>
>>>> Let me talk to Joseph tomorrow to see if we can release a patch of the
>>>> source code soon.
>>>>
>>>> I also cc Yang to see if he has already done that in a local version at
>>>> Princeton maybe (for instance to model a force for some of his oil
>>>> industry simulations in foothill regions?).
>>>>
>>>> Another difficulty is if the location of the force does not correspond
>>>> to a grid point; then there are two options:
>>>>
>>>> - select the closest grid point; i.e. slightly change the location of
>>>> the source (can be done automatically by the code)
>>>>
>>>> - use Christina Morency's way of adding a force source between grid
>>>> points; I think I remember she did that in the 2D code, but I am not
>>>> sure how; thus I cc her.
>>>>
>>>> (that second option would be better, if Christina can confirm that it
>>>> works fine)
>>>>
>>>> Best wishes,
>>>> Dimitri.
>>>>
>>>> ==================================================================
>>>>
>>>> ! ipoinsource is the global grid point at which we add the source
>>>> ! "3" means a vertical force (i.e. we add the force to the vertical
>>>> ! component of the force vector, which is later divided by the mass
>>>> ! matrix to get the acceleration vector)
>>>> ! factor_force is the norm of the force vector to add
>>>> ! ricker(t) is the Ricker wavelet at time t = (it-1)*deltat
>>>>
>>>>      accel(3,ipoinsource) = accel(3,ipoinsource) - factor_force * ricker(t)
>>>>
>>>> ==================================================================
>>>>
>>>> On 09/11/2012 12:59 PM, Yingzi Ying wrote:
>>>>> Hi all,
>>>>>
>>>>> I am trying to use specfem3d to simulate waves excited by a vertical
>>>>> force with user-defined source forms such as ricker wavelet.
>>>>>
>>>>> I tried to do it through adjoint simulation by setting 'SIMULATION_TYPE
>>>>> = 2' and put source position at STATIONS_ADJOINT and receivers at
>>>>> CMTSOLUTION. But I failed.
>>>>>
>>>>> Does anyone have an idea or suggestion how to do active simulations with
>>>>> specfem3d?
>>>>>
>>>>> Thanks in advance.
>>>>>
>>>>> Best regards,
>>>>> Yingzi
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> CIG-SEISMO mailing list
>>>>> CIG-SEISMO at geodynamics.org
>>>>> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>>>> --
>>>> Dimitri Komatitsch - komatitsch aT lma.cnrs-mrs.fr
>>>> CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics,
>>>> UPR 7051, Marseille, France    http://komatitsch.free.fr

-- 
Christina Morency
~~~~~~~~~~~~~~~~~
Geophysicist
Computational Geosciences Group
Atmospheric, Earth and Energy Division
Physical and Life Sciences Directorate
~~~~~~~~~~~~~~~~~
Lawrence Livermore National Laboratory
P.O.Box 808, M/S L-286
Livermore, CA  94551
~~~~~~~~~~~~~~~~~
Office: (925) 422-1686
Email: morency1 at llnl.gov
https://www-pls.llnl.gov/?url=about_pls-scientific_staff-morency_c