[CIG-SEISMO] 回复: Fwd: RE: suspicious bug of the gravity calculations

[Dimitri Komatitsch]

Hi Ming,

Thanks; regarding Z=-10m it depends how large your model is, the minimum 
Z value to use so that 1/z is not singular should probably be a value 
that is not too small compared to the typical size of your model, 
otherwise 1/z is OK mathematically but may be problematic from a 
numerical point of view (?).

Also, in most cases people compute the field outside of the model rather 
than inside, i.e. for positive values of Z. Please check e.g. 
http://komatitsch.free.fr/preprints/GJI_Roland_Martin_2013.pdf or 
references therein to see if the formulas remain OK for points inside 
the model rather than outside (I do not remember; Roland knows more 
about this than I do).

thanks
Cheers,
Dimitri.

On 06/04/2017 08:04 AM, 赵明 wrote:
> Hi Dimitri,Roland,
> 
> Thank you very much for your reply and suggestion. the observation can't 
> be at the surface of the model,I tried once again for two depths of the 
> observations plane:z=-10m and z=-122 m. The -122m seems to be quite 
> normal(toto_minus_122.ps), while z=-10m have abnormal values at the down 
> left corner.
> 
> Cheers
> 
> Ming
> --------- 原始邮件信息 ---------
> *发件人:* "Dimitri Komatitsch" <komatitsch at lma.cnrs-mrs.fr>
> *收件人:* <mzhao at cea-igp.ac.cn>, "Roland Martin" <roland.martin at get.omp.eu>
> *主题:* Fwd: RE:[CIG-SEISMO] suspicious bug of the gravity calculations
> *日期:* 2017/06/03 23:46:11 (Sat)
> 
> Hi Ming,
> 
> I am forwarding you this email from Roland, because when he sends it to
> you from his account he gets an error message.
> 
> Roland, attached is the file toto.ps
> 
> Cheers,
> Dimitri.
> 
> -------- Forwarded Message --------
> Subject: RE:[CIG-SEISMO] suspicious bug of the gravity calculations
> Date: Sat, 3 Jun 2017 14:21:15 +0000
> From: MARTIN Roland <Roland.MARTIN at Get.omp.eu>
> To: Dimitri Komatitsch <komatitsch at lma.cnrs-mrs.fr>, 赵明
> <mzhao at cea-igp.ac.cn>
> 
> Hi Ming,
> 
> It is not possible to put obsevation points on the topography solution
> will become usntable there.
> THis version is in revision and we need to work a little bit more on it
> before the release. I can not see the toto.ps because I did nit receive it.
> 
> Cheers
> Roland
> ________________________________________
> De : Dimitri Komatitsch [komatitsch at lma.cnrs-mrs.fr]
> Envoyé : samedi 3 juin 2017 16:09
> À : 赵明; cig-seismo at geodynamics.org; 赵明; MARTIN Roland; CHEVROT
> Sebastien; Vadim Monteiller
> Objet : Re: [CIG-SEISMO] suspicious bug of the gravity calculations
> 
> Dear Ming, dear all,
> 
> Thanks! Let me cc Roland Martin, who is the expert on that.
> However, it seems you used an observation surface that is located in z =
> 0 i.e. at elevation zero, but the gravity formulas are undefined there
> (they involve a term with 1/r , in which r is the distance to the
> surface). Thus, could you try again using a non-zero elevation?
> everything should work fine then.
> 
> Also, I think the results should be symmetric, but not necessarily
> identical at all points (since the mesh has a finite size, you should
> see different values near the edges than in the center, in a symmetric
> fashion (I think, but I am not an expert; Roland will confirm).
> 
> Thanks,
> Cheers,
> 
> Dimitri.
> 
> On 06/03/2017 02:05 PM, 赵明 wrote:
>> Hi Everyone,
>>
>> I tried the gravity field calculation of SPECFEM3D using the given
>> example:
>> homogeneous_halfspace_HEX8_elastic_absorbing_Stacey_5sides.However,I
>> found the results seem to be abnormal at some observations.
>>
>> Since it is a homogenous model,the gravity value should be the same to
>> the observation on the surface.However,look at the toto.ps,there are
>> some place (very likely to be the boundary of some elements) the gravity
>> is abnormal.
>>
>> Best regards
>> Ming Zhao
>>
>>
>>
>>
>> _______________________________________________
>> CIG-SEISMO mailing list
>> CIG-SEISMO at geodynamics.org
>> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>>
> 
> --
> Dimitri Komatitsch, CNRS Research Director (DR CNRS)
> Laboratory of Mechanics and Acoustics, Marseille, France
> http://komatitsch.free.fr
> 
> 

-- 
Dimitri Komatitsch, CNRS Research Director (DR CNRS)
Laboratory of Mechanics and Acoustics, Marseille, France
http://komatitsch.free.fr