[CIG-SHORT] questions about RELAX

[Sylvain Barbot]

Hi Beatriz,

This

# grid size (sx1,sx2,sx3)
32 32 32

will not work! Try:

# grid size (sx1,sx2,sx3)
512 512 512

or even more samples!

Cheers,
Sylvain

On Wed, Feb 24, 2016 at 6:16 AM, Beatriz COSENZA MURALLES
<cosenzamural at wisc.edu> wrote:
> Thank you very much for your answer.
>
> When dealing with a very simple fault, I chose an origin point with coordinates (lon/lat) -90.6648/14.8260. This would be a 300 km fault with a left superior tip at x1 = 0, x2 = -150 km, x3 = 0, and a right superior tip at x1 = 0, x2 =+150 km and x3 = 0. The lon/lat coordinates of these two points obtained with gmt mapproject are  -92.05676406/14.80200819 and -89.27035597/14.84166050. The fault trace in my map was drawn using the geographical coordinates of these three reference points. My input for relax looks like this:
>
> # grid size (sx1,sx2,sx3)
> 32 32 32
> # sampling size (in unit of length), smoothing (0-0.5) & nyquist (dx1,dx2,dx3,beta,nq)
> 20000 20000 20000 0.2 2
> # origin position & rotation
> 0 0 0
> # geographic origin (longitude and latitude), UTM zone and real length unit (usually m or km)
> -90.6648 14.8260 16 1
> # observation depth for displacement and for stress (stress in only exported in GRD)
> 0 0
> # output directory (all output written here)
> $WDIR
> # elastic parameters and gamma = (1-nu) rho g / mu = 8.33e-7 /m = 8.33e-4 /km
> 5.091e10 40e9 4.76e-10
> # integration time (in unit of time), step (negative for automatic) and scaling of computed value
> 40 1
> # number of observation planes
> 0
> # no x1 x2 x3 length width strike dip
> #   1  0 -3  0      6     6     90  90
> # number of observation points
> 0
> # number of Coulomb patches
> 0
> # number of prestress interfaces
> 0
> # number of linear viscous interfaces (where viscosity changes)
> 1
> # no depth gammadot0 cohesion (gammadot0 is shear modulus divided by viscosity)
> 1   25000   0.126144    0.0
> # number of linear ductile zones
> 0
> # number of powerlaw viscous interfaces
> 0
> # number of friction faults
> 0
> # number of interseismic loading strike-slip and opening
> 0
> 0
> # number of coseismic events (when slip distribution is prescribed)
> 1
> # number of shear dislocations (strike-slip and dip-slip faulting)
> 1
> # no  slip    x1     x2      x3  length width strike dip rake
> 001   1.59    0 -150000  0  300000 15000 90 90 0
> # number of tensile cracks
> 0
> # number of dilatation sources
> 0
> # number of surface traction
> 0
> EOF
> ___
>
> The results of the coseismic displacements are shown in the ps file. The fault I will work with has a much more complicated shape and I didn't noticed there was a problem with the coordinates until I started paying more attention to the coseismic to look for errors and then I tried to confirm the problem using the fault I'm showing you here. I still can't find where does the shift come from. Thanks again,
>
> Beatriz
>
>
>
> _____________________________________
> From: Sylvain Barbot <sylbar.vainbot at gmail.com>
> Sent: Monday, February 22, 2016 6:28 PM
> To: Beatriz COSENZA MURALLES
> Cc: cig-short at geodynamics.org
> Subject: Re: questions about RELAX
>
> Dera Beatriz,
>
> 1. The simplest way to go about coordinates is to locate all the items
> relevant to your computation in a local system of coordinates. If you
> take the epicenter 15.32°N 89.10°W as a reference coordinate, you can
> convert it to UTM as follows
>
> echo -89.10 15.32 | proj +proj=utm +zone=16
> 274529.84       1694811.97
>
> Then everything that comes your way can be converted UTM relative to
> this point. Then, keep in mind that the Relax coordinate system is
> x1,x2,x3 where x1 is north, x2 is east and x3 is depth. So a
> transformation following (x1-1694811.97), (x2-274529.84) is in order.
>
> 2. The simulation for an elastic layer over a viscoelastic substrate
> is presumably correct. And your parameterization is presumably fine,
> too. So if it doesn't produce the results intended, you are probably
> learning that the geometry is not adequate. An elastic layer over a
> viscoelastic half space is only relevant for the oceanic lithosphere.
> In the backarc of Guatemala, you are in a continental lithosphere,
> even though the details may be more complicated than that. So you can
> expect a viscoelastic deformation in the lower crust and another one
> in the asthenosphere. In this system, afterslip may also play an
> important role.
>
> Best,
> Sylvain
>
>
> On Tue, Feb 23, 2016 at 5:04 AM, Beatriz COSENZA MURALLES
> <cosenzamural at wisc.edu> wrote:
>> Dear Mr. Barbot,
>>
>>
>> I am a first year geophysics grad student at the Department of Geoscience of
>> UW Madison. I am using RELAX to obtain the viscoelastic response of the
>> Motagua 1976 earthquake (Guatemala) and I have encountering a few problems
>> regarding two issues:
>>
>>
>> 1. Coordinates: I am using 9 fault patches. All of them are strike-slip
>> vertical faults. I have the location of these patches in geological
>> coordinates, so I chose a point of reference to be the origin of the
>> coordinate system, which I gave to RELAX, and I obtained the relative
>> positions of everything else converting the geological coordinates to utm
>> (zone 16) and calculating the distances in km. The direction of the
>> coseismic displacements (files 000) seemed odd, so I tried to run RELAX with
>> a simpler one-patch 300 km strike-slip fault and I confirmed that the
>> displacements are shifted. The place where the fault should be is moved
>> upwards and to the right. I wonder if you have some ideas on what could I be
>> doing wrong to get this output.
>>
>>
>> 2. The magnitude of the viscoelastic response: I ran the software using a
>> viscoelastic layer that starts at a depth of 25 km, with a Maxwell time of
>> about 7.9 years (gamma dot of 0.126). I wanted yearly outputs, for the last
>> 40 years, so I wrote down for the integration time "40 1". I used the output
>> from the 039 and the 040 files to obtain the increment of the last year and
>> I'm seeing displacements of more than 2 mm for that last year, which is too
>> large and is not being picked up by gps stations, on the contrary, I would
>> expect the viscoelastic transient to be negligible by now. I can (and will)
>> run for viscoelastic layers with larger viscosities, but it seems odd that
>> the model produces such large displacement values after 40 years, so I might
>> be misinterpreting something. I am assuming that the output is in meters (as
>> is the input) and that the integration time is given in years.
>>
>>
>> Thank you in advance for your help.
>>
>>
>> Sincerely,
>>
>>
>> Beatriz Cosenza