[cig-commits] [commit] devel, master: fixed a comment in the attenuation routine, and added more details about how to use EXAMPLES/attenuation (3eb7f87)

[cig_noreply at geodynamics.org]

Repository : https://github.com/geodynamics/specfem2d

On branches: devel,master
Link       : https://github.com/geodynamics/specfem2d/compare/fc67e6fd7ad890705b2b72b4b3c509accb22249e...e9ca46c40131588d89d7b0883250bc6584ce6b4c

>---------------------------------------------------------------

commit 3eb7f8796cee6b8113b9fcf11de8339d4bfca9f3
Author: Dimitri Komatitsch <komatitsch at lma.cnrs-mrs.fr>
Date:   Tue Dec 27 23:12:43 2011 +0000

    fixed a comment in the attenuation routine, and added more details about how to use EXAMPLES/attenuation


>---------------------------------------------------------------

3eb7f8796cee6b8113b9fcf11de8339d4bfca9f3
 attenuation/README | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/attenuation/README b/attenuation/README
index 51675a9..3f9c9a5 100644
--- a/attenuation/README
+++ b/attenuation/README
@@ -19,7 +19,24 @@ TO RUN:
 4. execute script to run mesher and solver for the PSV case:
    ./process.sh
 
-5. check out the output files in the local directory OUTPUT_FILES; in particular, you can type "gnuplot plot_compare_to_analytical_solution.gnu" to compare the seismograms computed to the quasi-analytical solution of Carcione et al. (1988). Note that because our example has no absorbing conditions on the edges of the grid, there are extra (spurious) waves after the main P and S waves, reflected off the edges of the grid, which are not present in the quasi-analytical solution and which you can safely ignore. Beware that the reference solution is not exact, only quasi-exact (i.e., the formulation uses an approximation, and some integrals are computed numerically) therefore some small discrepancies can be noticed.
+5. check out the output files in the local directory OUTPUT_FILES; in particular, you can type "gnuplot plot_compare_to_analytical_solution.gnu" to compare the seismograms computed to the quasi-analytical solution of Carcione et al. (1988). 
+
+Note that because our example has no absorbing conditions on the edges of the grid, there are extra (spurious) waves after the main P and S waves, reflected off the edges of the grid, which are not present in the quasi-analytical solution and which you can safely ignore.
+
+Beware that the reference solution is not exact, only quasi-exact (i.e., the formulation uses an approximation, and some integrals are computed numerically) therefore some small discrepancies can be noticed.
+
+More importantly, in the example provided, "tau" relaxation times for attenuation memory variables are recomputed using approximate quality factor targets of Qkappa approximately equal to 27 and Qmu / Qs approximately equal to 20, read (roughly) from Figure 1 page 604 of the article of Carcione et al. (1988). This will NOT lead to the exact same "tau" values as in Table 1 of Carcione et al. (1988). Thus, in order to perform a far more precise comparison to the analytical solution, which is computed using the "tau" values from Table 1 of Carcione et al. (1988), in file "src/specfem2D/attenuation_model.f90" you should impose this instead for this test (only):
+
+  tau_epsilon_nu1(1) = 0.0325305d0
+  tau_sigma_nu1(1)   = 0.0311465d0
+  tau_epsilon_nu2(1) = 0.0332577d0
+  tau_sigma_nu2(1)   = 0.0304655d0
+
+  tau_epsilon_nu1(2) = 0.0032530d0
+  tau_sigma_nu1(2)   = 0.0031146d0
+  tau_epsilon_nu2(2) = 0.0033257d0
+  tau_sigma_nu2(2)   = 0.0030465d0
+
 
 References:
 -----------