forked from daniel-koehn/DENISE-Black-Edition
-
Notifications
You must be signed in to change notification settings - Fork 0
/
QUICKSTART_Marmousi.txt
107 lines (62 loc) · 3.2 KB
/
QUICKSTART_Marmousi.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
Quickstart to run the elastic Marmousi-II benchmark example in 14 easy steps
Details are described in the manual - https://danielkoehnsite.wordpress.com/software/
Minimum requirements:
---------------------
Intel Core i7 or similar AMD CPU with 8 cores
8 GB RAM
1-2 days computation time
C-Compiler: gcc, icc (recommended)
MPI-library: OpenMPI, Intel-MPI (recommended)
other libraries: Fastest Fourier Transform in the West (FFTW) - https://github.com/FFTW/fftw3
Seismic Unix and/or Numpy/Scipy/Matplotlib/Obspy for seismic data and model visualization
On NEC-cluster @ RZ Kiel activate libraries by typing
module load intel17.0.4 intelmpi17.0.4
1. Clone DENISE-Black-Edition on your local machine
git clone --depth=1 https://github.com/daniel-koehn/DENISE-Black-Edition
2. Compile the library cseife in /libcseife with
make
3. In /src adapt the compiler options in the Makefile to your system and compile the DENISE code with
make denise
4. Clone DENISE-Benchmark on your local machine
git clone --depth=1 https://github.com/daniel-koehn/DENISE-Benchmark
5. Copy model files for the Marmousi-II model from DENISE-Benchmark to DENISE-Black-Edition
cp DENISE-Benchmark/Marmousi-II/start/marmousi_II_* DENISE-Black-Edition/par/start/
6. In DENISE-Black-Edition/par/DENISE_marm_OBC.inp check if the following parameters are set correctly
MODE = 0
MFILE = start/marmousi_II_marine
7. Generate synthetic Marmousi-II data for an OBC acquisition geometry by running DENISE on e.g. 15 cores of
your CPU from the /par directory:
mpirun -np 15 ../bin/denise DENISE_marm_OBC.inp FWI_workflow_marmousi.inp
If you want to use a different number of cores change NPROCX and NPROCY in DENISE_marm_OBC.inp according to your
needs. The total number of cores = NPROCX * NPROCY.
8. Check if the modelling results are plausible, e.g. with Seismic Unix
sugain qbal=1 < su/DENISE_MARMOUSI_y.su.shot1 | suximage &
or the Jupyter notebook
/visu/plot_shot_gather.ipynb
9. In DENISE-Black-Edition/par/su generate the directory MARMOUSI_spike
mkdir MARMOUSI_spike
10. Move the data from /su to /su/MARMOUSI_spike
mv DENISE_MARMOUSI_* MARMOUSI_spike/
11. In DENISE-Black-Edition/par/DENISE_marm_OBC.inp change the following parameters
MODE = 1
MFILE = start/marmousi_II_start_1D
12. Start the 2D elastic FWI with
mpirun -np 15 ../bin/denise DENISE_marm_OBC.inp FWI_workflow_marmousi.inp
As defined in FWI_workflow_marmousi.inp the FWI runs a sequential frequency inversion using low-pass filtered data
with maximum frequencies of 2 Hz, 5 Hz, 10 Hz and 20 Hz, respectively.
The Vp, Vs and density models of the current iteration are saved in:
model/modelTest_vp.bin
model/modelTest_vs.bin
model/modelTest_rho.bin
13. Wait a few hours ...
The intermediate results after the FWI of each frequency range are saved seperatly in
model/modelTest_vs_stage_1.bin
model/modelTest_vs_stage_2.bin
model/modelTest_vs_stage_3.bin
model/modelTest_vs_stage_4.bin
You can visualize the results with Seismic Unix
ximage n1=174 < model/modelTest_vs_stage_2.bin
or the Jupyter notebook
/visu/FWI_marmousi.ipynb
14. Wait 1-2 days until the FWI is finished ...
Congratulations, you just finished your first time-domain elastic FWI with DENISE Black Edition.