def mogi_agu_elastic():
"""
Large domain mesh that exploited reflection symmetry of large domain for
simulating stacked Dipole source. Domain 600x200km (1/4 r=300km)
"""
#DEL NEGRO MESH
#surfaceh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output/surface.h5'
#pointsh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output/points.h5'
#points = '/home/scott/research/models/pylith/3d/delnegro2009/output_points.txt'
#UTURNCU MESH
surfaceh5 = '/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/apmb/points.h5'
points = '/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output_points.txt'
#Get analytic solution
#x = np.linspace(0,25e3,1e2)
#y = np.zeros_like(x)
#NOTE: in this case need radial distance b/c not on x-axis
x = np.sqrt(2) * 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(xoff=0, yoff=0, d=25e3, dP=50e6, a=2e3, mu=30e9, nu=0.25)
# Superposition of inflation and deflation source
#ur,uz = m.calc_mogi_dp(x,y,d=4e3,dP=100e6,a=700,mu=30e9,nu=0.25)
ur1, uz1 = m.calc_mogi_dp(x, y, **params)
ur2, uz2 = m.calc_mogi_dp(x, y, d=75e3, a=5e3, mu=30e9, dP=-10e6)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# NOTE: cludge fix for negative radial displacements in dipole model
ur_fem[x >= 60000] = -ur_fem[x >= 60000]
#void-space
#ur_fem[x >= 1000] = -ur_fem[x >= 1000]
#Convert units for plotting (done in plotting routine)
#x = x/1000.0
#ur1 = ur1 * 100.0
#uz1 = uz1 * 100.0
#ur2 = ur2 * 100.0
#uz2 = uz2 * 100.0
#ur_fem = ur_fem * 100.0
#uz_fem = uz_fem * 100.0
# Superposition of analytic solutions
ur = ur1 + ur2
uz = uz1 + uz2
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params, plotlos=True)
def agu_convergence():
"""
Mesh Refinement demonstration for 1/4 cylinder large domain mesh
"""
points = '/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output_points.txt'
outputs = [
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/supercoarse/points.h5',
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/coarse/points.h5',
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/medium/points.h5',
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/fine/points.h5',
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/test/points.h5',
'/home/scott/research/models/pylith/3d/fialko2012/model3_agu/output/elastic/superfine/points.h5',
]
#number of elements
#supercoarse 40000/1000/1000 = 300360
#coarse 30000/1000/500 = 445841
#medium 25000/750/300 = 642087
#fine 20000/500/250 = 832400
#test? 30000/300/150 = 1170025
#superfine 10000/250/100 = 1955701
#max? 20000/250/50 = 2840658 #NOTE: Fialko max # elements=2.5mil
nel = np.array([300360, 445841, 642087, 832400, 1170025, 1955701])
RMSE = []
for pointsh5 in outputs:
x = np.sqrt(2) * 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(xoff=0, yoff=0, d=25e3, dP=50e6, a=2e3, mu=30e9, nu=0.25)
# Superposition of inflation and deflation source
ur1, uz1 = m.calc_mogi_dp(x, y, **params)
ur2, uz2 = m.calc_mogi_dp(x, y, d=75e3, a=5e3, mu=30e9, dP=-10e6)
ur = ur1 + ur2
uz = uz1 + uz2
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
ur_fem[x >= 60000] = -ur_fem[x >= 60000]
rmseR, rmseZ = calc_stats(x, ur, uz, ur_fem, uz_fem)
RMSE.append(rmseR + rmseZ)
RMSE = np.array(RMSE)
plt.figure()
plt.plot(nel / 10000, RMSE * 100, 'k.-', lw=2)
plt.title('Benchmark Mesh Accuracy')
plt.xlabel('# Elements [x1e4]')
plt.ylabel('RMSE [mm]')
plt.grid()
def mogi_tomography():
"""
Compare homogeneous solution w/ Wa Domain 150x100 (cylinder), depth=25km. NOTE: also a version
of mesh that includes topography.
"""
#UTURNCU MESH
surfaceh5 = '/home/scott/research/models/pylith/3d/uturuncu_layered/output/step01/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/uturuncu_layered/output/step01/points.h5'
points = '/home/scott/research/models/pylith/3d/uturuncu_layered/output_points.txt'
#Get analytic solution
#x = np.linspace(0,25e3,1e2)
#y = np.zeros_like(x)
x = 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(
xoff=0.0, #-2150.0,
yoff=0.0, #250.0,
d=26e3,
dP=33.0e6,
a=2000.0,
mu=30e9,
nu=0.25)
#ur,uz = m.calc_mogi_dp(x,y,d=4e3,dP=100e6,a=700,mu=30e9,nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params)
def mogi_DN2009_elastic():
"""
Reproduce elastic model from Del Negro 2009 for viscoelastic model. Domain is
50x50x35, source at 4km depth. Then use same source in slightly bigger mesh
for uturuncu UPLIFT modeling (Domain 100x100x50)
"""
#DEL NEGRO MESH
#surfaceh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output/surface.h5'
#pointsh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output/points.h5'
#points = '/home/scott/research/models/pylith/3d/delnegro2009/output_points.txt'
#UTURNCU MESH
surfaceh5 = '/home/scott/research/models/pylith/3d/uturuncu/output_DN2009/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/uturuncu/output_DN2009/points.h5'
points = '/home/scott/research/models/pylith/3d/uturuncu/output_points.txt'
#Get analytic solution
#x = np.linspace(0,25e3,1e2)
#y = np.zeros_like(x)
x = 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(xoff=0, yoff=0, d=4e3, dP=100e6, a=700, mu=30e9, nu=0.25)
#ur,uz = m.calc_mogi_dp(x,y,d=4e3,dP=100e6,a=700,mu=30e9,nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params)
def mogi_currenti():
"""
Reproduce Gilda Currenti's Mogi source benchmark using a Cubit Mesh
instead of LaGrit Mesh. Domain 100x100x50km, source at 3km depth
"""
surfaceh5 = '/home/scott/research/models/pylith/3d/mogi_currenti/output/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/mogi_currenti/output/points.h5'
#Get analytic solution
#x = np.linspace(0,35e3,1e2) #100 points
#x = np.arange(36)*1e3 #simple output_points
x = 1e3 * np.loadtxt(
'/home/scott/research/models/pylith/3d/mogi_currenti/output_points.txt',
usecols=[0])
y = np.zeros_like(x)
params = dict(xoff=0, yoff=0, d=3e3, dP=10e6, a=500.0, mu=4e9, nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Print out statistics
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params)
def mogi_uturuncu_moat():
"""
Mogi benchmark for a deep source in a larger domain. Domain 200x200x100km,
source at 10km. Can be used for Dipole Source Model.
"""
surfaceh5 = '/home/scott/research/models/pylith/3d/uturuncu_moat/output/elastic1/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/uturuncu_moat/output/elastic1/points.h5'
#Get analytic solution
#x = np.linspace(0,35e3,1e2) #100 points
#x = np.arange(36)*1e3 #simple output_points
x = 1e3 * np.loadtxt(
'/home/scott/research/models/pylith/3d/uturuncu/output_points.txt',
usecols=[0])
y = np.zeros_like(x)
# Different Mogi formulations, same result
#params = dict(xoff = 0,
# yoff = 0,
# d = 10e3,
# dV = 41e6,
# nu = 0.25)
#ur,uz = m.calc_mogi(x,y,**params)
#params = dict(xoff = 0,
# yoff = 0,
# d = 25e3,
# dP = 53e6,
# a = 1e3,
# mu = 4e9, #volcanic regions
# nu = 0.25)
#ur,uz = m.calc_mogi_dp(x,y,**params)
params = dict(
xoff=0,
yoff=0,
d=10e3, #5e3
dP=100e6,
a=1e3, #0.7e3
mu=30e9, #volcanic regions
nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x,
ur,
uz,
ur_fem,
uz_fem,
params,
plotlos=True,
plotcomp=False)
def mogi_PS2004_layered(depth=12e3):
"""
Compare Uturuncu uplift source basic domain (100x100x50) to
Pritchard & Simons 2004 elastic layering calculations propagator matrix
calculations
"""
#UTURNCU MESH
if depth == 12e3:
h5elastic = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004/points.h5'
h5m1 = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004_M1/points.h5'
h5m2 = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004_M2/points.h5'
elif depth == 18e3:
h5elastic = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004_18/points.h5'
h5m1 = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004_M1_18/points.h5'
h5m2 = '/home/scott/research/models/pylith/3d/uturuncu/output_PS2004_M2_18/points.h5'
points = '/home/scott/research/models/pylith/3d/uturuncu/output_points.txt'
#Get analytic solution
x = 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(
xoff=0,
yoff=0,
d=depth, #12e3,18e3
dP=100e6,
a=1000.0,
mu=33e9,
nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
x_e, ur_e, uz_e = pp.extract_points(h5elastic)
x_m1, ur_m1, uz_m1 = pp.extract_points(h5m1)
x_m2, ur_m2, uz_m2 = pp.extract_points(h5m2)
# Statistics & Graph for elastic solution
calc_stats(x, ur, uz, ur_e, uz_e)
plot_result(x, ur, uz, ur_e, uz_e, params)
#redo figure 3
if depth == 12e3: norm = uz_m2.max()
elif depth == 18e3: norm = uz_m1.max()
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(x / 1000, uz_e / norm, 'k-', lw=2, label='homogeneous')
ax.plot(x / 1000, uz_m1 / norm, 'b-', lw=2, label='M1')
ax.plot(x / 1000, uz_m2 / norm, 'r-', lw=2, label='M2')
plt.xlabel('Distance [km]')
plt.ylabel('Normalized Displacement')
plt.title('Elastic Layer Test; Depth = {}km'.format(depth / 1e3))
plt.legend()
plt.show()
def mogi_LZ2003_layered():
"""
Fully 3D model for investigating Uplift source in the mid-crust. Used to
investigate layering proposed by Leidig & Zandt 2003, and seismic tomography
from Matt Haney. Domain 150x100 (cylinder), depth=25km. NOTE: also a version
of mesh that includes topography.
"""
#UTURNCU MESH
surfaceh5 = '/home/scott/research/models/pylith/3d/uturuncu_layered/output/step01/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/uturuncu_layered/output/step01/points.h5'
points = '/home/scott/research/models/pylith/3d/uturuncu_layered/output_points.txt'
#Get analytic solution
#x = np.linspace(0,25e3,1e2)
#y = np.zeros_like(x)
x = 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(
xoff=0.0, #-2150.0,
yoff=0.0, #250.0,
d=26e3,
dP=33.0e6,
a=2000.0,
mu=30e9,
nu=0.25)
#ur,uz = m.calc_mogi_dp(x,y,d=4e3,dP=100e6,a=700,mu=30e9,nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params)
def mogi_reservoir_depth(step, depth, dP):
"""
Show benchmark accuracy for variety of depths
"""
#UTURNCU MESH
surfaceh5 = '/home/scott/research/models/pylith/3d/uturuncu_cylinder/output/step0{}/surface.h5'.format(
step)
pointsh5 = '/home/scott/research/models/pylith/3d/uturuncu_cylinder/output/step0{}/points.h5'.format(
step)
points = '/home/scott/research/models/pylith/3d/uturuncu_cylinder/output_points.txt'
#Get analytic solution
#x = np.linspace(0,25e3,1e2)
#y = np.zeros_like(x)
x = 1e3 * np.loadtxt(points, usecols=[0])
y = np.zeros_like(x)
params = dict(
xoff=0.0, #-2150.0,
yoff=0.0, #250.0,
d=depth,
dP=dP,
a=2000.0,
mu=30e9,
nu=0.25)
#ur,uz = m.calc_mogi_dp(x,y,d=4e3,dP=100e6,a=700,mu=30e9,nu=0.25)
ur, uz = m.calc_mogi_dp(x, y, **params)
# Get FEM solution
#verts, data, tris = load_h5(surfaceh5)
#X, R, Z = extract_profile_axisym(verts, data, elmtsize=500)
x_fem, ur_fem, uz_fem = pp.extract_points(pointsh5)
# Statistics & Graph
calc_stats(x, ur, uz, ur_fem, uz_fem)
plot_result(x, ur, uz, ur_fem, uz_fem, params, plotcomp=True)
def mogi_DN2009_shell():
"""
DelNegro2009 Mogi source surrounded by viscoelastic shell in an elastic
halfspace, compared to analytic solution in Segall 2010 Ch.7
"""
surfaceh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output_shell2/surface.h5'
pointsh5 = '/home/scott/research/models/pylith/3d/delnegro2009/output_shell2/points.h5'
#Get analytic solution
x = 1e3 * np.loadtxt(
'/home/scott/research/models/pylith/3d/delnegro2009/output_points.txt',
usecols=[0])
y = np.zeros_like(x)
params = dict(
xoff=0,
yoff=0,
d=4e3, #m
dP=100e6, #Pa
a=700, #m
nu=0.25,
mu=30e9)
fig = plt.figure()
plt.suptitle('DelNegro Fig 4', fontsize=16)
# NOTE: 0,1,2,3 because output skip=3
#timesteps = [0,1,2,3]
timesteps = [0, 10, 20, 40]
# Loop through output points
for i, tstep in enumerate(timesteps):
print(i, tstep)
x_fem, ur_fem, uz_fem, time = pp.extract_points_visco(pointsh5, tstep)
tau0 = 2e16 / (0.5 * 30e9)
time_days = time / 8.64e4
time_maxwell = time / tau0
ur, uz = m.calc_mogi_dp(x, y, **params)
ax = fig.add_subplot(2, 2, i + 1)
ax.plot(x, uz, 'b-', lw=2, label='analytic')
ax.plot(x, ur, 'g-', lw=2)
ax.plot(x,
uz_fem,
marker='o',
ls='None',
lw=2,
mec='b',
mfc='None',
label='FEM')
ax.plot(x, ur_fem, marker='o', ls='None', lw=2, mec='g', mfc='None')
#ax.set_ylim(-0.01,0.07)
ax.set_ylim(0, 0.1)
ax.set_title('{:g} days, t/tau={:g}'.format(time_days, time_maxwell))
ax.grid(True)
if i == 2:
ax.set_xlabel('Radial distance [m]')
ax.set_ylabel('displacement [m]')
# Statistics & Graph
print("time={}\n".format(time_days))
calc_stats(x, ur, uz, ur_fem, uz_fem)
# Residual plots for last timestep
#plot_result(x,ur,uz,ur_fem,uz_fem)
#plt.xlim(0,35000)
plt.legend()
def mogi_dipole(track='west'):
"""
Based on Tom's inversion for dipole source
"""
outdir = '/home/scott/research/models/pylith/3d/agu2013_noapmb/output/step01'
pointsh5 = '/home/scott/research/models/pylith/3d/agu2013_noapmb/output/step01/points.h5'
pp.plot_profile(outdir,
comp2los=track,
adjustRadial=True,
yscale=1e-2,
xscale=1e3)
# Get FEM solution arrays (displacements should be exact at nodal points!)
x_fem, y_fem, z_fem, ux_fem, uy_fem, uz_fem = pu.extract_points(pointsh5)
r_fem = np.hypot(x_fem, y_fem)
ur_fem = np.hypot(ux_fem, uy_fem)
ur_fem = pu.radial2negative(ur_fem)
# Inflation Source
params = dict(
xoff=0,
yoff=0,
d=30e3, #increasing depth is positive
dP=20e6,
a=3e3,
mu=30e9,
nu=0.25)
ur1, uz1 = m.calc_mogi_dp(x_fem, y_fem, **params)
# Deflation Source
ur2, uz2 = m.calc_mogi_dp(x_fem, y_fem, d=70e3, a=4e3, mu=30e9, dP=-20e6)
# Superposition of analytic solutions
ur = ur1 + ur2
uz = uz1 + uz2
#pu.calc_stats(r_fem, ur,uz,ur_fem,uz_fem)
if track == 'west':
r_fem = -r_fem
ur = -ur
# Plot elastic benchmark solution on top of active plot!
#plt.figure()
#plt.gcf()
every = 2 #NOTE: this is a bit cleaners, avoid too much clutter...
plt.plot(r_fem / 1e3,
uz * 1e2,
marker='o',
markevery=every,
ls='None',
lw=2,
ms=5,
mew=1,
mec='b',
mfc='None',
label='analytic')
plt.plot(r_fem / 1e3,
ur * 1e2,
marker='s',
markevery=every,
ls='None',
lw=2,
ms=5,
mew=1,
mec='b',
mfc='None')
datalos = '/home/scott/research/models/pylith/scripts/collapsed_profile.txt'
xlos, ulos, stdlos = np.loadtxt(
datalos, unpack=True) #points in cm & in km from center!
if track == 'west':
xlos = -xlos
plt.plot(xlos, ulos, 'k.', ls='None', label='InSAR')
plt.show()