def predict_seismic(moho, lat, lon):
"""
Calculate the predicted Moho depth at the seismic points
from the estimated model.
Values are interpolated onto (lat, lon) from the given 'moho'.
"""
estimated = gridder.interp_at(moho.clat.ravel(), moho.clon.ravel(), moho.relief,
lat, lon, extrapolate=True)
return estimated
area = (lat.min(), lat.max(), lon.min(), lon.max())
sediments = {
'lon': lon,
'lat': lat,
'thickness': thickness,
'shape': shape,
'area': area
}
# Create top and bottom of sediments layer
# ----------------------------------------
# Regrid dem to sediments regular grid
sediments_top = gridder.interp_at(topography["lat"],
topography["lon"],
topography["topo"],
sediments["lat"],
sediments["lon"],
algorithm="linear")
sediments_top[np.isnan(sediments["thickness"])] = np.nan
# Compute the median of the topography on basin points
nans = np.isnan(sediments_top)
sediments_top[~nans] = np.median(sediments_top[~nans])
# Add top and bottom arrays to sediments dictionary
sediments["top"] = sediments_top
sediments["bottom"] = sediments_top - sediments["thickness"]
# Create Tesseroids model of the sediments layer
# ----------------------------------------------
bottom, top = sediments["bottom"].copy(), sediments["top"].copy()
top[nans] = 0
bottom[nans] = 0
p1, p2 = p
#%%
# find point position with respect to line equation defined by p1 and p2
U_a, p_a, bool_above, U_b, p_b = MALM.isabove(xp, yp, Uini, np.array(p1),
np.array(p2))
# Compute p1 and p2 line equation ax + by + c = 0
a, b, c = MALM.slope(p1, p2)
# Mirror points with respect to p1p2 line
Umirror, xy_mirror = MALM.mirrorU_alongLine(U_a, p_a, bool_above, a, b, c)
U_a_int = gridder.interp_at(xy_mirror[:, 0],
xy_mirror[:, 1],
Umirror,
xp,
yp,
algorithm='cubic',
extrapolate=True)
# U_mirror_int = np.copy(U_a_int)
U_mirror_int = np.copy(Uini)
U_mirror_int[np.where(bool_above == True)[0]] = U_a_int[np.where(
bool_above == True)[0]]
plt.figure()
plt.scatter(xp, yp, c=U_mirror_int, cmap='viridis', vmax=0.25)
plt.colorbar()
plt.axis('square')
plt.show()
#%% choose raw or mirrored field
p1, p2 = p
#%%
# find point position with respect to line equation defined by p1 and p2
U_a, p_a, bool_above, U_b, p_b = MALM.isabove(xp_r, yp_r, Uini, np.array(p1),
np.array(p2))
# Compute p1 and p2 line equation ax + by + c = 0
a, b, c = MALM.slope(p1, p2)
# Mirror points with respect to p1p2 line
Umirror, xy_mirror = MALM.mirrorU_alongLine(U_a, p_a, bool_above, a, b, c)
U_a_int = gridder.interp_at(xy_mirror[:, 0],
xy_mirror[:, 1],
Umirror,
xp_r,
yp_r,
algorithm='nearest',
extrapolate=True)
U = np.copy(Uini)
U[np.where(bool_above == True)[0]] = U_a_int[np.where(bool_above == True)[0]]
plt.figure()
plt.scatter(xp_r, yp_r, c=U, cmap='viridis', vmax=0.25)
plt.colorbar()
plt.axis('square')
plt.show()
plt.figure()
plt.scatter(xp_r, yp_r, c=U_a_int, cmap='viridis', vmax=0.25)
plt.colorbar()
plt.axis('square')
def load_MALM_LandfillPorto(path,
filename,
shape=None,
field=True,
interp=True,
radius=20):
if field == True:
file = open(path + 'PortoM.pkl', 'rb')
u = pickle._Unpickler(file)
u.encoding = 'latin1'
data_struct = u.load()
data_struct['HZ']
xA = (data_struct['HZ'][0][0] + data_struct['HZ'][0][1]) / 2
yA = data_struct['HZ'][0][2]
AnoPos = [xA, yA]
# x , y, z, U_raw = load_field_u_LandfillPorto(path, filename + '_uz0.dat')
print('load' + str(path + filename) + '_uz0_grid.dat')
x, y, z, U_raw = np.loadtxt(path + filename + '_uz0_grid.dat',
unpack=True)
# x , y, z = mesh_xyz
# ------------------------------- Plot the data
# plt.figure()
# plt.scatter(x, y,c=U_raw, cmap='viridis',vmin=None, vmax=1)
# plt.colorbar()
# plt.axis('square')
# plt.show()
else:
U_raw = load_obs(path, filename + '.txt') # load observation data
RemLineNb, Injection, coordE, pointsE = load_geom(
path) # find the geom file in the folder path
nb, x, y, z = np.array(coordE[:-3]).T
AnoPos = []
data_struct = []
# path2files="example_2add_later/Landfill_3d/Ano_0_E13/" # Test elec outside landfill
# # path2files="example_2add_later/Landfill_3d/Ano_0_E89/" # Test 2 elec outside landfill
# # path2files="example_2add_later/Landfill_3d/Ano_0_EA/" # Real A position without anomaly
# # path2files="example_2add_later/Landfill_3d/Ano_1_BH_EA/" # Real A position with big hole anomaly
# path2files="example_2add_later/Landfill_3d/Ano_1_SH_EA/" # Real A position with small hole anomaly
# # path2files="examples/Landfill_3d/RealData_Ano_0/" # Real data / without anomaly
# # path2files="examples/Landfill_3d/RealData_Ano_1_SH/" # Real data / with (small) anomaly
# -----------------------------------------------------------------------------------------
if shape is None:
shape = (31, 31)
else:
shape = shape
# print(AnoPosxAyA)
coords_liner, p1, p2, B = definep1p2(path, radius, AnoPos)
xnew, ynew = creategrid(coords_liner, B, shape)
# ------------- Raw data -----------------------------------------
# pEXP.plot_line(x, y, U_raw,p1,p2, title='U_raw', interp=interp) #, vmin=0.01, vmax=0.1,
print(len(x))
print(len(U_raw))
print(len(xnew))
# ------------- Interpolation -----------------------------------------
U_int = gridder.interp_at(x,
y,
U_raw,
xnew,
ynew,
algorithm='cubic',
extrapolate=True)
# xp,yp,U_int = gridder.interp(xnew,ynew,U,shape)
# pEXP.plot_line(xnew, ynew, U_int,p1,p2,title='U_int', interp=interp)
# ------------- correction of B + interpolation ----------------------
# remove influence of B
if field == False:
Ucor = uEXP.cor_field_B(x, y, z, U_raw, B, rho=10)
else:
Ucor = U_raw
Ucor_int = gridder.interp_at(x,
y,
Ucor,
xnew,
ynew,
algorithm='cubic',
extrapolate=True)
# xp,yp,Ucor_int = gridder.interp(xnew,ynew,Ucorint_tmp,shape)
# pEXP.plot_line(xnew, ynew, Ucor_int,p1,p2, title='U_cor&int', interp=interp)
# --------------- plot 2d maps -----------------------------------------
# plt.figure(figsize=(20,10))
# plt.subplot(2,2,1)
# plt.tricontourf(x, y, U_raw, 50, cmap='viridis', vmin = 0.01, vmax = 1)
# cbar = plt.colorbar(orientation='vertical', aspect=50)
# cbar.set_label('$u_{B}$ (V)')
# plt.plot(coords_liner[:,0],coords_liner[:,1],'*-')
# plt.axis('square')
# plt.subplot(2,2,2)
# plt.tricontourf(x, y, Ucor, 50, cmap='viridis', vmin = 0.01, vmax = 1)
# cbar = plt.colorbar(orientation='vertical', aspect=50)
# cbar.set_label('$u_{Bcor}$ (V)')
# plt.plot(coords_liner[:,0],coords_liner[:,1],'*-')
# plt.axis('square')
# plt.subplot(2,2,3)
# plt.tricontourf(xnew, ynew, U_int, 50, cmap='viridis', vmin = 0.01, vmax = 1)
# cbar = plt.colorbar(orientation='vertical', aspect=50)
# cbar.set_label('$u_{Binterp}$ (V)')
# plt.plot(coords_liner[:,0],coords_liner[:,1],'*-')
# plt.axis('square')
# plt.subplot(2,2,4)
# plt.tricontourf(xp, yp, Ucor_int, 50, cmap='viridis', vmin = 0.01, vmax = 1)
# cbar = plt.colorbar(orientation='vertical', aspect=50)
# cbar.set_label('$u_{Bint&cor}$ (V)')
# plt.plot(coords_liner[:,0],coords_liner[:,1],'*-')
# plt.axis('square')
# ---------------------------------------
zp = 0
z = zp
# ---------------------------------------
p = [p1, p2]
U = [U_raw, Ucor, U_int, Ucor_int]
coord_xyz = [x, y, z]
coord_xyz_int = [xnew, ynew, zp]
max_elevation = 50
return coord_xyz, coord_xyz_int, U, coords_liner, shape, max_elevation, p, data_struct
#%% Mirror field against p1p2
# find point position with respect to line equation defined by p1 and p2
U_af, p_af, bool_abovef, U_bf, p_bf = MALM.isabove(xf, yf, uf, np.array(p1),
np.array(p2))
# Compute p1 and p2 line equation ax + by + c = 0
af, bf, cf = MALM.slope(p1, p2)
# Mirror points with respect to p1p2 line
Umirrorf, xy_mirrorf = MALM.mirrorU_alongLine(U_af, p_af, bool_abovef, af, bf,
cf)
U_a_intf = gridder.interp_at(xy_mirrorf[:, 0],
xy_mirrorf[:, 1],
Umirrorf,
xf,
yf,
algorithm='cubic',
extrapolate=True)
# U_mirror_int = np.copy(U_a_int)
U_mirror_intf = np.copy(uf)
U_mirror_intf[np.where(bool_abovef == True)[0]] = U_a_intf[np.where(
bool_abovef == True)[0]]
xf_mirror = np.hstack([xy_mirrorf[:, 0], p_af[:, 0]])
yf_mirror = np.hstack([xy_mirrorf[:, 1], p_af[:, 1]])
uf_mirror = np.hstack([Umirrorf, U_af])
#%% choose raw or mirrored field
Uf = np.copy(uf)
plt.xlabel('x (m)')
plt.ylabel('y (m)')
prl = 60
# shape = shape (max(xp)-min(xp))/
# shape = (115,115)
xint_scipy, yint_scipy = gridder.regular(
(min(xp) - prl, max(xp) + prl, min(yp) - prl, max(yp) + prl), shape=shape)
#%% Solution 1
# extrapolate False and fill with 0 before derivative - mask them later on
U_int_scipy = gridder.interp_at(xp,
yp,
U,
xint_scipy,
yint_scipy,
algorithm='cubic',
extrapolate=False)
InterpData = np.array([xint_scipy, yint_scipy, U_int_scipy]).T
where_are_NaNs = np.isnan(InterpData)
InterpData[where_are_NaNs] = 0.0074
xint_scipy, yint_scipy, U_int_scipy = InterpData.T
plt.scatter(xint_scipy, yint_scipy)
plt.scatter(xp, yp)
#%% Solution 2
# Extrapolate = True
# U_int_scipy = gridder.interp_at(xp,yp,U, xint_scipy, yint_scipy, algorithm='cubic', extrapolate=True)