Extract a 1D profile from 2D and plot it#

Section author: Feliks Kiszkurno (Helmholtz Centre for Environmental Research GmbH - UFZ)

import matplotlib.pyplot as plt
import numpy as np

import ogstools as ogs
from ogstools import examples

Single fracture#

Define a profile line by providing a list of points in x, y, z coordinates and load an example data set:

mesh = examples.load_meshseries_HT_2D_XDMF().mesh(-1)

profile_HT = np.array([[4, 2, 0], [4, 18, 0]])
mesh_sp, mesh_kp = ogs.meshlib.sample_polyline(
    mesh, ["pressure", "temperature"], profile_HT
)

It has returned a pandas DataFrame containing all information about the profile and a numpy array with the position of the “knot-points”. Let’s investigate the DataFrame first:

mesh_sp.head(10)
x y z pressure temperature dist dist_in_segment
0 4.0 2.00 0.0 0.498440 79.850025 0.00 0.00
1 4.0 2.16 0.0 0.498725 79.850031 0.16 0.16
2 4.0 2.32 0.0 0.499150 79.850021 0.32 0.32
3 4.0 2.48 0.0 0.499663 79.849999 0.48 0.48
... ... ... ... ... ... ... ...
6 4.0 2.96 0.0 0.501203 79.849934 0.96 0.96
7 4.0 3.12 0.0 0.501717 79.849913 1.12 1.12
8 4.0 3.28 0.0 0.502230 79.849891 1.28 1.28
9 4.0 3.44 0.0 0.502936 79.849931 1.44 1.44

10 rows × 7 columns



We can see the spatial coordinates of points on the profile (“x”, “y”, “z” - columns), distances from the beginning of the profile (“dist”) and within current segment (“dist_in_segment”). Note, that since we defined our profile on only two points, there is only one segment, hence in this special case columns dist and dist_in_segment are identical. At the end of the DataFrame we can can find two columns with the variables that we are interested in: “temperature” and “pressure”. Each occupies one column, as those are scalar values. Using columns “dist”, “pressure” and “temperature” we can easily plot the data:

fig, ax = plt.subplots(1, 1, figsize=(7, 5))
ax = mesh.plot_linesample(
    x="dist",
    variable="pressure",
    profile_points=profile_HT,
    ax=ax,
    fontsize=15,
)
ax_twinx = ax.twinx()
ax_twinx = mesh.plot_linesample(
    x="dist",
    variable="temperature",
    profile_points=profile_HT,
    ax=ax_twinx,
    fontsize=15,
)
ogs.plot.utils.color_twin_axes(
    [ax, ax_twinx],
    [ogs.variables.pressure.color, ogs.variables.temperature.color],
)
fig.tight_layout()
plot sample mesh line

What happens when we are interested in a vector variable? We can see it in the following example using the Darcy velocity:

mesh_sp, mesh_kp = ogs.meshlib.sample_polyline(
    mesh, "darcy_velocity", profile_HT
)
mesh_sp.head(5)
x y z darcy_velocity_0 darcy_velocity_1 dist dist_in_segment
0 4.0 2.00 0.0 1.169737e-93 -2.231183e-94 0.00 0.00
1 4.0 2.16 0.0 1.196810e-93 -2.421097e-94 0.16 0.16
2 4.0 2.32 0.0 1.237164e-93 -2.637041e-94 0.32 0.32
3 4.0 2.48 0.0 1.285968e-93 -2.869550e-94 0.48 0.48
4 4.0 2.64 0.0 1.334773e-93 -3.102060e-94 0.64 0.64


Now we have two columns for the variable. The Darcy velocity is a vector, therefore “sample_over_polyline” has split it into two columns and appended the variable name with increasing integer. Note, that this suffix has no physical meaning and only indicates order. It is up to user to interpret it in a meaningful way. By the OpenGeoSys conventions, “darcy_velocity_0” will be in the x-direction and “darcy_velocity_1” in y-direction.

Elder benchmark#

In this example we will use a Variable object from the ogstools to sample the data. This allows “sample_over_polyline” to automatically convert from the “data_unit” to the “output_unit”:

profile_CT = np.array([[47.0, 1.17, 72.0], [-4.5, 1.17, -59.0]])
mesh = examples.load_meshseries_CT_2D_XDMF().mesh(11)
mesh_sp, mesh_kp = ogs.meshlib.sample_polyline(
    mesh, ogs.variables.saturation, profile_CT
)

As before we can see the profile parameters and propertiy values in a DataFrame:

mesh_sp.head(5)
x y z Si dist dist_in_segment
0 47.000 1.17 72.00 79.601252 0.000000 0.000000
1 46.485 1.17 70.69 73.478046 1.407595 1.407595
2 45.970 1.17 69.38 70.034378 2.815191 2.815191
3 45.455 1.17 68.07 67.833706 4.222786 4.222786
4 44.940 1.17 66.76 67.245709 5.630382 5.630382


This time we will prepare more complicated plot showing both the mesh and the profile.

fig, ax = mesh.plot_linesample_contourf(
    ogs.variables.saturation, profile_CT, resolution=100
)
plot sample mesh line

THM#

It is also possible to obtain more than one variable at the same time using more complex profiles. They can be constructed by providing more than 2 points. With those points:

profile_THM = np.array(
    [
        [-1000.0, -175.0, 6700.0],  # Point A
        [-600.0, -600.0, 6700.0],  # Point B
        [100.0, -300.0, 6700.0],  # Point C
        [3500, -900.0, 6700.0],  # Point D
    ]
)

the profile will run as follows:

\[\text{AB} \rightarrow \text{BC} \rightarrow \text{CD}\]

Point B will at the same time be the last point in the first segment AB and first one in second segment BC, however in the returned array, it will occur only once. For this example we will use a different dataset:

mesh = examples.load_meshseries_THM_2D_PVD().mesh(-1)
ms_THM_sp, dist_at_knot = ogs.meshlib.sample_polyline(
    mesh,
    [ogs.variables.pressure, ogs.variables.temperature],
    profile_THM,
    resolution=100,
)

Again, we can investigate the returned DataFrame, but this time we will have a look at its beginning:

ms_THM_sp.head(5)
x y z pressure temperature dist dist_in_segment
0 -1000.000000 -175.000000 6700.0 2.230181 8.832194 0.000000 0.000000
1 -969.230769 -207.692308 6700.0 2.551761 8.828883 44.894683 44.894683
2 -938.461538 -240.384615 6700.0 2.873396 8.825926 89.789366 89.789366
3 -907.692308 -273.076923 6700.0 3.195135 8.825316 134.684048 134.684048
4 -876.923077 -305.769231 6700.0 3.516919 8.821331 179.578731 179.578731


and end:

ms_THM_sp.tail(10)
x y z pressure temperature dist dist_in_segment
92 3075.000000 -825.000000 6700.0 0.0 32.327025 4366.176575 3020.968388
93 3122.222222 -833.333333 6700.0 0.0 35.663869 4414.128454 3068.920267
94 3169.444444 -841.666667 6700.0 0.0 40.358542 4462.080333 3116.872146
95 3216.666667 -850.000000 6700.0 0.0 44.984874 4510.032212 3164.824025
... ... ... ... ... ... ... ...
98 3358.333333 -875.000000 6700.0 0.0 34.045579 4653.887850 3308.679663
99 3405.555556 -883.333333 6700.0 0.0 31.436554 4701.839729 3356.631542
100 3452.777778 -891.666667 6700.0 0.0 28.867132 4749.791608 3404.583421
101 3500.000000 -900.000000 6700.0 0.0 26.706150 4797.743487 3452.535300

10 rows × 7 columns



Note, that unlike in the first example, here the columns “dist” and “dist_in_segment” are not identical, as this time profile consists of multiple segments. The following figure illustrates the difference:

plt.rcdefaults()
ax: plt.Axes
fig, ax = plt.subplots(1, 1, figsize=(7, 3))
ax.plot(ms_THM_sp["dist"], label="dist")
ax.plot(ms_THM_sp["dist_in_segment"], label="dist_in_segment")
ax.set_xlabel("Point ID / -")
ax.set_ylabel("Distance / m")
ax.legend()
fig.tight_layout()
plot sample mesh line

The orange line returns to 0 twice. It is because of how the overlap of nodal points between segments is handled. A nodal point always belongs to the segment it starts: point B is included in segment BC but not AB and point C in CD but not in in BC. The following figure shows the profile on the mesh:

plt.rcdefaults()

fig, ax = mesh.plot_linesample_contourf(
    [ogs.variables.pressure, ogs.variables.temperature],
    profile_THM,
    resolution=100,
)
plot sample mesh line

Total running time of the script: (0 minutes 2.372 seconds)