Issue with Spatial Random Field Grid and Contour Plotting + CSV Output for Kriging Input #366
-
|
Hi all, I'm computing a spatial random field using gstools on a 121-point grid (11x11) over a 100x100 area. However, in the resulting plot, I'm seeing sharp transitions and weird triangles instead of a smooth field. Is there something wrong with how I'm defining the grid? Also, I'm using matplotlib for contour plotting but have to transpose the field to match the output of SRF.plot(). Is this normal behavior, or am I missing something? Additionally, I want to use the generated field values at 121 points as input to a kriging model. Is there an easy way to convert the field into a CSV file with columns x, y, and variable? The input to the kriging model looks like this: |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments
-
|
Hey there, for me the plot looks as expected. Remember, that you only defined 11 points along each axis, so one every 10 meters (or whatever unit). For your second point: In order to export the field to CSV, you need to generate the points of the grid from the given axis and then flatten the field to export it. Using pandas, this could look like this: import gstools as gs
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
x = y = np.linspace(0, 100, 11)
model = gs.Exponential(dim=2, var=20, len_scale=25, nugget=5)
srf = gs.SRF(model, mean=50, seed=19841201)
field = srf.structured((x, y))
# Generate grid values from axes
pos = gs.tools.generate_grid((x, y))
# Create a DataFrame
data = pd.DataFrame({'x': pos[0], 'y': pos[1], 'field': field.flatten()})
# Export to CSV
data.to_csv('output.csv', index=False)This gives you: This could now be used for kriging like this: import gstools as gs
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
data = pd.read_csv('output.csv')
cond_pos = (data["x"].to_numpy(), data["y"].to_numpy())
cond_val = data["field"].to_numpy()
xx = yy = np.linspace(0, 100, 101)
model = gs.Exponential(dim=2, var=20, len_scale=25, nugget=5)
krig = gs.krige.Ordinary(model, cond_pos, cond_val)
krige_field, krige_var = krig.structured((xx, yy))
krig.plot(field="field")
krig.plot(field="krige_var")
plt.show()
Hope this helps. Cheers, Sebastian |
Beta Was this translation helpful? Give feedback.
-
|
But a CSV exporter could be a handy idea. Created an issue for this: #367 |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your response. It works as intended now. |
Beta Was this translation helpful? Give feedback.
Hey there,
for me the plot looks as expected. Remember, that you only defined 11 points along each axis, so one every 10 meters (or whatever unit).
If you want the plot to look smoother, you should increase the points along each axis.
For your second point: In order to export the field to CSV, you need to generate the points of the grid from the given axis and then flatten the field to export it. Using pandas, this could look like this: