Kriging for geotif image

[Empanag]

Hello, i m trying to use kriging to tiff image. the image is that below with number 1 that can be used as an array, it is regular grid. the 2nd image is the one that saga simple kriging module generates. I 'd like to produce a similar product using python. Is it possible somehow GSTools?

[MuellerSeb]

Hey there,

thanks for your interest in GSTools. This is of course possible. Could you provide dummy data, so I can help you with setting up the routines?

It should look something like this in the end:

import gstools as gs
import numpy as np

# your input data as x, y, data points
x, y, data = load_your_data()
cond_pos = [x, y]
mean = np.mean(data)
# simple gaussian model since your plot looks rather smooth
model = gs.Gaussian(dim=2)
krige = gs.krige.Simple(model, cond_pos, data, mean=mean, fit_variogram=True)

# output on a structured grid (image) with specified axis
xaxis, yaxis = get_your_axes()
field = krige.structured([xaxis, yaxis], return_var=False)

Cheers,
Sebastian

[MuellerSeb]

Sorry, but I don't fully understand your question. To make it "unstructured" you could simply reshape it, couldn't you?

[Empanag]

i m running the code as it is and the field is running without finish the process. i ll try to explain you.

i m using two 1-D arrays for x and y and one 2-D array for data to run gs.krige.Simple()
after this i am using the krige.structured() to create an image with x and y as axes

# output on a structured grid (image) with specified axis
xaxis, yaxis = x, y
field = krige.structured([xaxis, yaxis], return_var=False)

what type of data use the krige? can i take the new transformed data from directly krige, a 2-D array that is the same shape with data but with the new values? or i have to use the krige.structured() to take a structure image?
i let the script run over 30 minutes for krige.structured() and haven't finished yet.

inside the file i have a tiff with the data and a txt with the array of data if you can used them so you can figure out or can you send me any test data to see the format?

kriging.zip

[Empanag]

@MuellerSeb i finally had some result but i have two problems,

first the execution time of # output on a structured grid (image) with specified axis is over 2 hours

and the result (that is below) seems very different compared to input data and expected data that i have attached you above

[Empanag]

i m trying to find a solution to reduce the execute time. Is there any way to reduce the range of the points that the new values is calculated from?

[MuellerSeb]

Nearest Neighboorhood Kriging or local Kriging is not available at the moment, but is on the TODO list: GeoStat-Framework/GSTools-Core#13

One other option at the moment could be to use OpenMP (enable by default when installing via conda).