scikit-gstat

SciKit-GStat - Geostatistics

Quick Reference

import skgstat as skg
import numpy as np

# Create variogram
V = skg.Variogram(coordinates=coords, values=values, n_lags=15)

# Fit model
V.model = 'spherical'
print(f"Range: {V.parameters[0]:.2f}, Sill: {V.parameters[1]:.2f}")

# Kriging interpolation
ok = skg.OrdinaryKriging(V)
predictions = ok.transform(grid_coords)

Key Classes

Class	Purpose
Variogram	Empirical and theoretical variograms
OrdinaryKriging	Interpolation with spatial correlation
DirectionalVariogram	Anisotropic variograms
SpaceTimeVariogram	Spatio-temporal analysis

Essential Operations

Create and Fit Variogram

import skgstat as skg

V = skg.Variogram(
    coordinates=coords,      # (n, 2) array of x, y
    values=values,           # (n,) array of measurements
    n_lags=15,
    maxlag='median'          # or specific distance
)

# Fit model: 'spherical', 'exponential', 'gaussian', 'matern', 'stable'
V.model = 'spherical'

# Get parameters
print(f"Range: {V.parameters[0]:.2f}")
print(f"Sill: {V.parameters[1]:.2f}")
print(f"Nugget: {V.parameters[2]:.2f}")
print(f"RMSE: {V.rmse:.4f}")

Ordinary Kriging

import skgstat as skg
import numpy as np

V = skg.Variogram(coords, values, model='spherical')
ok = skg.OrdinaryKriging(V)

# Create prediction grid
x = np.linspace(0, 100, 50)
y = np.linspace(0, 100, 50)
xx, yy = np.meshgrid(x, y)
grid_coords = np.column_stack([xx.ravel(), yy.ravel()])

# Predict
predictions = ok.transform(grid_coords)
Z = predictions.reshape(xx.shape)

# Get variance
ok.return_variance = True
predictions, variance = ok.transform(grid_coords)

Directional Variogram

import skgstat as skg

DV = skg.DirectionalVariogram(
    coordinates=coords,
    values=values,
    azimuth=45,          # Direction in degrees
    tolerance=22.5,      # Angular tolerance
    bandwidth='q33'      # Perpendicular bandwidth
)

# Check anisotropy
for az in [0, 45, 90, 135]:
    DV.azimuth = az
    print(f"Azimuth {az}: Range = {DV.parameters[0]:.2f}")