Open Source Spatial Data Analytics in Python with GeostatsPy
Introduction to GeostatsPy
Lecture outline . . .
• Variogram Modeling
• Interactive Demo with GeostatsPy
• Workflow with GeostatsPy
Michael Pyrcz, The University of Texas at Austin
Motivation
• After we calculate / quantify spatial continuity we need to model it
for spatial prediction.
Variability in Trend 23%
0.72
Variability Stationary
77%
Michael Pyrcz, The University of Texas at Austin
Recorded
Lectures
Michael Pyrcz, The University of Texas at Austin
Open Source Spatial Data Analytics in Python with GeostatsPy
Introduction to GeostatsPy
Lecture outline . . .
• Variogram Modeling
Michael Pyrcz, The University of Texas at Austin
Reasons for
Variogram Modeling
1. Need to know the variogram for all possible h lags, distances
and directions – not just the ones calculated
All Distances All Directions
a specific lag
non-orthogonal offset may not
( ) ( ) exist in our data!
y (m)
between points
( )
lag distance, x (m)
2. Incorporate additional geological knowledge (such as analogue
information or information on directions of continuity …)
3. The variogram model must be positive definite (a legitimate
measure of distance), that is, the variance of any linear
combination must be positive
Michael Pyrcz, The University of Texas at Austin Slide modified from Clayton Deutsch’s MinE 612 Class
Common Variogram
Models
No spatial correlation Commonly
Should be a small encountered
component of the
overall variance variogram shape
Similar to Implies short scale
spherical but continuity; parabolic
rises more behavior at the
steeply and origin, instead of
reaches the sill linear
asymptotically
For periodic
For periodic variables, when
variables the period is not
regular
Variogram
Michael examples
Pyrcz, from Pyrcz
The University and Deutsch,
of Texas 2014.
at Austin
Variograms Models with
Nested Structures
( )
= + +
• The addition of
positive definite
variogram structures =
is positive definite. ( )
• Each structure covers
a proportion of the sill. Structure 1
• For each structure we
can change the: +
( )
– orientation
– range in major and Structure 2
minor
• We are explaining +
( )
the spatial
components of the Structure 3
variance!
Michael Pyrcz, The University of Texas at Austin
Variogram Modeling
Example
Major Direction Azimuth 090 Minor Direction Azimuth 000
1.0 1.0
( )
0.8 0.8
0.2 0.2
0 250 1000 0 400 1000
Structure Type c amaj amin
1 Nugget 0.2 - -
2 Spherical 0.8 250 400
3 Spherical 0.2 9999 400
4
5
ℎ = . + 0. ⋅ ℎ + 0. ⋅ ℎ
Michael Pyrcz, The University of Texas at Austin
Open Source Spatial Data Analytics in Python with GeostatsPy
Introduction to GeostatsPy
Lecture outline . . .
• Interactive Demo with GeostatsPy
Michael Pyrcz, The University of Texas at Austin
Interactive Variogram Calculation
Demonstration with GeostatsPy
Let’s model variograms:
• normal score transformed
(standard Gaussian) porosity
Some Hints:
• don’t change the variogram
calculation
• find lowest common nugget
effect
• add a structure to get to
pseudo sill
• model remainder with very
large range in major direction Interactive Python Jupyter variogram modeling
(Interactive_Variogram_Modeling.ipynb).
Michael Pyrcz, The University of Texas at Austin
Open Source Spatial Data Analytics in Python with GeostatsPy
Introduction to GeostatsPy
Lecture outline . . .
• Workflow with GeostatsPy
Michael Pyrcz, The University of Texas at Austin
Variogram
Modeling
Workflow
with
GeostatsPy
Let’s walkthrough a more
thorough variogram-based
spatial analysis workflow:
• calculate experimental
variograms
• formulate variogram models
Python Jupyter variogram calculation
(GeostatsPy_variogram_modleing.ipynb).
Michael Pyrcz, The University of Texas at Austin
Open Source Spatial Data Analytics in Python with GeostatsPy
Introduction to GeostatsPy
Lecture outline . . .
• Variogram Modeling
• Interactive Demo with GeostatsPy
• Workflow with GeostatsPy
Michael Pyrcz, The University of Texas at Austin
