Chapter Three

3. Geostatistical Analysis
 Chapter Three
3. Geostatistical Analysis

3.1. Principles of Geostatistical Analysis

3.2. Exploratory Spatial Data Analysis

3.3. Methods of Spatial Interpolation

3.4. Modelling Semivariogram

3.5. Creating Surface Using Geostatistical Methods

Chapter Objectives
• After completing this chapter, you will be able to:

>>> Understand the main principles of Geostatistical analysis.

>>> Discuss exploratory spatial data analysis

>>> know the different methods of spatial interpolation and semivariogram

modelling.

>>>Apply the different methods of creating surface using Geostatistical

techniques.
3. Geostatistical Analysis
Introduction
What is Geostatistical analysis? And/or why Geostatistics?

 It is a branch of statistics that deals with spatially correlated data based on

the theory of the regionalized variables (Hosseini et.al, 2018).

 >>> focusing on spatiotemporal datasets to model, predict and simulate

multivariable geospatial information.

 It aims to estimate the statistical parameters that determine the spatial

and/or temporal distribution and dependence of the relevant variables.

Cont…
• Geostatistical analysis are based on statistical models that include
autocorrelation (statistical relationships among the measured points).

• It has the capability of surfaces prediction/ continuous creation ( eg.

interpolation) and testing the accuracy of predictions.

Why geostatistical analysis?

• To assess patterns, trends, and relationships
• Better understanding of geographic phenomena
• Make decision with high level of confidence
• Summarization of the distribution data (spatial) in a simple and understandable
ways
• Predication of uncertainties
Cont…

• In general, it can be used;

 to describe and model spatial patterns (variography),

 to predict values at unmeasured locations (kriging), and

 to assess the uncertainty associated with a predicted value at the

unmeasured locations (kriging).

3.1. Principles of Geostatistical Analysis
• It relies on both statistical and mathematical methods, which can be used to create
surfaces and assess the uncertainty of the predictions.

The main assumption and principles of Geostatistical analysis are:

1) The analysis is based on both mathematical functions and statistical

models/theories (spatial autocorrelation >>> based on spatial variability analysis.
2) The sample point data has normal distribution
3) Randomization of measured sample points - at least some of the spatial variation
observed in natural phenomena can be modeled by random processes.
4) Prediction/creation of surfaces based on random points and testing the accuracy of
predictions
5) Assessing uncertainty associated with a predicted value at unmeasured locations.
6) Estimation of predication error
The geostatistical workflow
 Cont…

• There are three main components of Geostatistical Analyst:

I. A set of exploratory spatial data analysis (ESDA)

II. The Geostatistical Wizard

III. The Geostatistical Analyst toolbox, which houses geoprocessing tools

specifically designed to extend the capabilities of the Geostatistical
Wizard and allow further analysis of the surfaces it generates
3.2. Exploratory Spatial Data Analysis (ESDA)

• ESDA >>>> used to assess the statistical properties of the data.

• It helps to create a variety of output map types

For example, prediction, error of prediction, probability, and quantile) using
many variants of kriging and cokriging algorithms (ordinary, simple, universal,
indicator, probability, disjunctive, and empirical Bayesian).

• ESDA tools allows to examine the data in more quantitative ways than
mapping it and
• >>> helps for deeper understanding of the phenomena so that it helps
to make more informed decisions.
Cont…
The main steps involved in ESDA / in creating geostatistical analysis are:

1) Examining the data (distribution, trends, directional components, outliers).

2) Calculating the empirical semivariogram or covariance values.

3) Fitting a model to the empirical values.

4) Generating the matrices of kriging equations.

5) Solving them to obtain a predicted value and the error (uncertainty)

associated with it for each location in the output surface >>> Make a
prediction >>>
 The ESDA tools are:
Cont… Histogram
Normal QQ Plot and General QQ Plot
Trend Analysis
Voronoi Map
Semivariogram/Covariance Cloud
Crosscovariance Cloud
Tools for exploring a single dataset
 Examples of univariate ESDA tools >>> analyzing one dataset at a time
 Using Histogram, Normal QQPlot, Trend Analysis, Voronoi map and semivariogram or covariance
Using Histogram
Using Normal QQPlot
Using Voronoi map
Tools for exploring relationships between datasets and,
 Examples of bivariate ESDA tools >>> to examine relationships between two
datasets.
Cont…

The most common tasks to explore the data are:

1. Examine the distribution of your data
2. Look for global and local outliers
3. Look for global trends
4. Examine local variation
5. Examine spatial autocorrelation
 1. Examine the distribution of your data

 Most of the interpolation methods do not require the data to be normally distributed,
although in this case the prediction map may not be optimal.

 However, certain kriging methods require the data to be approximately normally

distributed (close to a bell-shaped curve).

Purpose
• To maximize insights in to the data
• To detect outliers
• To explore the distribution of the data

Techniques
• Data visualization
• Data plotting ( histogram, QQ plot)
 Cont…
• The Histogram and Normal QQ plot tools are designed to explore the distribution of the
data.
• >>> they also include different data transformations (Box-Cox and logarithmic) that help to
assess the effects on the data.

Using ‘log’

Using ‘Box-Cox’
 2. Outliers
• Explicitly identify both global and local outliers.
A global outlier is a measured sample point that has a very high or a very
low value relative to all the values in a dataset.

A local outlier is a measured sample point that has a value within the
normal range for the entire dataset, but if you look at the surrounding
points, it is unusually high or low.

Examples of local outlier

 Cont…
Identifying outliers through Semivariogram/Covariance cloud

• If there are global outlier (unusually high value in

your dataset), all pairings of points with that outlier
will have high values in the Semivariogram cloud, no
matter what the distance is.

• This can be seen in the semivariogram cloud.

Semivariogram
Using histogram in a single dataset
Using Normal QQPlot in a single dataset
 3. Trend analysis
• The trend analysis tool provides a three-dimensional perspective of the
data.

4. Local variation >>> the tool allows to examine local variation and distribution of
the data
 5. Spatial autocorrelation
 Examine spatial autocorrelation using (eg. Semivariogram modelling)
 Spatial autocorrelation >>> based on first law of Geography
 Spatial autocorrelation >>>> extending the concept of correlation by incorporating space/spatial
concepts.
3.3. Methods of Spatial Interpolation
• The Geostatistical Wizard provides access to a number of interpolation techniques, which
are divided into two main types: deterministic and stochastic/geostatistical.

1) Deterministic techniques >>> such as IDW, spline

• It use mathematical functions for interpolation
• >>> not based on statistical theory.
• >>> are not based on a random spatial process model, and there is no explicit measurement
or modeling of spatial autocorrelation in the data.
• >>> not able to estimate predication error

2) Stochastic such as kriging

• >>> based on both mathematical functions and statistical models/theories (spatial
autocorrelation).
• >>> assuming that the data has normal distribution
• >>>> Randomization
Mathematically, IDW can be described as:

Where;
 Z(x) is the estimated value at a predicted point,
 Zi is the observed value at point I,
 Wi is the weight value assigned at point i,
 di is the distance between point i and the
• Wu et.al. (2019) predicted point,
 k is the power variable.
 The power variable decides how surrounding
points affect the estimated value.
 A lower power results in higher influence from
distant points.
 Mathematically, kriging can be described as:

• Wu et.al. (2019)

• All the parameters have the same meanings as those in Equations.

• However, unlike the IDW method, the weighting coefficient (Wi) is assigned by not
only considering the distance between unsampled and sampled points, but also the
spatial correlation between these points.
Geostatistical Wizard
• The Geostatistical Wizard is a dynamic set of pages that is designed to guide you
through the process of constructing and evaluating the performance of an
interpolation model.

• It offers several types of kriging, which are suitable for

different types of data and have different underlying
assumptions, like;

 Cokriging >>> for spatial auto- and cross-correlation

 Areal interpolation >>> takes into account polygonal
input of various shapes.
 Empirical Bayesian >>> accounts for uncertainty in
semivariogram estimation by simulating many
semivariograms from the input data.
A simple workflow in the Geostatistical Wizard
 Cont…
• The Geostatistical Wizard is a dynamic set of pages that is designed to
guide you through the process of constructing and evaluating the
performance of an interpolation model.

o Cokriging >>> for spatial auto- and cross-correlation

o Areal interpolation >>> takes into account polygonal input of various shapes.
o Empirical Bayesian >>> accounts for uncertainty in semivariogram estimation by simulating many
semivariograms from the input data.
Geostatistical Analyst toolbox

• The Geostatistical Analyst toolbox includes tools for:

analyzing data,
producing a variety of output surfaces,
examining and transforming geostatistical layers to other formats,
 performing geostatistical simulation and sensitivity analysis, and
aiding in designing sampling networks.
Cont…
The Geostatistical Analyst tools have been grouped into five toolsets:
1) Interpolation - contains geoprocessing tools that perform interpolation
(as does the Geostatistical Wizard) that can be used as stand-alone
tools or in ModelBuilder and Python.

2) Sampling Network Design - has tools that aid in designing or modifying

an existing sampling design/monitoring network.

3) Simulation - extends kriging by performing geostatistical simulation and

permits extraction of the simulated results for points or polygonal
areas.
Cont…
4) Utilities - general use tools to:
 extract subsets of a dataset,
perform cross-validation to assess model performance,
examine sensitivity to variation in semivariogram parameters,
and visually represent the neighborhoods used by the
interpolation tools

5) Working with Geostatistical Layers – this tools helps to:

generate predictions for point locations,
export geostatistical layers to raster and vector formats,
retrieve and set interpolation model parameters (in an XML
parameter file), and
generate new geostatistical layers (based on an XML parameter
file and datasets)
Geostatistical Analyst toolbox
 3.4. Modelling Semivariogram
• >>>To analysis surface properties
• Semivariogram >>>> depicting the spatial autocorrelation of the measured sample points.
• The Semivariogram/Covariance Cloud tool can be used to investigate spatial autocorrelation in your
dataset.
• Spatial autocorrelation >>> based on first law of Geography
• Spatial autocorrelation >>>> Extending the concept of correlation by incorporating space.
Modelling Semivariogram(autocorrelation)
Exploring spatial structure through the Semivariogram/ Covariance Cloud tool

• The Semivariogram/Covariance Cloud tool can be used to investigate

autocorrelation in the dataset.

Autocorrelation
Looking for directional influences with the Semivariogram/ Covariance Cloud tool

• When you looking at the semivariogram surface, there might be directional

differences in the semivariogram values.
• >>> i.e using ‘show search direction”
• The locations linked together have very similar values because the
semivariogram values are relatively low.

Directional variation
 Cross validation techniques

Cross validation using statistics >>>

Error = ( Predicted value – true value)

1. Root-mean-square (RMS) >>> root of average squared errors

2. Root-mean-square standardized (RMSS) >>> standardized RMSE

3. Mean error >>> average of the errors

4. Mean standardized error >>> standardized mean errors

Cross validation …
3.5. Creating Surface Using Geostatistical Methods

• >>>> “Practical Session”

• Use all the above discussed Geostatistical methods and tools to

create surface >>>>

Also, please read more on:

1. Bivariate Arithmetic Operations
2. Multivariate Arithmetic Operations