## Univariate analysis

Univariate analysis is a fundamental statistical technique focused on

examining a single variable within a dataset to describe and
summarize it. The primary goal of univariate analysis is to understand
the statistical properties of individual variables, which is crucial
before conducting any spatial or multivariate analysis. In the context
of geostatistics, for example, understanding variables like mineral
concentration, porosity, or depth is essential. The simplicity and
foundational role of univariate analysis make it an important tool
across various fields such as finance, engineering, and environmental
science.
Types of Univariate Analysis
Univariate analysis can be categorized into two major types:
1. Descriptive Analysis: This involves summarizing datasets
using numerical measures. Key metrics include:
Central Tendency:
• Mean: The average of the data points.
• Median: The middle value when the data is arranged in
ascending or descending order.
• Mode: The most frequently occurring value in the dataset.
Dispersion:
• Describes the spread of values around the central tendency.
• Range: Difference between the maximum and minimum
values.
• Variance: The average of the squared differences from the
mean.
• Standard Deviation (SD): The square root of the variance,
showing how spread out the data is around the mean
2. Graphical Analysis: Visual representations such as
histograms, boxplots, and probability plots provide a more
intuitive understanding of the distribution, patterns, and
outliers within a dataset.
o Histograms and Density Plots: These are used to visualize
the frequency and probability density of variables, helping
to assess distributions and identify potential outliers.
o Boxplots: It summarizes data by showing the middle 50%
(the box), the median (a line inside the box), and any
outliers (dots outside the whiskers). It's a great way to
quickly compare data spreads and spot outliers.

o Probability Plots: These help in determining whether the

data follows a particular distribution, such as normal or
log-normal.
Question 1:
Given the following data set:
12, 14, 15, 14, 17, 18, 16, 14, 15, 19
Perform a univariate analysis and calculate the following:
Mean, Variance, Standard Deviation, Median, Mode, Range,
Skewness

Metric Description Calculation/Value

Mean (μ) Average of all data (12 + 14 + 15 + 14 + 17 +

points 18 + 16 + 14 + 15 + 19) /
10 = 15.4
Variance (σ²) Measure of data Variance = 4.24
dispersion from the
mean
Standard Square root of the √(4.24) = 2.06
Deviation (σ) variance

Median Middle value when Median = 15

data is sorted

Mode Most frequently Mode = 14

occurring value

Range Difference between 19 - 12 = 7

max and min values
Question 2:
A factory produces steel rods, and the lengths (in cm) of 8
randomly selected rods are:
100, 102, 99, 101, 100, 98, 103, 101
Perform a univariate analysis to calculate the following:
• Mean deviation
• Coefficient of variation

Metric Description Calculation/Value

Mean (μ) Average of all rod (100 + 102 + 99 + 101 +

lengths 100 + 98 + 103 + 101) / 8
= 100.5 cm

Mean Deviation Average of absolute (0.5 + 1.5 + 1.5 + 0.5 +

deviations from the 0.5 + 2.5 + 2.5 + 0.5) / 8
mean = 1.25 cm

Coefficient of Standard deviation (σ / μ) * 100 = 1.72%

Variation as a percentage of
the mean
Advantages of Univariate Probability Theory
• Simplicity:
o Univariate analysis is simple to apply because it focuses on
a single variable, making it easier to interpret and model.
• Foundational for Complex Analysis:
o It serves as a building block for more complex multivariate
analyses. Understanding one variable well sets the stage for
understanding multiple variables.
• Wide Applications:
o Applicable across a wide range of fields like finance,
healthcare, and engineering, providing insight into
outcomes and decision-making.
• Predictive Power:
o Helps make accurate predictions based on historical data or
known probabilities, such as calculating expected outcomes
and managing risks.

Disadvantages of Univariate Probability Theory

• Limited to One Variable:
o It only considers one variable at a time, so it doesn’t
capture the interaction between multiple factors. For
complex systems, multivariate analysis is needed.
• Oversimplification:
o Real-world problems often involve several interacting
variables, and focusing on one variable may lead to
oversimplified conclusions.
 • Assumptions of Independence:
o Many univariate methods assume that observations or
events are independent, which may not always hold
true in practical scenarios.
• Not Suitable for Complex Problems:
o Complex real-world scenarios often require
multivariate approaches to account for
correlations and dependencies between
variables.

## BIVARIATE ANALYSIS TECHNIQUES

3.1 Covariance

Covariance measures how two variables move together. The formula

for covariance between two random variables

3.2 Correlation Coefficient

The Pearson correlation coefficient is a normalized measure of the

linear relationship between two variables. It is calculated as:
 3.3 Scatter Plots

Scatter plots are a graphical representation of the relationship

between two variables. The plot allows one to visualize any patterns,
trends, or correlations between variables in a dataset.
 NUMERICAL PROBLEMS

Problem 1: Mining Ventilation

Scenario: A mining engineer is analyzing the relationship between

ventilation rate and air quality in an underground mine. The ventilation
rate is measured in cubic meters per second, and air quality is
determined by the percentage of harmful gases present.

Data:
• Ventilation Rate (m³/s): 5, 10, 15, 20, 25
• Air Quality (% harmful gases): 12, 9, 7, 5, 3

Objective: Calculate the Pearson correlation coefficient to determine if

there is a significant relationship between ventilation rate and air
quality.

Solution:
Compute the means of the ventilation rate and air quality data.
Use the Pearson correlation formula to find the correlation coefficient.
Interpretation: A negative correlation would imply that increasing
ventilation improves air quality (reduces harmful gases).

Problem 2: Grade Variability

Scenario: In a specific area of a mine, distance from the center of an

ore body and the grade variability are measured as follows:

Grade Variability
Distance (meters)
(%)
10 4.0

20 3.6

30 3.0
 40 2.9

50 2.5

Compute the correlation.

Solution

## Multivariate Analysis
3.3 Techniques of Multivariate Analysis

Multivariate analysis includes several techniques, each designed for specific

types of data and research goals. Some of the key techniques include:
1. Multiple Regression: Examines the relationship between a single
dependent variable and multiple independent variables.
2. Discriminant Analysis: Used to classify data into different groups by
identifying which variables best differentiate the groups.
3. MANOVA (Multivariate Analysis of Variance): Evaluates whether there
are differences between groups on multiple dependent variables.
4. Factor Analysis: Identifies underlying relationships between variables
by reducing the dimensionality of the data.
5. Cluster Analysis: Groups data points based on similarities without
predefined categories.
 1 Numerical Problem 1: Multiple
Regression

Problem:
We want to predict sales (Y) based on two independent variables:
advertising spending (X₁) and price discount (X₂). The relationship is
modeled as follows:

Solution:
Using the method of least squares, the multiple regression equation can
be derived. After computing the coefficients using statistical software, the
estimated regression equation might look something like this:

Interpretation:

For every 1 unit increase in advertising spending, sales increase by 0.2

units.
For every 1 unit increase in price discount, sales increase by 1.5 units.

2 Numerical Problem Covariance Matrix

 Problem Statement

Consider two random variables, height (X₁) and weight (X₂), of students in a
university. The goal is to compute the covariance matrix based on the
following data:
• Heights (X₁): [170, 160, 180, 175, 165]
• Weights (X₂): [65, 55, 75, 70, 60]

Solution

## Central Limit Theorem

STATEMENT
Central limit theorem states that whenever a random sample of size n is taken
from any distribution with mean μ and variance σ², then the sample mean 𝐗 # will
be approximately a normal distribution with 𝛍𝐗" = 𝛍 and standard deviation 𝛔𝐗" =
𝛔
. The larger the value of the sample size, the better the approximation of the
√ 𝐧
normal.
From the theorem we can conclude that,
# will, as n increases, approach a normal distribution.
1) The distribution of sample 𝐗
2) The mean of the sample means is population mean i.e., 𝛍𝐗" = 𝛍 .
 '
3) The standard deviation of all sample means is σ&" = .
√(
• If the sample is normal, then the sampling distribution of means (XO ) will also
be normal, no matter what the sample size.
• When the sample population is approximately symmetric, the distribution
becomes approximately normal for relatively small values of n.
• When the sample population is skewed, the sample size must be at least 30
before the sampling distribution of XO becomes approximately normal.
Graphical examples:
• For normal population distribution

• For exponential population distribution

ASSUMPTIONS
• The sample should be drawn randomly following the condition of
randomisation.
• The samples drawn should be independent of each other. They should not
influence the other samples.
• When the sampling is done without replacement, the sample size shouldn’t
exceed 10% of the total population.
• The sample size should be sufficiently large. Generally, n >= 30 is considered
sufficient, especially for skewed distributions.

CONDITIONS
The central limit theorem states that the sampling distribution of the mean will
always follow a normal distribution under the following conditions:
• The sample size is sufficiently large. This condition is usually met if the
sample size is n ≥ 30.
• The samples are independent and identically distributed (i.i.d.) random
variables. This condition is usually met if the sampling is random.
• The population’s distribution has finite variance. Most distributions have
finite variance.

FORMULA FOR STANDARD NORMAL VARIABLE

Population must be normally distributed in order to get the value of Z.

NUMERICALS
1) The average weight of a water bottle is 30 kg, with a standard deviation of 1.5 kg.
If a sample of 45 water bottles is selected at random from a consignment and their
weights are measured, find the probability that the mean weight of the sample is
less than 28 kg.
Solution:
Population mean, μ = 30 kg
Population standard deviation: σ = 1.5 Kg
Sample size: n = 45 (which is greater than 30)
# &.(
The sample standard deviation, σ!
" = = = 6.7082
√% √)(
Z-score for the raw score of x = 28 kg is
*+,- (12,34 )
Z= # = = -0.2981
.++++/ 7.8421
√%
Using the z-score table or normal CDF function on a statistical calculator,
P(z < -0.2981) = 0.3828
Thus, the probability that the mean weight of the sample is less than 28 kg is
38.28%.

2) The salaries at a very large corporation have a mean of $62,000 and a standard
deviation of $32,000. If 100 employees are randomly selected, what is the
probability their average salary exceeds $66,000?
Solution:
Population mean, μ = $62,000
Population standard deviation: σ = $32,000
Sample size: n = 100
# 31444
The sample standard deviation, σ!
" = = = 3200
√% √&44
Z-score for the raw score of x = $66,000 is
*+,- (66,000,62,000)
Z= # = = 1.25
.++++/ 3144
√%
Using the z-score table or normal CDF function on a statistical calculator,
P(z > 1.25) = 0.106
Thus, the probability that the average salary exceeds $66,000 is 10.6%.

## Spatial Continuity Analysis

Numerical Problem 1: Estimating Spatial Continuity in

 Soil Contamination

Problem Statement
A region affected by industrial pollution has soil samples collected from five locations. The concentration of
a contaminant (in mg/kg) is recorded as follows:

• Location A (0, 0): 30 mg/kg • Location B (0, 5): 35 mg/kg

• Location C (5, 0): 28 mg/kg

• Location D (5, 5): 33 mg/kg

• Location E (10, 0): 25 mg/kg

Calculate the semivariance for a lag distance of 5 units and fit a simple spherical model to the
Semivariogram.

Solution
1. Calculate Semivariance:

• Calculate pairs with a lag distance of h = 5:

• Between C (5, 0) and D (5, 5):

• Average semivariance for h = 5:

## Spatial Continuity Analysis: Experimental
Variogram

EXAMPLE - 1 SPATIAL VARIABILITY IN SOIL MOISTURE

EXAMPLE 2: SPATIAL VARIABILITY IN TEMPERATURE

 ## Spatial Analysis: Variogram versus
Univariate Statistics
3. Variogram Models
Several theoretical models are commonly used to fit empirical variograms:
• Spherical Model: The semi-variance increases rapidly and then flattens out after reaching the
range.

• Exponential Model: The semi-variance increases gradually but never fully levels off.

• Gaussian Model: The semi-variance increases smoothly, particularly over short distances.

4. Comparison with Univariate Statistics

Univariate statistics provide an overall summary of the dataset without considering spatial relationships,
while variograms are designed to capture and model the spatial dependence between data points.

Example: Mineral Concentration in a Mining Area

Consider a mining company analysing copper concentrations across multiple sampling points in a mine. The
goal is to understand how mineral concentration changes over space to guide efficient extraction.

Univariate Statistics: The average copper concentration across the sampling points is found to be 2.5%,
with a standard deviation of 0.4%. While this gives a sense of the overall distribution, it does not reveal any
spatial patterns, such as whether copper concentrations are higher in certain areas of the mine.

Variogram: By plotting the variogram, the mining company notices that the semi-variance increases
rapidly at short distances (up to 50 meters), indicating strong spatial dependence. Beyond 150 meters, the
semi-variance levels off, suggesting that copper concentrations are no longer spatially correlated beyond
this distance (range = 150 meters).
In this case, the variogram provides valuable insight into the spatial distribution of copper, allowing the
company to identify zones with higher concentration and optimize their extraction strategy.

5. Example Problem

Problem Statement: Copper Concentration in a Mining Region

You are tasked with studying copper concentration in a mining region. Samples are collected at various
distances, and the copper content (measured as a percentage by weight) is recorded. The table below
shows the distances between sampling locations and the corresponding semi-variance values:

You need to fit a spherical variogram model to the data and calculate the nugget, range, and sill.
Step-by-Step Solution:

1. Plot the Empirical Variogram: The semi-variance increases rapidly up to 6 km and then flattens out,
suggesting a spherical model is appropriate.

2. Choose the Spherical Model: Since the semi-variance increases sharply and then stabilizes, the
spherical model is a good fit.

3. Fit the Nugget (C0): The semi-variance at short distances (0.5 km) is 0.03, indicating small-scale
variability or measurement noise. Set the nugget at C0=0.03.

4. Fit the Range (a): The semi-variance flattens out around 6 km, so the range is a=6 km. Beyond this
distance, there is no spatial correlation between data points.

5. Fit the Sill (C): The semi-variance stabilizes at 0.52, so set the sill at C=0.52C, representing the total
variance in the data.

6. Final Fitted Model:

 ## SPACIAL CONTINUITY ANALYSIS
EXPLORING ANISOTROPY

Numericals Problems:

Numerical Example 1: Isotropic Variogram

Given a dataset from a mineral field with ore grades measured at several
locations, let's compute the isotropic variogram.
Data:
• Ore grades at distances (in meters):

Z(0) = 1.5
Z(10) = 2
Z(20) = 2.5
Z(30) = 3
Z(40) = 2.8

Variogram Calculation:
gamma(10) = 1/2 * [(2 - 1.5) ^ 2 + (2.5 - 2) ^ 2 + (3 - 2.5) ^ 2 + (2.8 - 3) ^ 2]
gamma(10) = 1/2 * [0.25 + 0.25 + 0.25 + 0.04]
gamma(10)= 0.395

Numerical Example 2: Anisotropic Variogram

Suppose we have variograms computed along two directions (N-S and E-
W) from a dataset of soil strength values in a region.
Data for N-S direction: Z(0) = 20 Z(10) = 18 Z(20) = 16 Z(30) = 14
Z(40) = 12 Data for E-W direction: Z(0) = 20 Z(10) = 19.5 Z(20) = 19
Z(30) = 18.5 Z(40) = 18
Variogram Calculation:
For the N-S direction:
YNS (10) = [(18-20)2 + (16-18)2 + (14-16)2 + (12-14)²] = 4
↓pmergies↑models
For the E-W direction:
YEW(10)= =[(19.5-20)² + (19.0-1 19.5)2+(18.5-19.0)² + (18.0-18.5)²] = 0.125
The results show greater variability in the N-S direction compared
to the E-W direction, indicating Anisotropy
 ## Geostatistical Estimation: Random
Function Models
3. Case Study: Estimating Ore Grades in a Mineral Deposit
3.1. Background
A mining company seeks to estimate the distribution of ore grades across a newly
discovered deposit. They have taken several core samples at various locations and
wish to predict the ore grade at unsampled locations using geostatistics.
3.2. Data Collection
Core samples are taken at regular intervals across the deposit. These samples
provide data on the grade (metal content) of the ore at specific locations.
3.3. Applying the Random Function Model
The random function model assumes that the ore grade is a spatially continuous
random variable that varies across the deposit. The variability is captured using the
variogram, which describes how the grade correlation between locations decreases
as the distance between them increases.
3.4. Kriging
Kriging, a widely used geostatistical method, is employed to estimate the ore grade
at unmeasured locations. Based on the variogram, kriging provides not only the best
linear unbiased estimate of the grade but also a measure of uncertainty associated
with the estimate.
1. Constructing the Variogram: A variogram is constructed by calculating the
squared differences of measured values at different locations, plotted against
the distance between the points.
4. Mathematical Example 1
4.1. Problem Setup

Consider a scenario where we have measurements of a spatial variable Z(x) (e.g., soil
contamination levels) at three locations:

• Z(x1)=10
• Z(x2)=15
• Z(x3)=12

We want to estimate the value at an unsampled location Z(x4), which is located

between x1 and x2.

4.2. Step-by-Step Calculation Using Ordinary Kriging

Step 1: Compute the Variogram

Let’s assume we’ve already computed the variogram and found that the following
relationships hold for distances between points:

Step 2: Form the Kriging System

Step 4: Estimate
The predicted value is 13.2, with an associated kriging variance that quantifies the
uncertainty of the estimate.

1. Mathematical Example 2
Step 1: Collect Data
Assume we have a dataset of sample points with their corresponding values
(e.g., pollutant concentrations, mineral grades, etc.):

Step 2: Variogram Calculation

The next step is to calculate the variogram, which describes the spatial
continuity of the data. The empirical variogram γ(h) can be calculated
using:

Where:
• N(h) is the number of pairs of points separated by a distance h
• Z(xi) and Z(xj) are the values at locations xi and xj.
For simplicity, let’s assume we calculate the variogram for h=1(the distance
between points).

Calculate the differences for each pair of points separated by 1 unit:

• Pair (1, 2) and (2, 3): |10 - 15|^2 = 25

• Pair (2, 3) and (3, 1): |15 - 20|^2 = 25

 • Pair (3, 1) and (4, 4): |20 - 25|^2 = 25

1. For h=1, there are 3 pairs, so:

γ(1)=1/2⋅3(25+25+25)=75/6=12.5
Step 3: Kriging Estimation
To estimate the value at a new location (2,2) we can use the kriging formula:

For simplicity, assume we have determined the weights λ1=0.3, λ2=0.5and λ3=0.2

Using the known values:

Z(2,2)=(0.3×10)+(0.5×15)+(0.2×20)

Calculating this:

Z(2,2)=3+7.5+4=14.5

## Ordinary Kriging

Application of the Ordinary Kriging

Ordinary kriging (OK) is a spatial estimation method that can be
used in a variety of applications, including:
• Geochemical data interpolation: OK can be used to
interpolate geochemical data in different types of deposits,
such as porphyry ores.
• Data with a trend: OK can be used for data that seems to
have a trend.
 • Data that is more variable on the left and becomes
smoother on the right: OK can be used for data that is more
variable on the left and becomes smoother on the right.
OK is a spatial estimation method that minimizes error
variance, also known as the kriging variance. It is based on
the configuration of the data and the variogram. OK is one
of the most commonly used kriging techniques.

## Co-Kriging

1 Variogram and Cross-Variogram

A variogram is a function that defines the spatial dependence of a
variable. It expresses the expected squared difference between values
of the variable as a function of the distance between locations. In
cokriging, two types of variograms are used:
• Variogram γ(h): Describes how the similarity between two values of the same variable
changes with distance h. For the primary variable Z1, this is written as:
1
γ11 (h) = E (Z1 (x) − Z1 (x + h))2
2
Where h is the distance between locations x and x + h.
 • Cross-Variogram γ12(h): Describes the correlation between the primary variable Z1
and the secondary variable Z2. It defines how the similarity between these two variables
changes with distance:
1
γ12(h) = E [(Z1(x) − Z1(x + h))(Z2(x) − Z2(x + h))]
2

The variograms and cross-variograms are essential for calculating

the weights used in the cokriging estimation process. These weights
are optimized to minimize the estimation variance, thus providing
the most accurate predictions for the unsampled locations.

2 Cokriging Equations
The cokriging estimation at an unsampled location x0 is a weighted
linear combination of both the primary and secondary variables
from sampled locations:
N M
Σ Σ
Z1∗ (x0 ) = λi Z1 (xi ) + µj Z2 (xj )
i=1 j=1

Where:
• Z1∗ (x0 ) is the estimated value of the primary variable at x0 .
• λi are the cokriging weights for the primary variable.

• µj are the cokriging weights for the secondary variable.

• Z1(xi) and Z2(xj) are the observed values of the primary and secondary variables at
sampled locations.

The weights λi and µj are determined by solving a system of

equations based on the variogram and cross-variogram models.
These weights ensure that the estimation variance is minimized,
which leads to more accurate predictions.
The cokriging system is set up in a matrix form, where the variogram
and cross-variogram values between the known points and the
unsampled location x0 are used to solve for the weights. Once the
weights are calculated, they are applied to the known values of the
variables to estimate the value of the primary variable at x0.
 3 Numerical Example
We will now present a simple numerical example to illustrate how
cokriging is applied for spatial estimation. Suppose we want to
estimate the temperature Z1 at an unsampled loca- tion x0 using
both temperature data Z1 (primary variable) and elevation data Z2
(secondary variable) from two sampled locations x1 and x2.

5.1 Problem Setup

Data:

Location Z1 (Temperature) Z2 (Elevation)

x1 10°C 100 m
x2 12°C 90 m

Variogram and Cross-Variogram Models:

• Variogram for Z1 (Temperature): γ11(h) = 0.5h

• Variogram for Z2 (Elevation): γ22(h) = 0.4h

• Cross-variogram between Z1 and Z2: γ12(h) = 0.3h

Distances:

• d(x0, x1) = 1 km

• d(x0, x2) = 2 km

• d(x1, x2) = 1 km

5.2 Compute Variogram and Cross-

Variogram Values
Using the variogram models, we calculate the necessary variograms
and cross-variograms:
Variogram for Z1 (Temperature):

γ11(x0, x1) = 0.5 × 1 = 0.5, γ11(x0, x2) = 0.5 × 2 = 1.0, γ11(x1, x2) = 0.5 × 1 = 0.5
Variogram for Z2 (Elevation):

γ22(x0, x1) = 0.4 × 1 = 0.4, γ22(x0, x2) = 0.4 × 2 = 0.8, γ22(x1, x2) = 0.4 × 1 = 0.4
Cross-variogram between Z1 and Z2:

γ12(x0, x1) = 0.3 × 1 = 0.3, γ12(x0, x2) = 0.3 × 2 = 0.6, γ12(x1, x2) = 0.3 × 1 = 0.3

5.3 Set Up the Cokriging System

The cokriging system involves solving for the weights λ1, λ2, µ1, µ2.
The system of equations is based on the variogram and cross-
variogram relationships. The objective is to minimize the variance of
the prediction by solving the system for these weights.

5.4 Prediction of Z1(x0)

Now, using the calculated variograms and cross-variograms, we
estimate the temperature
Z1(x0) at location x0 using the cokriging equation:
Z1∗ (x0 ) = λ1 Z1 (x1 ) + λ2 Z1 (x2 ) + µ1 Z2 (x1 ) + µ2 Z2 (x2 )
Substitute the known values:
Z1∗ (x0 ) = 0.5 × 10 + 0.5 × 12 + 0.5 × 100 + 0.5 × 90

Z1∗ (x0 ) = 5 + 6 + 50 + 45 = 106

Thus, the estimated temperature at location x0
is 106◦C.

5.5 Interpretation and Discussion

From this example, we observe how cokriging combines both the
primary variable (tempera- ture) and the secondary variable
(elevation) to provide an estimate at an unsampled location. In this
case, the predicted temperature is 106◦C. This result is influenced by
the choice of weights, and in practical applications, these weights
would be computed more rigorously using the variogram and cross-
variogram models.
Key Points:
• Cross-Variogram’s Role: The cross-variogram defines the relationship between the
primary and secondary variables. A strong correlation between these variables, indi-
cated by a high cross-variogram, improves the accuracy of the cokriging estimation.

• Weight Calculation: The weights λ1, λ2, µ1, µ2 are optimized to minimize the estima-
tion error. The correct determination of these weights is critical for accurate cokriging
predictions.

• Applicability: Cokriging is useful in situations where secondary variables can be

easily measured and are strongly correlated with the primary variable. For example, in
environmental monitoring, elevation data can improve the estimation of temperature,
especially in areas where direct temperature measurements are sparse.

4 Applications of Cokriging
Cokriging is widely used in various fields where spatial data is
available, and multiple corre- lated variables can improve
predictions. Some common applications include:
• Mining and Resource Estimation: In mining, cokriging is used to estimate the
concentration of minerals in unsampled locations by considering both the primary
variable (e.g., ore grade) and secondary variables (e.g., geophysical data). This helps
in optimizing resource extraction and planning.
• Environmental Science: Cokriging is employed to predict pollutant levels in the en-
vironment. For example, air pollution levels (primary variable) can be estimated using
meteorological data like temperature, humidity, and wind speed (secondary variables),
which are easier to measure and highly correlated with pollution.

• Agriculture: It is used for predicting soil properties such as nutrient content (pri-
mary variable) by incorporating other measurable factors such as moisture content or
elevation (secondary variables). This helps farmers in better land management and
crop yield predictions.

• Meteorology: Cokriging is used to estimate temperatures or rainfall in unsampled

regions by leveraging topographical or satellite data as secondary variables.

• Hydrology: In water resource management, cokriging can be used to estimate water

table depth or contamination levels by incorporating secondary variables like elevation
or land use data.
5 Advantages of Cokriging
Cokriging provides several advantages over traditional kriging
techniques:
• Increased Accuracy: By using correlated secondary variables, cokriging improves
the accuracy of estimates at unsampled locations compared to simple kriging, which
only uses one variable.

• Utilizes Secondary Information: Secondary variables that are easier or cheaper to

measure can be used to inform the predictions of the primary variable, thus maximizing
the use of all available data.

• Reduces Uncertainty: The incorporation of secondary variables reduces estimation

variance, leading to more robust predictions and lower uncertainty, particularly in
regions with sparse data points.

• Flexibility: Cokriging can be adapted to a wide range of situations where multiple

variables are spatially correlated, making it a flexible tool for spatial interpolation.

6 Limitations of Cokriging
Despite its advantages, cokriging has certain limitations:
• Complexity: The cokriging system involves more complex mathematical models and
requires variogram and cross-variogram modeling for both the primary and secondary
variables, making it computationally more intensive and harder to implement than
simple kriging.
 • Requires Extensive Data: To build accurate cross-variograms, a significant
amount of data is required for both the primary and secondary variables. If
secondary data is scarce or weakly correlated with the primary variable, cokriging
may not offer signifi- cant advantages.

• Modeling Assumptions: Cokriging assumes linearity in the relationships

between the variables. If the relationship between the primary and secondary
variables is non- linear or if variogram models are poorly fitted, the results can
be inaccurate.

• Sensitivity to Correlation: The effectiveness of cokriging relies heavily on the

strength of the correlation between the primary and secondary variables. If the
corre-lation is weak, the benefits of using cokriging over kriging diminish.

• Time-Consuming: The variogram modeling, system setup, and solution process for
cokriging are more time-consuming, which can be a drawback for large datasets or
when quick estimations are needed.

## Indicator Kriging

2. Theoretical Framework
Indicator Variables:
In Indicator Kriging, the continuous data values Z(x) at location x are transformed into indicator
variables I(x), based on a cutoff or threshold value z_c:
I(x) = 1 if Z(x) >= z_c, 0 if Z(x) < z_c
The indicator function transforms the spatial data into binary form, facilitating the calculation of
conditional probabilities at unsampled locations.
Variogram in Indicator Kriging:
The experimental variogram γ(h) is computed for the indicator data. It quantifies the spatial
autocorrelation between two locations separated by distance h. This variogram is key to determining
the kriging weights.
Assumptions and Limitations:
1. The model assumes stationarity of the indicator variable.
2. Kriging does not provide absolute certainty; it provides a probabilistic estimate of exceeding a
threshold.

3. Example Problem 1: Estimation of Gold

Concentration
Problem:
A geologist is studying gold concentration in a region, where gold concentration varies between 0 and
15 grams per ton. The interest is to estimate the probability that the concentration at unsampled
locations exceeds a threshold of 5 grams per ton. The following data from 5 locations is available:
Location Gold Concentration (g/t)
A 6.0
B 4.5
C 7.2
D 2.8
E 10.5
 Solution:
Step 1: Define Indicator Variable
For each sample, the indicator variable is defined as follows, using the threshold z_c = 5.0 g/t:

I(x)=1 if Z(x)≥5g/t
I(x)=0 if (x)<5g/t
Location Gold Concentration (g/t) Indicator Value
A 6.0 1
B 4.5 0
C 7.2 1
D 2.8 0
E 10.5 1

Step 2: Calculate Experimental Variogram

1. Compute the experimental variogram based on the distances between each pair of sample
locations. This step involves calculating the squared differences between the indicator values
and averaging them over different distances.

Distances between locations (in km):

• A−B=2km
• A−C=3km
• A−D=4km
• A−E=5km
2. Compute squared differences:
Pair Indicator Difference I1−I2 (I1 − I2) 2
A-B 1− 0 = 1 1
A-C 1− 1 = 0 0
A-D 1− 0 = 1 1
A-E 1− 1 = 0 0

Step 3: Kriging Weights

Using the distances and variogram, set up a linear system to solve for the kriging
weights λ1,λ2,...,λn.

For a location X, the system of equations is:

λ1⋅γ(h1x)+λ2⋅γ(h2x)+...+λn⋅γ(hnx)=γ(hx)

Distance (km) Variogram γ(h)γ(h)

2 0.8
3 0.6
4 0.4
5 0.3
The kriging system would look like this (simplified):

λ1⋅0.8+λ2⋅0.6+λ3⋅0.4+λ4⋅0.3=1

Step 4: Final Estimation

Once we have the kriging weights, we can calculate the final estimated indicator value for an
unsampled location X using the formula:

I((X)=λ1⋅IA+λ2⋅IB+λ3⋅IC+λ4⋅ID+λ5⋅IE

Suppose the weights turn out to be:

λ1=0.3,λ2=0.2,λ3=0.25,λ4=0.15,λ5=0.1λ1=0.3,λ2=0.2,λ3=0.25,λ4=0.15,λ5=0.1

So, the probability that the gold concentration at location X exceeds 5 grams per ton is 65%.

4. Example Problem 2: Estimation of Soil

Contamination
Problem:
An environmental scientist is studying soil contamination due to heavy metals. The goal is to estimate
the probability that soil contamination exceeds 100 ppm at unsampled locations in a contaminated
field. Data from 6 locations is available:
Location Contamination (ppm)
F 120
G 80
H 150
I 90
J 60
K 130
Solution:
Step 1: Define Indicator Variable
The indicator variable is defined using the threshold z_c = 100 ppm:

I(x)=1 if Z(x)≥100ppm
I(x)=0 if Z(x)<100ppm

Location Contamination (ppm) Indicator Value

F 120 1
G 80 0
H 150 1
I 90 0
J 60 0
K 130 1
 Step 2: Construct Indicator Variogram
Similar to the previous example, compute the indicator variogram using the spatial separation
distances between the sample locations and the corresponding indicator values.
Compute squared differences:

Pair Indicator Difference I1−I2 (I1−I2)2

F-G 1−0= 1 1
F-H 1−1= 0 0
F-I 1−0= 1 1

Step 3: Solve the Kriging System

Determine the kriging weights by solving the kriging system using the variogram model.

We now solve the linear system of equations for the weights λ1,λ2,...,λ6.

Using a solver (e.g., matrix inversion or Gaussian elimination), we get the following
kriging weights:

λ1=0.4,λ2=0.2,λ3=0.3,λ4=0.05,λ5=0.05,λ6=0.1
Step 4: Final Estimation
The estimated indicator value at the unsampled location X is a weighted sum of the
indicator values at the known locations, using the kriging weights:

I(X)=λ1⋅IF+λ2⋅IG+λ3⋅IH+λ4⋅II+λ5⋅IJ+λ6⋅IK

Substituting the known values:

I(X)= (0.4×1)+(0.2×0)+(0.3×1)+(0.05×0)+(0.05×0)+(0.1×1)
Simplifying:
I(X)=0.4+0+0.3+0+0+0.1=0.8

Thus, the estimated probability that the contamination at

location X exceeds 100 ppm is 80%.

## Block Kriging
Key Differences Between Block Kriging and Point Kriging

• Point Kriging estimates the value of a variable at a single location.

• Block Kriging estimates the average value over a block (or region).
• Block kriging involves the integration of the point kriging process over an area, while point
kriging operates at specific, individual points.
• Block kriging generally results in smoother estimates because the block average inherently
reduces variability.
Applications of Block Kriging

Block kriging is used in various fields to estimate the distribution of resources or other variables over
larger areas:

• Mining: In mining operations, block kriging is often used to estimate the average grade of
minerals over mining blocks, helping in the efficient extraction of resources.
• Environmental Science: Block kriging is applied to estimate average pollutant concentrations
over a defined region, aiding in environmental impact assessments and monitoring.
• Agriculture: It is used to estimate soil properties or crop yields over agricultural fields,
facilitating better decision-making for land management and crop planning.

By applying block kriging, researchers and industry professionals gain valuable insights into the spatial
distribution of variables across large areas, enabling more effective decision-making and resource
management.
2. Mathematical Formulation of Block Kriging
Variogram and Covariance Function

The variogram and covariance function are fundamental tools in kriging, as they describe the spatial
correlation between data points. The variogram is a function that quantifies how the difference
between values at two locations increases as the distance between them increases. The variogram is
defined as:

where Z(x) and Z(x+h) are values at two locations separated by a distance h. The covariance function,
on the other hand, describes how similar values at two locations are, based on the distance between
them. The covariance function is related to the variogram as:

where C(0) is the variance of the variable.

In block kriging, these functions play a key role in capturing the spatial correlation between the data
points and the block. The block kriging model integrates the variogram over the block to account for
the spatial variability within it, allowing for a more accurate prediction of the block’s average value.
Ordinary Kriging Equations
The basic form of the kriging estimator is:

where:
• Z^(x0) is the estimated value at location x0,
• Z(xi)are the known values at surrounding data points,
• λi are the kriging weights determined from the spatial correlation between data points.
To find the weights λi we solve the ordinary kriging system:
where C(hij) represents the covariance between data points, and C(hi0) represents the covariance
between a data point and the location of interest.

Block Kriging Adaptation

In block kriging, instead of predicting a value at a point, we aim to estimate the average value over a
block. The kriging system is modified by integrating the covariance function over the block to account
for the spatial relationships within the block. The block kriging estimator becomes:

where Z^(B) is the estimated average value over the block, and the weights λi are determined by
solving a system that incorporates the covariances between the block and the data points

The block-to-point covariances account for the average spatial correlation between the block and the
surrounding data points.

3. Example Problem 1: Estimating the Average Concentration

of a Pollutant
Problem Description

Problem Description

Suppose we are monitoring the concentration of a pollutant in a river basin. The goal is to estimate the
average pollutant concentration in a specific 1 km² block. We have concentration measurements at
several nearby locations (given as coordinates in kilometers), and we want to apply block kriging to
estimate the average concentration over this block.

• Dataset: The concentration of the pollutant Z(x) is measured at the following locations (in
mg/L):
o Z(0,0)=10 mg/L
o Z(1,0)=12 mg/L
o Z(0,1)=8 mg/L
o Z(1,1)=11 mg/L
•
• Block: The block for which we want to estimate the average concentration is the square area
with vertices at (0.5,0.5) to (1.5,1.5)
 • Variogram Model: For simplicity, let’s use a spherical variogram model:

• where c0=0.5 (nugget effect), c1=1.5 (sill), and a=2 (range).

Solution

Step 1: Calculate Covariances Between Data Points

We start by calculating the covariances between the data points based on the variogram model.

1. Distance between (0,0)and (1,0):

Substituting h=1into the spherical variogram model:

1. Covariance:

C(1)=1.5+0.5−1.625=0.375

Repeat this process to calculate all pairwise covariances C(hij) between the known data points.

Step 2: Compute Block-to-Point Covariances

Now we compute the covariances between the block and each data point. Since the block is centered
at (1,1) we calculate the average covariance over the block for each point. For simplicity, we'll
approximate the covariance between the block and the point at (0,0)by taking the covariance at the
block’s center (1,1) and the point.

For example, the distance between the center of the block (1,1)and the point at (0,0) is:

Substitute h=sqrt{2}≈1.414 into the variogram to find the covariance C(B,x1).

Step 3: Set Up the Kriging System

Using the covariances calculated above, set up the kriging system:

Solve this system to obtain the kriging weights λ1,λ2,λ3

Step 4: Estimate the Average Block Value

Finally, use the kriging weights to estimate the average pollutant concentration in the block:

4. Example Problem 2: Estimating Mineral Content in a

Mining Block
Problem Description

In this example, we want to estimate the average gold concentration in a 100 m x 100 m mining block
using block kriging. Gold concentrations are measured at several drill holes located near the mining
block. The dataset, block size, and variogram model are as follows:

• Dataset (gold content in g/ton):

o Z(0,0)=2.5
o Z(50,0)=3.1
o Z(0,50)=2.8
o Z(50,50)=3.3
• Block: A 100 m×100 m block centered at (25,25).
• Variogram Model: Exponential variogram:

• where c0=0.2 (nugget), c1=1. 0 (sill), and a=150 meters (range).

Solution

Step 1: Calculate Covariances Between Data Points

Using the exponential variogram, calculate the covariances between the data points:

For example, the distance between (0,0)and (50,0) is:

h=50 m

Substitute h=50 into the exponential variogram:

Repeat this for all other pairs.

Step 2: Compute Block-to-Point Covariances

As in the first example, compute the covariances between the block and each point. For simplicity,
approximate the covariance using the distance between the block's center (25,25) and the data points.

For example, the distance between the center of the block and the point (0,0) is:

Substitute this distance into the variogram to find the covariance C(B,x1).

Step 3: Set Up the Kriging System

Using the calculated covariances, set up and solve the kriging system to find the weights λ1,λ2,λ3.

Step 4: Estimate the Average Block Value

Use the kriging weights to estimate the average gold content in the mining block:

1. ## Cholesky Decomposition

Introduction
Cholesky Decomposition is a mathematical technique used to
decompose a positive-definite matrix into a product of a lower triangular
matrix and its transpose. In geostatistics, this method is widely used
for solving large systems of equations, such as those encountered in
spatial interpolation methods like Kriging. The decomposition
simplifies matrix inversion and provides numerical stability, making it
an essential tool in spatial data modeling and prediction.

Relevance in Geostatistics
Geostatistics involves the study and modeling of spatially distributed
data, where covariance matrices play a crucial role in capturing the
relationships between data points across space. Cholesky
Decomposition is particularly valuable because these covariance
matrices are typically large and dense. This decomposition allows for
efficient computation, making it feasible to handle large datasets
common in geostatistical problems.
 2. Properties of Cholesky Decomposition
• Symmetry Preservation: Since the covariance matrix is symmetric, Cholesky
Decomposition preserves this structure by decomposing itinto L× L^T.
 • Positive-Definiteness: The covariance matrix is positive-definite, which
guarantees that the diagonal entries of L are positive.
• Efficiency: The decomposition is computationally efficient, with a
complexity of O(n^3), where n is the size of the matrix. This is faster than
other matrix decomposition methods like LU Decomposition.
• Numerical Stability: Cholesky Decomposition is numerically stable and
less prone to rounding errors, making it ideal for large geostatistical
models.

3. Problem 1 : Estimating Ore

Concentration in a Mine Problem
Statement:
A mining company is estimating the ore concentration at a new
location within a mine. They have collected data from three nearby
locations, where the ore concentrations and their spatial coordinates
are known. The covariance matrix of the observed data is given by:

The covariance between the new location and known locations is:

The ore concentrations at the observed locations are:

Estimate the ore concentration at the new location using Kriging.

Solution Using Cholesky Decomposition:
Step 1: Decompose the Covariance Matrix Perform Cholesky Decomposition on
the covariance matrix C:
 C=L× L^T
After decomposition, we get:

Step 2: Solve for the Kriging Weights www To estimate the ore concentration
at the new location, we solve for the Kriging weights by solvingthe system:

C × w= c^T
Using forward substitution and then back-substitution, we solve for
w1,w2,w3. This gives us the Kriging weights:

Step 3: Compute the Estimated Ore Concentration The estimated ore

concentration at the new location is given by:

Substituting the values of the weights and ore concentrations, we get:

The estimated ore concentration at the new location is 4.15 g/t.

Problem 2: Estimating Soil

Contamination Levels Problem
Statement:
An environmental scientist is tasked with estimating the level of
contamination in a new location based on soil samples collected from
three nearby locations. The observed contamination levels (in mg/kg)
and their
spatial covariance matrix are known. The covariance matrix of
theobservations is:

The covariance between the new location and the observed locations is:

The contamination levels at the observed locations are:

Estimate the contamination level at the new location using Kriging.

Solution Using Cholesky Decomposition:
Step 1: Decompose the Covariance Matrix Perform Cholesky
Decomposition of the covariance matrix C:

C=L× L^T
After performing the decomposition, we get:

Step 2: Solve for the Kriging Weights www Using the Cholesky factor L,solve the
system:

Using forward substitution and back-substitution, we find the

Krigingweights:
Step 3: Compute the Estimated Contamination Level The estimated
contamination level z^\hat{z}z^ at the new location is given by:

Substituting the values of the weights and contamination levels, we get:

Thus, the estimated contamination level at the new location is 37.1 mg/kg.

4. Advantages of Cholesky Decomposition

1. Efficient for Large Matrices: Cholesky Decomposition is computationally
efficient compared to other matrix decomposition methods (such as LU
decomposition) when dealing with large, sparse, and symmetric positive-
definite matrices, making it ideal for applications like geostatistics and
machine learning.
2. Numerically Stable: It is more numerically stable and avoids many issues
related to round-off errors that can arise with direct matrix
 inversion, providing accurate results, especially for ill-conditioned
matrices.
3. Memory Efficient: Since it only operates on half of the matrix (due to symmetry), it
requires less memory, which is crucial in solving high- dimensional problems efficiently.

5. Disadvantages of Cholesky Decomposition

1. Requires Symmetric Positive-Definite Matrices: Cholesky Decomposition can only
be applied to symmetric positive-definite matrices. If the matrix does not meet this
criterion, the decompositioncannot be performed.
2. Sensitive to Small Numerical Errors: Although numerically stable, Cholesky
Decomposition can still be sensitive to small numerical errors if the matrix is near
singular or not perfectly positive definite.
3. Not General-Purpose: Unlike other matrix decompositions (such as LU
decomposition), Cholesky Decomposition is limited to a specific class of matrices,
reducing its flexibility for use in general matrixproblems.

## Geostatistics Field Applications

Applications of geostatistics:
The development of geostatistics is maturing, and geostatistics has been applied in many
domains, including soil science, hydrology, geology, zoology, agriculture, ecology, forestry,
computer science, mechanical engineering, medicine, environmental engineering and
management, etc.

1) Mining Industry:
● Application:
Geostatistics is used for ore reserve estimation and mineral deposit evaluation. It
predicts the concentration of minerals at different points using sparse data from
exploratory drilling.
● Applied:
○ Techniques like kriging and variogram analysis help in creating spatial models
of ore bodies.
○ These models estimate mineral grades and determine the most promising
locations for drilling.
 ● Benefits:
○ Reduces the number of necessary drill holes, cutting exploration costs.
○ Provides more accurate estimates of mineral resources, improving financial
forecasting for mining operations.
● Used in
Geostatistics was used in India coal mines to predict ore distribution, reducing
unnecessary excavation

India targets to increase its coal production to 1,200 million metric tons (1,300 million
short tons) by 2023–24.

2) Environmental Science:
● Application:
Used for environmental monitoring, pollution assessment, and analyzing spatial
patterns of environmental variables such as air pollution, temperature, and soil
quality.
● Applied:
○ Geostatistical methods like interpolation create pollution maps that show
concentration levels of pollutants across different regions.
○ Models predict changes in air quality or temperature over time.
● Benefits:
○ Identifies pollution hotspots, aiding in policy and regulatory actions.
○ Helps in long-term environmental monitoring for climate change studies.
● Used in
Geostatistical models have been used to track air quality in large cities like Beijing, helping
in pollution control strategies.
 3) Hydrogeology:
● Application:
Geostatistics is used in Hydrogeology to model groundwater systems, assess aquifer
properties, and predict groundwater flow patterns. These methods are essential in
water resource management, pollution control, and environmental studies.
● Applied:
○ Models help track the spread of contaminants in groundwater, predicting
areas at risk for water pollution.
○ Geostatistics supports the prediction of groundwater recharge zones and
discharge areas, improving water resource management.
● Benefits:
○ Helps in management of water supplies by identifying areas where water
extraction can occur without further future depletion.
○ Ensures proper management of groundwater resources.
● Used in
In California’s Central Valley, geostatistics has been used to predict groundwater recharge
rates and manage water use during drought conditions, ensuring water availability for
agricultural and urban needs.

4) Hydrology:
 ● Application:
Used for groundwater modeling, flood prediction, and analyzing the spatial
distribution of hydrological variables.
● Applied:
○ Spatial models of rainfall, runoff, and groundwater levels help predict water
resource availability and flood risks.
○ Geostatistics assists in managing water resources, especially in arid regions.
● Benefits:
○ Helps in efficient water resource management, flood control, and drought
mitigation.
○ Supports sustainable urban planning by predicting water demands.
● Used in
Geostatistics was applied in the Nile Basin to model groundwater recharge and predict
flood risks.

5) Meteorology:
● Application:
Study of atmospheric phenomena, including weather forecasting and climate
analysis.
● Applied:
○ Meteorological models use satellite data and ground observations to predict
weather patterns.
 ○ Advanced forecasting techniques assess the impact of atmospheric conditions
on various sectors.
● Benefits:
○ Informs disaster response efforts and enhances public safety during severe
weather events.
○ Supports agricultural planning by providing forecasts for optimal planting and
harvesting times.

● Used in
The National Hurricane Center uses meteorological models to track and predict hurricanes,
providing crucial information for coastal communities.

6) Oceanography:
● Application:
In oceanography, geostatistical methods are applied to study ocean currents,
temperatures, and salinity levels. This helps in understanding marine ecosystems,
predicting climate impacts, and managing marine resources sustainably.
 ● Applied:
○ Oceanographic models analyze ocean currents, temperature, salinity, and
biological productivity.
○ Used in climate change studies, marine resource management, and pollution
tracking.
● Benefits:
○ Supports sustainable fishing practices and marine conservation efforts.
○ Enhances understanding of climate systems through ocean-atmosphere
interactions.
● Used in
Research on the India Stream's variability has helped understand its role in regulating
climate patterns across the Indian ocean.

7) Geochemistry
● Application:
Geostatistics is crucial in geochemical analysis to assess the distribution of chemical
elements and contaminants in soils and waters. This application aids in
environmental monitoring, resource exploration, and remediation strategies.
● Applied:
○ Geochemical analysis identifies contaminants in soils, water, and sediments.
○ Used in mineral exploration and environmental remediation efforts.
● Benefits:
○ Helps in assessing the health of ecosystems and the impact of human
activities.
○ Supports the identification of mineral resources for sustainable extraction.
● Used in
Geochemical surveys in mining regions have been crucial for assessing environmental
impacts and guiding remediation efforts.

8) Geography
● Application:
Geostatistics supports geographic analyses by mapping and interpreting spatial data.
It enhances understanding of spatial relationships, informing urban planning,
resource management, and policy development based on geographic patterns.
● Applied:
 ○ Geographic Information Systems (GIS) analyze spatial data for urban planning,
environmental management, and disaster response.
○ Utilizes mapping techniques to visualize demographic and environmental
data.
● Benefits:
○ Informs land use planning and resource allocation.
○ Enhances understanding of human-environment interactions.
● Used in
GIS has been used in urban planning to identify suitable locations for infrastructure
development while minimizing environmental impacts.

9) Soil Sciences:
● Application:
Geostatistics is used in soil science to analyze spatial variability in soil properties,
aiding in precision agriculture and land management practices. It informs
assessments of soil fertility, contamination, and erosion risks.
● Applied:
○ Soil analysis assesses fertility, contamination, and erosion.
○ Techniques like soil mapping and profile analysis inform agricultural practices
and land use.
● Benefits:
○ Enhances agricultural productivity through informed soil management
practices.
 ○ Supports environmental conservation efforts by understanding soil health.

● Used in
Soil health assessments guide conservation practices in agricultural regions, promoting
sustainable farming.

10) Forestry:
● Application:
In forestry, geostatistics aids in the assessment of forest health and biomass
distribution. It supports sustainable management practices by analyzing spatial
patterns of tree species and their ecological impacts.
● Applied:
○ Forest inventories assess tree species, health, and biomass.
○ Remote sensing techniques monitor deforestation and forest cover changes.
● Benefits:
○ Supports sustainable timber production and conservation of forest resources.
○ Enhances biodiversity through habitat management practices.

● Used in
Monitoring forest cover changes in the Amazon rainforest helps inform conservation
strategies and combat deforestation.
 11) Landscape Ecology
● Application:
Geostatistics helps analyze spatial patterns in landscapes, assessing habitat
fragmentation and connectivity. This application informs conservation efforts and
land use planning by understanding ecological dynamics and their implications for
biodiversity.
● Applied:
○ Landscape models assess habitat fragmentation and connectivity.
○ Used in conservation planning to maintain biodiversity and ecosystem
functions.
● Benefits:
○ Informs land use planning to minimize ecological impacts.
○ Supports habitat restoration efforts by identifying critical areas for
conservation.

● Used in
Landscape ecological studies have guided the restoration of degraded habitats in urban
areas, improving biodiversity.