C H A P T E R

1
Multivariate Data

LEARNING OBJECTIVES

After careful consideration of this chapter, you should be able:

• To describe the concept of environmental statistics.
• To understand the importance of environmental statistics for making intelligent conclusions.
• To describe the concept of multivariate analysis.
• To know the advantages of employing multivariate methods for analyzing environmental data.
• To organize the outcomes of multivariate data to be prepared for analysis.
• To explain the distinction between univariate and multivariate notions.
• To explain how and where to employ multivariate data.
• To understand the univariate normal distribution.
• To understand the multivariate normal distribution.

1.1 THE CONCEPT OF ENVIRONMENTAL STATISTICS

Environmental statistics is the application of various statistical methods, including procedures and techniques in the
field of environmental science and environmental engineering, such as weather, air, water quality, climate, soil, fish-
eries and other environmental activities. Statistical methods are used in designing environmental projects and for the
analysis and interpretation of environmental data to help and guide scientists draw useful and meaningful conclusions
in various aspects of the environment. Environmental statistics can help in describing environmental problems in
terms of mathematical modeling to understand the impact of the chosen variables under study and show the direction
of change—increase or decrease—or the nature of the relationship—positive or negative. Furthermore, statistical tech-
niques can identify the general trend and entangle the hidden relationships as well, which would help scientists under-
stand the process and have a clear picture regarding all relationships to avoid risk and guide management to properly
plan environmental projects.
Environmental statistics can help in understanding the importance of variability and oscillation in the data, employ-
ing various measures and methods to show the influence of variability, and lead scientists to search for scientific expla-
nations. Thus, scientists should at least learn and understand basic statistics to help understand the importance of the
results and the analysis that guides informative conclusions.

1.2 THE CONCEPT OF MULTIVARIATE ANALYSIS

The concept of univariate statistical analysis covers statistical techniques for testing a data set with one variable.
However, most research projects need to measure several variables for each research unit or individual (sampling units
or experimental units) in one or more samples.
For example, consider assessing the water quality of a river based on monitoring certain parameters such as pH,
dissolved oxygen (DO), electrical conductivity (EC), turbidity, biological oxygen demand (BOD), chemical oxygen

Applied Statistics for Environmental Science with R 1 © 2020 Elsevier Inc. All rights reserved.
[Link]
2 1. MULTIVARIATE DATA

demand (COD), and total suspended solids (TSS), to make a decision about the pollution status of the river. In this case,
there are seven variables to be measured for each sample, which are generally highly correlated. If one variable at a
time is considered to analyze the results of multivariate data, the relationships between the variables would be ignored
and a different picture would be reflected regarding the true behavior of the chosen parameters (variables) in the pres-
ence of other parameters (variables). Thus, we should use a method that takes into account the correlation between
chosen variables to untie the overlapping input (information) by the correlated variables to understand the behavior of
the chosen variables properly.
Therefore, data sets with several variables can be analyzed, employing multivariate methods that consider the
relationship between the chosen variables. Multivariate methods are a collection of techniques that can serve several
purposes in the field of environmental science and engineering, which include cluster analysis for recognizing
groups of similar observation (e.g., individuals, objects); principal components analysis and factor analysis as data
reduction methods to reduce the number of variables to a smaller number of dimensions, called components
(factors), which are uncorrelated without losing valuable information; discriminant analysis, which is applied
to separating the data into various groups based on the measured variables; multivariate analysis of variance
(MANOVA), employed to perform statistical hypothesis testing based on multivariate data (several variables);
and multivariate multiple regression analysis, which is employed for making predictions based on the relation
among the variables.

1.3 CONFIGURATION OF MULTIVARIATE DATA

We can organize multivariate data for various variables measured from a number of samples (items) in a table.
The number of samples (items) specify the number of rows and the number of variables specify the number of columns
of the table.
In general, Table 1.1 shows the configuration of n samples (representing the rows) and k variables (representing the
columns) measured for each sample.
Note: The total number of variables measured for each sample (k) is usually smaller than the total number of
samples, n.

1.4 EXAMPLES OF MULTIVARIATE DATA

We can illustrate the concept of multivariate data by providing some real examples of multivariate data with regard
to environmental science that may be useful in more easily explaining the associated computations. Furthermore, the
interpretation of these examples will be given, including the scenario of each case.

TABLE 1.1 The Configuration of n Samples and k Variables in Multivariate Form

Variable
Sample Y1 Y2 … Yj … Yk

1 Y11 Y12 … Y1j … Y1k

2 Y21 Y22 … Y2jj … Y2k
. . . … . … .
. . . … . … .
. . . … . … .
i Yi1 Yi2 … Yij … Yik
. . . … . … .
. . . … . … .
. . . … . … .
n Yn1 Yn2 … Ynj … Ynk
 1.4 EXAMPLES OF MULTIVARIATE DATA 3

EXAMPLE 1.1 ASSESSMENT OF SURFACE WATER

An assessment of water quality in the Juru and Jejawi Rivers in the Penang State of Malaysia was conducted by monitoring
10 physiochemical parameters; namely, the electrical conductivity (EC) and temperature were measured employing an HACH
portable pH meter, and dissolved oxygen (DO) was measured with a YSI 1000 DO meter. The chemical oxygen demand (COD),
biochemical oxygen demand (BOD), total phosphate concentrations, and total nitrate were analyzed employing a spectropho-
tometer (HACH/2010). The turbidity was measured employing a nephelometer. The total suspended solids (TSS) were ana-
lyzed gravimetrically in the laboratory. The APHA Standard Methods for the Examination of Water and Wastewater were
applied to analyze the concentration of the abovementioned parameters. The data obtained from 10 different sites are each given
in Table 1.2.

TABLE 1.2 Physiochemical Parameters for the Juru and Jejawi Rivers
Location Temperature PH DO BOD COD TSS EC Turbidity Total phosphate Total nitrate

Juru 28.15 7.88 6.73 10.56 1248.00 473.33 42.45 13.05 0.88 12.45
Juru 28.20 7.92 6.64 10.06 992.50 461.67 42.75 14.11 0.55 15.05
Juru 28.35 7.41 5.93 6.01 1265.00 393.34 29.47 25.95 0.88 15.80
Juru 28.40 7.84 6.23 14.57 1124.00 473.34 42.35 22.00 1.26 12.45
Juru 28.30 7.86 6.29 7.36 1029.50 528.34 40.70 12.36 0.61 12.90
Juru 28.30 7.89 6.36 5.27 775.00 458.33 42.50 17.90 1.01 19.25

Juru 28.55 7.40 5.67 4.81 551.00 356.67 29.51 27.70 1.02 25.55
Juru 28.75 7.41 5.26 4.96 606.00 603.33 28.45 29.35 0.84 17.90
Juru 29.30 7.25 4.18 6.61 730.00 430.00 26.25 35.35 0.73 12.85
Juru 29.55 7.18 5.14 5.11 417.00 445.00 25.79 27.75 0.72 17.15
Jejawi 27.80 7.40 4.96 36.65 82.35 815.00 29.55 1.01 1074.00 10.51
Jejawi 27.70 7.53 5.86 37.20 122.90 858.34 21.80 0.77 947.50 6.31

Jejawi 27.85 7.45 5.52 41.50 46.35 823.34 27.55 1.12 1194.50 7.36
Jejawi 27.85 7.41 5.27 36.75 52.65 821.67 16.45 0.46 988.50 5.87
Jejawi 27.65 7.27 5.71 37.95 99.95 736.67 21.10 1.01 653.00 16.36
Jejawi 27.85 7.36 5.16 34.85 36.30 373.34 11.50 0.73 911.00 11.26
Jejawi 27.85 7.37 5.08 37.05 33.50 385.00 13.35 0.71 632.00 3.75
Jejawi 27.75 7.40 5.23 37.10 36.35 403.34 10.85 0.57 787.00 3.31

Jejawi 27.70 7.38 5.30 38.30 30.00 550.00 13.90 0.75 879.50 3.15
Jejawi 27.65 7.36 5.17 38.40 26.15 363.34 12.65 0.62 1339.00 8.56

In this example, 20 samples (rows) are collected from the Juru and Jejawi Rivers (10 different sites each), and 10 physiochem-
ical parameters (columns) are measured from each sample. It is very difficult for the researchers to make a decision about the
pollution status of the rivers or about the behavior of the different parameters based on the chosen parameters under study
because there are many different values, and the differences (fluctuations) between the values of the parameters from sample
to sample will mislead the researcher, making it difficult to make a correct decision. Thus, these data should be analyzed
employing an appropriate multivariate method that meets the objective of the study to untangle the information and to under-
stand the body of the phenomenon. The objectives of this research were to determine the range of similarity among the sampling
sites, to recognize the variables responsible for spatial differences in river water quality, to determine the unobserved factors
while demonstrating the framework of the database, and to quantify the effect of possible natural and anthropogenic sources of
the chosen water parameters of the rivers.
4 1. MULTIVARIATE DATA

EXAMPLE 1.2 ASSESSMENT OF LANDFILL LEACHATE TREATMENT

A researcher wishes to investigate the concentrations of 12 heavy metals (magnesium (Mg), calcium (Ca), sodium (Na),
iron (Fe), zinc (Zn), copper (Cu), chromium (Cr), cadmium (Cd), lead (Pb), arsenic (As), cobalt (Co), and manganese (Mn))
and eight physiochemical parameters (chemical oxygen demand (COD), biochemical oxygen demand (BOD), total dissolved
solids (TDS), total suspended solids (TSS), electrical conductivity (EC), pH, ammoniacal-N (NH3N), and dissolved oxygen
(DO)) in three ponds (collection, aeration, and stabilization) of a landfill leachate. The data for the three ponds are given in
Table 1.3.

TABLE 1.3 The Results of Physiochemical and Heavy Metals Obtained From Three Ponds of Landfill Leachate
Parameter
Pond pH EC TDS TSS COD BOD NH3H DO Mg Ca

Collection 7.19 6242.00 4530.00 43.26 944.67 9.57 81.66 6.90 12.58 14.28

Collection 7.36 6062.00 4392.00 29.45 845.33 8.77 61.33 6.69 17.91 13.17
Collection 7.41 6164.00 4476.00 31.52 897.00 9.21 71.00 6.48 14.44 15.54
Collection 7.92 6710.33 4710.00 39.36 810.00 8.66 52.43 6.85 19.99 88.25
Collection 7.84 6690.00 4709.83 36.46 686.67 9.66 55.80 6.85 20.50 81.78
Collection 7.83 6600.33 4654.83 37.00 1365.00 10.33 52.80 7.23 21.18 85.18
Collection 8.46 6893.00 4849.00 47.33 692.00 8.47 110.67 2.34 27.23 24.94

Collection 8.47 6850.00 4860.00 50.67 967.00 9.27 93.00 2.82 19.01 14.38
Collection 8.48 6873.00 4859.00 49.00 833.00 8.87 103.00 2.67 24.66 19.99
Aeration 8.98 2501.00 1792.00 101.67 502.67 76.33 3.27 7.88 10.98 17.04
Aeration 8.97 2438.00 1750.00 89.00 543.67 93.67 8.90 7.62 10.99 9.84
Aeration 9.07 2422.00 1762.00 92.00 526.67 71.00 10.80 8.31 10.98 15.00
Aeration 8.21 2255.00 1594.00 148.83 566.67 4.66 4.30 7.23 18.51 54.30

Aeration 8.38 2092.83 1577.83 151.00 618.33 15.33 3.80 7.37 20.14 54.08
Aeration 8.32 2240.67 1578.83 162.67 573.33 10.00 4.40 7.55 21.77 55.30
Aeration 9.97 1513.00 1065.00 51.00 339.00 9.10 20.08 8.08 7.96 5.25
Aeration 10.18 1497.00 1049.00 42.00 322.00 9.32 23.27 8.01 6.85 4.44
Aeration 10.21 1532.00 1084.00 38.00 324.00 8.95 23.26 8.12 8.53 6.78
Stabilization 8.67 1491.00 1087.00 115.00 521.33 17.33 1.37 6.95 20.82 19.99

Stabilization 8.60 1521.00 1104.00 105.33 504.00 29.67 7.82 6.42 20.50 23.10
Stabilization 8.58 1545.00 1123.00 108.33 532.00 47.33 6.00 5.71 20.48 27.13
Stabilization 8.82 1321.83 932.67 98.00 253.67 13.66 6.60 7.55 11.92 66.96
Stabilization 8.92 1352.88 965.00 89.39 520.00 15.00 6.20 7.23 12.46 65.28
Stabilization 8.93 1314.83 942.83 101.61 395.00 15.00 6.20 7.23 12.34 62.13
Stabilization 9.09 652.00 462.00 57.00 248.00 4.98 2.60 6.74 4.47 1.96

Stabilization 9.19 669.00 476.00 51.00 257.00 5.61 2.59 6.71 4.11 2.31
Stabilization 9.43 680.00 480.00 47.00 273.00 4.92 1.93 6.66 3.87 2.30
 1.4 EXAMPLES OF MULTIVARIATE DATA 5
TABLE 1.3 The Results of Physiochemical and Heavy Metals Obtained From Three Ponds of Landfill Leachate—cont’d
Parameter
Pond Na Fe Zn Cu Cr Cd Pb As Co Mn

Collection 978.37 0.35 14.46 9.80 31.50 0.31 2.71 32.28 6.34 5.44
Collection 978.37 0.23 17.89 9.03 22.37 0.32 3.35 20.55 5.14 9.55

Collection 978.37 0.62 11.50 9.59 10.70 0.01 2.16 9.75 3.95 16.60
Collection 284.44 0.56 23.80 57.10 18.12 0.85 8.03 11.52 6.26 52.72
Collection 273.47 0.56 24.00 57.40 17.04 1.73 7.60 14.41 7.43 51.82
Collection 283.80 0.56 23.10 56.80 17.58 1.71 7.04 10.09 6.61 50.92
Collection 673.45 1.34 61.01 64.92 42.36 0.04 0.70 25.21 21.30 109.34

Collection 633.35 1.98 49.15 53.45 59.06 0.56 3.91 14.81 18.78 133.59
Collection 665.61 0.79 27.48 58.75 52.27 0.06 2.08 17.07 23.67 122.59
Aeration 980.00 0.25 151.10 0.43 47.82 0.01 2.82 10.72 6.74 22.94
Aeration 975.55 0.73 259.08 2.83 3.67 0.39 2.72 11.12 2.61 32.44
Aeration 970.55 1.21 203.88 5.23 91.97 0.08 2.62 10.92 10.81 41.94
Aeration 525.57 0.87 425.46 69.42 86.18 0.56 7.17 21.66 16.18 4.97

Aeration 516.21 0.93 424.43 66.68 89.81 0.35 7.57 21.36 15.99 5.32
Aeration 517.13 0.85 423.45 63.63 85.56 0.16 7.37 20.68 15.78 5.96
Aeration 676.98 0.77 28.05 13.66 27.88 0.32 2.58 6.73 5.73 8.52
Aeration 630.31 0.94 18.69 15.97 37.11 0.08 4.90 10.12 5.61 10.34
Aeration 665.12 0.57 34.44 16.75 30.65 0.11 3.86 7.66 6.75 9.52
Stabilization 975.37 1.33 117.47 3.90 55.15 0.30 3.80 22.23 15.01 11.50

Stabilization 978.37 0.81 278.47 5.21 70.29 0.01 4.12 18.10 20.84 21.61
Stabilization 973.37 0.29 175.47 6.51 54.57 0.00 2.90 12.15 17.93 30.51
Stabilization 411.60 1.37 229.90 23.93 48.47 0.28 6.45 16.55 5.19 6.44
Stabilization 439.04 1.43 227.39 25.23 49.26 0.61 5.25 15.75 4.75 6.02
Stabilization 494.26 1.23 221.89 25.03 48.83 0.01 5.89 17.35 4.62 6.82
Stabilization 141.00 0.33 149.11 6.42 12.88 0.02 2.48 5.71 1.28 1.63

Stabilization 103.81 0.87 240.77 7.91 8.76 0.14 4.63 3.32 1.54 1.93
Stabilization 81.20 0.51 190.68 6.64 11.00 0.04 3.78 2.37 1.62 1.66

The leachate samples were collected from collection, aeration, and stabilized ponds in the ATLS leachate collection system.
The leachate samples were collected three times during the period between August 2017 and January 2018, with three sampling
points at each pond. The samples were manually gathered and placed in 500 ml polyethene containers. The samples were
immediately transported to the laboratory and cooled to 4°C to reduce biological and chemical reactions (Japan International
Cooperation Agency (JICA)).
In this example, 27 samples (rows) were collected from the three ponds (collection, aeration, and stabilization), and
20 parameters (columns) were measured from each sample. It is not easy for the scientist to make a decision about the
6 1. MULTIVARIATE DATA

treatment process of the landfill or about the behavior of the different parameters under study because there are many dif-
ferent values (27 samples
20 parameters ¼ 540 values). Thus, the relationship among the different chosen parameters
should be investigated and studied properly to understand the differences (fluctuations) in the parameters from one pond
to another. These data should be analyzed employing an appropriate technique to achieve the objective of the project. The
first objective was to assess whether the treatment process of the landfill leachate worked properly, and the relationship
among the chosen parameters should be investigated for more information on the behavior of each variable in the presence
of other chosen variables, which would help to identify the source of the variation. The second objective was to assess the
effect of the landfill on the groundwater and surface water in the chosen area (the data for groundwater and surface water are
not presented to save space). Furthermore, the contribution of each chosen variable (parameter) in illustrating the total var-
iation in the collected data was identified employing multivariate methods. This research may help in estimating the impact
of the landfill on groundwater and surface water in the chosen area.

EXAMPLE 1.3 INORGANIC ELEMENTS IN THE PARTICULATE MATTER

IN THE AIR
A researcher wishes to investigate the concentrations of nine inorganic elements in the particulate matter (PM10) in the air of
an equatorial urban coastal location. In 2009, air pollution levels were studied during the summer and winter monsoon seasons
employing high-volume sampling techniques. Atomic absorption spectrophotometry was employed to collect PM10 samples,
with an average time of 24 h. The parameters were the particulate matter PM10, aluminum (Al), zinc (Zn), iron (Fe), copper (Cu),
calcium (Ca), sodium (Na), manganese (Mn), nickel (Ni), and cadmium (Cd). The data are given in Table 1.4.

TABLE 1.4 The Results of Inorganic Elements in Particulate Matter in the Air (μg/m3)
Season PM10 Al Zn Fe Cu Ca Na Mn Ni Cd

Summer 37.71 0.01 2.32 0.97 0.00 1.37 7.94 0.05 0.01 0.01
Summer 48.03 0.01 2.05 0.75 0.00 1.56 8.46 0.05 0.01 0.04
Summer 67.87 0.01 3.43 0.44 0.00 1.41 9.98 0.06 0.10 0.05
Summer 39.01 0.01 4.24 0.45 0.00 1.12 12.93 0.01 0.07 0.00
Summer 38.33 0.01 3.51 0.50 0.02 1.60 9.92 0.02 0.00 0.05

Summer 29.70 0.01 2.78 0.58 0.02 1.38 12.13 0.04 0.06 0.02
Summer 53.66 0.01 2.40 0.73 0.02 1.59 7.81 0.23 0.11 0.02
Summer 132.28 0.01 2.34 0.55 0.02 1.59 9.01 0.03 0.00 0.01
Summer 66.31 0.01 2.13 0.45 0.02 1.23 8.76 0.02 0.00 0.05
Summer 69.20 0.01 2.13 0.50 0.03 1.95 8.66 0.03 0.08 0.05
Summer 78.17 0.01 1.96 0.56 0.04 1.35 8.88 0.03 0.02 0.07

Summer 31.63 0.01 2.21 0.42 0.04 1.24 9.02 0.02 0.08 0.05
Summer 66.73 0.01 1.46 0.51 0.03 1.33 8.53 0.02 0.13 0.03
Summer 113.56 0.01 2.07 0.60 0.03 1.51 7.65 0.02 0.06 0.05
Summer 123.40 0.01 1.75 0.57 0.03 1.75 7.45 0.02 0.12 0.04
Summer 72.39 0.01 1.61 0.57 0.03 1.54 8.40 0.02 0.07 0.02

Summer 51.85 0.01 1.16 0.51 0.18 1.72 7.44 0.01 0.09 0.00
Summer 77.59 0.01 1.66 0.81 0.04 2.04 7.89 0.03 0.08 0.03
Summer 30.30 0.01 1.09 0.61 0.03 1.80 6.28 0.02 0.05 0.03
Summer 100.40 0.01 1.82 0.47 0.04 1.70 7.05 0.02 0.05 0.02
Summer 132.98 0.01 1.86 0.52 0.03 1.74 7.69 0.01 0.10 0.03
Summer 126.38 0.01 0.34 0.02 0.00 0.01 1.36 0.01 0.18 0.00
 1.4 EXAMPLES OF MULTIVARIATE DATA 7
TABLE 1.4 The Results of Inorganic Elements in Particulate Matter in the Air (μg/m3)—cont’d
Season PM10 Al Zn Fe Cu Ca Na Mn Ni Cd

Summer 31.82 0.01 1.41 0.52 0.03 1.93 7.04 0.02 0.04 0.00
Summer 110.53 0.01 1.25 0.67 0.04 2.29 7.39 0.02 0.03 0.00
Summer 38.41 0.01 1.55 0.79 0.12 2.21 7.74 0.02 0.06 0.03

Summer 124.26 0.01 2.24 0.00 0.00 1.56 9.65 0.07 0.06 0.01
Summer 53.62 0.01 1.97 0.01 0.00 1.39 7.07 0.03 0.04 0.03
Summer 23.30 0.01 1.28 0.02 0.00 1.80 6.39 0.03 0.00 0.00
Summer 67.34 0.01 0.91 0.02 0.00 0.80 7.14 0.00 0.08 0.00
Summer 22.58 0.03 1.29 0.02 0.01 0.92 7.21 0.02 0.05 0.02

Summer 54.52 0.03 1.37 0.04 0.01 1.48 7.98 0.01 0.03 0.03
Summer 112.83 0.01 1.88 0.03 0.00 1.46 8.44 0.01 0.02 0.00
Summer 175.28 0.01 1.68 0.03 0.00 1.38 8.13 0.00 0.03 0.00
Summer 47.30 0.01 2.20 0.03 0.00 1.13 11.39 0.01 0.03 0.01
Summer 57.08 0.01 1.69 0.04 0.00 1.04 8.34 0.02 0.07 0.00
Summer 15.16 0.01 1.58 0.04 0.01 1.77 6.69 0.05 0.08 0.02

Summer 272.98 0.01 1.88 0.04 0.01 1.60 7.05 0.04 0.08 0.00
Summer 101.82 0.01 1.86 0.02 0.01 1.79 6.02 0.05 0.13 0.00
Summer 59.90 0.01 1.73 0.02 0.01 1.70 7.25 0.05 0.00 0.02
Summer 31.07 0.01 1.92 0.02 0.02 1.40 4.92 0.05 0.03 0.04
Summer 107.43 0.01 2.03 0.02 0.02 1.48 5.19 0.04 0.06 0.00
Summer 30.51 0.01 2.42 0.03 0.03 2.08 5.70 0.07 0.00 0.04

Summer 84.02 0.01 2.10 0.22 0.01 1.61 5.85 0.03 0.00 0.00
Summer 7.37 0.01 2.63 0.35 0.02 1.24 3.90 0.03 0.00 0.00
Summer 7.65 0.01 1.38 0.34 0.01 1.11 4.20 0.04 0.00 0.00
Summer 15.05 0.01 1.56 0.21 0.02 0.35 2.47 0.03 0.00 0.00
Summer 84.13 0.01 1.89 0.04 0.02 0.41 4.44 0.02 0.00 0.02
Winter 82.87 0.01 1.66 0.15 0.02 0.64 4.79 0.02 0.06 0.00

Winter 138.25 0.01 2.53 0.30 0.03 0.42 3.02 0.02 0.00 0.03
Winter 82.67 0.01 2.66 0.31 0.04 0.03 3.16 0.03 0.00 0.00
Winter 46.08 0.01 3.39 0.52 0.02 0.21 2.34 0.03 0.00 0.00
Winter 15.33 0.01 2.63 0.23 0.02 0.06 4.38 0.03 0.05 0.00
Winter 30.78 0.01 1.98 0.17 0.01 0.40 3.68 0.01 0.00 0.00
Winter 155.00 0.01 3.72 0.15 0.02 1.21 4.24 0.02 0.00 0.00

Winter 61.75 0.01 2.93 0.00 0.03 0.92 4.29 0.01 0.00 0.00
Winter 88.90 0.01 4.91 0.11 0.02 0.93 6.65 0.02 0.00 0.00
Winter 38.47 0.01 3.46 0.15 0.02 0.73 4.25 0.01 0.01 0.00
Winter 24.46 0.01 1.97 0.19 0.01 0.75 1.96 0.01 0.00 0.00
Winter 14.29 0.01 2.76 0.11 0.01 0.67 9.55 0.03 0.00 0.00
Winter 65.52 0.01 2.56 0.32 0.01 0.54 2.70 0.04 0.00 0.00

Winter 63.33 0.01 1.21 0.16 0.02 0.53 3.40 0.04 0.03 0.00
Winter 62.28 0.01 3.52 0.43 0.01 0.80 7.78 0.02 0.11 0.00
Continued
8 1. MULTIVARIATE DATA

TABLE 1.4 The Results of Inorganic Elements in Particulate Matter in the Air (μg/m3)—cont’d
Season PM10 Al Zn Fe Cu Ca Na Mn Ni Cd

Winter 65.87 0.01 3.48 0.25 0.02 0.99 11.06 0.01 0.03 0.00
Winter 122.83 0.01 3.24 0.41 0.03 1.09 11.47 0.01 0.06 0.00
Winter 50.18 0.00 2.40 0.49 0.02 1.16 7.90 0.02 0.11 0.00

Winter 60.19 0.01 3.02 0.38 0.02 0.84 9.51 0.02 0.06 0.01
Winter 28.28 0.01 3.68 0.48 0.06 1.18 12.21 0.01 0.09 0.00
Winter 42.50 0.01 1.86 0.46 0.06 1.34 10.51 0.03 0.07 0.00
Winter 49.68 0.01 2.29 0.32 0.04 1.57 11.02 0.02 0.00 0.04
Winter 39.91 0.01 3.30 0.45 0.05 1.46 8.61 0.03 0.08 0.02

Winter 37.94 0.01 3.57 0.31 0.08 1.59 9.47 0.04 0.00 0.05
Winter 27.97 0.01 3.14 0.09 0.06 1.46 12.08 0.03 0.00 0.05
Winter 45.34 0.01 2.59 0.26 0.06 1.59 11.43 0.03 0.06 0.00
Winter 58.45 0.01 3.46 0.12 0.06 1.66 8.98 0.04 0.10 0.00
Winter 43.30 0.01 1.58 0.09 0.05 1.46 7.59 0.03 0.00 0.00

Table 1.4 provides a huge and complex data set. It can easily be observed that the table does not provide helpful data (infor-
mation) for making a decision. The objective of the study was to assess the air quality of Penang, Malaysia, in terms of PM10 and
inorganic elements, and to recognize the main sources of PM10 and inorganic elements, whether crustal or noncrustal. The goals
included investigating the relation between the different chosen inorganic elements and PM10 during the summer and winter
monsoons and determining the similarities between the chosen parameters.

EXAMPLE 1.4 HEAVY METALS IN SEDIMENT (mg/L)

A researcher wishes to investigate the concentrations of eight heavy metals (cadmium (Cd), iron (Fe), copper (Cu), zinc (Zn),
chromium (Cr), mercury (Hg), manganese (Mn), and lead (Pb)) in sediments obtained from two sites. The two sites, with 10 sam-
pling points at each site, were Kuala Juru (the Juru River) and Bukit Tambun (the Jejawi River) in the Penang State of Malaysia.
The sediment samples were gathered at low tide with an Eijkelkamp gouge auger from each of the 20 chosen sampling points (10
sampling points from each river estuary). A flame atomic absorption spectrometer (FAAS; Perkin Elmer HGA-600) was
employed for the analysis of Cu, Pb, Zn, Cd, Mn, Fe, and Cr, and a cold vapor atomic absorption spectrometer (CV-AAS)
method was used for Hg analysis after sample digestion in an acid solution. The data are given in Table 1.5.

TABLE 1.5 The Concentration of Heavy Metals in Sediment (mg/L) for Juru and Jejawi Rivers
Cu Zn Cd Cr Fe Pb Hg Mn

Juru 0.63 1.89 1.95 0.15 26.17 0.48 0.10 0.01

Juru 0.73 1.79 1.99 0.17 22.95 0.35 0.12 0.02

Juru 0.35 0.90 1.94 0.09 25.41 0.38 0.12 0.03
Juru 0.76 2.19 1.98 0.20 24.13 0.17 5.19 0.01
Juru 0.60 1.74 1.94 0.14 23.16 0.36 0.49 0.02
Juru 0.36 3.38 1.95 0.14 27.20 0.17 0.14 0.02
Juru 0.63 3.12 1.98 0.12 26.84 0.61 0.12 0.03

Juru 0.52 3.42 1.93 0.14 25.98 0.32 0.12 0.03

 1.5 MULTIVARIATE NORMAL DISTRIBUTION 9
TABLE 1.5 The Concentration of Heavy Metals in Sediment (mg/L) for Juru and Jejawi Rivers—cont’d
Cu Zn Cd Cr Fe Pb Hg Mn

Juru 0.55 2.86 1.97 0.19 25.60 0.30 0.12 0.02

Juru 0.47 3.71 1.92 0.13 26.26 0.80 0.11 0.03
Jejawi 0.38 3.27 1.60 0.13 38.72 0.16 0.17 1.63

Jejawi 0.43 2.47 1.69 0.12 38.66 0.18 0.23 1.68

Jejawi 0.39 1.59 1.62 0.17 38.94 0.15 0.15 1.75
Jejawi 0.30 2.12 1.64 0.20 37.74 0.18 0.15 1.75
Jejawi 0.49 2.28 1.66 0.28 39.52 0.28 0.56 1.90
Jejawi 0.45 2.24 1.68 0.11 38.49 0.28 0.24 1.67

Jejawi 0.48 1.76 1.76 0.16 38.22 0.15 0.17 1.71

Jejawi 0.46 2.51 1.75 0.23 39.01 0.42 0.22 1.77
Jejawi 0.23 1.80 1.71 0.21 39.61 0.41 0.24 1.85
Jejawi 0.20 1.45 1.73 0.17 38.80 0.15 0.18 1.70

In this study, the researcher wanted to investigate and understand the interrelationship between the chosen parameters and
to extract information about the resemblance or differences between the different sampling sites, identification of the variables
(heavy metals) accountable for the spatial differences in river estuaries, and the effect of the possible sources (natural and
anthropogenic) on the chosen heavy metals of the two river estuaries.

1.5 MULTIVARIATE NORMAL DISTRIBUTION

The normal distribution is the most important distribution in statistics. The normal distribution is very important
because most of the tests used in statistics require that the assumption of normality be met; the data are gathered from a
normally distributed population.
A brief explanation of the univariate and multivariate normal distributions is given below.

1.5.1 Univariate Normal Distribution

Suppose Y is a random variable that follows the normal distribution. Then, the probability distribution function of
the univariate normal distribution is given in Eq. (1.1).
1 2
f ðYÞ ¼ pffiffiffiffiffiffiffiffiffiffi eðYμÞ =2σ 2 ,  ∞ < Y < ∞ (1.1)
2πσ 2

where
μ is the mean; and
σ 2 is the variance.
The univariate normal distribution can be written as Y  N(μ, σ 2).
The normal distribution (bell-shaped) curve is presented in Fig. 1.1. R statistical software was used to generate
the curve. The commands and built-in functions for creating a normal distribution curve are presented in the
Appendix.
10 1. MULTIVARIATE DATA

FIG. 1.1 Normal distribution plot.

1.5.2 Multivariate Normal Distribution

Suppose Y is a random vector of k random variables that follows the multivariate normal distribution. Then, the
multivariate normal distribution is given in Eq. (1.2).
1 X1
ðYμÞ
g ð YÞ ¼ X1=2 e ðY  μÞ=2 (1.2)
 
ð2π Þk=2  

where
k is the number of variables;
μ
Pis the mean vector;
is the
Pcovariance matrix; and
ðY  μÞ 1 ðY  μÞ is the Mahalanobis distance (statistical distance).
P
The multivariate normal distribution can be denoted as Y  Nk( μ, ).
Note:
• A suitable transformation of the data should be used if the normality assumption is violated by one or more
variables under investigation, as when the data are highly skewed with several outlier (extreme) values (high or
low) or repeated values.
• If all the individual variables follow a normal distribution, then it is supposed that the combined (joint) distribution
is a multivariate normal distribution.
• In practice, the real data never follow a multivariate normal distribution completely; however, the normal density
can be employed as an approximation of the true population distribution.

Further Reading
Alkarkhi, A.F.M., Alqaraghuli, W.A.A., 2019. Easy Statistics for Food Science with R, first ed. Academic Press.
Alkarkhi, A.F.M., Ismail, N., Ahmed, A., Easa, A.M., 2009. Analysis of heavy metal concentrations in sediments of selected estuaries of Malaysia—a
statistical assessment. Environ Monit Assess 153, 179–185.
Banch, T.J.H., Hanafiah, M.M., Alkarkhi, A.F.M., Amr, S.S.A., 2018. Statistical evaluation of landfill leachate system and its impact on groundwater
and surface water in Malaysia.
Blogger, 2011. R graph gallery: A collection [Online]. Available, [Link]
Bryan, F.J.M., 1991. Multivariate Statistical Methods: A Primer. Chapman & Hall, Great Britain.
Daniel & Hocking, 2013. Blog Archives, High Resolution Figures in R [Online]. R-bloggers. Available, [Link]
daniel-hocking/.
Johnson, R.A.W., W, D., 2002. Applied Multivariate Statistical Analysis. Prentice Hall, New Jersey.
Rencher, A.C., 2002. Methods of Multivariate Analysis. J. Wiley, New York.
Yusup, Y., Alkarkhi, A.F.M., 2011. Cluster analysis of inorganic elements in particulate matter in the air environment of an equatorial urban coastal
location. Chemistry and Ecology 27, 273–286.