Introduction

SAMPLING DESIGN

Sampling design is a fundamental part of data collection

It plays a critical role to ensure that there are sufficient

data to draw conclusion

A crucial step in collecting appropriate and defensible

data that accurately represent the investigated problem
 Why is selecting an appropriate
sampling design important?
To obtain a sound, science based decision thus the basis is
collecting accurate information

To generate accurate information, we must consider:

The appropriateness and accuracy of the sample collection
and handling method
The effect of measurement error
The quality and appropriateness of the laboratory analysis
The representativeness of the data with respect to the study
objective
 Purpose of a sampling design
1. To support a decision about whether contamination levels
exceed a threshold of unacceptable risk
2. To determine whether certain characteristics of two
populations differ by some amount
3. To estimate the mean characteristics of a population or its
proportion
4. To identify the location of hot spots areas having high
levels of contamination
5. To characterise the nature and extent of contamination at a
site
6. To monitor trends in environmental conditions or indicators
of health
What does a sampling design consist of?
1. The number of samples
2. Identifies the particular samples (e.g: geographic position
or time points of sample collection)
3. Explanation and justification for the number and the
positions/timings of the sample
Soil sample longitude and attitude
Air /water sample geographical location and time
Air - Note the time the air sampling would begin
and when it would end
4. Measurement protocol would specify when the sampler
would be retrieved and how the sample will be analysed
Concept and term : Representativeness
Measure the degree to which data accurately and precisely
represent
Population characteristic
Parameter variations at a sampling point
Process condition
Environmental condition

Moreover, a well-planned sampling design is intended to

ensure data are adequately representative of the target
population and defensible
Example:
Soil sample A, B and C may
be representative if the
objective is to determine
whether the pipe has
released a particular
contaminant.
But its not representative if
the objective is to estimate
the average concentration
level of the entire lagoon.
For that estimation,
random sampling locations
should be generated from
the entire site (include D, E
and F).
 Sampling design concepts and terms
Target population the set of all units that comprise
the items of interest in a scientific study

Sampled population that part of the target

population that is accessible and available for sampling

Examples:
Target population: surface soil in residential yard
Sampled population : areas of soil in that yard not
covered by structures or vegetation
 Sampling unit member of population that may be
selected for sampling, such as individual trees, or a
specific volume of air or water

Sampling frame a list of all the possible sampling

units from which the sample can be selected

Sample a collection of some of these sampling units

Sample support represent that portion of the

sampling unit such as an area, volume, mass or other
quantity
 Measurement protocol
Once sampling unit is selected, a measurement protocol
is applied.
A specific procedures for making observations or
performing analyses to determine the characteristics of
interest for each sampling unit

The sampling design specifies the number, type and

location (spatial/temporal) of sampling units to be selected
for measurement

Conceptual model
Describe the expected source of the contaminant and
the size and breadth of the area of concern
Identifies the relevant environmental media and the
relevant fate and transport pathways
Defines the potential exposure pathways
Example: water sampling - A study to measure E. coli and enterococci
level in a specific swimming area of a lake
Target Water flowing through the area from May 1 until Sept 15
population
Sampled Water in the swimming area at 7 am and 2 pm at
population approximately 6 inches below the surface
Sampling 1-liter volume of water at particular locations in the
units swimming area
Sample Equal to sampling unit, 1-liter of water
support
Measureme Use of a 2-liter beaker, held by a 6-inch handle. Non-
nt protocols motorised boat to minimise water disturbance. Sample is
collected in the standard manner and poured into a 2-liter
sample jar, up to 1-liter line. Sample is taken to the lab for
analysis within 6 hours.
Sampling Collect a minimum of 2 samples on each sampling day at 7
design am and 2 pm or up to 3 times a day if theres heavy rainfall
 Where, when, what, how, and
how many
The choice of where (spatially), and when
(temporally), to take samples should be based on
sound statistics (simple random sampling, stratified
random sampling, systematic sampling, composite
sampling).
 General sampling rules
1. Sample must be collected from the least to the most
contaminated sampling locations within the site
2. Disposable latex gloves should be worn when sampling
3. New, unused gloves must be used for each separate
sampling point.
4. For sampling hazardous materials, rubber gloves are
recommended
5. For compositing or mixing samples, use a bowl and
spatula. For trace organic and metal analysis, the material
of these tools should be stainless steel, glass or teflon
 Categories of sampling design:
Probabilistic and judgmental sampling designs
Types of sampling
Judgmental sampling
Simple random sampling
Stratified sampling
Systematic and grid sampling
Ranked set sampling
Adaptive cluster sampling
Composite sampling
 Types of sampling
Judgmental sampling

Selection of sampling units (e.g. number, location,

timing) is based on knowledge of the condition to be
investigated and on professional judgment.

Limited conclusion on target population

Depends entirely on the validity and accuracy of

professional judgment
 Simple random sampling

Selection of sampling units (e.g. number, location,

timing) is selected using random numbers
All possible selections of a given number of unit have
equal chance to be selected
E.g: simple random sample of a set of trees can be taken by
numbering all the trees and randomly selecting numbers from
that list OR by using pairs of random coordinates
Most useful when population of interest is relatively
homogenous (no hot spots)
Advantages:
1. Provide statistically unbiased results
2. Easy to understand and implement
3. Straightforward sample size calculation and data analysis
 Stratified sampling

Target population is separated into non-overlapping strata,

or subpopulations that are known/thought to be more
homogenous
Less variation among sampling units in the same stratum
than among sampling units in different strata.
Choose strata on the basis of spatial or temporal proximity,
based on professional judgment.
Advantages
Has greater potential for precision in estimating mean or variance
Greater precision if measurement of interest is strongly
correlated with the variable used to make the strata.
Stratified on the basis of
information about how
the contaminant is
present based on:
Wind pattern
Soil type
Surface soil texture

Useful for estimating a

parameter when the
target population is
heterogenous

When the area can be

subdivided based on
expected contamination
levels.
 Systematic and grid sampling

Samples taken at regularly spaced intervals over space or time

Initial location is chosen at random
Then, remaining sampling locations are defined so that all
locations are at regular intervals over an area (grid) or time
(systematic)
Used to search for hot spots
To infer means, percentiles or other parameters
To estimate spatial patterns or trends over time
Provide a practical and easy method for designating sample
locations
Ensure uniform coverage of a site, unit or process
 Ranked set sampling

Incorporate professional judgment of a field investigator or a

field screening measurement to pick specific sampling
location
1. Identify sets of field location using random sampling
2. Locations are ranked independently within each set using
professional judgment, or using surrogate measurement
(inexpensive, fast)
3. One sampling unit from each set is then selected (based on
observed rank) for subsequent measurement (accurate, reliable
hence expensive)
Compared to random sampling, this type of sampling is
more representative
 Adaptive cluster sampling

N samples are taken using simple random sampling

Additional samples are taken at locations where
measurements exceed some threshold level
It tracks the selection probabilities for later phases of
sampling so that an unbiased estimate of the population
mean can be calculated despite oversampling
Advantages:
Estimating rare characteristics of population
Inexpensive
Rapid measurement
Delienate boundaries of hot spots
 Composite sampling

Volumes of material
from several sampling
units are physically
combined and mixed to
form a single
homogenous sample,
then analysed.

Cost effective
 Types of sample
Grab samples
An individual sample collected at a particular time and
place
Represent conditions at the time it was collected
Should not be used as a basis for a decision about
pollution abatement
Some sources are quite stable in composition, and may
be represented well by single grab sample
 Types of sample
Composite samples
A mixture of grab samples collected at the same point at
different times
A series of smaller samples are collected in a single container
and blended for analysis
The mixing process averages the variations in sample
composition and minimises analytical effort and expense
Reflect the average characteristics during the sampling period
and in most cases a 24 hour is standard
The volume of samples should be constant in constant time
interval and mixed well at the end of the composite period
Not suitable for: VOC, oil and grease, total recoverable
petroleum hydrocarbon, microbiology testing
 Types of sample
Duplicates
Collected for checking the preciseness of the sampling
process

Split samples
Collected for checking analytical performance
Sample is taken in one container, mixed thoroughly and
halved into another properly cleaned container. Preserve
both samples as needed
Both halves are now samples that represent the same
sampling point
 Requirement of sampling container
according to parameters
Inorganic parameters if do not need preservation;
plastic, glass, teflon, stainless steel, aluminium
Nutrients plastic, glass, teflon, stainless steel,
aluminium
Trace metals plastic, stainless steel, teflon
Extractable organics glass, aluminium, brass,
stainless steel, teflon
VOC glass, stainless steel, teflon
Microbiological presterilised sample container
Systematic planning process of sampling design
Sampling design process
Steps in sampling design process
Choosing the appropriate sampling design