&EPA
EPA 600/R-14/159 | June 2014 | www.epa.gov/ord
United States
Environmental Protection
Agency
Air Sensor Guidebook
\;
Office of Research and Development
National Exposure Research Laboratory
-------
EPA/600/R-14/159 June 2014 www.epa.gov/ord
Air Sensor Guidebook
Ron Williams and Vasu Kilaru
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC, USA
Emily Snyder
National Homeland Security Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC, USA
Amanda Kaufman
ASPPH Environmental Health Fellow hosted by EPA
Association of Schools and Programs of Public Health
Washington, DC, 20036
Timothy Dye, Andrew Rutter, Ashley Russell, and Hilary Hafner
Sonoma Technology, Inc.
Petaluma, CA 94954
-------
Air Sensor Guidebook Disclaimer
Disclaimer
The development of this document has been funded in part by the U.S. Environmental
Protection Agency to Sonoma Technology (EP-D-09-097). It has been subjected to the
Agency's peer and administrative review and has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
• • •
Hi
-------
Air Sensor Guidebook Acknowledgements
Acknowledgments
The U.S. EPA would also like to acknowledge the contributions from other Sonoma Technology,
Inc staff in the development of this document. Sonoma Technology was responsible for
developing an initial version of this document ultimately revised by the U.S. EPA to meet its
purpose. Kristen Benedict of the U.S. EPA's Office of Air Quality Planning and Standards is
acknowledged for her assistance in gathering and incorporating review comments from that
organization. In addition, we acknowledge the following people for their invaluable technical
contributions to this document:
• Wayne Cascio (Environmental Protection Agency / Office of Research and
Development)
• Ron Cohen (University of California, Berkeley)
• Mark Fairbank (Paso Robles High School - Science Teacher and Awardee of the
Presidential Award for Excellence in Mathematics and Science, 2009)
• Phil Fine (South Coast Air Quality Management District)
• Michel Gerboles (European Commission Joint Research Center)
• Robert J. Griffin (Rice University)
• Michael Heimbinder (HabitatMap)
• Paul Roberts (Sonoma Technology, Inc.)
• Jamie Schulte (PM-Air.net)
• Gary Secrest (Environmental Protection Agency Office of Enforcement and Compliance
Assurance)
• Jill Teige (University of California, Berkeley)
• Matthew Viens (Environmental Protection Agency, student contractor, ORD Innovation
Team)
• Holly Wilson (Environmental Protection Agency Region 10, Community Air Programs)
• Participants of Air Sensor 2013 Breakout Session C, including EPA staff, state air quality
managers, academics / health researchers, and sensor manufacturers
• Stacey Katz and Gail Robarge (Environmental Protection Agency)
• • •
iv
-------
Air Sensor Guidebook Table of Contents
Table of Contents
Section Page
Executive Summary vii
1. I ntroduction 1
1.1 About This Document/Intended Audience 1
1.2 Air Quality 1
1.3 Air Pollution Monitoring 2
1.4 Uses for Air Sensors 4
2. Air Quality 101 6
2.1 Overview 6
2.2 Pollutant-Specific Effects on Health and the Environment 7
2.3 Important Air Quality Concepts and Characteristics 11
2.4 Atmospheric Pollutants, Their Sources, and Concentration Ranges to Expect 12
2.5 Health Implications of Air Quality Measurements 15
3. Before You Purchase a Sensor 18
3.1 What to Look for in a Sensor 18
3.2 What to Look for in a User Manual 27
4. How to Collect Useful Data Using Air Sensors 28
5. Sensor Performance Guidance 33
5.1 Application Areas 33
5.2 Suggested Performance Goals for Each Application 37
6. Maintaining Your Sensing Device 41
7. Additional Resources 42
Appendix A: Potential Questions from State and Local Officials 44
Appendix B: Air Quality Concepts and Characteristics 46
Appendix C: Technical Considerations 49
C. 1 Considerations for Air Sensor Users and Developers 49
• • •
V
-------
Air Sensor Guidebook Lists of Figures and Tables
List of Figures
Figure Page
Figure 1-1. Example of the interactive My Environment map on EPA's website 4
Figure 2-1. The Air Quality Index (AQI) levels of health concern, numerical values, and
meanings 15
Figure 3-1. Graphics illustrating accuracy, precision, and bias 20
Figure C-1. Comparison of a true value of NO2 and biased measurements of NO2 50
Figure C-2. Time series showing measurements of 1-minute and 15-minute averaged
ozone measurements 52
Figure C-3. Graphical representation of a detection limit 57
Figure C-4. Response time (tso and t90) of an instrument to a calibration gas 58
Figure C-5. Examples of sensor responses as a function of concentration 59
Figure C-6. Illustration of Drift 63
List of Tables
Table Page
Table 1-1. Descriptions of potential uses for low cost air sensors 5
Table 2-1. Health, environmental, and climate effects of common air pollutants 8
Table 2-2. Summary of some common air pollutants 13
Table 3-1. Performance characteristics of a few commercially available portable, low-cost
air pollution sensors 23
Table 3-2. Performance characteristics of commercially available and emerging sensors
for continuous measurements of PM mass and physical properties 24
Table 5-1. Summary of Suggested Performance Goals for Sensors for 5 Types of Citizen
Science Applications in Comparison to Regulatory Monitoring Requirements 39
Table B-1. Air quality topics, discussion, and relevance 46
• • •
vi
-------
Air Sensor Guidebook Executive Summary
Executive Summary
This Air Sensor Guidebook has been developed by the U.S. EPA to assist those interested in
potentially using lower cost air quality sensor technologies for air quality measurements. Its
development was in direct response to a request for such a document following a recent
scientific conference (Apps and Sensors for Air Pollution-2012). Low cost air quality sensors
($100-$2500) are now commercially available in a wide variety of designs and capabilities. This
is an emerging technology area and one that is quickly evolving. Even so, their availability has
resulted in questions from many as to how they might be used appropriately. This document
attempts to provide useful information concerning some of those questions.
The use of sensors to meet a variety of needs ranging from educational programs to
professional research data collections is described. A select market survey is provided here to
inform the reader about the cost range and performance capability of commercially available air
quality sensors. The document provides background information on common air pollutants such
as those defined as "criteria pollutants" as well as select others. Useful information is provided
in the guidebook relative to key considerations about selecting the most appropriate sensor for
one's need concerning these pollutants.
Professional air quality researchers are trained to look for various attributes in monitoring
technologies. While this document is limited in its scope concerning this area, basic information
is provided that should assist citizen scientists and others in making the most appropriate
choices. A major component of this guidebook is a discussion about data quality
considerations. Such topics as the need to calibrate sensors, determining the precision of the
device's response, its response bias, and other performance characteristics are explained in
practical terms. Examples of such performance characteristics determinations are provided to
assist the user in understanding these important concepts.
This guidebook does not attempt to answer every question the U.S. EPA has received about the
selection and use of various sensor technologies. Sensor use must be considered on an
individual basis and only following careful consideration of why the data is being collected and
for what purpose. Extensive resources, nearly all easily obtained free through the internet, are
highlighted in the document to assist potential sensor users in obtaining useful information as
they consider the incorporation of sensor technology to meet a variety of applications.
• • •
vii
-------
Air Sensor Guidebook Introduction
1. Introduction
This Document
This document guides people who develop and use low cost, highly portable sensors
through:
• Background information about air quality
• Uses for air quality data
• Considerations when using or developing air sensors
• Ways to identify the best technology for a specific application
• How to collect useful data
• Performance characteristics to consider
1.1 About This Document/Intended Audience
This document provides useful information for individuals who are interested in air quality
monitoring using commercially available lower cost sensors. In this case, such a sensor is one
typically costing <$2500 and capable of estimating air pollutant concentrations in a continuous
fashion (on the order of seconds to a few minutes in elapsed time). Potential users include
individuals such as sensor developers, citizen scientists, teachers, and students; community
organizations such as neighborhood alliances and environmental justice groups; and federal,
tribal, state, and local air quality agencies.
The document was developed with input from experts across many different disciplines within
the air quality community, including air sensor developers, users, and potential user groups.
The U.S. EPA and its technical experts have provided input on this document and the
technology being discussed.
1.2 Air Quality
Air quality affects our health and our environment. Numerous scientific studies have linked air
pollution to a broad range of health and welfare effects. Potential health effects associated with
air pollution exposures include decreased lung function, aggravation of respiratory and
cardiovascular diseases, and increased asthma incidence and severity among a variety of
others1. The U.S. EPA routinely reports on several key pollutants relative to their sources and
known ecological and human health effects (www.epa.qov/ncea/isa). As defined in these
reports, air pollutants have the potential to impact our lives including damaging vegetation,
causing health issues, decreasing visibility, and affecting global climate conditions.
1 U.S. EPA (2009) Integrated science assessment for particulate matter. EPA/600/R-08/139F; U.S. EPA (2006) Air
quality criteria for ozone and related photochemical oxidants (2006). EPA/600/R-05/004aF-cF; Long-term ozone
exposure and mortality, N Engl J Med 360:1085-1095, March 12, doi: 10.1056/NEJMoa0803894; U.S. EPA (2008)
Integrated science assessment for oxides of nitrogen - health criteria. EPA/600/R-08/071; U.S. EPA (2010)
Integrated science assessment for carbon monoxide. EPA/600/R-09/019F.
• • •
1
-------
Air Sensor Guidebook Introduction
Air pollution consists of a complex mixture of different chemical compounds in the form of solid
particles (in a range of sizes), liquid droplets, and gases. Some of these pollutants are short-
lived in the atmosphere (i.e. hours to days), while others are long-lived (i.e. years). The amount
of time that a particular pollutant remains in the atmosphere depends on its reactivity with other
substances and its tendency to deposit on a surface; these factors are governed by the pollutant
form (i.e., chemical compound) and weather conditions including temperature, sunlight,
precipitation, and wind speed.
Pollutants are emitted by a wide variety of man-made and naturally occurring sources.
Examples of man-made sources include electricity-generating power plants, automobiles, and
oil and gas production facilities. Natural pollutant sources include wildfires, dust storms, and
volcanic eruptions, among others. Some pollutants, called primary pollutants, are emitted
directly from a source (including particulate matter [PM], carbon monoxide [CO], nitrogen
dioxide [NO2], sulfur dioxide [SO2], and lead [Pb]). Others also known as secondary pollutants
are formed by chemical reactions and are often found downwind from the source. This group
includes ozone [O3] and some forms of particulate matter. Airborne pollutant concentrations vary
significantly over space and time because of variations in local emissions, proximity to pollutant
sources, and weather conditions.
1.3 Air Pollution Monitoring
The Environmental Protection Agency (EPA) has identified six "criteria pollutants" as pollutants
of concern because of their impacts on health and the environment2. The criteria pollutants are
ozone3 (O3), particulate matter4 (PM), carbon monoxide5 (CO), nitrogen dioxide6 (NO2), sulfur
dioxide7 (SO2), and lead8 (Pb). Under the Clean Air Act, the EPA has established primary and
secondary National Ambient Air Quality Standards (NAAQS) for these six pollutants. Primary
standards are designed to protect public health, particularly sensitive populations, while
secondary standards are designed to protect the public welfare which includes the environment.
If a geographical area does not meet one or more of the NAAQS, it is designated as a non-
attainment area and must design a plan to meet the standard9. NAAQS concentration limits are
shown in Table 2-2.
The current monitoring network for criteria pollutants is comprised of monitors that meet Federal
Reference Method (FRM) or Federal Equivalent Method (FEM) requirements. Monitors are
operated by state, local and tribal air pollution agencies across the United States to assess
pollutant concentrations in relation to the NAAQS; a variety of instruments and techniques are
needed to measure specific pollutants. Regulatory monitoring generally requires very
sophisticated and well-established instrumentation to meet measurement accuracy
requirements and an extensive set of procedures to ensure that data quality is sufficient. These
requirements (e.g., calibration, maintenance, audits, data validation)10 help ensure the collection
2 http://www.epa.gov/airqualitv/urbanair/
3 http://www.epa.gov/air/ozonepollution/
4 http://www.epa.gov/air/particlepollution/
5 http://www.epa.gov/airgualitv/carbonmonoxide/
6 http://www.epa.gov/air/nitrogenoxides/
7 http://www.epa.gov/air/sulfurdioxide/
8 http://www.epa.gov/air/lead/
9 http://epa.gov/oagps001/greenbk/
1U 40 CFR Part 53 and Part 58. See www.epa.gov/ttnamti1/40cfr53.html
• • •
2
-------
Air Sensor Guidebook Introduction
of high-quality data. Refer to 40 CFR Parts 50, 53, 58, and the QA Handbook Volume II for
activities/criteria for monitoring network data. The overall quality and credibility of
measurements are determined by both the type of instrument and how it's operated.
National Air Toxics Trends Stations (NATTS) are set up across the United States to monitor air
toxics. These stations ensure that quality data is collected in a consistent manner.11
Under the Clean Air Act, EPA also regulates a list of 187 hazardous air pollutants (HAPs),
commonly referred to as "air toxics." Starting in 2003, the EPA worked with state and local
partners to develop the NATTS program to monitor several air toxics. The principal objective of
the NATTS network is to provide long-term monitoring data across representative areas of the
country for priority pollutants, including benzene, formaldehyde, 1,3-butadiene, hexavalent
chromium and polycyclic aromatic hydrocarbons (PAHs) such as naphthalene, in order to
establish overall trends. Additionally, some regulated industrial sources are required to submit
air toxics emissions information to the EPA. The quality and completeness of emissions data
varies significantly by region and source. NATTS-related information can be found at
http://www.epa.gov/ttnamti1/natts.html.
11 http://www.epa.gov/ttnamti1/natts.html
• • •
3
-------
Air Sensor Guidebook
Introduction
To learn more about air pollutants in your neighborhood, you can access EPA's My
Environment page (www.epa.gov/enviro/myenviro). Information accessed on this site represents
reported air pollutant data from industrial and other major sources. Sensor data, like that
described in this report, are not reported on this website. In the box marked "Location," enter
your zip code, and you will be able to view a wealth of environmental data (including information
on air, water, land, energy, and health) specific to your location. The image below provides an
example of the "My Environment" map, and the information it contains. The dark blue squares
are air emission sources and the light blue squares are toxic releases to air.
Contact Us @Shai
Select 3 new Location: How to use this page
"""«"""«•••"=»•» 63101, MO
a AIR
30 ' Air Emissions (ARS/AFS!r96)
HE)
U
[BE] • Cfean Diesel Programs
HD • NAAPM2.5 24-hr (2006 stt)
NAA Ozone 3-hr (20Q3SU)
+: WATER
±1 LAND
3 COMMUNITY
±1 HEALTH
+i ENERGY
+! OTHER
Figure 1-1. Example of the interactive My Environment map on EPA's website.
1.4 Uses for Air Sensors
The new generation of low-cost, highly portable air quality sensors is providing an exciting
opportunity for people to use this technology for a wide range of applications beyond traditional
regulatory or regulatory-equivalent monitoring. Air pollution sensors are still in an early stage of
technology development, and many sensors have not yet been evaluated to determine the
accuracy of their measurements. EPA has specific guidelines it must use in establishing
regulatory-grade air monitors. No lower cost sensors currently meet these strict requirements or
have been formally submitted to EPA for such a determination. Table 1-1 summarizes some
potential non-regulatory application areas for air sensors and provides brief descriptions and
examples. These application areas are described in more detail in Section 5.
• • •
4
-------
Air Sensor Guidebook
Introduction
Table 1-1. Descriptions of potential uses for low cost air sensors.
Application
Research
Description
Scientific studies aimed at
discovering new information
about air pollution.
Example
A network of air sensors is used to
measure particulate matter variation
across a city.
Personal Exposure Monitoring
Monitoring the air quality that a
single individual is exposed to
while doing normal activities.
An individual having a clinical
condition increasing sensitivity to air
pollution wears a sensor to identify
when and where he or she is
exposed to pollutants potentially
impacting their health.
Supplementing Existing
Monitoring Data
Placing sensors within an
existing state/local regulatory
monitoring area to fill in
coverage.
A sensor is placed in an area
between regulatory monitors to
better characterize the
concentration gradient between the
different locations.
Source Identification and
Characterization
Establishing possible emission
sources by monitoring near the
suspected source.
A sensor is placed downwind of an
industrial facility to monitor
variations in air pollutant
concentrations overtime.
Education
Using sensors in educational
settings for science, technology,
engineering, and math lessons.
Sensors are provided to students to
monitor and understand air quality
issues.
Information/Awareness
Using sensors for informal air
quality awareness.
A sensor is used to compare air
quality at people's home or work, in
their car, or at their child's school.
Sensor performance requirements differ according to the application. The quality of a
measurement is dictated by the basic performance of the sensor, the way the sensor is
operated, and the way its measurements are analyzed. Understanding the strengths and
limitations of an air sensor is important if that sensor is to collect information that is useful for a
specific purpose.
• • •
5
-------
Air Sensor Guidebook Air Quality 101
2. Air Quality 101
This Section
Introduces factors affecting air quality (such as type of pollutant, weather, and data
collection location)
Discusses how these factors relate to each other
Explains how these factors may influence the way you use sensors
Summarizes typical air pollutants, sources, health effects, and concentrations
2.1 Overview
"Air quality" is a term used to relate how much pollution is present in the air - good air quality
means there is less air pollution, while poor air quality means there is more pollution. The U.S.
EPA has developed a general guide for citizens called the Air Quality Index
(http://airnow.qov/index.cfm?action=aqibasics.aqi) where pollutant concentrations and health
concerns have been established for a number of common pollutants. We care about air quality
because air pollutants can affect our health and our environment. An increasing number of
studies link air pollution to a range of health problems. In 2010, with respect to the "criteria
pollutants" (see section 1.3), over 120 million Americans lived in counties where concentrations
exceeded the levels of one or more National Ambient Air Quality Standards (NAAQS)12. It must
be pointed out that measuring a concentration value above a respective level of a standard
does not necessarily mean an air pollutant violation has occurred. Multiple factors must be
considered before a true violation of air quality can be established. In addition to causing
adverse health effects, these pollutants can also cause adverse ecological effects such as
impaired visibility or significant damage to plant life. Acidic pollutants deposited on the ground,
predominantly in rain, harm both land and water ecology. Furthermore, many pollutants -
including greenhouse gases (which trap heat in the atmosphere) and particles - also affect the
Earth's energy balance, impacting climate13.
Air pollution is a complex mixture of many different chemical compounds, which are emitted
through human activity as well as natural events like wildfires and volcanos. Pollutants of
concern in ambient (outdoor) air include ozone (O3), sulfur dioxide (SO2), oxides of nitrogen,
carbon monoxide (CO), lead (Pb), ammonia (NH3), volatile organic compounds (VOCs),
mercury (Hg), and other toxic air pollutants. Also of concern are airborne particles, commonly
referred to as particulate matter (PM). While these particles can range in size, they are typically
characterized into one of two groups. A standard for PM10 particles has been established to
provide protection for effects associated with thoracic coarse particles having diameters up to
10 micrometers. Such particles are commonly found near roadways and dusty industries. "Fine
particles" (up to 2.5 micrometers in diameter, also known as PM2.s) are emitted or formed
12 www.epa.qov/airtrends/2011/
13 http://www.epa.gov/climatechanqe/science/causes.html
• • •
6
-------
Air Sensor Guidebook Air Quality 101
through chemical reactions, fuel combustion (e.g., burning coal, wood, diesel), industrial
processes, agriculture (plowing, field burning), and unpaved roads. As noted above in Section
1.3, ambient levels of particulate matter and other certain pollutants in the air (O3, PM, SO2,
NO2, CO, and Pb) are regulated by the EPA through the NAAQS. The term "ambient" relates to
outdoor air used to identify air quality conditions in select locations identified by the EPA as
being representative of a given geographical location. Typically such locations are not in close
proximity to major air pollution sources.
2.2 Pollutant-Specific Effects on Health and the Environment
A broad range of health and environmental effects have been seen following exposures to air
pollutants. Many air pollutants, can remain in the environment for long periods of time and are
carried by the wind hundreds of miles from their origin. The effects resulting from various air
pollutants may be seen/associated after short-term (hours to weeks) or long term exposures
(months to years). Air pollution can also cause environmental harms, including climate change,
acid rain, smog and haze. Table 2-1 summarizes health, environmental, and climate effects of
common air pollutants.
• • •
7
-------
Air Sensor Guidebook
Air Quality 101
14.
Table 2-1. Health, environmental, and climate effects of common air pollutants: quality status and trends through 2008 ).
Pollutant Health Effects
Breathing ozone can trigger a variety of health
problems including chest pain, coughing, throat
irritation, and congestion. It can worsen bronchitis,
Ozone (O3) emphysema, and asthma. Ground level ozone can also
reduce lung function and inflame the linings of the
lungs. Repeated exposure may permanently scar lung
tissue.
Environmental and Climate Effects
Damages vegetation by injuring leaves, reducing photosynthesis,
impairing reproduction and growth, and decreasing crop yields.
Damage to plants may alter ecosystem structure, reduce
biodiversity, and decrease plant uptake of carbon dioxide (CO2).
Ozone is a greenhouse gas that contributes to the warming of the
atmosphere.
Particulate
Matter
(PM includes
PM2.5 and PM10)
Breathing particulate matter can cause premature death
in people with heart or lung disease, nonfatal heart
attacks, irregular heartbeat, aggravated asthma,
decreased lung function, and increased respiratory
symptoms, such as irritation of the airways, coughing or
difficulty breathing.
Long- and short-term exposures to fine particles cause
premature death and adverse cardiovascular effects,
including increased hospitalizations and emergency
department visits for heart attacks and strokes. Fine
particle exposures are also linked to respiratory effects
including increased hospital admissions and emergency
department visits for respiratory effects, such as asthma
attacks, as well as increased respiratory symptoms such
as coughing, wheezing, and shortness of breath as well
as reduced lung development in children.
Short-term exposures to thoracic coarse particles are
linked to premature death and hospital admissions and
emergency department visits for heart and lung disease.
Impairs visibility, affects ecosystem processes, and can deposit
onto surfaces damaging materials. Climate impacts: most particles
are reflective and lead to net cooling, while some (especially black
carbon) absorb energy and lead to warming.
www.epa.qov/airtrends/2010/
• • •
8
-------
Air Sensor Guidebook
Air Quality 101
Pollutant
Health Effects
Lead (Pb)
Damages the developing nervous system, resulting in
IQ loss and negative impacts on children's learning,
memory, and behavior. In adults, causes cardiovascular
and renal effects and early effects related to anemia.
Environmental and Climate Effects
Lead is persistent in the environment and accumulates in soils
and sediments through deposition from air sources, direct
discharge of waste streams to water bodies, mining, and
erosion. Ecosystems near point sources of lead demonstrate a
wide range of adverse effects including losses in biodiversity,
changes in community composition, decreased growth and
reproductive rates in plants and animals, and neurological effects
in vertebrates.
Sulfur Dioxide
(S02)
Aggravates pre-existing respiratory disease in
asthmatics leading to symptoms such as cough,
wheeze, and chest tightness. Asthmatics are most at-
risk, but very high levels can cause respiratory
symptoms in people without lung disease. Exposures
over longer time periods can result in hospital
admissions and ED visits in the general population.
Contributes to the acidification of soil and surface water. Causes
injury to vegetation and losses of local species in aquatic and
terrestrial systems. Increases the bioavailability of mercury in
surface waters which impacts fish and other wildlife. Contributes
to particle formation, which has a net cooling effect on the
atmosphere.
Nitrogen
Dioxide (NO2)
Aggravates respiratory symptoms, increases hospital
admissions, and ED visits, particularly in asthmatics,
children, and older adults; increases susceptibility to
respiratory infection.
Contributes to the acidification and nutrient enrichment
(eutrophication, nitrogen saturation) of soil and surface water.
Leads to oxygen depletion in waters, losses of plants and animals,
and changes in biodiversity losses. Impacts levels of ozone,
particles, and methane with associated environmental and climate
effects.
Carbon
Monoxide
(CO)
Reduces the amount of oxygen reaching the body's
organs and tissues; aggravates heart disease, leading to
hospital admissions and ED visits.
Contributes to the formation of CO2 and ozone, greenhouse
gases that warm the atmosphere.
Volatile Organic
Compounds
(VOCs)
Some are toxic air pollutants that cause cancer and/or
other serious health problems. Contribute to ozone
formation with associated health effects.
Contribute to ozone formation with associated environmental and
climate effects. Also, contribute to the formation of CO2 and
secondary organic aerosols that can warm and cool the
atmosphere, respectively.
• • •
9
-------
Air Sensor Guidebook
Air Quality 101
Pollutant
Mercury
(Hg)
Health Effects
Concerns center on bioaccumulation and methylation in
fish consumed by humans. Methyl mercury poisoning
causes neurological and developmental damage.
Environmental and Climate Effects
Deposits onto soil and into rivers, lakes, and oceans, where it
accumulates in fish, resulting in harmful levels of exposure to
humans and predatory wildlife.
Other Toxic Air
Pollutants (e.g.,
Benzene)
May cause cancer; immune system damage; and
neurological, reproductive, respiratory, developmental,
and/or other health problems. Some contribute to ozone
and particle pollution with associated health effects.
Harmful to wildlife and livestock. Some toxic air pollutants
accumulate in the food chain. Some toxic air pollutants contribute
to ozone and particle pollution with associated environmental and
climate effects.
• • •
10
-------
Air Sensor Guidebook Air Quality 101
2.3 Important Air Quality Concepts and Characteristics
Air quality is a complex, multi-faceted topic with many nuances. Although it can take years to
gain a comprehensive understanding of air pollution, there are some basic concepts and
characteristics that can be learned quickly and provide a good foundation on which to build.
Below is an overview of the most important concepts and characteristics to be aware of when
making air quality measurements. For additional information on these concepts and
characteristics, please consult Appendix B.
Choosing a location: Many pollutants have high spatial variability, that is, their concentration
varies over long or even short distances. This makes sensor location an important consideration
in the design of any monitoring study. Concentrations for most pollutants will almost always be
highest near the source, and will decrease rapidly within the first few hundred feet of the source.
If multiple sources are widely distributed within a given area, pollutant concentrations may be
more similar but will still experience change from location to location. Other factors, including
pollutant type and local atmospheric conditions (discussed below) will also influence the
concentration variability of a given pollutant. Carefully locating your sensor will play a significant
role in determining whether the data you have collected are representative and useful. Where
and how to properly locate sensing devices is discussed further in Section 4.
Factoring in pollutant type: As discussed in Section 1.2, some pollutants may be emitted
directly by a source (primary pollutants), while others may be formed as the products of
chemical reactions in the air (secondary pollutants). Primary pollutants are often more localized
(i.e. near the source) and may have a greater variability over distances than secondary
pollutants. It is important to consider whether a pollutant of interest is primary or secondary
when deciding where and how to collect monitoring data.
Whether a pollutant comes from man-made or natural sources (or both) is also an important
consideration. While measurements typically focus on man-made sources of pollution, all known
sources should be considered when designing a monitoring study. Pollutants coming from
unknown sources can compromise the utility and accuracy of conclusions drawn from data.
Considering wind and atmospheric conditions: Meteorological processes - including
sunlight, temperature, humidity and clouds - can affect pollutant concentrations. For example,
stagnant air can lead to pollutant concentrations that gradually increase, whereas strong winds
can decrease concentrations by spreading pollutants over a larger geographic area.
Understanding how weather conditions can influence pollution concentration and data collection
is important in gathering accurate information and interpreting trends in data.
Factoring in pollutant variation overtime: Pollutant concentrations may vary significantly
depending on the time of day, the day of the week, and the season. These differences can be
attributed to changes in emissions patterns, temperature, the activity schedule of the source
(weekly traffic patterns, for example), and differences in formational processes. Daily, weekly,
and seasonal variations are important considerations when developing a measurement plan,
and will guide the time and conditions under which measurements should be taken.
• • •
11
-------
Air Sensor Guidebook Air Quality 101
Sensor response time: This is a key attribute in determining whether true pollutant fluctuations
can be captured. For capturing a quickly changing or short-lived pollutant plume, detection on
the order of seconds may be important. In other applications, such as monitoring general
outdoor air quality trends, detection at tens of minutes may be sufficient.
2.4 Atmospheric Pollutants, Their Sources, and Concentration
Ranges to Expect
Table 2-2 summarizes select pollutants and information relevant to detecting these pollutants in
air. Please note the following about the information provided in the columns of this table:
Air Pollutant of Interest: Pollutants in the table includes the gases: SO2, NOX, ozone, CO,
CO2, methane, VOCs, and benzene. Solid particle pollutants are: PM2.5, PM10, lead, and black
carbon.
Type: Pollutants may be directly emitted (primary pollutants) or formed in the atmosphere by
chemical reactions (secondary pollutants). CO, emitted directly from combustion processes
(such as car exhaust), is an example of a primary pollutant. Ozone, formed by the reaction of
NOX and VOCs in the presence of sunlight, is a secondary pollutant. Some pollutants, such as
particulate matter (PM), can have both primary (e.g., black carbon - the most strongly light-
absorbing component of PM, formed by incomplete combustion15) and secondary (e.g., sulfate,
nitrate) components.
Useful Detection Limits: A detection limit is the lowest concentration of a pollutant in the
environment that a particular sensor or other instrument can routinely detect. The detection
limits in the table are provided to inform citizen scientists of what sensor detection limits would
be practically useful. Explanation of units and averaging periods:
ug/m3 = microgram per cubic meter
ppm = parts per million
ppb = parts per billion
(1 hr) = one hour averaging time period
(8 hr) = one eight hour averaging time period
(24 hr) = one 24 hr averaging time period
(3 mo) = one three month averaging time period
(1 yr) = one year averaging time period
Range to expect: The table indicates average concentration ranges to expect in ambient air in
the United States. Concentrations near sources (adjacent to and downwind of a major power
plant or roadway, for instance) may at the upper end of the range, or even higher. On the other
hand, pollution in an area that is not close to a specific source is more likely to be at the low end
of the range.
15 http://epa.gov/blackcarbon/basic.html
-------
Air Sensor Guidebook
Air Quality 101
Level: A level is the airborne pollutant concentration that has been identified where concerns
exist if exposure occurs for a defined period of time. These examples are provided for
comparison purposes only. For air toxics, there are no NAAQS, but instead the table provides
examples of exposure levels of concern. A complete list of air toxic values of concern is
available at http://www.epa.gov/ttn/atw/hlthef/hapindex.html.
,16
Table 2-2. Summary of some common air pollutants - See text on page 12 for
explanation of column headings.
Air
Pollutant of
Interest17
Ozone (O3)
Carbon
monoxide
(CO)
Sulfur dioxide
(S02)
Nitrogen
dioxide (NO2)
Carbon
dioxide
(C02)19
Methane
(CH4)20
Volatile
organic
compounds
(VOCs)21
Type
Secondary
Primary
Primary
Primary and
Secondary
Primary
Primary
Primary and
Secondary
Source Example
Formed via UV (sunlight)
and pressure of other
key pollutants
Fuel combustion -
mobile sources,
industrial processes
Fuel combustion -
electric utilities, industrial
processes
Fuel combustion -
mobile sources, electric
utilities, off-road
equipment
Fuel combustion -
electric utilities, mobile
sources
Industry (e.g., natural
gas operations),
agriculture, and waste
management
Fuel combustion (mobile
sources, industries)
gasoline evaporation;
solvents
Useful
Detection
Limits
10 ppb
0.1 ppm
10 ppb
10 ppb
100 ppm
500 ppb
1 ug/m3
Range to
Expect
0-150 ppb
0-0.3 ppm
0-100 ppb
0-50 ppb
ppm
1500-2000
ppb
5-100 ug/m3
(total VOCs)
Level18
75 ppb (8 hr)
9 ppm (8 hr)
35 ppm (1 hr)
75 ppb (1 hr)
0.5 ppm (3 hr)
100 ppb (1 hr)
53 ppb (1 yr)
None
None
None
16 Table adapted from EPA's Draft Roadmap for Next Generation Air Monitoring
http://www.epa.qov/research/airscience/docs/roadmap-20130308.pdf.
Information on pollutants is available at http://www.epa.gov/airqualitv/ and http://www.epa.gov/airtrends/
18 See http://epa.gov/air/criteria.html for additional information on a select number of the pollutants listed here. If must
be recognized that multiple factors must be considered in establishing a pollutant concentration of concern in addition
to just the averaging time. Various statistical data treatments are often required and the information in this column is
not fully descriptive of these issues.
19
21
http://www.epa.gov/climatechange/ghgemissions/gases/co2.html.
http://epa.gov/climatechange/ghgemissions/gases/ch4.html.
http://www.epa.gov/ttn/amtic/pamsmain.html.
• • •
13
-------
Air Sensor Guidebook
Air Quality 101
Air
Pollutant of
Interest17
Benzene (an
example of a
VOC and air
toxic)22
particulate
matter (PM2 5)
Particulate
matter (PM10)
Lead (Pb)
Black carbon
(BC)23
Type
Primary
Primary and
Secondary
Primary and
Secondary
Primary
Primary
Source Example
Gasoline, evaporative
losses from above
ground storage tanks
Fuel combustion (mobile
sources, electric utilities,
industrial processes),
dust, agriculture, fires
Dust, fuel combustion
(mobile sources,
industrial processes),
agriculture, fires
Smelting, aviation
gasoline, waste
incinerators, electric
utilities, and lead-acid
battery manufacturers
Biomass burning, diesel
engines
Useful
Detection
Limits
0.01 -10
ug/m3
5 ug/m3
(24-hr)
1 0 ug/m3
(24-hr)
0.05 ug/m3
(24-hr)
0.05 ug/m3
Range to
Expect
0-3 ug/m3
0-40 ug/m3
(24-hr)
0-100 ug/m3
(24-hr)
0-0.1 ug/m3
(24-hr)
0-1 5 ug/m3
Level18
None
35 ug/m3 (24 hr)
12ug/m3(1yr)
1 50 ug/m3
(24 hr)
0.15 ug/m3
(3 mo)
None
It should be recognized that there are certain circumstances in which concentrations above
those discussed in Table 2.2 are permissible, especially under occupational settings. The
National institute of Occupational Safety and Health (NIOSH) has established guidelines for
such circumstances that sensor users also need to consider in determining data collection
plans. For example, the NAAQS indicates that CO levels at the 35 ppm level for a 1 hour period
warrants potential health concerns (ambient air), the NIOSH has established guidelines
indicating that occupational exposures of 35 ppm (8 hour exposure time period) with a 200 ppm
maximum (at any time) must be considered. Therefore, sensor users should be aware that just
because a measurement exceeds the EPA ambient standard that a violation of air quality might
not exist. The circumstances must be considered. In other words, there are legitimate
circumstances where airborne concentrations higher than the NAAQS are permitted. Sensor
users are suggested to view information on the NIOSH web link provided below that provides
specific guidelines on occupational settings and allowable concentrations for a wide variety of
air pollutants.
http://www.cdc.gov/niosh/npg/npgd0105.html
23
http://www.epa.gov/ttnatw01/hlthef/benzene.html.
http://www.epa.gov/blackcarbon/basic.html.
• • •
14
-------
Air Sensor Guidebook
Air Quality 101
2.5 Health Implications of Air Quality Measurements
As described in Table 2-1, air pollution can have a number of serious health impacts. For all of
the criteria pollutants except lead, EPA has established the Air Quality Index (AQI) as a means
to translate pollution measurements into the potential for effects in individuals. The AQI is an
index for reporting daily air quality. It tells you how clean or polluted your air is, and what
associated health effects might be a concern for you. The AQI focuses on health effects you
may experience within a few hours or days after breathing polluted air. EPA calculates the AQI
for five major air pollutants regulated by the Clean Air Act: ground-level ozone, particle pollution
(also known as particulate matter), carbon monoxide, sulfur dioxide, and nitrogen dioxide. For
each of these pollutants, EPA has established national air quality standards to protect public
health. Ground-level ozone and airborne particles are the two pollutants that pose the greatest
threat to human health in this country. EPA uses information based on ambient ozone
concentrations in the determination of the daily AQI. On-line formulas incorporate information
such as observed 24-hr average concentrations and established minimum and maximum
pollutant ranges. Easy to use on-line calculators are available that allow individuals to either
calculate a local AQI or back calculate the reported AQI to estimated pollution levels.24
Air Quality
Index Levels
of Health
Concern
Moderate
Numerical
Value
OtoSO
51 to 100
Meaning
\ir quality is considered satisfactory, and air pollutio
poses little or no risk
Air quality is acceptable; however, for some pollutants there
may be a moderate health concern for a very small number
of people who are unusually sensitive to air pollution.
Unhealthy for
Sensitive
Groups
101 to 150
Members of sensitive groups may experience health effects.
The general public is not likely to be affected.
Everyone may begin to experience health effects; members
Unhealthy 151 to 200 of sensitive groups may experience more serious health
effects.
.. ... ... __„ . ,__ Health warnings of emergency conditions. The entire
Very Unhealthy 201 to 300 popu|atjon js more |jke|y9to J affected
Hazardous
301 to 500
Health alert: everyone may experience more serious health
effects
Figure 2-1. The Air Quality Index (AQI) levels of health concern, numerical values, and meanings.
24
http://airnow.gov/index. cfm?action=aqibasics.aqi
• • •
15
-------
Air Sensor Guidebook Air Quality 101
However, it's important to understand that the AQI is calculated based on air pollution data that
is averaged over 1, 8, or 24 hours, depending on the pollutant (see Table 2-2). The reason for
the different averaging times is that different pollutants affect the human body in different ways.
For example, SO2 can cause difficulty breathing and respiratory symptoms such as cough,
wheeze, and chest tightness within 5 minutes of exposure. This is because SO2 affects the parts
of your lungs that communicate with your central nervous system, triggering a reflex response
that can quickly cause a narrowing of the airways (called bronchoconstriction). Importantly, the
SO2 NAAQS has already taken into account the speed at which SO2 can cause these
respiratory effects. That is, the level of the 1-hourSO2 NAAQS was set low enough to provide
substantial protection against the much higher 5-minute concentrations that can cause these
effects. Because the AQI for SO2 is based on the 1-hour NAAQS, it can effectively
communicate when the air is healthy or unhealthy to breath. Those most susceptible to the
effects of SO2 are asthmatics who exercise, work, or play outdoors.
While SO2 can affect the respiratory system within minutes, the respiratory effects of O3 can
happen in an hour or so, or may not occur until the next day. Again, this is because of the
specific way that O3 affects the body. Ozone can reduce lung function, but based on a person's
antioxidant status and the O3 dose inhaled, it may take an hour or more for symptoms such as
bronchoconstriction to occur. Ozone can also inflame and damage the lining of the lung, but
this effect may not be most obvious until the day after exposure. Some scientists have
compared ozone's inflammatory effect on the lining of the lung to the inflammatory effect of
sunburn on the skin. Ozone damages the cells that line the air spaces in the lung. Within a few
days, the damaged cells are replaced and the old cells are shed- much in the way that skin
peels after a sunburn. Also like a sunburn, the effects can be worse later the same day or the
next day, and it can take your lungs a while to recover. And, if this kind of damage occurs
repeatedly, the lung may change permanently in a way that could cause long-term health effects
and a lower quality of life. Similar to the SO2 NAAQS, the 8-hour O3 NAAQS provides
substantial protection against these respiratory effects by taking into account the way O3
interacts with the body.
With respect to my health, what do my sensor readings mean?
Some citizen scientists may become concerned if they measure levels of a pollutant higher than
the health benchmarks provided in Table 2-2. However, it is very important for the user to
consider the time period over which the pollutant level was measured. For example, the daily
(24 hour) PM2.5 standard is 35 ug/m3. Because the standard is based on the average of hourly
monitoring measurements over a 24-hour period, it does not mean that a single PM2.5
measurement taken over a few minutes, or even hours, above 35 ug/m3 is cause for immediate
concern. By using the AQI calculator on the EPA website, you can learn that a 24-hour average
measurement of PM2.5of 35 ug/m3 is "yellow," or moderate air quality, and a 24-hour average
measurement of 50 ug/m3 is "orange," or unhealthy for sensitive groups. Again, for the AQI, it is
very important to remember that the concentration that you enter into the calculator is meant to
be an average value over a longer time period, (in this example, over 24 hours) not just a single
reading taken over the span of a few minutes or hours.
The increasing use of sensors is expected to provide more data on air pollution than has
previously been available, and in shorter time increments. For example, it will be much easier
16
-------
Air Sensor Guidebook Air Quality 101
to track minute-by-minute changes in pollution levels. As a result, we will become more aware
of short-term, peak levels of some pollutants. However, the actual health effects of very short
term elevated levels of most pollutants are not well understood and EPA has not established
health information defining such short-term pollutant exposures.
This document does not provide detailed guidance on health-based interpretation of sensor
measurements. Much research has to be performed before it is understood how health
messaging for short periods of data collections should be communicated.
• • •
17
-------
Air Sensor Guidebook Before You Purchase a Sensor
This Section
• Discusses how to identify the sensor technology best suited for your intended
application
• Outlines the key questions to consider when purchasing a sensing device
3. Before You Purchase a Sensor
It is important to "ask questions" before you begin collecting data. This will help you have a clear
concept of what it is you are hoping to accomplish via the collection of air pollution data prior to
beginning a monitoring project. It is also an important step to take when purchasing an air
sensing device. Defining the questions you hope to answer will help identify the pollutant of
interest, the field conditions you are likely to encounter, the duration of data collection, and the
type of measurements needed (i.e. short-term, mobile measurements vs. long-term, stationary
measurements) and the quality of these measurements. All of these data collection
characteristics will determine the sensing equipment that is best suited for your data collection
purposes.
Here are a few examples of the types of well-defined questions that users and developers
should ask to help in the identification of an appropriate sensing device. These examples will be
used to illustrate the concepts and choices that are important in using sensors.
a. How can I teach my students about air quality and integrate hands-on data collection
into the lesson plan?
b. What is my exposure to air pollution during my usual walking route?
c. Is the nearby oil and/or gas facility creating an air pollution concern in my
neighborhood?
3.1 What to Look for in a Sensor
There are several things to think about before purchasing a sensor. First and foremost is to
determine a target pollutant. This will guide each of your subsequent decisions, and ultimately
lead you to the most appropriate types of sensors for your application. Also important to
consider are device specifications like detection range and detection limit, precision and bias,
calibration procedures, and others, each of which is discussed below, and in greater detail in
Appendix C.
Selecting a target pollutant: What pollutants do you need your sensor to measure? The
answer to this will depend on the question you have decided to ask. Be sure to consult Table 2-
2, which identifies common sources of various pollutants, for guidance. It is important to
remember that your decision will be further influenced by what sensors are available within your
price range.
• • •
18
-------
Air Sensor Guidebook Before You Purchase a Sensor
We have used questions a, b, and c from above to provide three examples of how to choose a
target pollutant. While this list does not address all potential applications/pollutants of interest, it
does illustrate how the design of a project can influence choices about key pollutants, and the
sensors to measure them.
a. Education project: Projects like this are primarily concerned with building a basic
understanding of the pollutants in question. Accordingly, you may choose to use a low-
cost sensor that can detect a common pollutant. Ozone is one example of a common
outdoor pollutant that can be detected by inexpensive sensors. If ozone is an air
pollutant of concern in your area, this might be a good choice. A good way to tell
whether ozone is common in your area is to visit the links provided at
http://www.epa.gov/airquality/greenbook/hindex.html. If you are in or near a
"nonattainment area" then you can expect ozone to be present in higher levels (note:
these levels will vary seasonally - visit http://www.epa.gov/air/ozonepollution/ for more
information on ground-level ozone and ozone pollution).
b. Daily walk: If you walk near roads or highways, you may choose to measure NO2, or
particulate matter (PM), both of which often indicate air pollution emissions from traffic.
For more information on NO2 and PM, visit http://www.epa.gov/air/nitrogenoxides/ and
http://www.epa.gov/airguality/particlepollution/.
c. Nearby oil and/or gas facility: Emissions from oil and gas facilities typically include
volatile organic compounds (VOCs), particularly benzene, so this may be a good choice
for measurement. It is important to note that sensors selective enough to detect for
individual VOCs (i.e. benzene) are often quite expensive, so choosing one that
measures VOCs generally can be a low-cost alternative. Such a sensor will respond to
a wide variety of VOCs as a general indicator of their presence.
Consider detection range and detection limit: Environmental pollutants can often be present
in very low concentrations, particularly when measurements are being made far from the source
of the pollution. A sensor will be most useful when it is able to measure a target pollutant over
the full range of concentrations commonly found in the atmosphere (consult Table 2-2 under
"Range to Expect" for each pollutant). Depending on how close you are to a pollution source,
the ability of the sensor to be accurate at either very low or very high concentrations must be
understood before you collect any measurements.
To ensure that concentrations at the low end (see ranges in Table 2-2) do not go undetected,
you will need to determine the detection limit of the sensor you are looking to purchase. The
detection limit is the lowest concentration of a pollutant in the environment that a sensor can
detect, and may or may not be provided by the manufacturer.
For example, carbon monoxide (CO) in outdoor air often occurs at background levels in the
range of 40-200 parts per billion (as shown in Table 2-2). The carbon monoxide detector sold for
use in personal residences can detect levels above -75 parts per million, which is appropriate
for its purpose, but not sensitive enough for measuring outdoor CO for most environmental
studies. For many environmental uses, very sensitive detectors are needed that measure CO
concentrations well under 5 parts per million (ppm). Depending on the pollutant, ambient levels
may be somewhat higher in urban areas, so sensors with detection limits above the background
• • •
19
-------
Air Sensor Guidebook Before You Purchase a Sensor
concentrations shown in Table 2-2 may still be useful in some settings. Similar pollutant ranges
and needed sensitivity issues exist with all monitoring devices. You will need to either establish
these characteristics with the manufacturer or evaluate them yourself as part of your using
them.
Consider precision and bias: Precision and bias are terms that refer to the accuracy of a
sensor measurement (see Figure 3-1). Accuracy is the overall agreement of a sensor's
measurement to the true value. Precision refers to how well the sensor reproduces the
measurement of a pollutant under identical circumstances. Bias refers to measurement error; for
example, a sensor may always measure a little higher or lower than the true concentration.
Before purchasing a sensor, you should consult the manufacturer's specifications about
reported precision and bias for the sensor. In addition, the user should then conduct their own
precision and bias measurements as defined in Appendix C to further qualify the value of the
data they are collecting. Table 5-1 identifies acceptable ranges of precision and bias for various
sensor applications.
Accuracy= how close Precision= being able Bias- a systematic
to "true" concentration to consistently predict (common) error of
the same reporting a value
concentration higher or lower than
the true value
Figure 3-1. Graphics illustrating accuracy, precision, and bias
You should also be aware that accuracy, precision, and bias of a sensor can change overtime.
For example, exposure to warm temperatures or humid air may lead to a gradual increase in
bias (also known as drift). The sensor may experience interference from other chemicals in the
atmosphere that could lead to erroneous concentration estimates. Some sensors may come
with an "expiration date" after which its measurements are no longer likely to be accurate. For
more detailed information on precision, bias, and related concepts, please consult Appendix C.
• • •
20
-------
Air Sensor Guidebook Before You Purchase a Sensor
Identify calibration requirements: Calibration is the process of checking and adjusting an
instrument's measurements to ensure that it is reporting accurate data25. Calibration compares
the response of the instrument to a known reference value. Calibration is important because
sensor performance can change over time. If at all possible, sensors should be calibrated for
their response before, during, and after a set of data collections. Guidance on this issue is
provided in Appendix C. Before you purchase a sensor, find out if it has been calibrated by the
manufacturer. Also be sure to check the user manual for information on specific instructions
concerning how to calibrate your sensor, including how long the calibration will last once the
sensor is being used.
Understand Response Time. A sensor may be quick or slow to measure a pollutant in the air.
A sensor that responds quickly may be useful for mobile monitoring and for observing very rapid
changes in pollutant concentrations. A sensor that responds slowly may be more suited to
stationary monitoring of pollutants that vary in concentration gradually. Your specific data
collection goals and intentions will determine which type of sensor is best. It is desirable for a
sensor to respond in less than 1 minute if it is to be used in any mode other than stationary
monitoring.
Verify durability and quality of construction: Durability is referring to a sensor's ability to
endure wear and tear and continue to perform. Sensors are likely to experience such effects
during normal use. For example, sensors that are carried by the user or are used for mobile
monitoring on vehicles might be jostled, shaken, hit against other objects, or dropped. All
sensors measuring outdoor air quality are likely to be exposed to variable weather conditions
such as heat, moisture, and dust. The sensor manufacturer should be able to describe the
general durability of the device. Even so, they are often not able to describe specifics about
how many times you can drop the device or other events before it will fail.
Packaging: Packaging refers to the material used to contain the sensor system components.
Packaging can be used to provide protection from water, light, temperature variations (by
adding heaters or cooling fans), and electromagnetic noise. However, the air sampled by
sensors often comes into contact with sensor packaging. Because of this contact, packaging
may interfere with or actually contribute to pollutant concentration levels. Sensors with strong,
waterproof, non-reactive packaging will be more durable over time. Reactive materials might
include certain types of plastics and coatings that might react with the pollutant of interest or
even release the pollutant, interfering with collecting data accurately.
Consider sensor usability: Usability refers to the ease of use of a sensor - is it straight-
forward to operate? Air sensors are used by a wide variety of people, ranging from those with
no formal training in air quality science to researchers who have many years of advanced
training and expertise. Intuitive and easy-to-use sensors will be more attractive for projects that
rely on community involvement and citizen scientists, while sensors that provide more detailed
information may be preferable for more advanced users.
Other important considerations include how the sensor is powered; if the sensor relies on a
battery, how long is the battery life? Is it a rechargeable or replaceable battery pack? Also
consider how the data is stored, processed and transmitted - is the data wirelessly transmitted?
25 http://nepis.epa.qov/Exe/ZyPURL.cqi?Dockev=20001 QWV.txt
-------
Air Sensor Guidebook Before You Purchase a Sensor
Or is manual processing required? Experience tells us that establishing wireless communication
between a sensor and a web service is not always easy and may even depend on the type of
cellular provider for a given geographical area.
Cost: The cost of sensor technology may vary greatly depending on the pollutant to be
measured and the degree of accuracy and sensitivity one needs26. In general terms, costs often
range from $100 to $2500 for what might be considered "consumer-based" air quality sensors.
Even within a given pollutant, the cost range might be very large depending on the features of
the device.
Potential red flags: Just as the characteristics above can help find a sensor that is appropriate
for your specific applications, their absence can be cause for caution. Next generation air
monitoring technology is part of an emerging market, and as such there is likely to be a wide
range in the quality and reliability of available devices. Some sensor devices have been tested
for measurement performance, durability, and usability, but many others have not. While the
EPA is beginning to test some currently available sensor technologies27, as of the writing of this
manual there is no formal process for verification. We suggest that you use the information
provided here to carefully review a sensor, including its user manual, before purchasing. When
investigating a sensor for purchase, it is also important to consider demonstrated performance,
measurement repeatability, and feedback from past users.
Table 3-1 provides examples of a few commercially available portable, low-cost air pollution
sensors.
Table 3-2 provides examples of performance characteristics of commercially available and
emerging sensors for continuous measurements of PM mass and physical properties. There are
many lower cost sensors now available. The examples provided here are reported solely to
share the types of sensors being developed and some of their stated capabilities. A recent
report has attempted to define the current market status of a wide variety of high performance to
citizen science type air quality sensors27. Sensor users need to carefully consider all available
information in selecting the right sensor for any specific purpose.
26 Mobile Sensors and Application for Air Pollutants (NTIS PB 2014 105955), 2014. EPA/600/R-1 4/051.
27 More information on this initiative can be found at http://www.epa.qov/airscience/air-sensor.htm
22
-------
Air Sensor Guidebook
Before You Purchase a Sensor
Table 3-1. Performance characteristics of a few commercially available portable, low-cost air pollution sensors.
Analyzer
CO (8 h: 9 ppm;
Langan
DataBear, I
T15d
Aeroqual
Series 500
N02 (1 h: 100
Aeroqual
Series 500
Sensor „ . _, . . Env ronmenta We ght Res.P°nse
T . . Range Accuracy Precision . . .. .. M. Time
Technology M y Limits (kg) . .
1 h: 35 ppm)
Electrochemical 2_2QQ ppm Q5 ppm Q5 ppm 23 {Q 4Q OG Q 43 > 1
Metal Oxide <±2 from 0-20 ppm; 040°C5t 95°/
Semiconductor 2-1 00 ppm <±10%from20- 0.1 ppm " ',, ° < 0.46 < 150
(MOS) 100 ppm
ppb; annual average: 53 ppb)
<±0.01 from 0-0.1
ppnr
MOS ' " <±10% from 0.1- 1 ppb 7r)0/ ' < 0.46 <180
0.2 ppm
Price USD
1.5K
1.5K
2K
03 (8 h: 75 ppb)
2B
Technologies,
202; FEM
EQOA-0410-
190
Aeroqual
Series 500L
UV absorption 10cnPPbt° 1.5 ppb or 2% 0.1 ppb 0 to 50 °C 0.70 10
^ou ppm
MOS 8-500 ppb 8 ppb 1 ppb "5 *° fP .?' 5 to < 0.46 < 60
yo /o r\n
5K
2K
23
-------
Air Sensor Guidebook
Before You Purchase a Sensor
Analyzer
OMC-1108
Sensor
Technology
Electrochemical
Cell
Range Accuracy Precision
°-01t°10 +10% 10ppb
ppm
Environmental Weight
Limits (kg)
' 0 5
Response
Time
(s)
<70
Price USD
1.2K
Adapted from Snyder et al .
Table 3-2. Performance characteristics of commercially available and emerging sensors for continuous measurements of PM mass and physical
properties.
Reference
Sampler/
Sensor
831 Aerosol
Mass Monitor
Personal
DataRAM,
Model pDR-
1500
Measurement .. , .
_, . . . Manufacturer
Principle
Light
scattering; MetOne
Mass Instruments
concentration
Light
Scattering; Thermo
Mass Scientific
concentration
Accuracy Precision
±10% to
calibration -b
aerosol
±5% of ±0.2% of
reading ± reading or ±0.5
precision ug/m3 60-s avg
Limit of
Detection More |nformatjon**
(ug/m j or weight (kg) and -Cost ($, when
ower a ice available) as of May 2014
(Mm)
_ ,. Range: 0-1 ,000 ug/m3; 0.8 kg;
u.s urn <$2,000
Size Range: 0.1-10 urn; Cone
1 to 4x1 0 ug/m3
0.1 urn Precision (2a); 10-s avg
1.2 kg; $5500 with PM2.5 and
cyclones
Range:
'PM10
28
The Changing Paradigm of Air Pollution Monitoring. Emily G. Snyder, Timothy H. Watkins, Paul A. Solomon, Eben D. Thoma, Ronald W. Williams, Gayle S. W.
Hagler, David Shelow, David A. Hindin, Vasu J. Kilaru, and Peter W. Preuss. Environmental Science & Technology 2013 47 (20), 11369-11377
• • •
24
-------
Air Sensor Guidebook
Before You Purchase a Sensor
Reference
Sampler/
Sensor
DCHOOAir
Quality Monitor
Meparindpn|1eent Manufacturer Accuracy
Light
scattering; _ . _ b
Laser particle Dylos Corp'
counter
Limit of
Detection
Precision (M9/m3) or
precision Lower Partjc|e
Size Detected
(Mm)
colfocaled- °'5 ^m
More Information**
Weight (kg) and ~Cost ($, when
available) as of May 2014
Size ranges: Pro: >0.5 urn, >2.5 urn or
Household: >1 urn, >5 urn, difference
between size ranges equals reported
counts; Linear up to ~106 pt/mL with
<10% coincidence**; -0.4 kg;< $300
microAeth®
Model AE51
Light
absorption,
880 nm
AethLabs;
Black Carbon
no standard
for comparison
±0.1 ug BC/m3
60-s avg**
<0.16 ug/m3,
2.5 mL/s,
60-s avg
Precision at 2.5 mL/s flow rate; Range:
1-1 000 ug BC/m3
Resolution 1 ng BC/m3; 0.3 kg; $6,000
Conversion from light scattering, particle number or size distribution, requires estimates of particle density and shape factors; No data.
Performance capabilities are from manufacturers' datasheets except where noted with a **. Text in bold type represents a typical fixed-site higher-
cost monitor for comparison purposes only to the sensors that follow in that category. Adapted from Snyder et al29.
The list provided in Tables 3-1 and 3-2 are not intended as recommended sensors. These represent some whose performance has
been better established at the time this report was developed. The list of commercially-available sensors is expanding. The U.S.
EPA has plans of sharing reports in the future (2014-2015) where sensors that have undergone specific laboratory and/or real-world
evaluations shall be described. Even so, sensor users should perform their own market surveys to determine which sensors might
best fit their budget and data collection needs.
The Changing Paradigm of Air Pollution Monitoring. Emily G. Snyder, Timothy H. Watkins, Paul A. Solomon, Eben D. Thoma, Ronald W. Williams, Gayle S. W.
Hagler, David Shelow, David A. Hindin, Vasu J. Kilaru, and Peter W. Preuss. Environmental Science & Technology 2013 47 (20), 11369-11377
• • •
25
-------
Air Sensor Guidebook Before You Purchase a Sensor
Designing a Real-World Monitoring Initiative: The Village Green Project
EPA's Village Green Project was initiated in an effort to develop a self-powered, low-maintenance
monitoring system to measure air quality in Durham, NC. The project, intended as a community
research and educational effort, incorporates a series of sensors into a public park bench
structure. While these sensors collect and stream local air quality and weather data, people can
interact with the system to learn more about air pollutants and pollution trends. The prototype
system, installed outside of the Durham County Library South Regional Branch, has collected
5,664 hours of data since June 20, 2013, all of which is now available on the project website
(www.villaqeqreen.epa.qov).
Before the Village Green Project could launch and data could be made available, the project
team needed to identify the most appropriate sensor technology for this particular application.
Team members worked together to identify the research question, and then used this question to
guide the identification of constraints and the selection of instrumentation. This real-world
example helps illustrate the design and selection process that sensor users should go through
before purchasing a device and collecting data.
Research Question: How can we measure outdoor air quality in a community setting with a
minimal footprint, high quality data, and in support of community engagement?
Instrumentation Constraints:
S Efficient energy usage to support a self-powered system
S Long-term accuracy and precision of measurements without significant infrastructure to
maintain (avoid costly auto-calibration systems and power-hungry heating or cooling)
S Real-time active monitoring technology with capability to directly output measurements to
a database (enable public access to and interaction with real-time data)
S Detection sensitivity at background to urban levels (capture a range that accurately
reflected local air quality trends)
S Data provided by the instrument to monitor performance
S Past record of proven performance in similar measurement environments (ensure ability
of sensor to withstand expected monitoring conditions)
Based on the above criteria, only a few instruments were available at the time of selection (Fall
2011). The research team decided to utilize a light-scattering method to measure particulate
matter (PM), which showed promise in correlations with higher-cost, mass-based PM
measurements. This instrument met the criteria for being low power and low maintenance, and
was sensitive down to very low concentrations. The other instrument applied to the monitoring
set-up was an ozone monitor that measures using ultraviolet light-absorbance. Similarly to the
PM instrument, the ozone monitor is low power, low maintenance, has long-term stability, and is
sensitive at ambient concentrations. The constraints imposed by the research question led the
project team to implement higher cost instruments (~$5,000 USD) for this particular application
than would be used by most people, but which were ideal for the intended project.
• • •
26
-------
Air Sensor Guidebook Before You Purchase a Sensor
3.2 What to Look for in a User Manual
It is also important that the sensor technology being purchased has a complete and informative
user guidebook/manual. This manual will serve as your roadmap of operation, outlining the key
operation requirements and characteristics of your monitoring device. Look for effective manuals
to include:
• General operation (i.e. how to turn on and off, how to charge or change batteries);
• How to store and recover data;
• Conditions of operation;
• Sensor expiration date (if there is one);
• Directions for calibration (if the sensor has that capability);
• Expected performance (precision and bias);
• Maintenance requirements;
• Response time (how quickly does the sensor respond to changing conditions);
• Target pollutants;
• Support information (i.e. company representative, customer support number);
• Technical specifications (i.e. type of sensor used, data storage capabilities);
• Known interferences;
• Demonstrations of sensor performance in real-world applications (ideally in the form of
scientific articles reporting on sensor tests)
Sensors that come without a user manual, or with a user manual that is incomplete, may be more
difficult to use and maintain. It is strongly advised that you read through or inquire about a user
manual before purchasing a particular technology.
• • •
27
-------
Air Sensor Guidebook How to Collect Useful Data Using Air Sensors
This Section
• Describes how to plan your data collection, review, and analysis
• Lists basic steps for collecting useful data
4. How to Collect Useful Data Using Air Sensors
The five basic steps to collecting useful data with air sensors are (1) ask a question, (2) develop
an approach, (3) determine sensor location, (4) collect measurements, and (5) understand and
communicate results. This section, while not meant to be a comprehensive guide, briefly
describes these five steps, which apply to all sensor applications.
1. Ask a Question. It is important to take the time to clearly establish and document what
question you would like to answer before you begin developing your plan to collect
measurements. A simple question, such as "Are ozone concentrations higher in the
afternoon than in the morning in my neighborhood?" can help you get started.
2. Develop an Approach. Once you have established the question you would like to
answer, it is time to plan how the measurements will be made. Think about these issues:
- The dimension of the problem. Section 2.3 and Appendix B provide some overview
of how pollutants may change over time and/or location.
- The who, what, where, when, and how. Who will take the measurements? What
measurements are needed? Where and when should measurements be taken? For
how long should measurements be conducted, and how should samples be taken?
- The number and quality of sensors needed. For example, will your question be better
answered by one powerful sensor (i.e., very accurate) or multiple inexpensive
sensors collecting less accurate data.
- The resources (e.g., funding, knowledge) and labor required.
- How the data will be collected and stored.
- How you will ensure your data are of good quality, and the degree of quality you
need. Section 5 and Appendix C provide further guidance on this topic.
- What additional data (e.g., meteorology, other pollutants, site information) are
needed to answer your question? See Appendix B for more information.
- How you might address the potential questions that other parties may ask, a
sampling of which is provided in Appendix A. You should be able to answer the
questions in this list if you wish to communicate the value of the data you have
collected to others.
3. Determine Sensor Location. Careful planning of your measurement approach is an
incredibly valuable process that can help minimize complications later. Developing a
specific monitoring plan will also allow you to share your ideas and project design with
others before you have invested time and/or money. Whenever possible, share your
plan with those you hope to communicate with about your future findings, and/or with
experts in the field who are willing to give you constructive feedback. Such information
• • •
28
-------
Air Sensor Guidebook
How to Collect Useful Data Using Air Sensors
sources might include local and state air quality officials, university researchers, and
trained community environmentalists. This may help you identify potential problems at
an early stage. Such an effort is worthwhile, because it is likely that you will need some
or all of this information to answer questions about your measurements when presenting
your results. To ensure these results are as accurate as possible, a sensor or instrument
should be placed in a location where it can measure the atmosphere or source of
interest with minimal interference. A well-placed site would yield data that are
representative of the area being monitored.
Why it matters: Air pollution concentrations can be considerably affected by local sources,
buildings and other structures, among other factors. You will want to consider potential
effects when you choose a monitoring location. If emissions from sources close to the
site are not of interest, then the site may not be suitable for monitoring. The data will be
most useful if you can measure the pollutant of interest with little impact from other
sources at your site. Here are some important considerations:
Allow free air flow to the sensor
by making sure it is far enough
away from the ground (1-2
meters above the surface) and
away from building surfaces, if
possible (ideally at least 1 meter
away).
Avoid local pollutant emission
sources if you are trying to
measure more general
community levels of pollutants.
Avoid sources of gases that can
react with your pollutant of
interest (e.g. ozone is depleted
very quickly by certain organic
compounds, as well by nitric
oxide from tail pipes) (see Table
2.1).
The inlets to personal exposure
monitors must have access to
the air the person is inhaling. For
example, a personal exposure
sensor will not make
representative measurements if
kept in a purse or pocket. Inlets
for personal samplers can be
close to your body or clothing as
long as they are sampling air outside of your clothes. PM is a special case, since
clothing is a source of PM.
Taking some preliminary measurements can help identify a good monitoring location.
Examples of Sensor Placement
Considerations
Location of sources relative to the pollutant
of interest. Some pollutants may have higher
concentrations closer to a particular source, while
others may not. Let's say you want to measure
how a freeway influences ozone and particle
(PM) concentrations. Ozone concentrations may
be depleted next to a roadway because it reacts
with freshly emitted nitric oxide (NO). PM is likely
to be highest directly adjacent to the roadway,
and then decrease with distance. Therefore, you
could choose to monitor for both pollutants close
to the road, and also some distance downwind of
the roadway, to determine the change in
concentrations that occurs near the roadway.
Location of the sensor relative to the exposed
population. If your study objective is to measure
the impact of an industrial source on the pollutant
concentrations in a community, the sensor might
be placed in a neighborhood. In contrast, if you
want to measure the emissions each day from
the same industrial facility, the sensor might be
used very close to the source and moved around
to map the pollutant plume.
• • •
29
-------
Air Sensor Guidebook How to Collect Useful Data Using Air Sensors
4. Collect Measurements. With your measurement approach clearly defined and your
sensor properly located, it is time to collect your data. This is not as easy as just turning
on your sensor and collecting measurements; you will need some additional preparation
before and during data collection. Preparations may include:
- Quality Control - It is advisable that you calibrate a sensor before collecting
measurements, and at periodic intervals during measurement collection, to test
instrument response to changes in concentrations. A calibration procedure checks
an instrument's response by comparing them to a standard or reference value30.
Sensor calibration is vital for producing accurate data. Ideally, calibrations are carried
out under the same conditions (temperature and humidity ranges, concentration
ranges, background air, etc.) as those in which the instrument will collect
measurements, because many sensors are strongly influenced by these conditions.
Sensor manuals often include information on how to calibrate a device (if necessary
- some devices can be purchased pre-calibrated).
Likewise, sensors should be evaluated for precision by testing them multiple times
with "clean" air containing none of the pollutant. Such testing is then followed up by
testing the unit multiple times with an air source having a known concentration of the
pollutant. Data at the "zero" and higher concentration will allow you to determine
how well the sensor repeats itself under various conditions.
Bias, an error in the measurement that is repeatable, can be determined by taking
multiple measurements with the sensor and comparing these data with the "true"
concentration. The true concentration can be established by a reference monitor
located in close proximity to the sensor. For more information on these concepts,
refer to Appendix C.
- Sensor Maintenance - Some actions may be required to maintain sensor
performance over the measurement period. Sensor maintenance processes include
regularly cleaning internal surfaces (especially optics) to prevent the buildup of bugs
or dust, replacing filters and/or batteries, and examining site features to ensure that
no significant changes to the landscape have occurred. Sensor maintenance
processes over both the short- and long-term are discussed in further detail in
Section 6 and Appendix C.
- Data Review - A data review is a technical evaluation of the data collected by a
monitoring device31. It is a good idea to evaluate the quality of your data during the
collection phase to identify and correct potential problems that may arise. In order to
do this, analyze data to look for seasonal, day/night, or weekday/weekend patterns.
An absence of expected patterns may indicate a problem with your sensor or with
your measurement approach.
- Data validation - Data validation is the process of evaluating collected data against
established acceptance criteria to determine data quality and usability31. As you are
collecting data, it is important to visually screen for odd patterns, decreases in overall
response, and other unusual features. If you wait until your study is complete, it will
30http://nepis.epa.qov/Exe/ZyPURL.cqi?Dockev=20001QWV.txt
31 http://www.epa.gov/esd/factsheets/data-auditinq.pdf
• • •
30
-------
Air Sensor Guidebook How to Collect Useful Data Using Air Sensors
be too late to fix these issues, which tend to produce data that look too regular or
change too abruptly to be caused by natural atmospheric phenomena. Some specific
problems which may occur during data collection include:
• Interferences - factors that hinder, obstruct, or impede the ability of a
sensor to make accurate measurements. Interferences may have a
positive or negative effect on sensor response, and can include anything
from pollutants or other chemical compounds that are not of interest to
weather conditions and dirt/dust/insects. It is possible for a sensor to
respond to several different interferences simultaneously. Manufacturers
usually disclose pollutants and weather conditions that may impact
sensor performance, but may not describe how severely the sensor will
be affected. Before using a sensor to monitor air quality, consider
possible sensor interferences, test for them, and minimize them if
possible.
• Drift - refers to a gradual change in a sensor's response characteristics
over time. Instrument drift may lead you to wrongly conclude that
concentrations have increased or decreased over time. Drift can be
positive or negative, and it may occur due to a variety of reasons. One
way to reduce drift is to calibrate the sensor frequently so that the
instrument only drifts a small amount between each recalibration. The
frequency of calibration needed will depend on how much drift occurs.
5. Analyze, Interpret, and Communicate Your Results. The way you present your
results to your audience is critical to successfully sharing your understanding of the data
and achieving the objectives of sensor-based air quality data collections. Common ways
of visualizing data are: graphs of pollutant concentrations over time to show daily,
weekly, seasonal, or yearly variation in concentrations; charts of wind direction and/or
pollution to identify sources, and maps plotting data from several sensors to illustrate
patterns in concentrations32. Generally, simply showing the measurements that you have
collected will not be sufficient; your audience will want to know about all the steps that
you took to ensure data quality:
- Quality Assurance - Adequate planning to ensure that sensor design and use met
the performance requirements of your specific application. Depending on your
intended use of the data you collect, you might consider data quality assurance at
various levels (Section 5.1). For instance, data intended for a direct comparison with
State or Federal monitoring would require significantly more quality assurance than a
general survey of pollutant concentrations for informational purposes only (such as
an educational event for a grammar school).
- Quality Control - Sensor calibrations, precisions and bias checks, maintenance, and
data audits required for your application during data collection to identify and correct
potential issues such as sensor degradation, problems with sensor location, etc.
32 See Appendix B of the EPA's Data Quality Assessment: A Reviewer's Guide (EPA QA/G-9R), available at
www.epa.qov/QUALITY/qs-docs/q9r-final.pdf, for some helpful tips on data displays.
31
-------
Air Sensor Guidebook How to Collect Useful Data Using Air Sensors
- Quality Assessment - Determination of the quality of your measurements and
sufficient analysis of the data prior to reaching final conclusions.
Regardless of whether you present your results as a written report, a presentation, or in
conversation, you should be able to describe your approach, the measurements you made, the
quality checks you had in place (calibrations, etc.), and your interpretation of the data. If any one
of these components in missing or not well executed, the usefulness of your data will be
compromised.
Keep in mind that using sensors to answer a question about air quality is often an iterative step
by step process. You may find that your measurements do not satisfactorily answer your
question, or you may find yourself with many more questions after analyzing your data.
Reevaluate your approach and repeat the steps described above as needed.
• • •
32
-------
Air Sensor Guidebook Sensor Performance Guidance
5. Sensor Performance Guidance
This Section
• Helps you select sensor performance metrics appropriate for your application
• Describes different types of applications for sensors
• Presents a table of performance metrics
The performance of an air sensor or instrument describes its overall ability to measure air
pollution. This section provides initial guidelines on how well a sensor or instrument needs to
perform in order to be used for different types of air pollution applications. Specifically, we define
each application, provide performance metrics for a range of different applications, and provide
several real-world examples.
5.1 Application Areas
We have defined five application areas of interest to sensor users. These are: I) Information and
Education, II) Hotspot Identification and Characterization, III) Supplementary Network
Monitoring, IV) Personal Exposure Monitoring, and V) Regulatory Monitoring, as discussed
below. Several real-world examples from organizations using sensors are provided to illustrate
these application areas. For your reference, Appendix C provides a detailed discussion of a
number of technical considerations, including how to find the precision and bias of a specific
sensing device. It must be stated that no low cost sensors meet the Regulatory Monitoring
requirements and the discussion here is for informational purposes only.
Tier I. Education and Information. Educational applications use sensors as teaching tools.
These applications center around informational measurements, which are intended to foster
informal and qualitative awareness. For example, an instrument might indicate the presence or
absence of a pollutant by a signal such as a light going on or off. Or, a device may use
qualitative indicators, such as colors, to communicate a general sense of air quality. Such
measurements can be used for relative comparisons between air pollution levels in two
locations or at different times, rather than for measurement of absolute levels. Measurements
like these may help address questions such as: Is air pollution on my daily commute to work
higher or lower than at home? Is air pollution higher today compared to yesterday? Where
would it be best for me to run or bike today? Sometimes these sensors may not report air quality
in traditional concentration units. However, users may still find measurements made by these
unitless scales or colors to be useful for making relative comparisons.
The uncertainty in these types of measurements is quite large. The expectation would be that
the air pollution sensor is directionally consistent with pollutant trends; for example, a light might
consistently turn on if a PM concentration is 50-100 ug/m3 or higher, and consistently turn off if
measurements are below this threshold. Even if the "estimated" concentration is appreciably
higher or lower than the "true" concentration when using such a device and results in an
accuracy error of 50% (e.g. 10 ppb versus 15 ppb), the device might still be useful in educating
others where data quality needs are less important.
33
-------
Air Sensor Guidebook Sensor Performance Guidance
Sensor technology and sensor use fits within the Next Generation Science Standards (NGSS) -
new K-12 science standards that work provide students with an internationally benchmarked
science education33. The use of sensor technology in an education setting can also help
advance science, technology, engineering and mathematics (STEM) learning at various grade
levels. Exposing students to STEM education and hands-on science projects (i.e. the
development and deployment of air pollution sensors) can improve classroom learning and help
support the President's goal to provide students at every level with the skills they need to excel
in the fields of science, technology, engineering and math34.
Tier II. Hotspot Identification and Characterization typically uses fixed location and/or mobile
sensor systems to map pollutants and determine emission sources. For example, this can be
done by clustering a network of sensors downwind of an industrial facility or shipping port;
placing a network of near-road sensors along an urban interstate freeway; or placing sensors in
a vehicle for industrial fence line surveys or on an aircraft that flies in and out of a power plant
emissions plume. In most cases, the sensors will be making measurements close to the
emission location, where pollutant concentrations are usually high. One example of
sophisticated hot spot identification is the U.S. EPA's Geospatial Measurements of Air Pollution
(GMAP) monitoring vehicle (www.epa.gov/nrmrl/appcd/emissions/sec gmap.html). For Tier II
applications, a bias and precision of ±30% might be reasonable.
Tier III. Supplementary Network Monitoring (also referred to as "exploratory monitoring") is
the use of air sensor systems to complement an existing network of air quality monitors. This is
done by supplementing the regulatory network with many lower-cost devices, filling in spatial
gaps. These additional sensors may be at a permanent fixed location, or on mobile platforms,
depending on network objectives. The data from supplemental monitoring may not be sufficient
for regulatory purposes, but may help you identify potential pollution sources of interest. A
selection of state and regional officials said in interviews that if they were presented with
community group data that had a precision and bias of 20% or better, they would be willing to
investigate the findings further (provided the project design and execution seemed reasonable).
This general consensus must not be considered as representative of all state and federal air
quality officials or their opinions on this subject. Likewise, these descriptions of precision and
bias error ranges are application dependent and probably highly conditional to the pollutant
being monitored and the circumstances involving the data collections. European guidance
suggests a precision and bias range of 30-50% might be applicable. Note that there currently
exists no U.S.-based defined role for supplemental monitoring requirements and the discussion
here is solely for informational purposes.
33 http://www.nextqenscience.org/
34 http://www.whitehouse.gov/issues/education/k-12/educate-innovate
• • •
34
-------
Air Sensor Guidebook
Sensor Performance Guidance
Education Application: AirCasting Queens Vocational & Technical High School
In 2012, HabitatMap received an Environmental Justice Community Impact Grant from the
New York State Department of Environmental Conservation to develop and teach a course
on air quality monitoring at the Queens Vocational & Technical High School (QVT) in New
York City. The course objectives were: (1) Equip students with the skills, knowledge, and
tools to record, analyze, interpret, and communicate air quality information; (2) furnish
students with hands-on learning experiences that encourage them to engage with their
environment, participate in community life, and understand why science, technology,
engineering, and mathematics are important in the context of solving real-world problems;
and (3) provide the public and policymakers with meaningful and accessible air quality
information that would lead to more informed decision-making and improved air quality.
26 students from QVT participated in the course and built their own air monitors using open-
source information and instructions from AirCastinq.org. The students learned the basics of
air quality, constructed their own PM sensors using parts that are widely available from online
vendors, and used a 3D printer to create enclosures to house and protect the electronics
assembly. The total cost of electronics for each air monitor was approximately $120.
The students toured the neighborhood around their school and recorded thousands of
particulate matter measurements and dozens of observations related to air pollution hot
spots and air pollution incidences (e.g., idling vehicles, visible smokestack emissions, black
exhaust from diesel trucks). They presented their findings to their peers, members of the
Newtown Creek Alliance, and staff from the New York State Department of Environmental
Conservation. Their aggregated measurements are available on the AirCasting CrowdMap.
Because these devices were not calibrated and the quality of the data was not verified, this
fits more in the Education and Information application of showing qualitative data.
By learning how to build air monitors, understand air quality in local context, and use the
AirCasting platform to record, map, and share their findings, these QVT students learned
how science, technology, engineering, and mathematics can be used to inform the possible
solutions to real-world problems.
The map at the right shows student-
recorded PM levels throughout Sunnyside,
Queens, New York City. Colored points
indicate sampling locations, with different
colors representing difference levels of air
quality:
• Green: below detection limit
• Yellow: low
• Orange: medium
• Red: high
CrowdMap showing all of the combined measurements, http://bit.lv/15d7gbk.
• • •
35
-------
Air Sensor Guidebook
Sensor Performance Guidance
Supplemental Monitoring in the European Union (EU)
The EU directive on "ambient air quality and cleaner air for Europe" provides for the use of
"indicative measurements." These measurements can be used to supplement "fixed" (or,
"regulatory") measurements to provide information on the spatial variability of pollutant
concentrations. These supplementary measurements have less stringent requirements for
data quality. The performance requirements for the fixed and indicative measurements are
defined below. There are several differences between U.S. and EU efforts:
• The directive 2008/50/EC is a regulatory document, and makes provisions for the use
of indicative measurements to supplement fixed measurements in the regulatory
process. However, no provision is made for them to be used in isolation for
regulatory purposes. Currently, in the U.S., there is no defined role for supplementary
sensor measurements in regulatory monitoring, and this document does not provide
one; the performance metrics in Table 5-1 are suggestions only.
• The EU performance requirements relate only to the use of indicative measurements
to supplement fixed measurements for regulatory purposes. Therefore, while the
performance requirements of the indicative measurements vary by pollutant, there
are no performance requirements for the other application areas discussed in this
document.
• EU performance requirements are listed below (from Table A of Annex I of the
directive 2008/50/EC). Note: the EU requirements specify a maximum uncertainty,
and do not address precision and bias separately.
Type of
Measurement
Regulatory
(fixed)
Supplemental
(indicative)
Maximum Uncertainty Allowable in Pollutant Measurement
SO2, NOX, CO
15%
25%
Benzene
25%
30%
PM & Lead
25%
50%
Ozone
15%
30%
EU directive 2008/50/EC can be seen at http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001:0044:EN:PDF
• • •
36
-------
Air Sensor Guidebook Sensor Performance Guidance
Tier IV. Personal Exposure Monitoring encompasses any application where a person's
exposure to air pollution is monitored, often to evaluate the impact of air pollution on health. This
may include measurements taken to protect an individual whose health might be impaired by
elevated air pollution, or an epidemiological research study to help understand the effects of air
pollution on a group of people. An example of one such effort was the U.S. EPA's Detroit
Exposure Research Study (DEARS), where participants were involved in wearing portable
sensors to document their exposures (http://www.epa.gov/dears/). Personal exposure studies
have historically been research projects where people wear devices that measure air quality as
they go about their daily routines. In the future, people may monitor their own exposure to air
pollution to help make medical decisions. Personal exposure is currently estimated using EPA's
Air Quality Index (AQI), which communicates health risks from air pollution using a color-coded
scale. For this application, a bias and precision of 30% or better might be a goal for such air
quality monitoring scenarios.
Tier V. Regulatory Monitoring includes monitoring for criteria pollutants to determine if an area
is in compliance with the National Ambient Air Quality Standards (see section 1.3 for more
information). In the U.S., regulatory monitoring is performed by air quality agencies and
governed by the performance requirements specified by the Code of Federal Regulations35.
Instruments or technologies which are used to comply with requirements for regulatory
monitoring must meet the requirements of Federal Reference Methods or Federal Equivalent
Methods. Requirements include meeting stringent measurement quality objectives and
substantial operational requirements36, and are therefore considered the "gold standard." In
contrast, there are no such written requirements for measurements in Tiers I-IV. No low cost
sensors have been approved to collect regulatory monitoring data.
The U.S. EPA also regulates air quality associated with a select number of air toxics. One
example of such a pollutant is benzene, an air toxic widely distributed in our environment.
Sensor users are encouraged to review (www.epa.gov/ttn/amtic/airtox.html) for specific
information on data quality for these pollutants. Often precision error of no more than 15% is
required to ensure adequate measurements of these air pollutants.
5.2 Suggested Performance Goals for Each Application
As outlined above, sensor systems have the potential to be used across specific air quality
measurement applications, which can range from those requiring relatively high-performing
measurements to informal projects with minimal data quality requirements.
Table 5-1 provides information about how well your sensor must perform so that the data you
collect will be useful. Which of the tiers above best describes the specific purpose of your
monitoring effort? Once you have identified the appropriate tier, consult Table 5-1, which has
columns for four performance characteristics. Additional data quality indicators and associated
35 See 40 CFR Part 50 Appendices L and N, 40 CFR Part 58 Appendix A. Also refer to Appendix D of the EPA
Quality Assurance Handbook for Air Pollution Measurement Systems Volume II,
www.epa.qov/ttnamti1/files/ambient/pm25/qa/QA-Handbook-Vol-ll.pdf.
Appendix D of EPA Quality Assurance Handbook for Air Pollution Measurement Systems Volume II is a synthesis
of the requirements specified in the CFR and guidance provided by members of a working group formed by the EPA.
37
-------
Air Sensor Guidebook Sensor Performance Guidance
performance characteristics are required for regulatory monitoring and may be required for other
applications requiring higher data quality.
• Bias
• Precision
• Data averaging time
• Data completeness
Bias and precision are discussed above in Section 3 as metrics for assessing the accuracy
of sensor.
Data averaging time is a key performance metric because precision can be improved when
more data from a particular measurement system are averaged together. Data are often
averaged, or aggregated, to facilitate comparison to measurements from another
instrument, health-based benchmarks, or environmental standards. Data averaging helps
improve the quality, usefulness, and manageability of your data. The exact type of averaging
will depend on your application and the question you are trying to answer.
Table 5-1 provides appropriate averaging times over which data should be averaged for
various uses. For example, if you are interested in observing a pollutant concentration trend
over the course of a month, you may want to analyze your data to 1-hour or 24-hour
patterns. You will be able to see how the concentrations change, but averaging will reduce
the amount of data you are working with to a manageable size. It will also minimize the
effects of outliers (those individual data points that stray far from the average). On the other
hand, if you would like to identify a pollution hotspot, you may prefer to use a shorter
averaging period, such as a few minutes, to capture the precise location of the hotspot. A
shorter averaging period would allow you to detect the hotspot in your data as the sensor
moved around the area of interest.
Data completeness refers to the amount of data that was actually obtained, compared to
the amount that was expected (for example, a sensor operating correctly and providing data
for 4 days out of a 5 day monitoring test would have 80% data completeness). See
Appendix C for more information on these topics.
Detection limit is another important performance metric to consider, but because detection limit
needs can vary between projects it is better to assess requirements on a case-by-case basis
(see Section 3 for more details). As discussed in Sections 3 and 4, and Appendix C, a wide
range of factors influence sensor performance, including interferences from other gases and
particles and methods of operating the sensor.
As shown in Table 5-1, the suggested performance goals are different for each of the five
application areas (tiers). Tier V is the highest quality level discussed, representing the regulatory
monitoring application. Applications in lower tiers have less stringent performance goals.
The performance goals presented in Table 5-1 were developed based on expert interviews,
group meetings and discussions, and peer-reviewed and government literature. These
performance goals are an initial guideline that will be refined over time as technology, the
community's collective experience, and sensor systems evolve and improve.
• • •
38
-------
Air Sensor Guidebook
Sensor Performance Guidance
Table 5-1. Examples of Suggested Performance Goals for Sensors for 5 Types of Citizen Science Applications in Comparison to
Regulatory Monitoring Requirements
Tj Application
'ier Area
Education and
Information
Pollutants
All
Precision
and Bias
Error
<50%
Data
Completeness*
> 50%
Rationale (Tier I-IV)
Measurement error is not as important as simply
demonstrating that the pollutant exists in some wide
range of concentration.
Hots pot
Identification and
Characterization
All
<30%
> 75%
Higher data quality is needed here to ensure that not
only does the pollutant of interest exist in the local
atmosphere, but also at a concentration that is close
to its true value.
Supplemental
Monitoring
Criteria pollutants, Air
Toxics (incl. VOCs)
<20%
> 80%
Supplemental monitoring might have value in
potentially providing additional air quality data to
complement existing monitors. To be useful in
providing such complementary data, it must be of
sufficient quality to ensure that the additional
information is helping to "fill in" monitoring gaps rather
than making the situation less understood.
IV
Personal
Exposure
All
<30%
> 80%
Many factors can influence personal exposures to air
pollutants. Precision and bias errors suggested here
are representative of those reported in the scientific
literature under a variety of circumstances. Error rates
higher than these make it difficult to understand how,
when, and why personal exposures have occurred.
• • •
39
-------
Air Sensor Guidebook
Sensor Performance Guidance
Tier
V
Application
Area
Regulatory
Monitoring 7
Pollutants
03
CO, S02
N02
PM2.5, PM10
Precision
t t WVslOI VI 1
and Bias
Error
<7%
<10%
<15%
<10%
Data
Completeness*
> 75%
Rationale (Tier I-IV)
Precise measurements are needed to ensure high
quality data is being obtained to meet regulatory
requirements
Note: These are guidelines only (Tier I- Tier IV), and are likely to evolve overtime as technology continues to develop and the state of the
science continues to advance. *The amount of data needed for any air quality purpose is highly specific to that purpose and could range from
minutes to even years of data measurements.
Precision, bias, and data completeness requirements in part were taken from Appendix D of the EPA Quality Assurance Handbook for Air Pollution
Measurement Systems Volume II (May 2013 edition). Refer to 40 CFR Parts 50, 53, 58, and the QA Handbook Volume II for activities/criteria for monitoring
network data.
• • •
40
-------
Air Sensor Guidebook Maintaining Your Sensing Device
6. Maintaining Your Sensing Device
Air monitoring technology, like most other forms of technology, requires careful care and
maintenance to ensure proper functionality and reliable performance. These preventative
actions are necessary in both the short- and long-term, and may vary with the specific
monitoring technology being utilized. By properly caring for a monitoring device you can reduce
errors in data collection, extend the shelf-life of the device, and save money that would
otherwise be spent on replacement parts and repair services.
Maintenance Processes
Maintenance processes are the actions required to maintain sensor performance over an
extended period of time. Good maintenance processes can help maximize and sustain sensor
performance. Typical maintenance processes include regularly:
• Calibrating with pollutant standards and flow meters as described under the Calibration
section of Appendix C.
• Cleaning internal and external surfaces and components to prevent the buildup of bugs,
dust, etc.
• Replacing filters and consumables.
• Replacing the sensor when it has failed or reached its lifespan of service.
• Replacing rechargeable batteries.
• Reviewing (visually inspecting) data for odd patterns, a decrease in overall response,
drift in the baseline, and other unusual features. Instrument problems tend to produce
data that often look too regular and repeatable, or that change too abruptly, to be due to
natural atmospheric phenomena.
• Inspecting sensor placement to ensure that no significant changes have occurred (e.g.,
tree growth, building changes, etc.).
Developing a set of maintenance processes relevant to your sensor helps the user consider
how best to deploy and maintain the sensor. Developing and maintaining a logbook to ensure
maintenance occurs at regular intervals is helpful.
Learn more:
http://www.epa.gov/ttn/amtic/contmont.html. See the Standard Operating Procedures for
regulatory monitors on this website for examples of maintenance done on PM2.s
monitors.
• • •
41
-------
Air Sensor Guidebook Additional Resources
7. Additional Resources
Air Quality
• EPA criteria pollutants: http://www.epa.gov/air/urbanair/
• Criteria pollutants overview and standards: http://www.epa.gov/airtrends/sixpoll.html
• Air pollutant information: http://www.epa.gov/air/airpollutants.html
• Black carbon health effects: http://www.epa.gov/research/airscience/air-blackcarbon.htm
• Carbon dioxide emissions page:
http://www.epa.gov/climatechange/ghgemissions/gases/co2.html
• Sources of greenhouse gas emissions:
http://www.epa.gov/climatechange/ghgemissions/sources.html
• Air guality trends: http://www.epa.gov/airtrends/agtrends.html
• Weather effects on trends in ozone pollution: http://www.epa.gov/airtrends/weather.html
• Local area trends for criteria air pollutants: http://www.epa.gov/airtrends/where.html
• Atmospheric science and the formation of pollutants: http://www.epa.gov/airscience/air-
atmosphericscience.htm#chemistry
• EPA toxics website: http://www.epa.gov/air/toxicair/newtoxics.html
Sensors
EPA's Air Sensors 2013 and Next Generation Air Monitoring Workshop Series
homepage: https://sites.google.com/site/airsensors2013/final-materials
EPA Next Generation Air Monitoring website:
http://www.epa.gov/research/airscience/air-sensor-research.htm
A forum for the air sensors community to share and collaborate: http://citizenair.net/
Citizen science opportunities for monitoring air guality fact sheet:
http://www.epa.gov/research/priorities/docs/citizen-science-fact-sheet.pdf
Data Sources
• Multiple links to air guality data sources: http://www.epa.gov/air/airpolldata.html
• Access to real-time air guality maps and forecasts from EPA's AirNow system:
http://www.airnow.gov
• AirNow Gateway for obtaining real-time data via files and web services:
http://airnowapi.org/
• Access to historical air guality data from EPA's Air Quality System (AQS):
http://www.epa.gov/airdata/
• Portal to download detailed AQS data:
http://www.epa.gov/ttn/airs/airsags/detaildata/downloadagsdata.htm
Health Effects
• EPA's Air Quality Index: A Guide to Air Quality and Your Health:
http://www.epa.gov/airnow/agi brochure 08-09.pdf
• • •
42
-------
Air Sensor Guidebook Additional Resources
EPA's Guide to Particle Pollution and Your Health:
http://www.epa.gov/airnow/particle/pm-color.pdf
EPA's Guide to Ozone and Your Health: http://www.epa.gov/airnow/ozone-c.pdf
EPA's Risk Assessment for Toxic Air Pollutants: A Citizen's guide:
http://www.epa.goV/ttn/atw/3 90 024.html
Other
General Air Research/Air Science: http://www.epa.gov/research/airscience/
EPA's "plain English guide" to the Clean Air Act: http://www.epa.gov/air/caa/peg/
Near roadway research and pollutant effects: http://www.epa.gov/airscience/air-
highwayresearch.htm
Role of vegetation in mitigating air guality impacts of air pollution:
http://www.epa.gov/nrmrl/appcd/nearroadway/workshop.html
Air Pollution Training Institute (APTI) Learning Management System: http://www.apti-
learn.net
CDC Agency for Toxic Substances and Disease Registry (ATSDR) Toxic Substance
FAQ (ToxFAQ): http://www.atsdr.cdc.gov/toxfags/index.asp
CDC The NIOSH Pocket Guide to Chemical Hazards: http://www.cdc.gov/niosh/npg/
• • •
43
-------
Air Sensor Guidebook Appendix A
Appendix A: Potential Questions
If and when you decide to share your data with others, it is likely they will have a number of
questions regarding the data you've collected and the techniques you've employed. Below we
have tried to provide a list of the types of questions to expect. While this list is by no means
exhaustive, it gives a general outline of the information you are likely to be asked for.
Basics
• What do you want to find out or show with your measurements?
• What pollutants did you measure?
• Do you consider this a nuisance or a health hazard? Is this a recurring problem?
• Do you know the normal levels for the pollutant, including seasonal and day/night
profiles?
• Do you have Standard Operating Procedures (SOPs) (detailed written instructions so
that measurements are taken in a consistent way)?
• Did you receive adequate training in how to operate the device and maintain it?
Monitor
• What instrument/sensor did you use?
• How were the measurements taken?
• When did you make your measurements? (i.e. time of day/night, day of week, season)
• How long was the period during which you collected measurements?
• How did you ensure that quality measurements were collected? (How did you calibrate
your sensor? How did you estimate precision and bias?)
• Did you co-locate your sensor near regulatory monitors or other approved measurement
systems to evaluate their performance?
• How were samples identified and their identity recorded and tracked as they were
transferred to others or analyzed?
• What, if any, additional data were collected? (e.g., wind measurements, site photos,
GPS, activity logs, event logs, health info)
Location/Surrounding Environment
• Where were the measurements collected?
• Were there other emission sources near the location you were measuring that could
have mixed with the pollutants coming from the source of interest? (e.g., roadways,
other industrial facilities, etc.)
• Was anyone, including you, smoking nearby when you collected the measurements?
• Did you take the measurements while in a moving vehicle or were you stationary?
• How were you holding the sensor, or was it attached to a vehicle or stationary object?
• What were the weather conditions?
• Were you indoors or outdoors while taking measurements?
• • •
44
-------
Air Sensor Guidebook Appendix A
Data Analysis
• How will the data be analyzed? (e.g., compare with meteorological measurements, other
site data)
• How will you differentiate the source you are trying to measure from the background?
• Did you average your measurements and if so, how?
Other
• Is this an anonymous report, or will you provide contact information for follow up?
• Did you have any interaction with the people creating the emissions? Have you in the
past?
• • •
45
-------
Air Sensor Guidebook
Appendix B
Appendix B: Air Quality Concepts and Characteristics
Table B-1 outlines various concepts and characteristics that provide a good foundation for
understanding air pollution. This table expands on the discussion in Section 2.3, and includes
an examination of how each concept may influence the development and use of air sensing
devices. Concepts defined below are generalizations of those reported in the scientific literature
for a variety of pollutants and general air quality discussions (http://www.epa.gov/airquality/).
Table B-1. Air quality topics, discussion, and relevance.
Topic
Primary vs.
secondary
pollutants
Discussion
Pollutants may be emitted directly by a source
(primary pollutants) or may be formed as a product
of a chemical reaction in the air (secondary
pollutants). Primary pollutants that contribute to the
formation of secondary pollutants are also called
precursors.
Spatial difference in primary pollutants can be
large, especially if there are no other nearby
sources of the pollutants. Spatial differences may
be smaller for secondary pollutants.
Relevance to Sensor
Development/Use
Consider whether a pollutant
of interest is primary or
secondary pollutant to help
select a monitoring location.
In some cases, it may be
easier to determine the
source of a primary pollutant
than the source of a
secondary pollutant.
Short-lived vs.
long-lived
pollutants
The atmospheric lifetime of a pollutant is the
average amount of time the pollutant resides in the
atmosphere before it is removed by reacting to
form a new molecule or depositing onto a surface.
This lifetime varies significantly for each pollutant
according to its likelihood of reacting with other
species (reactivity) or depositing.
Species with longer atmospheric lifetimes tend to
be more uniformly distributed in the atmosphere,
while concentrations of species with shorter
atmospheric lifetimes may be more variable in
space and time. Atmospheric lifetime of some
chemicals may be affected by seasonal
temperatures. Short-lived pollutants that react
quickly after they have been emitted may be highly
variable in space and time. Long-lived pollutants
typically show less variation over distances or time.
Detecting a short-lived
pollutant requires a sensor
that responds quickly.
A slower sensor response
may be used for detecting
long-lived pollutants,
especially if the sensor is not
moving.
• • •
46
-------
Air Sensor Guidebook
Appendix B
Topic
Local vs. regional
vs. global scale
Discussion
The atmospheric lifetime of a pollutant governs
whether a source will affect air quality on a local,
regional, or global scale. While some air pollution
problems are limited to the local area where
pollution is emitted (e.g., 1,3-butadiene, a short-
lived pollutant), others are transported and impact
air quality across cities or entire regions of the
country (e.g., ozone, PM25). For some pollutants,
emissions from everywhere on earth contribute to
a global problem (e.g., CO2).
Relevance to Sensor
Development/Use
Knowing the lifetime and
emission sources of the
pollutant of interest helps
you understand whether
concentrations are
influenced by a local source
or distant sources.
Weather (e.g.,
sunlight, winds,
temperature)
Concentrations of pollutants are also controlled by
weather, including sunlight, temperature, humidity,
clouds, precipitation, and winds. Concentrations
can increase more rapidly when winds are
stagnant.
Higher winds typically dilute pollutant
concentrations, but may lead to increased
concentrations of other pollutants (such as dust).
Air quality and weather are
linked. Weather can affect
both air pollution
concentrations and sensor
performance. Therefore, it is
very important that you know
how weather conditions can
influence your sensor
measurements.
Time of day
Some pollutants have strong day/night patterns
due to source patterns or meteorological changes.
Sensor performance may
vary throughout the day due
to changes in source
patterns and weather.
Day of week
Concentrations of some pollutants vary according
to the activity schedule of the source (e.g., traffic
patterns, industrial schedule).
When developing a
measurement plan, consider
the day-of-week pattern in
emissions from the sources
you are trying to measure.
Season
Some pollutants display a strong seasonal
variation because of differences in emissions
patterns, formation processes, and atmospheric
longevity. For example, wildfires emit particles,
VOCs, and NOX, and are more prevalent in dry,
warm conditions; residential wood burning,
however, may be more important in the winter.
Consider the seasonal
variation of the pollutant of
interest to inform your study
design. Sensor systems may
need to work in particularly
adverse conditions such as
extreme heat, humidity, or
cold.
Near-source
concentrations
Concentrations of primary pollutants are typically
highest very close to their emissions source.
Concentrations generally decrease rapidly within
the first few hundred feet of a source as the
pollutants are transported and dispersed.
Consider concentration
gradients in your study
design. More than one type
of sensor (or a sensor
system with more than one
operational mode) may be
needed, depending on the
range of concentrations that
will be measured.
• • •
47
-------
Air Sensor Guidebook
Appendix B
Topic
Multiple, well-
distributed
sources
Discussion
When there are many widely distributed sources in
an area, such as gasoline stations in an urban
environment, concentrations may be very similar
across the area.
Relevance to Sensor
Development/Use
A network of sensors
(upwind, near-source, and
downwind) may be needed
to identify these sources.
Man-made vs.
naturally
occurring
pollutant sources
Typically, measurements focus on human
influenced sources, but there are natural sources
such as fires, lightning, windblown dust, and
volcanic activity.
Consider all sources of the
pollutant of interest when
designing your study.
Pollutant transport
The distance a pollutant may be transported is
governed by atmospheric chemistry (formation and
depletion reactions), weather (air mass movement
and precipitation), and topography (mountains and
valleys that affect air movement). The longer a
pollutant stays in the atmosphere, the farther it can
be transported and the harder it becomes to
identify its source.
Understanding how
pollutants are transported
can help you identify the
source.
• • •
48
-------
Air Sensor Guidebook Appendix C
Appendix C: Technical Considerations
The quality of data collected with sensors can vary greatly depending on sensor design and
your deployment strategy. You must consider the following factors carefully while planning for,
making, and processing measurements if you are to produce quality data and useful results.
You should be able to show that the quality of the data you collected is sufficient to meet the
performance requirements of the application and of the intended audience.
This section describes several factors to consider in order to collect quality measurements from
air sensors, regardless of the intended application. This builds upon the discussion in Section
3.1, which introduces considerations relevant to air sensor users and developers.
C.1 Considerations for Air Sensor Users and Developers
The following considerations cover a broad range of performance-related characteristics. We
provide a technical description of each topic and explain its relevance to low-cost sensor
applications. This section covers the following factors affecting air quality measurements:
• bias
• precision
• calibration
• detection limit
• response time
• linearity of sensor response
measurement duration
measurement frequency
data aggregation
selectivity
• interferences
• sensor poisoning and expiration
• concentration range
• drift
• accuracy of timestamp
• climate susceptibility
• data completeness
• response to loss of power
Bias
Bias means an average systematic or persistent distortion of a measurement process that
causes errors in one direction38. Bias can be thought of as a fixed value that is always added or
subtracted from the true value of the pollutant by the sensor (Figure C-1).
38 http://www.epa.qov/qualitv/qs-docs/q5-final.pdf.
• • •
49
-------
Air Sensor Guidebook
Appendix C
Why it matters: Biased measurements consistently misrepresent the true concentration of a
pollutant, usually by producing data that are either regularly higher or lower than the true
value of the pollutant by a fixed amount. Bias is usually caused by a characteristic of the
sensor, by a problem with the overall measurement method, or by a persistent mistake
that the operator inadvertently makes with each measurement. A bias is considered a
determinate error (the cause is known) and may be corrected by recalibrating the
sensor, altering the method, or correcting operating procedures.
How to calculate it: There is no one correct method of estimating bias. One example of a bias
calculation is as follows:
/ „ \
-1
B =
where 6 is the bias, C is the average of the measurements, and CR is the reference
concentration, or true value, of the pollutant. Confidence in a calculated bias generally
increases with the number of measurements. Zero bias is ideal, but low values for bias
may also be acceptable. The bias may change as a function of environmental conditions
(e.g., with temperature and humidity), lifespan of the sensor, or other factors. Therefore,
consider checking your sensor for bias routinely, with frequent calibrations and/or inter-
comparisons with other sensors. Comparisons with high-performance instruments, or
sensors that work by another measurement principle, may be valuable.
Learn more: A more in-depth discussion of bias is presented in the text box below. See also
www.epa.gov/ttn/amtic/files/ambient/monitorstrat/precursor/07workshopmeaning.pdf.
Figure C-1 shows a comparison of a true value of NO2 (blue line) and biased measurements of
NO2 (red line). The consistent offset between the two time series is the bias.
•True NO2 concentration
•Biased NO2 measurements
0 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19 20 21 22 23
Hour of the Day
Figure C-1. Comparison of a true value of NO2 and biased measurements of NO2.
• • •
50
-------
Air Sensor Guidebook Appendix C
Precision
Precision measures the agreement among repeated measurements of the same property under
identical or substantially similar conditions. The more frequently data are collected over a given
period the more confidence one has in the concentration estimate. Precision can be expressed
in terms of standard deviation39. Precision can be thought of as the scatter introduced into data
by random (indeterminate) errors when an instrument attempts to measure the same
concentration of a pollutant multiple times.
Why it matters: The precision of a sensor will determine the quantity of data needed to achieve
a quality level that is suitable for your needs. The precision of an instrument can be
improved by averaging more of the raw data together. For example, if 1-second data are
subject to significant random error, the data can be grouped into 5-minute averages so
the random errors cancel each other out (Figure C-2). Grouping data often results in
fewer individual data points, but the grouped data will be more precise (i.e., a lower
standard deviation) and potentially a better representation of the true value of the
pollutant, provided the measurements are unbiased. Even so, one must understand how
critical the time period is when grouping data. If one wishes to estimate 15-minute data
concentrations, grouping data on a one hour basis to determine precision would not be
an acceptable practice.
The precision of an instrument can also be improved by averaging the data from multiple
sensors operating at the same location. It is conceivable that a number of sensors
measuring the same pollutant could be used at the same site and averaged together to
increase the precision of the combined measurement.
How to calculate it: There is no one correct method of estimating precision. Precision can be
estimated by various statistical techniques using some derivation of the standard
deviation. For example, P = Cs / Cm (where P is the precision, Cs is the standard
deviation of the measurements, and Cm is the measurement mean at a given
concentration).
Learn more:
www.epa.gov/ttn/amtic/files/ambient/monitorstrat/precursor/07workshopmeaning.pdf
Figure C-2 is a time series showing measurements of 1-minute (red) and 15-minute (green)
averaged ozone measurements. The line showing the 15-minute measurements is
significantly more representative of the sensor's true response.
39 www.epa.qov/qualitv/qs-docs/q5-final.pdf.
• • •
51
-------
Air Sensor Guidebook
Appendix C
50
5/18/2013 9:00
5/18/2013 11:00 5/18/2013 13:00 5/18/2013 15:00 5/18/2013 17:00 5/18/2013 19:00
•1-min ozone measurements
-15-min averaged ozone measurements
Figure C-2. Time series showing measurements of 1-minute and 15-minute averaged
ozone measurements.
• • •
52
-------
Air Sensor Guidebook Appendix C
Understanding Accuracy, Bias, and Precision
Accuracy is a measure of the overall agreement of a measurement with a known value. Accuracy
includes a combination of systematic error (bias) and random error (precision). Reducing systematic
and random errors will improve measurement accuracy.
Accuracy is sometimes confused with bias; you may see the terms used interchangeably. EPA
recommends using the terms "precision" and "bias," rather than "accuracy," to convey the
information usually associated with accuracy (see http://www.epa.gov/emap/html/pubs/docs/
resd ocs/mg lossary. htm I).
Accuracy can be calculated by comparing the mean of a small data set to the true value:
A = x — xt
where x is the mean of a small set of data and xt is the accepted true value. The measurement
accuracy of an instrument typically improves with more data, as long as the instrument is unbiased.
In contrast, bias is the consistent, systematic difference between the measurements and the true
value. A bias can either be higher or lower than the actual value, and is caused by systematic errors
(instrumental, method, or operator errors). A bias cannot be minimized with additional data.
There are several ways to define and explain accuracy, bias, and precision. The National Institute of
Standards and Technology (NIST) and the International Union of Pure and Applied Chemistry
(IUPAC) provide useful alternatives to those presented here (http://www.itl.nist.gov/ and
http://goldbook.iupac.org), and instrumental and chemical analysis text books are also great
resources (e.g., Skoog D.A., Holler F.J., and Nieman D, 1998, Principles of instrumental analysis,
5th ed., Saunders College Publishing, Philadelphia, PA). The EPA's guideline is available at
epa.gov/ttnamti1/files/ambient/monitorstrat/precursor/07workshopmeaning.pdf.
• • •
53
-------
Air Sensor Guidebook Appendix C
Calibration
A calibration procedure checks and adjusts an instrument's measurements by comparing them
to a standard, reference, or value40.
Why it matters: Sensor calibration is vital for producing accurate, usable data. Calibration
relates the response of the instrument to the true concentration of a pollutant. Ideally,
calibrations are carried out under the same conditions (temperature and humidity
ranges, concentration ranges, background air, etc.) as those in which the instrument will
collect measurements, because many sensors are strongly influenced by these
conditions.
How to perform: Here is the basic approach to calibrating a sensor:
1. Compare the response of the air sensor with the response of a reference
instrument.
• There are two main approaches to calibrating an instrument. The first is to do a
calibration with standards, in which you introduce some widely accepted reference
standard to the sensor. The second is to do a comparison against a reference
instrument that has been calibrated with a recognized standard. This can be done by
locating the sensor near an air quality station managed by your local authority. This
is typically referred to as "collocation." If you decide to collocate, consider doing so
for a few days prior to the start, during, and after your measurement period. Locate
your sensor as close to the air quality monitor as possible, so that the two devices
are measuring the same air quality.
• Sensor calibrations may also involve using a flow meter to measure air flow through
the device if it is a device that pulls air into it.
• Here are some additional tips regarding calibrations:
- Calibrations are best done with a reference standard. Such standards are
available from many science product vendors.
- A gas standard is typically delivered from a compressed gas cylinder. However,
very reactive gases have to be made at the time of calibration because they
degrade in a gas cylinder. Ozone is a good example. An ozone generator is
needed to produce known ozone concentrations, and these are expensive. It
may be better to use the collocation calibration option for this pollutant. NO2 can
be delivered from a gas cylinder, but it degrades fairly quickly. Small bottles of
NO2 test gas have one-year expiration dates.
- Examples of particle standards are Urban Particulate Matter (UPM), Arizona
Road Dust (AZRD), and Polystyrene Latex Spheres. The UPM and AZRD are
sold as powders that would be blown ("resuspended") into a large volume (i.e., a
tank or bag) using a clean compressed-air source and then introduced to the
40 http://nepis.epa.qov/Exe/ZvPURL.cqi?Dockev=20001QWV.txt.
• • •
54
-------
Air Sensor Guidebook Appendix C
instrument. A reference measurement will be needed to determine the mass
concentration of the resuspended standard. Polystyrene latex spheres (PSLs)
are solids emulsified in water that are aerosolized using a nebulizer or atomizer
(a sprayer) and then dried using dilution air or a diffusion drier. PSLs have a very
narrow size range and are used for calibration sensor sizing performance. Even
so, developing and delivering such particles requires very sophisticated
laboratory equipment and therefore one unlikely available for citizen scientists to
perform themselves.
- Sometimes standards have to be mixed with a clean air source. Medicinal grade
breathing air or industrial grade nitrogen may be enough for your purposes, but
there are other, more expensive, options. You can buy cylinders of high purity
(HP) and ultra-high-purity (UHP) gases, as well as "Zero Air," from specialty gas
suppliers. You can also buy Zero Air generators.
- Calibrations done under very controlled environments, where contaminants and
environmental conditions (temperature and relative humidity) are known and held
constant, may not be directly relevant to real-world applications. It is important
that any laboratory calibrations be complemented with field calibrations. For
example, ozone sensors calibrated in ambient air were shown not to suffer from
the temperature and relative humidity effects that were observed in these same
sensors during a laboratory-based calibration41.
2. Create a calibration curve that relates the responses of the air sensor to the
reference instrument.
• The idea behind a calibration is to convert a raw instrument response, which is
usually some sort of electronic signal, into useful units (e.g., concentration). This is
done by creating a scatter plot comparing measurements made by your sensor
device to the standard concentrations or measurements of the reference instrument,
and then relating them using a mathematical equation.
• The amount of data needed to develop a good calibration curve (i.e., the sensor
response compared to the target concentration) depends on the linearity of sensor
response (see Sensor Response discussion below) to the target pollutants. For
example, an initial standard calibration for an ozone sensor may consist of a
calibration point collected with no ozone being available, followed by between 4 and
6 calibration events across the range of concentrations that you expect to see during
the measurement period.
3. Repeat the calibration periodically and track the changes in the calibration curve
with time.
• Subsequent calibrations should be done periodically (e.g., daily, weekly, quarterly,
semi-annually, annually) The timing and need for these events will be highly
dependent upon the sensor being used and the purpose for how it is being used. It is
41 Michel Gerboles and Daniela Buzica (2009) Evaluation of micro-sensors to monitor ozone in ambient air. Joint
Research Centre, Institute for Environment and Sustainability, Transport and Air Quality Unit, Via E. Fermi, I -21027
Ispra (VA) http://publications.irc.ec.europa.eu/repository/bitstream/1 1111111 1/10477/1/eur23676.pdf.
55
-------
Air Sensor Guidebook Appendix C
important that sensors are calibrated regularly to address changes in performance
over time. Pre-calibrated sensors are available from some manufacturers.
• Instruments that drift or change in performance quickly need more frequent
calibrations than very stable instruments.
• Track your calibrations to see how the sensor response is changing. You will pick up
on problems more quickly by doing this.
Learn more:
• http://www.epa.qov/fem/pdfs/calibration-quide-ref-final-oct2010.pdf
• NIST Standard Reference Materials can be found at http://www.nist.gov/srm/
Detection Limit
The lowest concentration that can be determined as being above zero by a single measurement
at a stated level of certainty is the detection limit42. There are many types of detection limits.
One often used is referred to as the Method Detection Limit (MDL)43, and it is typically defined
as 99% confidence that the measurement is not instrument noise (Figure C-3).
Why it matters: Environmental pollutants can often be present in very low concentrations,
particularly when measurements are being made far from the source of the pollution. To
be useful, sensors must be able to measure pollutants over the ranges of concentrations
typically seen in the atmosphere. One instrument with a higher MDL may be appropriate
near a source location, but an instrument with a much lower MDL (more able to measure
lower concentrations) may be needed far away from sources, in locations where
pollutant concentrations have become diluted. Typical pollutant ranges are shown in
Table 2-2, although depending on the situation, an instrument may or may not need to
measure well at the lower end of the concentration range.
The detection limit is usually provided by the manufacturer. You may want to ask how
the manufacturer determined the detection limit. A sensor's detection limit may vary over
time, so if you routinely measure very low concentrations, consider measuring the
detection limit frequently. This can be done by diluting the calibration gas until the
instrument cannot reliably detect the pollutant of interest anymore. If such equipment is
not available, comparing your data to those of a reference instrument measuring low,
regional background concentrations can be useful.
Learn more:
• http://www.epa.gov/fem/calibration.htm
• http://www.epa.gov/fem/pdfs/Env Measurement Glossary Final Jan 2010.pdf
42 http://www.epa.qov/qualitv/qs-docs/q5-final.pdf.
43 See 40 CFR Part 50 Appendices L and N, 40 CFR Part 58 Appendix A. Also refer to Appendix D of the EPA
Quality Assurance Handbook for Air Pollution Measurement Systems Volume II,
www.epa.qov/ttnamti1/files/ambient/pm25/qa/QA-Handbook-Vol-ll.pdf.
• • •
56
-------
Air Sensor Guidebook
Appendix C
Method detection limit
E
E
Many sensor
measurements
averaged to
give a norma
distribution
99% confidence limit
Concentration (ppb)
Figure C-3. Graphical representation of a detection limit, (source: EPA's Air Toxics
Workbook)44.
Response Time
Response time is the amount of time required for a sensor to respond to a change in
concentration.
Why it matters: A sensor that responds quickly may be useful for mobile monitoring and for
observing very rapid changes in pollutant concentrations. A sensor that responds slowly
may be more suited to stationary monitoring of pollutants that vary in concentration
gradually. The measurement duration and frequency are governed by the sensor
response time.
Learn more: Most manufacturers characterize sensor response times as a means to compare
the specifications between sensors. They typically use tgo for fast-responding sensors
and t50 for sensors with slower responders. The tgo is the time taken by the sensor
response to get to 90% of the pollutant or standard concentration that is being
measured. It is measured by first delivering zero air to the sensor and then suddenly
switching on a flow of the pollutant or standard of interest. Similarly, the tso is the time
taken by the sensor response to get to 50% of the pollutant or standard concentration
that is being measured. These concepts are illustrated in Figure C-4.
Hafner H.R., Charrier J.G., and McCarthy M.C. (2009) Air toxics data analysis workbook. Prepared for the U.S.
Environmental Protection Agency, Research Triangle Park, NC, by Sonoma Technology, Inc., Petaluma, CA, STI-
908304-3651-WB, June. Available at http://www.epa.qov/ttnamti1/files/ambient/airtox/workbook/AirToxicsWorkbook6-
09.pdf.
• • •
57
-------
Air Sensor Guidebook
Appendix C
100 - -
_Q
a
Q_
(0 50- -
o
u
Calibration gas
Calibration gas
turned on at 0
Time
at 0 * tg0
Figure C-4. Response time (t50 and t90) of an instrument to a calibration gas.
Sensor Response
A useful sensor response is composed of a unique response for each concentration measured.
Such a response is called a monotonic increase.
Why it matters: Sensor responses to pollutant concentrations are normally related using a
mathematical equation, and they are typically single valued (i.e., unique to each pollutant
concentration) in the region of interest. The sensor response does not need to be linear,
but it needs to be quantifiable through an equation; polynomial, power law, or
exponential equations are all acceptable. Figure C-5 shows examples of sensor
response functions.
• • •
58
-------
Air Sensor Guidebook
Appendix C
O
1/1
c
0)
Ideal
reference
Curved, but the sensor output
becomes constant even though
reference concentrations are still
increasing
To be most useful, a calibration
curve must only increase, or only
decrease, and not do both. This
calibration curve both increases and
decreases, causing the calibration
curve to be difficult to use properly.
The dashed line shows that one
sensor value can be interpreted as
three concentrations.
• • •
59
-------
Air Sensor Guidebook
Appendix C
Only the linear (straight) region in the
middle of the calibration curve is
useful in this example because
curves at the end start to curve
inappropriately.
Figure C-5. Examples of sensor responses as a function of concentration.
Measurement Duration
Measurement duration is the length of time over which a measurement is collected (e.g., 1
minute, an hour).
Why it matters: Shorter measurement times allow you to see more rapidly changing
concentrations. The minimum measurement duration depends on the sensor response
time and other factors. There are situations in which you might want to average
measurements over longer time durations to:
• Improve the precision of measurements from less precise sensors, or
• Reduce the size of a data set to make it more manageable during processing. For
example, you might average 1-second data to 1-minute or 5-minute data if these
measurement durations will still give enough detail to meet your study objectives. It is
important to ensure that the measurement duration of your sensor is compatible with
your application. In order to capture variations in concentration by location, a sensor
on a mobile platform (e.g., walking) may require a shorter measurement duration
than a stationary sensor would require. It is very difficult to accurately collect data
from fast-moving vehicles; this is not really practical with current sensor technology.
Measurement Frequency
Measurement frequency describes the number of measurements collected per unit of time.
Why it matters: The measurement frequency will be dictated by your study objectives and will
affect the sample collection and precision aspects of your data quality objectives45.
• Sample Collection: measurement frequency refers both to how often you make
measurements (i.e., one hour per week) and how often measurements are made during
this time (i.e., one measurement per minute). This will affect how much data coverage
Data quality objectives describe when, where, how often, and to what precision you need samples to be taken to
answer the scientific question you are interested in.
• • •
60
-------
Air Sensor Guidebook Appendix C
you have to describe the problem or process you are looking at. If you intend to evaluate
a long-term trend in concentrations at a particular location, you may choose to collect
measurements every five minutes for an hour, on a different day each week for a year.
On the other hand, if you would like to evaluate how concentrations change over the
course of a day next to a source location, you may want to collect measurements once
every minute for 24 hours over several consecutive days.
• Precision: the more frequently data are collected over a given time period, the more the
data's precision increases, because there are more data to cancel out random errors in
the measurements. However, there is a point at which collecting data more frequently
produces diminishing returns on improving precision and instead gives you too much
data to manage. Imagine collecting 60 data points each hour for 24 hours for one week.
That means reviewing and analyzing more than 10,000 pollutant concentration
measurements!
Data Grouping
Data grouping involves averaging data over time and/or space.
Why it matters: Data are often grouped to facilitate comparison to measurements from another
instrument, health-based benchmarks, or environmental standards. For example, the
NAAQS listed in Table 2-2 represent limits on concentrations that have been grouped
over a range from 1 hour to 1 year. The NAAQS for ozone is 0.075 ppm averaged over
eight hours.
Data grouping helps improve the quality, usefulness, and manageability of your data.
The exact type of grouping will depend on your application and the question you are
trying to answer. For example, if you are interested in observing a pollutant
concentration trend over the course of a month, you may want to group your data in 1-
hour or 24-hour averages. You will be able to see how the concentrations change, but
averaging will reduce the amount of data you are working with to a manageable size. On
the other hand, if you would like to understand how a plume of gas coming from an
industrial facility moves over your community, you may prefer to use a shorter averaging
period, such as 1-minute, to capture it's movement.
Selectivity
The ability of a sensor to respond to a particular pollutant, and not to other pollutants, is called
selectivity.
Why it matters: Sensors are most useful when they only respond to a single pollutant or several
pollutants of interest. However, air is composed of a wide variety of chemical
compounds, and some sensors may respond simultaneously to pollutants of interest as
well as other substances in the air. For example, some air quality sensors that measure
ozone also respond to changes in nitrogen oxides and sulfur dioxide concentrations,
providing a deceptively high signal. On high quality instrumentation, the manufacturers
have developed techniques for eliminating or reducing such concerns.
• • •
61
-------
Air Sensor Guidebook Appendix C
Interferences
Interferences are factors that hinder, obstruct, or impede the ability of a sensor to make
accurate measurements.
Why it matters: As mentioned under the Selectivity consideration, an ideal sensor would only
respond to the pollutant or pollutants of interest. However, sensors may respond
significantly to other pollutants in a way that is indistinguishable from the response to the
target pollutant. Specifically, sensor readings may be affected by:
• pollutants or other chemical compounds that are not of interest
• weather conditions (e.g., fluctuations in wind speed, humidity, and temperature)
• radio frequencies
• power fluctuations
• vibration
• dirt, dust, and insects
Interferences may have a positive or negative effect on a sensor signal. Also, it is
possible for a sensor to respond to several different interferences simultaneously.
Manufacturers usually disclose pollutants and meteorological parameters that may
impact sensor performance but not the response factor, which would be useful to
determine the importance of the interference. Before using a sensor to monitor air
quality, consider possible sensor interferences, test for them, and minimize them if
possible.
Sensor Decay and Expiration
Sensor decay and expiration refer to a permanent decline in sensor performance due to any
number of factors. In general it means the sensor loses its ability to take meaningful
measurements.
Why it matters: Some chemical compounds in the atmosphere can react with and damage
sensors in a non-reversible way, limiting the ability of the sensor to respond as well to
the pollutant of interest as it did initially. Note that some sensors have an expiration date,
even if they are never used and are in their original packaging.
Dynamic Range
An instrument's dynamic range is the concentration range from minimum to maximum values
that the instrument is capable of measuring46.
Why it matters: Concentration ranges vary by pollutant and by proximity of the sensor to the
source. In some cases, concentrations may be either too low or too high for a sensor to
detect. It is important to consider the range of concentrations you expect to monitor and
whether your sensor will be able to collect measurements throughout this range. Typical
ambient concentration ranges for each pollutant discussed in this document are provided
in Table 2-2.
46 http://nepis.epa.qov/Exe/ZyPURL.cqi?Dockev=20001 QWV.txt
• • •
62
-------
Air Sensor Guidebook
Appendix C
Drift
A gradual change in instrument response to a constant, quantitative characteristic (i.e., a
standard concentration or zero air) is called drift.
Why it matters: Instrument drift may lead a user to inaccurately conclude that concentrations
have increased or decreased over time. Drift can be positive or negative, and it may
occur due to a variety of reasons. For example, the sensor may respond to changes in
weather conditions, to sensor poisoning, or, in the case of optical sensors, to light
sources becoming less powerful or less efficient over time. Figure C-6 shows an
example of an NO2 measurement that drifted. One way to overcome drift is to calibrate
the sensor frequently so that the instrument only drifts a small amount between each
recalibration. The frequency of calibration needed will depend on how much drift occurs.
Figure C-6. graphically illustrates drift using a time series of measurements from a sensor (blue
line) experiencing significant drift compared with the true concentration of NO2 (green
line).
30 -r
-5
3/1/20130:00 3/2/20130:00 3/3/20130:00 3/4/20130:00 3/5/20130:00 3/6/20130:00 3/7/20130:00
True NO2 concentration
-Measured NO2 concentration
Figure C-6. Illustration of Drift.
Accuracy of Timekeeping
Timestamp accuracy describes the correctness and reliability of the time value recorded as
each measurement is collected.
Why it matters: Time keeping accuracy is most important when you need to compare
measurements made by different instruments. This type of accuracy becomes more
• • •
63
-------
Air Sensor Guidebook Appendix C
critical for comparing data showing large, rapid changes in concentration or data from
instruments with high measurement frequencies.
Climate Susceptibility
Climate susceptibility is a measure of an instrument's ability to endure variation in
meteorological conditions, including changes in temperature, humidity, and sun exposure.
Why it matters: Air quality instruments are expected to operate in a wide range of atmospheric
conditions. A sensor is most useful if it can operate robustly in many different
environments, but it needs to operate well in the intended use environment at the very
least. It is important to consider which sensor is best suited for the climate of your study
location. For example, relative humidity and temperature influence the performance of
electrochemical sensors. Consider whether the instrument enclosure would benefit from
being air-conditioned, or whether environmental effects on the measurements can be
corrected after data collection.
Learn more: Michel Gerboles and Daniela Buzica (2009) found that ozone sensors calibrated in
a laboratory reactor suffered from temperature and relative humidity effects, although
these problems were removed when calibrations were done at the field sites using
ambient air. See Evaluation of Micro-Sensors to Monitor Ozone in Ambient Air, 2009.
Michel Gerboles and Daniela Buzica, Joint Research Centre, Institute for Environment
and Sustainability, Transport and Air Quality Unit, Via E. Fermi, I - 21027 Ispra (VA)
http://publications.irc.ec.europa.eu/repository/bitstream/111111111/10477/17eur23676.
pdf.
Data Completeness
The amount of valid data obtained from a measurement system, compared to the amount that
was expected to be obtained under correct, normal conditions, is called data completeness47.
Why it matters: Data completeness is a key to producing high-quality, representative data.
Missing data can significantly hinder analyses, minimizing the strength of conclusions
drawn. EPA's guidance for regulatory data includes a requirement to achieve 75% data
completeness over the required period of time (hourly, daily, quarterly, annually).
Commonly, reductions in data completeness are due to data transmission problems;
data storage errors; power loss and the time required for subsequent restart (see
Response to Loss of Power discussion below); the need for frequent or long-duration
calibrations; and time the instrument is offline for repair. For data transmission, if data
will be transferred using a wireless connection, the reliability of the connection is very
important. Onsite data storage may also be considered so that data are not lost if the
wireless connection is interrupted.
Learn more: http://www.epa.gov/ttn/oarpa./t 1/memoranda/pmfinal.pdf; 40 CFR 50 contains
Appendix A-N (NAAQS data completeness)
47 http://www.epa.qov/ttn/amtic/files/ambient/monitorstrat/precursor/07workshopmeaninq.pdf.
• • •
64
-------
Air Sensor Guidebook Appendix C
Response to Loss of Power
This refers to the amount of time that an instrument requires after shutdown to warm up and
resume measurement, as well as the consistency of the sensor response prior to and after
shutdown.
Why it matters: If a sensor requires a large amount of time to warm up and resume
measurement after a loss of power, data continuity and completeness can be
significantly affected. Once the sensor resumes collecting measurements, its response
should ideally be the same as before the loss of power.
• • •
65
-------
&EPA
PRESORTED STANDARD
POSTAGE & FEES PAID
United States
Environmental Protection EPA
Agency
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |