EPA Handbook: Optical and Remote Sensing
for Measurement and Monitoring of Emissions
Flux of Gases and Particulate Matter
-------
EPA 454/B-18-008
August 2018
EPA Handbook: Optical and Remote Sensing for Measurement and Monitoring of Emissions
Flux of Gases and Particulate Matter
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC
-------
EPA Handbook: Optical and
Remote Sensing for
Measurement and Monitoring
of Emissions Flux of Gases and
Particulate Matter
Informational Document
9/1/2018
This informational document describes the emerging technologies
that can measure and/or identify pollutants using state of the science
techniques
-------
Forward
Optical Remote Sensing (ORS) technologies have been available since the late 1980s. In the
early days of this technology, there were many who saw the potential of these new instruments
for environmental measurements and how this technology could be integrated into emissions
and ambient air monitoring for the measurement of flux. However, the monitoring community
did not embrace ORS as quickly as anticipated. Several factors contributing to delayed ORS use
were:
Cost: The cost of these instruments made it prohibitive to purchase, operate and
maintain.
Utility: Since these instruments were perceived as "black boxes." Many instrument
specialists were wary of how they worked and how the instruments generated the
values.
Ease of use: Many of the early instruments required a well-trained spectroscopist who
would have to spend a large amount of time to setup, operate, collect, validate and
verify the data.
Data Utilization: Results from path integrated units were different from point source
data which presented challenges for data use and interpretation.
Over the years, the air monitoring community has come to accept both the challenges and
overall utility of ORS technologies and applications. The emissions monitoring community and
monitored sources have been employing ORS for several years and are using these technologies
to answer questions that traditional instrumentation could not address. In addition, ORS
technology has been applied to ambient and fenceline monitoring, including near-roadway
monitoring. Therefore, application of ORS technology has an expanding place with other air
measurement tools.
-------
The EPA staff and other scientists and engineers in the monitoring community recognized that a
compilation of ORS material was needed to encourage wider use and understanding of ORS.
Questions on how instruments generate data and how an agency or source validates and verifies
data are universal, whether the instrument is optically remote or an extractive instrument on
site or within the stack. With this in mind, the EPA developed this Handbook to assist the "non-
spectroscopist" in understanding and using data and information generated by ORS. This
Handbook is divided into five sections:
Section 1: Discusses what ORS means and how this technology can be used. It also has
several tables that have a "crosswalk" between the different technologies and their use
(i.e., techniques).
Section 2: Describes the different technologies or "hardware" that are currently
available that are considered "optically remote."
Section 3: Explains how to use the "hardware" with different techniques and how to
calculate emission flux.
Section 4: Discusses the "other" data that needs to be collected to understand and
better validate and verify the ORS data.
Section 5: Provides a very brief overview of how to validate and verify this data once it
is collected.
-------
Disclaimer of Endorsement
Mention of, or referral to, commercial products or services and/or links to non-EPA internet sites
does not imply official EPA endorsement of, or responsibility for opinions, ideas, data, or
products presented at those locations, or guarantee the validity of the information provided.
Mention of commercial products/services and non-EPA websites is provided solely as a
reference to information on topics related to environmental protection that may be useful to
EPA staff and the public.
-------
Acknowledgement
A document such as this requires the work and dedication of many people. This section
acknowledges those that have provided their time and effort to create this document.
Team Lead: Dennis K. Mikel, EPA-OAQPS
Reviewers:
David Nash, EPA-OAQPS
Raymond Merrill EPA-OAQPS
Jason Dewees, EPA-OAQPS
Comments and questions can be directed to:
Steffan Johnson EPA-OAQPS
109 Alexander Drive
Research Triangle Park, NC 27711 Email: johnson.steffan@epa.gov
-------
Table of Contents
1.0 Introduction 1
1.1 Purpose of the Handbook 1
1.2 Contents and Overview of the Handbook 3
1.3 Stationary Sources and Emissions Points 6
Ducted or Vented Emissions 7
Area or Fugitive Emissions Sources 8
1.4 Why Remote Measurement? 9
Criteria Pollutant Gases, HAPs and GHGs 11
1.5 Knowledge and Advancement of Remote Sensing to Emissions Measurement 15
Active 15
Passive 16
Backscatter 17
Mobile 18
Advantages Over Closed Path Techniques 18
1.6 General Discussion of the EPA Quality System 20
Data Quality Objectives 21
Measurement Quality Objectives 22
1.7 Choosing the Right Tool for the Right Job 24
What is the nature of the source? 24
What is the required time resolution? 25
What are the QA/QC requirements for the measurement? 26
How easy are the data to process? 27
1.8 Future Evolution and Updates of this Handbook 28
1.9 References 28
2.0 Optical Remote Sensing Technologies 1
2.1 FourierTransform Infrared Spectroscopy 2
Basic Operation 2
Extractive (Closed) Cell Measurement Applications 5
Open-Path Measurement Applications 6
Pollutants and Relative Levels That Can Be Measured 8
Typical QA/QC 11
Calibration Spectra 11
QA/QC for OP-FTIR Instrumentation 13
Data Quality Indicators for Precision and Accuracy for OP-FTIR 14
Example Applications and Vendors 14
Strengths and Limitations 16
References 19
2.2 Tunable Diode Laser 21
Basic Operation 22
Pollutants and Relative Levels That Can Be Measured 26
Typical QA/QC 27
Selection of the Laser and Absorption Line 27
Calibration 29
Quality Control Procedures 30
-------
Example Applications and Vendors Applications 30
Vendors 31
Strengths and Limitations 31
References 33
2.3 Ultraviolet Differential Optical Absorption Spectroscopy 35
Basic Operation 35
UV-DOAS Field Implementation 38
Passive UV-DOAS 40
Pollutants and Relative Levels That Can Be Measured 40
Typical QA/QC 41
Record Keeping 42
Instrument Performance 42
Example Applications and Vendors 43
Vendors 44
Strengths and Limitations 44
References 46
2.4. Differential Absorption Light Detection and Ranging Systems 47
Basic Operation 48
Pollutants and Relative Levels That Can Be Measured 54
Typical QA/QC 55
Record Keeping 56
Instrument Performance 56
Example Applications and Vendors Applications 56
Vendors 58
Strengths and Limitations 58
References 60
2.5 Thermal Infrared Cameras 62
Basic Operation 63
Pollutants and Relative Levels That Can Be Detected 69
Typical QA/QC 70
Example Applications and Vendors 74
Vendors 75
Strengths and Limitations 75
References 76
2.6 Cavity Ring-Down Spectroscopy 79
Basic Operation 79
General Experimental Design 81
Current Method Developments 85
Pollutants and Relative Levels That Can Be Measured 86
Typical QA/QC 88
Example Applications and Vendors Applications 89
Vendors 90
Strengths and Limitations 90
References 92
2.7 Particulate Matter LIDAR 94
Basic Operation 95
Pollutants and Relative Levels That Can Be Detected 101
Typical QA/QC 102
-------
Example Applications and Vendors 102
Vendors 102
Strengths and Limitations 104
References 104
3.0 Measurements Applicable to Emissions Flux 1
Reference 2
3.1 Radial Plume Mapping: Other Test Method 10 3
General Description of Approach 3
Horizontal RPM Algorithm 4
Vertical RPM algorithm 6
One-dimensional RPM algorithm 9
RPM-ORS Technologies 11
Verification/Validation Studies 12
OP-FTIR and OP-TDLAS comparison studies 12
OP-FTIR and UV-DOAS comparison: Colorado Springs field study 13
VRPM plume capture validation study 13
Typical QA/QC 14
Siting Concerns 15
Strengths and Limitations 16
References 17
3.2 Range Resolved Measurements using Differential Absorption LIDAR 19
General Description of Approach 19
Basic DIAL algorithm to calculate backscatter 20
Verification/Validation Studies 22
Verification of DIAL for Gas Species Measurements 22
Typical QA/QC 24
Siting Concerns 25
Strengths and Limitations 25
References 26
3.3 Solar Occultation Flux Measurement 28
General Description of Approach 29
Verification/Validation Studies 34
Typical QA/QC 37
Siting Concerns 39
Strengths and Limitations 39
References 40
3.4 Tracer Gas Correlation 42
General Description of Approach 42
Verification/Validation Studies 45
Typical QA/QC 48
Sitting Concerns 48
Strength and Limitations 49
References 50
3.5 Backward Lagrangian Stochastic Inverse-Dispersion model 51
General Description of Approach 51
Backward LS Dispersion Model for Calculating (C/Qsim) 53
bLS Model Output Units 55
-------
Verification/Validation Study 55
Flesch et al Conditions and Setup 56
Flesch et al Results and Conclusions 58
Typical QA/QC 59
Siting Concerns 60
Strengths and Limitations 61
References 63
3.6 Geospatial Measurement of Air Pollution, Remote Emission Quantification - Other
Test Method 33 64
General Description of Approach 66
Verification/Validation Studies 69
Typical QA/QC 69
Siting Concerns 70
Strength and Limitations 71
References 73
3.7 Geospatial Measurement of Air Pollution, Remote Emission Quantification: Direct
Assessment - Other Test Method 33A 75
General Description of Approach 76
Verification/Validation Studies 82
Typical QA/QC 91
Siting Concerns 93
Strengths and Limitations 95
References 96
3.8 Hyperspectral Imaging 98
General Description of Approach 99
Operating Principles 100
Example Instruments 102
Physical Sciences' AIRIS 103
Rebellion Photonics' Gas Cloud Imager 104
Telops' HyperCam and MS-IR 106
Additional Considerations 108
Verification/Validation Studies 110
Typical QA/QC 113
Summary of Thermal IR Camera QA/QC 114
Siting Concerns 115
Strengths and Limitations 115
References 116
3.9 Fenceline Passive Sampling - Method 325 A/B 118
General Description of Approach 118
Passive Tube Samplers 119
Sampler Deployment 122
Passive Sampler Recovery 123
Meteorological Data Collection 124
Passive Sampler Analysis 124
Recordkeeping and Data Analysis 126
Verification/Validation Studies 126
Flint Hills West Refinery 126
South Philadelphia PS and Sensor Study 128
-------
Typical QA/QC 129
Field Sampling QA/QC 129
Laboratory QA/QC 130
Siting Concerns 131
Strength and Limitations 132
References 133
3.10 Method to Quantify Particulate Matter Emissions from Windblown Dust - Other
Test Method 30 134
General Description of Approach 134
Instrumentation 136
Dispersion Modeling and K-factors 139
Sample Collection 140
Owens (dry) Lake, Inyo County California 141
Mono Lake, California 141
Typical QA/QC 142
Siting Concerns 143
Strengths and Limitations 144
References 144
3.11 Determination of Emissions from Open Sources by Plume Profiling - Other Test
Method 32 146
General Description of Approach 146
Sampling Equipment 147
Sample Deployment 149
Sample Analysis 151
Verification/Validation Studies 151
United Taconite and U.S. Steel Minntac 151
Typical QA/QC 153
Quality Control Samples 153
Quality Assurance 153
Precision 153
Siting Concerns 154
Strengths and Limitations 155
References 156
3.12 Method to Quantify Road Dust Particulate Matter from Paved and Unpaved Roads -
Other Test Method 34 157
General Description of Approach 157
Instrumentation 158
Calibration 161
Data Collection 162
Data Analysis 162
Verification/Validation Studies 163
Clark County, Nevada 163
Typical QA/QC 165
Screening Criteria 165
Collocation 165
Siting Concerns 166
Strengths and Limitations 166
References 167
-------
4.0 Meteorological Measurements 1
4.1 Meteorological Station Siting 2
Horizontal Wind Speed and Direction 3
Vertical Wind Speed and Lateral Turbulence 4
Relative Humidity 4
Temperature 4
Net Solar Radiation 5
Atmospheric Pressure 5
Differential Global Positioning for Tracking Monitoring Locations 5
Collection of Process Information 6
Attribution of Emissions to Source of Intent 9
4.2 References 9
5.0 Data Validation and Verification 1
5.1 General Approach 2
5.2 Data Validation Methods 2
Levels of Data Quality Review 3
5.3 Data Verification Methods 4
Data Review and Verification 8
Validation of Primary and Ancillary Measurements 9
Final Validation and Evaluation of Measurements for Data Users 11
5.4 References 12
-------
Figures
Number Title Section
1-1 Types and sources of air pollutants 1.3
1-2 Example of Ducted Stationary Source Stack 1.3
1-3 Example of Potential Fugitive Source 1.5
1-4 Example of Potential Fugitive Sources 1.5
1-5 Block Diagram of bi-static ORS 1.5
2-1 Diagram Showing Beam Path and Major Components of FTIR 2.1
2-2 FTIR Absorption Spectrum Recorded at 1075ฐ K 2.1
2-3 FTIR Closed Cell Unit Used to Monitor Stack Gas 2.1
2"4 Basic Setup Used to Make Monostatic Open-path FTIR
Measurements 2.1
2-5 The Corner-cube Reflector 2.1
2-6 Typical Telescopic FTIR Transmitting and Detection Unit 2.1
Corner Cube Reflector 2.1
2-8 Calibration Plot of Absorbance vs. Concentration 2.1
2-9 TDL Bistatic Configuration 2.2
2-10 TDL Monostatic Configuration 2.2
2-11 The Corner-cube Reflector 2.2
2-12 Calibration Data for an OP-TDL System 2.2
2-13 Opsis DOAS Unit 2.3
2-14 Bistatic Configuration of UV-DOAS 2.3
2-15 The Corner-cube Reflector 2.3
Basic Setup Used to Make Mono-static Open-path UV- DOAS
Measurements 2.3
2-17 Beam Path and Major Components of DIAL Unit 2.4
2-18 Monostatic Coaxial and Biaxial Configuration for DIAL 2.4
2-19 Bistatic Configuration for DIAL 2.4
2-20 Illustration of DIAL unit mapping an emission plume 2.4
2-21 Mobile DIAL Unit 2.4
-------
Figures
Number Title Section
2-22 Overview of Thermal IR Camera Technology Basics 2.5
2"23 Image of a Controlled Gas Release where the Gas is Warmer than
the Background 2.5
2-24 IR Spectrum for Propane with the Molecule Bond Structure 2.5
2-25 Spectral Curves for an IR Camera Window of Transmittance 2.5
2-26 Calibration/Verification Examples Configuration 2.5
2-27 Example of Quality Control Daily Operations Check Chart 2.5
2-28 Anticipated Change in Pixel Intensity for various concentrations 2.5
2-29 Flare Detection by Thermal IR Camera 2.5
2-30 The essential components of any CDRS experimental set-up 2.6
2-31 Schematic representation of the expected rate of decay 2.6
2-32 Comparison of Pulsed and Continuous Wave Laser Light 2.6
2-33 Optical components schematic of a cavity ring-down spectrometer 2.6
2-34 Differences PMT and charge couples devices 2.6
2-35 Simplified Schematic of LIDAR system 2.7
2-36 Illustration of LIDAR System Geometry 2.7
2-37 Illustration of Monostatic Coaxial and Biaxial LIDAR 2.7
2-38 Illustration of the USU AGLITE LIDAR System 2.7
2-39 Illustration of the USU CELis LIDAR System 2.7
3-1 Horizontal RPM setup 3.1
3-2 Vertical RPM setup 3.1
3-3 One-dimensional RPM setup 3.1
3-4 Examples of RPM algorithm outputs. 3.1
3.5 Conceptual picture on the operation of DIAL. 3 2
3-6 Contour profile of S02 concentration measured 2.1 km downwind of 3.2
3.7 Solar tracker configurations 3 3
3-8 Rough overview of a mobile SOF system 3.3
3-9 SOF cartoon 3.3
3-10 An example of path-integrated calculations determined from
SOF measurements 3.3
-------
Figures
Number Title Section
3-H Tracer gas/SOF experiment measuring SF emissions in an open
field over time of day 3.3
3-12 Wind velocity profiles by height 3.3
3-13 Tracer Gas release setup cartoon 3.4
3-14 Methane concentration verses acetylene concentration
Tracer gas characterization using FTIR 3.4
3-15 Illustration of an inverse-dispersion model for estimating
emission rate Q 3.5
3-16 Illustration of the WindTrax bLS Modeling Software
Graphical Interface 3.5
3-17 Map of the laser paths used in the 2004 Flesch et al
experiment 3.5
3-18 GMAP Operational Regime 3.6
3-19 GMAP (OTM-33) Limitations 3.6
3-20 GMAP (OTM-33) Overlapping Plume Sources 3.6
3-21 GMAP (OTM-33) Limitations 3.7
3-22 GMAP-REQ-DA (OTM 33A) Concentration Mapping Survey 3.7
3-23 An Illustration of a Stationary EQ Observation 3.7
3-24 Effect of varying atmospheric conditions 3.7
3-25 Results from controlled release experiments 3.7
3-26 Comparison of methane measurements by basin 3.7
3-27 Density and cumulative density of methane emission measurements 3.7
3-28 Comparison of onsite measurements and remote measurements 3.7
3.29 Source emission strength calculations for each controlled release 3 7
3-30 Source strength calculation averages by controlled release study 3.7
3-31 Example of Differences between Multi-, Hyper-and Ultraspectral 3.8
3-32 Radiative Transfer Model Illustration for IR Remote Sensing 3.8
3-33 Multispectral Imaging using Spectral Differentiation of Methane Gas 3.8
3-34 AIRIS Interferometer Design 3.8
3.35 Illustration of the IMS Optical Layout. 3 8
3-36 Example Datacube with Waveband Elements 3.8
-------
Figures
Number Title Section
3-37 Telops HyperCam (Left) Hyperspectral Imager and MS-IR Infrared 3.8
3-38 Example Telops Software Output 3.8
3-39 Detector Methods for DataCube Acquisition. 3.8
3-40 Snapshot Detection with FPA Divided into smaller collections 3.8
3-41 High/Low Controlled Gas Release Platform 3.8
3-42 ERG Controlled Leak Simulation Platform 3.8
3"43 Example Releases as Seen from the GCI 3.8
3-44 Cross-Section of the PS Tube 3.9
3-45 Sorbent Tube Protection Cover 3.9
3-46 Saltation and Dust Production Process for Windblown Dust 3.10
3-47 Schematic of CSC Placement for Sampling. 3.10
3-48 Cut-out of a CSC and Construction Specifications 3.10
3-49 CSC Placement in the Field with a Height Adjustment Tool 3.10
3-50 BSNE Sampler Shown Sampling in the Field 3.10
3-51 Monitoring Network at Mono Lake, CA. 3.10
3-^2 Traditional PM Sampler Configuration 3.11
3-^3 Illustration of Fixed Point Source Sampling Array 3.11
3-54 Illustration of Moving Point Source-Unpaved Road Dust 3.11
3-55 Example Sampling Array for Moving Point Source 3.11
3-56 Deployment of Collocated Plume Towers at Roadside Location 3.11
3-57 Sample inlet installation behind the front tire 3.12
3-58 Sample inlet installation on a trailer pulled behind test vehicle 3.11
3-59 Coefficient of Variation determined from data in Clark County 3.11
5-1 Generalized Data Verification and Validation Process Flow 5.1
5-2 Level 0 Verification Checks the Most Fundamental Quality
Requirements 5.2
5-3 Example of Optical Remote Measurement Visual Check List 5-2
5-4 Level 1 Verification Ensures Quality Requirements are met in
the Field 5.2
5-5 Level 2 Quality Checks Start the Data Validation Process 5.3
5-6 Level 3 Quality Checks Ensure the Data is Usable for the
Purpose Intended 5.3
-------
Tables
Number Title Section
1-1 Gas Phase Source Type 1.7
1-2 Particle Phase Source Type 1.7
1-3 Gas Phase Time Resolution Matrix 1.7
1-4 Particle Phase Time Resolution Matrix 1.7
Gas Phase QA/QC Matrix 1-7
1-6 Particle Phase QA/QC Matrix \j
1-7 Gas Phase Ease of Operation Matrix 1.7
1-8 Particle Phase Ease of Operation Matrix 1.7
2-1 Example List of Compounds Measured by FTIR Open-path Systems 2.1
2-2 Typical Applications for OP-FTIR. 2.1
2-3 FTIR Supply Vendors 2.1
2-4 Summary Table of the OP-FTIR's Strengths 2.1
2-5 Summary Table of the OP-FTIR's Limitations 2.1
2-6 Example List of Gaseous Compounds Measured by Near IR
OP-TDL Systems 2.2
2-7 Near-IR Laser Types Available for OP-TDL Systems 2.2
2-8 Potentially Usable Mid-IR Lasers 2.2
2-9 Typical Applications for OP-TDL 2.2
2-10 Near-IR OP-TDL Vendors 2.2
2-11 SummaryTable of the TDL's Strengths 2.2
2-12 SummaryTable of the TDL's Limitations 2.2
2-13 Species Measured with UV-DOAS Systems 2.3
2-14 Approximate Detection Limits for UV-DOAS 2.3
2-15 Typical Applications for UV-DOAS 2.3
2-16 UV-DOAS Vendors 2.3
2-17 Summary Table of the UV-DOAS's Strengths 2.3
2-18 Summary Table of the UV-DOAS's Limitations 2.3
2-19 Reported Species Measured with DIAL Systems 2.4
2-20 Approximate Detection Limits for DIAL 2.4
-------
Tables
Number Title Section
2-21 Typical Applications for LIDAR 2.4
2-22 DIAL Vendors 2.4
2-23 SummaryTable of the DIAL's Strengths 2.4
2-24 SummaryTable of the DIAL's Limitations 2.4
2-25 Example List of Gaseous Compounds that can be Detected 2.5
2-26 Typical Applications for Thermal IR Camera 2.5
2-27 Thermal IR Camera Vendors 2.5
2-28 SummaryTable of the IR Camera's Strengths 2.5
2-29 SummaryTable of the IR Camera's Limitations 2.5
2-30 Example list of detectable pollutants by CRDS 2.6
2-31 Typical Applications for CRDS 2.6
2-32 CRDS Vendors 2.6
2-33 Summary Table of CRDS Strengths 2.6
2-34 Summary Table of CRDS Limitations 2.6
2-35 Typical Applications for LIDAR Systems 2.7
2-36 LIDAR Systems Vendors 2.7
2-37 Table of LIDAR Strengths 2.7
2-38 Table of LIDAR Limitations 2.7
3-1 Data quality indicators for the QA/QC process 3.1
3-2 SummaryTable of the VPRM's Strengths 3.1
3-3 SummaryTable of the VPRM's Limitations 3.1
Results from the comparison of DIAL and plant 3.2
measurements ofSC>2 mass emissions
3-5 DIAL Strengths 3.2
3-6 DIAL Limitations 3.2
3-7 SOF technique VOC emissions compared to DIAL 3.3
3-8 Summary for measurements on the Aby field, 2002 3.3
3-9 SOF traverse done on day 24-June 2002 3.3
3-10 Estimation of statistical errors for the SOF measurements 3.3
3-11 Feature strengths of using the SOF method 3.3
-------
Tables
Number
Title
Section
3-12
Feature limitations of using the SOF method
3.3
3-13
Tracer Gas Correlation Strengths
3.4
3-14
Tracer Gas Correlation Limitations
3.4
3-15
bLS Model Strengths
3.5
3-16
bLS Model Limitations
3.5
3-17
Summary of mobile measurement approaches
3.6
3-18
Strengths of the General OTM 33 Approach
3.6
3-19
Limitations of the General OTM 33 Approach
3.6
3-20
Summary of GMAP-REQ-DA Studies
3.7
3-21
Pearson Correlation Coefficients of Emission and Production
3.7
3-22
Strengths of the OTM 33 A Approach
3.7
3-23
Limitations of the OTM 33A Approach
3.7
3-24
Detection Method by Instrument.
3.8
3-25
Summary of Rebellion GCI EPA Controlled Release Results
3.8
3-26
Strengths and Limitations of Hyperspectral Imaging
3.8
3-27
Passive Sampler Manufacturers
3.9
3-28
Pollutants that can be Collected with Passive Diffusive Samplers
3.9
3-29
GC/MSTuning Criteria
3.9
3-30
Analytical QA/QC Procedures for EPA Method 325B
3.9
3-31
Strengths and Limitations of EPA Method 325A/B
3.9
3-32
Strengths and Limitations of the OTM 30 Approach
3.10
3-33
Strengths and Limitations of the OTM 32 Approach
3.11
3-34
Strengths and Limitations of the OTM 34 Approach
3.12
-------
List of Acronyms
Acronym Description
Hg/m3 micrograms per cubic meter
pirn micrometer
2-D two dimensional
3-D three dimensional
ANSI American National Standards Institute
AOM acousto-optic modulator
ASQC American Society for Quality
bLS Backward Lagrangian Stochastic
BTX benzene, toluene, elemental mercury and p-xylene
C2H2 acetylene
CCD charge-coupled device
CEA cavity-enhanced absorption
CEMs continuous emissions monitors
CFCs chlorofluorocarbons
CH4 methane
CO carbon monoxide
C02 carbon dioxide
COSPEC correlation spectrometer
CRDS cavity ring-down spectroscopy
CRLAS cavity ring-down laser absorption spectroscopy
CW continuous wave
DGPS differential global positioning system
DIAL Differential absorption LIDAR
DIAL/LIDAR differential absorption light detection and ranging
DOAS differential optical absorption
DQI data quality indicators
DQO data quality objective
EPA Environmental Protection Agency
ETV environmental technology verification
FID flame ionization detectors
FTIR Fourier transform infrared
g/s grams per second
GPS global positioning system
HAP hazardous air pollutants
HITRAN high-resolution transmission molecular absorption
database
HRPM horizontal RPM
ICOS integrated cavity output spectroscopy
IR Infrared
LASER light amplification by stimulated emission of radiation
MAXDOAS Multi-axis Differential Absorption Spectroscopy
-------
List of Acronyms
Acronym Description
SBFM
smooth basis function minimization
SF6
sulfur hexafluoride
S02
sulfur dioxide
SOF
Solar Occultation Flux
SOx
sulfur oxides
SSE
sum of squared errors
TDC
tracer dilution correlation
TDL
tunable diode laser
TDLAS
tunable diode laser absorption spectroscopy
UV
ultraviolet
VOCs
volatile organic compounds
VRPM
vertical radial plume mapping
US
Microseconds
-------
ORS Handbook
Section 1.0
Page 1-1
1.0 Introduction
This document is intended as an introductory handbook for those planning to use or review remote
emissions measurement and monitoring approaches for emission sources or for data users building
their expertise about current information concerning the technologies and application in these
types of measurements. For the purposes of this handbook, "remote measurement" is defined as
any measurement of air emissions conducted away from the point or area where the pollutant is
released. This definition includes optical remote sensing (ORS), as well as other approaches such as
those coupling point measurements with a mobile measurement platform. As the nation's air
quality management programs evolve, we need more measurements of non-point or unvented
sources, often referred to as fugitive sources or fugitive emissions. Remote measurement
technologies offer approaches that have been otherwise unavailable to measure emissions from
these challenging sources.
The information presented in this document is written to be generally informative, as well as more
"user friendly" than technical papers or review articles found in the open literature (i.e., peer
reviewed literature and articles in periodicals that are available on the internet). Practical
information is provided for those who need to understand the principles behind the use of
spectroscopy or other remote measurement technologies, but who may not be trained in these
technologies and their applications. This document is intended to aid readers in understanding the
uses and limitations of data generated by remote measurement approaches. In this document, you
will find discussion of the practical uses and operation of remote sensing equipment and
applications of these and other technologies to produce emissions data. Some of the complex
technical information has been provided in summary form with illustrations.
1.1 Purpose of the Handbook
The purpose of this handbook is to describe the primary remote measurement technologies and
current approaches to use these technologies. This handbook also describes how potential users
can assess the applicability of remote measurements and the resulting data to their emissions
measurement needs. We designed this handbook for EPA, state, local, and tribal measurement
-------
ORS Handbook
Section 1.0
Page 1-2
project leads, measurement contractors, industry managers planning measurements to create
emission factors, and those reviewing test plans and test reports. When the term "measurement" is
used in this handbook, it is referring to short-term studies (e.g., emission fluxes assessment). The
term monitoring is used for long-term studies (i.e. spatial and temporal trend assessment).
Optical remote measurement techniques are most typically designed and used to measure
concentrations and, when paired with meteorological data, allow calculation of mass fluxes of
pollutants downwind of fugitive and non-point emission sources. Optical remote techniques
provide opportunities to measure sources that are not conducive to measurement using more
traditional stack testing or single point ambient techniques. Actual application, however, needs to
be determined on a case-by-case basis.
This handbook describes the more prevalent and technologically demonstrated open-path, cell-
based, and point measurement technologies used to make remote measurements and it provides a
background for the application of remote measurement techniques for emissions measurements.
Viable applications for qualitative and quantitative measurements of constituents in air are also
described as examples of different ways remote measurement technologies can be applied to
meet measurement and monitoring requirements. Quantitative emissions data from remote
measurements may then be used for multiple purposes including possible development of
emission factors, evaluation of exposure levels, compliance with ambient regulatory limits, and
identification of sources of air pollution. Examples of several pollutant detection and quantification
methods are provided to show the focus of current monitoring applications. Applications of ORS
are relatively new, but maturing rapidly. For example, Differential Optical Absorption (DOAS) and
Fourier Transform Infrared (FTIR) systems have been commercially available since the early 1990s.
These earlier instruments, although designed for both background ambient and higher stationary
source emission-related monitoring applications, have mostly been employed to measure
stationary and fugitive source emissions. Some of the more technical information have been
simplified, and illustrations have been updated and clarified to make them more understandable.
Internet links and references have been added throughout the document to allow the reader to
-------
ORS Handbook
Section 1.0
Page 1-3
quickly research more detailed information.
1.2 Contents and Overview of the Handbook
This handbook discusses remote measurement technologies, applications of those technologies,
ancillary data necessary to use the remote measurement data, potential issues with using remote
measurement data for emission factors development, models, and other atmospheric process
needs. Each chapter contains information that is split into two areas. The first area focuses only on
the technologies including the specific hardware, scientific principles, how the pollutant
concentrations are measured, pollutant and performance capabilities. The second area discusses
the current vendors of the instrument or technology, general strengths and limitations.
What's New?
This is the second iteration of this handbook, which was first posted in 2011. Since that first writing,
this handbook has several new chapters that appear in this edition. Below, is a list of the new
features and chapters that appear in this edition:
Decision tables that illustrate techniques and technology. These are based on ease of use,
cost and time frame and quality assurance concerns;
Chapter 2.5 has been rewritten to include Optical Gas Imaging. This section was limited to a
description of Thermal Infrared Camera technology;
Chapter 2.7 discusses ORS instrumentation that can measure PM, with size ranges to the
UFPs up to PMio.
Chapter 3.6 and 3.7 describes Other Test Methods (OTM) 33 and 33a which describe
Geospatial Measurements of Air Pollution, Remote Emission Quantifications;
Chapter 3.8, Hyperspectral Monitoring;
Chapter 3.9, Fenceline Passive Sampling - Method 325 A/B;
-------
ORS Handbook
Section 1.0
Page 1-4
Chapter 3.10, Method to Quantify Particulate Matter Emissions from Windblown Dust;
Chapter 3.11, Determination of Emissions from Open Source by Plume Profiling, and;
Chapter 3.12 Method to Quantify Road Dust Particulate Matter Emissions from Vehicular
Travel on Paved and Unpaved Roads.
Numerous figures and diagrams are scattered within these new sections that illustrate the
techniques and technology that are utilized.
Structure of the Handbook
This second edition addresses how the technologies can and are being used to measure and monitor
stationary source emissions including measuring mass emissions flux, monitoring emissions
concentrations, detecting fugitive emissions leaks and measuring PM from remote sources. The
handbook also includes examples of remote measurement projects and readily available test
reports.
Section 1.0 introduces the handbook including background information that is necessary to
understand the more detailed sections to follow. In this section you will also find a description of
the EPA Quality System (QS) and how it can be used to create a data collection system that gathers
data of sufficient quality for its intended use. The Measurement Quality Objectives (MQOs) in
Section 1.6 will be useful to organizations planning remote measurement programs. The tables will
help users to quickly review the requirements of a particular program.
Section 2.0 provides an overview of the remote measurement detection technologies that are
currently available for remotely measuring pollutant emissions concentrations. Included in this
overview are discussions of FTIR, Tunable Diode Laser (TDL), UV-DOAS, and LIDAR (both gas and PM)
spectroscopy technologies. In addition, qualitative ORS technologies including Thermal Infrared
Imaging are described. Each of these technology descriptions includes information on the basic
principles of operation, the pollutants that can be monitored, typical quality control (QC) and quality
-------
ORS Handbook
Section 1.0
Page 1-5
assurance (QA) for the technology, strengths, limitations, example vendors, and applications.
Section 3.0 describes the predominant remote measurement applications used to deploy the
detection technologies addressed on Section 2.0 and to quantify emissions concentrations and flux
measurements. This section also describes how the different technologies are applied to
measurement methods, which is an extremely important section in this handbook. The application
descriptions briefly summarize the activity and explain how the application is verified or validated in
field tests, and details typical QA/QC associated with the application. Section 3.0 also describes
siting considerations or information. Each application in Section 3.0 includes a table of strengths
and limitations that must be considered during the planning, implementation, and interpretation of
field study results. Applications covered in Section 3.0 include RPM using EPA Other Test Method 10
(OTM-10), Differential Absorption LIDAR (DIAL), Tracer Dilution Correlation (TDC), Solar Occultation
Flux (SOF), and bLS emissions modeling. Section 3.0 provides examples of how these applications
are used in fugitive emissions and area source emissions flux and concentration measurement, site
remediation, plant fenceline monitoring, fugitive leak detection, and ambient air measurement. In
addition, several methods have been added that quantify large area and mobile generated PM.
Section 4.0 presents the ancillary measurements and data that may be needed for each ORS
application. Ancillary data is defined as meteorological measurements, industrial process
information and source activity necessary to translate ORS results generated from the detection
technique and measurement application combinations described in Sections 2.0 and 3.0,
respectively, into emission data that meet project specific data quality objectives.
Section 5.0 addresses various methods to validate and verify remote measurement data.
-------
ORS Handbook
Section 1.0
Page 1-6
1.3 Stationary Sources and Emissions Points
Stationary sources are one of the major contributors of pollution to the atmosphere. They are
fixed-site (i.e., stationary), producers of air pollution such as power plants, chemical plants, oil
refineries, manufacturing facilities, and other industrial facilities. Air pollution from stationary
sources is produced by two primary activities: (1) stationary combustion of fuel such as coal, oil,
wood, or natural gas, and (2) pollutant losses from industrial processes. Examples of industrial
processes include petroleum wells, refineries, chemical manufacturing facilities, coating
operations and smelters.
; OM DA FIV
ROLLU TAM Tฃ
SC3 a-
HNOS OS KH
r
-------
ORS Handbook
Section 1.0
Page 1-7
Ducted or Vented Emissions
A process vent is basically an opening where substances (mostly in gaseous form) are "vented" into
the atmosphere. Common process vents in a chemical plant are distillation columns and oxidation
vents, for example.
Figure 1-2. Example of a Ducted Stationary Source Stack
-------
ORS Handbook
Section 1.0
Page 1-8
Historically, ducted or vented stationary source emissions have been measured in the ducts or
stacks before release into the atmosphere. These sources are also often referred to as point
sources because the final release of emissions can be traced to a single or multiple defined duct or
stack exhaust. Ducted sources permit emission stream parameters, such as flow rates,
temperature, pressure, and other physical characteristics, to be recorded within the accuracy
requirements for end data use because they are confined under relatively steady conditions.
Area or Fugitive Emissions Sources
Those stationary facilities or activities whose individual air emissions do not qualify them as point
sources are called area (e.g., a landfill) or fugitive sources (e.g., leaking valves at a gas or oil
processing plant). Area and fugitive sources are often collections of a multitude of minor sources
with individually small emissions that are impractical to consider as separate point sources. Area
sources, including fugitive emissions, are those emissions that could not reasonably pass through a
stack, chimney, vent, or other ducted line that could easily be characterized with conventional
point source or stack sampling methods. Measurement of emissions from these sources
traditionally requires total enclosure techniques, or a combination of point measurements and
modeling using upwind-downwind or exposure-profiling methods.
Area sources represent numerous facilities and activities, including various unintended or irregular
emissions. Fugitive and area sources may release small amounts of a given pollutant individually,
but collectively can release significant amounts of a pollutant. For example, dry cleaners, vehicle
refinishing, animal feeding operations and gasoline storage facilities do not typically qualify as point
sources, but collectively, the various emissions from these sources are classified as area sources.1
Fugitive emissions from storage tanks are due to pollutants that can leak through the roofs and
through tank openings when liquids expand or cool because of outdoor temperature changes. In
addition, air pollutants can escape during the filling and emptying of a storage tank. Air pollution is
also produced when wastewater containing volatile chemicals comes in contact with the air.
-------
ORS Handbook
Section 1.0
Page 1-9
Both stationary point source and fugitive/area source emissions measurements have traditionally
been performed using single point sampling that accumulates and integrates sampled gas for a set
period of time followed by analysis for target components. Continuous point source instrumental
methods have also been applied to stationary source emissions and area source measurements.
Instrumental methods collect samples from a single point and provide information on the
concentration of a target component of interest over relatively small increments of time. A critical
issue with traditional air measurements is collection and reporting of data from a single point that is
assumed to be representative of the air or emission being monitored. This assumption is verifiable
when ducts or stacks are sampled but much less certain for area source and ambient
measurements.
1.4 Why Remote Measurement?
Fugitive emissions are emissions not contained or caught by a capture system and are often caused
by equipment leaks, evaporative processes, and windblown disturbances. These emissions may
occur from breaks or small cracks in seals, tubing, valves, or pipelines, as well as when lids or caps
on equipment or tanks have not been properly closed or tightened. When natural gas escapes via
fugitive emissions, methane, volatile organic compounds (VOCs), and any other contaminants in
the gas (e.g., hydrogen sulfide) are released to the atmosphere. Other examples of area sources
with significant fugitive emissions include landfills or waste lagoons.2
-------
ORS Handbook
Section 1.0
Page 1-10
9/3W8 12,47 23PM
Figure 1-3. Example of Potential Fugitive Source
Area and fugitive emissions sources are especially challenging to monitor because the pollutants of
interest are not contained within a duct or stack before release. The development of emission
factors for area sources is equally difficult due to the measurement challenges. In contrast,
stationary stack emissions and their related emission factors' determination are easier to measure
and determine. Over the past 20 years, remote measurement approaches, including ORS methods,
have been improving technologically and gaining greater use as an emissions estimation tool,
especially for stationary area sources and some on-road/near-road mobile sources. A significant
number of remote measurement activities have been performed for open area sources such as
landfills, wastewater treatment plant ponds, agricultural waste, wastewater lagoons, oil and gas
field production sites, waste ponds for mining operations, and ambient fenceline concentrations
surrounding large chemical and refinery facilities.
-------
ORS Handbook
Section 1.0
Page 1-11
Blast Furnace Upset
Other Fugitive Emissions
Stack/Ducted
ssions
Figure 1-4. Examples of Ducted and Fugitive Sources
These types of sources are prime candidates for the application of remote measurement
techniques because the ORS technology and techniques are small, mobile, do not take lengthy setup
time and can return data quickly to the staff collecting the data and to the operators of the
facilities.
Criteria Pollutant Gases, HAPs and GHGs
The measurement technologies and approaches addressed in this handbook are focused on five
groups of pollutants currently regulated or on the regulatory horizon under the Clean Air Act.
These four groups are:
criteria pollutants,
hazardous air pollutants (HAPs)
greenhouse gases (GHGs),
-------
ORS Handbook
Section 1.0
Page 1-12
ozone-depleting substances, and
Particulate Matter (PM).
Gaseous criteria pollutants are those inorganic pollutants (e.g., carbon monoxide, sulfur oxides,
nitrogen oxides and ozone) that are commonly found all over the United States. The EPA uses
these "criteria pollutants" as indicators of air quality. Each of the criteria pollutants is discussed in
detail below.3
Carbon monoxide (CO) is a colorless, odorless gas formed when carbon in fuel is not burned
completely. Motor vehicle exhaust contributes about 60 percent of all CO emissions nationwide.4
Other sources of CO emissions include industrial processes (such as metals processing and chemical
manufacturing), residential wood burning, and natural sources such as forest fires.
Sulfur oxides (SOx) are colorless gases formed when fuel containing sulfur, such as coal and oil, is
burned, and when gasoline is extracted from oil or metals are extracted from ore. Sulfur dioxide
(SO ) is the criteria pollutant that is the indicator of SO concentrations in the ambient air. Other
2 x
sources of SO^ are industrial facilities that derive their products from raw materials like metallic
ore, coal, and crude oil, or that burn coal or oil to produce process heat. Examples are petroleum
refineries, cement manufacturing, sulfuric acid plants, and metal processing facilities. Also,
locomotives, large ships, and some non-road diesel equipment currently burn high sulfur fuel and
release SO^ emissions into the air in large quantities.
Nitrogen oxides (NOx), is the generic term used to describe the sum of nitric oxide (NO), nitrogen
dioxide (NO^), which is a criteria pollutant, and other oxides of nitrogen. NOx is a group of highly
reactive gases that play a major role in the formation of ozone. NOx form when fuel is burned at
high temperatures, as in a combustion process. The primary sources of NOx are motor vehicles,
electric utilities, and other industrial, commercial, and residential sources that burn fuels.
Ozone (O3) is a gas composed of three oxygen atoms. It is a unique criteria pollutant in that it is
-------
ORS Handbook
Section 1.0
Page 1-13
exclusively a secondary pollutant. It is not usually emitted directly into the air, but at ground level
is created by a chemical reaction between NOx and VOCs in the presence of heat and sunlight. 0^
has the same chemical structure whether it occurs miles above the earth or at ground level and can
be "useful" or "damaging" to the environment depending on its location in the atmosphere. Useful
0 occurs naturally in the stratosphere and forms a layer that protects life on earth from the sun's
3
harmful rays or ultraviolet radiation. In the earth's lower atmosphere, or troposphere, ground-
level 0^ is considered damaging or destructive. 0^ is the most prevalent chemical found in
photochemical air pollution, or smog.3
HAPs or air toxics are those pollutants that are known or suspected to cause cancer, respiratory
problems or other serious health effects, or are thought to have adverse environmental or
ecological effects. The presence of HAPs in the air can be more localized than criteria pollutants
and the highest levels are usually found close to the emission sources. Examples of air toxics
include benzene, found in gasoline; mercury from coal combustion; perchloroethylene from some
dry-cleaning facilities; and methylene chloride used as a solvent by many industries. Most air toxics
originate from man-made sources including mobile sources (e.g., cars, trucks, construction
equipment), stationary sources (e.g., factories, refineries, power plants), and indoor sources (e.g.,
some buildings materials and cleaning solvents).4
GHGs are those compounds that enhance the retention of the sun's heating of the earth. Clouds
and a natural layer of atmospheric gases absorb a portion of earth's heat and prevent it from
escaping into space. This keeps our planet warm enough for life and is known as the natural
"greenhouse effect." Scientific evidence shows that the greenhouse warming effect is being
increased by the release of certain gases into the atmosphere that cause the earth's temperature
to rise. This rise in temperature caused by greenhouse gases is called "global climate change."
Carbon dioxide (CO^), methane (CH4), nitrous oxide (N2O), sulfur hexafluoride (SFe), ammonium
trifluoride (NF3), and hydro and per-fluorinated compounds, are the major compounds contributing
to global climate change. CO^ emissions account for about 81 percent of greenhouse gases
-------
ORS Handbook
Section 1.0
Page 1-14
released in the United States and are largely due to the combustion of fossil fuels in electric power
generation, motor vehicles, and industries. Methane emissions, which result from agricultural
activities, landfills, and other sources, are the next largest contributors to greenhouse gas
emissions in the United States and worldwide 5
Ozone-depleting substances are compounds such as chlorofluorocarbons (CFCs), halons, carbon
tetrachloride, methyl bromide, and methyl chloroform. The stratosphere contains a layer of 0^ gas
that protects living organisms from harmful Ultraviolet-B (UV-B) radiation from the sun which has
been linked to many harmful effects, including various types of skin cancer, cataracts, and harm to
crops, materials, and marine life.6
Particulate matter (PM) has been shown to be deleterious to humans and the environment. In the
early days of the EPA, the Agency created a NAAQS standard for total suspended particulate matter
(TSP). The size cut was approximately 50 |j,m. However, research in the 1980 through the 1990 led
the agency to adopt a smaller cut size: aerodynamic size of less than 10 |j,m (PMio). Some particles,
such as dust, dirt, soot, or smoke, are large or dark enough to be seen with the naked eye. Others
are so small they can only be detected using an electron microscope. These include:
PMio: inhalable particles, with diameters that are generally 10 micrometers and smaller;
PM2.5 : fine inhalable particles, with diameters that are generally 2.5 micrometers and
smaller; and
UFP: ultrafine Particle. These are extremely fine inhalable particles that are generally in the
less than 100 nanometer (nm) range.
As stated above, PM comes in many sizes and shapes and can be made up of hundreds of different
chemicals. Some are emitted directly from a source, such as construction sites, unpaved roads,
fields, smokestacks or fires. Most particles form in the atmosphere as a result of complex
reactions of chemicals such as sulfur dioxide and nitrogen oxides, which are pollutants emitted from
power plants, industries and automobiles.
-------
ORS Handbook
Section 1.0
Page 1-15
1.5 Knowledge and Advancement of Remote Sensing to Emissions Measurement
There are four major sensing approaches that will be described in more detail in later sections of
this handbook. They include the following:
Active. Open-path ORS techniques typically use optical telescopes to transmit and receive
energy beams, such as UV, infrared (IR), or visible wavelength range.
Passive. Open-path ORS techniques receive light energy from pollutants activated by an
external uncontrolled source such as combustion gases (e.g., Passive FTIR radiation) or the
sun (e.g., Solar Occultation and mobile DOAS).
Backscatter. ORS techniques use energy reflected from pollutants after activation from a
controlled source of light energy (e.g., Differential Absorption Light Detection and Ranging
(DIAL/LIDAR) systems).
Mobile. Measurement methods do not have to be optically based. However, optical
technologies have been engineered to be rugged enough to allow stable operation from a
moving vehicle. Typically, these optical techniques sample the gas and PM into a confined
cell while moving along a path to be measured (e.g., cavity ringdown, white cell and FTIR
tracer release systems, particle counters and filter based systems).
Active
ORS techniques use the light generated under controlled conditions from one of many sources
including heated glow bars for IR light, quartz lamps filled with deuterium or xenon gas, or laser
light. The light energy is broadcast over relatively long distances (up to 1,000 meters) in an open-
air setting. A simplified schematic of an open-path ORS technique to measure emissions from an
open source is provided in Figure 1-4. In general, open-path ORS test methods involve the
transmission of an energy beam across a path (straight line or two- dimensional plane) located
downwind of the emission source to be measured (e.g., wastewater lagoon). The pollutant
concentration along the line or plane is determined by evaluating certain qualities of the energy
-------
ORS Handbook
Section 1.0
Page 1-16
beam (e.g., the amount of light absorbed) after it has passed through the sample path and is
captured by a receiver. Chemical compound reference spectra and computational algorithms are
used to translate the instrument signal into a pollutant path-integrated concentration (e.g., parts
per billion (ppb) benzene per meter). Additionally, a mathematical calculation routine, combined
with meteorological data (wind speed, wind direction) collected during the sampling event, is
needed to convert the ORS instrument output (e.g., a path-integrated concentration or a flux
measurement) to an emission flux rate (e.g., milligrams per second). Open-path ORS methods can
be designed and applied in several different ways to capture area source emissions in both vertical
and horizontal planes. The predominant measurement applications that use ORS technologies in
the open-path mode include line of sight monitoring, Radial Plume Mapping7 (RPM), and Backward
Lagrangian Stochastic (bLS) Modeling.
Retroreflector
Energy Beam Source
and
Detector
Computational
computer
Figure 1-5. Block Diagram of a bi-static ORS
Passive
Passive techniques use the same technology as active without the need for a controlled source of
energy. The PFTIR technique is an example of this technology. PFTIR can be used to measure
-------
ORS Handbook
Section 1.0
Page 1-17
infrared spectra in air at elevated temperatures because hot gases emit radiation with the same
infrared signature as their absorption spectrum. Hot gases above the flame zone in an industrial
flare contain combustion products such as CO2, CO, and vapor phase organic material resulting from
products of incomplete combustion. For example, hot gases emitted by the flare can be identified
and quantified using the radiant FTIR absorption measurements.
The primary difference that must be considered between optical remote infrared absorption (e.g.,
FTIR) and hot gas radiance measurements using PFTIR is the temperature dependence of the FTIR
spectral measurements. The results of PFTIR are both temperature and concentration dependent.
Knowing the source temperature at the location where the gas concentrations are measured is
necessary to quantify the compounds of interest.
Backscatter
Open-path optical measurement approaches used in this handbook refer to the use of Light LIDAR
technology. DIAL is an application of LIDAR using powerful lasers directed into the atmosphere to
measure reflected light energy from aerosols, dust, and gases. The DIAL measurement is achieved
by the direct impingement of the laser beam on these materials and its subsequent reflection and
scattering. Because the target substances vary in concentration along the axis (optical path) of the
transmitted beam, the receiving telescope equipment analyzes the strength of the returning
(reflected) beam continually during its reception.9 The reflected beam strength is reduced from the
original transmission strength by a measurable amount that is proportional to the concentration of
the target matter.
LIDAR technology can also be deployed to measure PM. LIDARs that operate in the Raman and
Rayleigh size ranges (i.e., wavelengths in the nanometer to micrometer range) can be utilized to
measure the amount of backscatter from particles in the UFP range up to PM10. These instruments
are discussed in Section 2.7.
-------
ORS Handbook
Section 1.0
Page 1-18
Mobile
Optical monitoring approaches use optical techniques to measure gas samples pumped into
measurement cells where pressure and temperature are controlled. Unlike stationary monitoring
techniques, mobile optical techniques allow the user to move along and between the emission
plumes generated by area or fugitive sources. A tracer ratio application of mobile monitoring can
be used to simulate emissions from a source through the release of a tracer gas at or near the
center of the area source with subsequent measurement of the tracer and emission compound(s)
concentrations downwind of the source. Measurements must be conducted at a distance from the
source that is sufficient for the plume (e.g., from a landfill or wastewater lagoon) and tracer gas to
be well mixed and close enough that emission plume is measurable well above background
concentrations. These distances can range from 1 to 5 km to achieve proper mixing.9
In addition to measurements of gases, PM plume measurements techniques have become available
using the current technology, such as the fast version of particle counter that are now on the
market. These devices, used in conjunction with probes and plenums, can measure PM on roads
instantaneously.
Advantages Over Closed Path Techniques
Although open-path ORS techniques have been used for 20 years and are well-established, they are
constantly improving and gaining use to characterize and quantify pollutant emissions from sources
that are not conducive to traditional point source testing methods, such as large area sources.
Improvements often include changes to technologies that improve detection limits or the types of
compounds detected. For large area sources, ORS methods have distinct advantages when
compared with traditional single point measurement techniques, such as photo-ionization
detectors (PID), PID/flame ionization detectors (FID), Summa canisters, various sorbent methods,
and flux boxes. Specific advantages and disadvantages of the ORS measurement technologies and
applications are addressed in Chapters 2 and 3 of this Handbook. The advantages of ORS
applications should be determined on a case-by case-basis tailored to specific measurement goals
-------
ORS Handbook
Section 1.0
Page 1-19
and objectives. Some of the general ORS advantages are as follows:
More likely to identify emissions "hot spots" because measurements are collected over a
large area,
Achieve better spatial and temporal emissions resolution,
No sample shipping costs,
Perform direct, measurement-based emission calculations, and
Represent personal exposure better than fixed point monitoring.
Some general issues that require consideration when ORS methods are used include the following:
Costly initial sampling instrumentation investment,
Experienced manpower and higher site preparation cost more to deploy,
Dependent on weather conditions (e.g., heavy rain, fog, dust), and
Dependent on chemical interferences (e.g., water, oxygen, 0^ and CO^).
As the use of open-path ORS technologies to quantify emissions from area sources has advanced,
the desire to use ORS data in the development of atmospheric models and to support air quality
standards has increased. However, use of remote sensing presents some challenging issues.
Classical point measurement technologies and their associated results are typically based on the
size of the stack or leak, flow data, moisture, bulk gas molecular weight and stack pollutants to be
measured. Performance tests for emission masses are usually snapshots of short duration and not
continuous. Using ORS data, unlike point sources wherein emissions measurements are typically
straightforward, a more critical evaluation of the ORS method application, the emission mechanism
of the source, and the source activity is needed of the emissions developer to ensure that the
resulting data provides an accurate representation of average emissions from the source. While
-------
ORS Handbook
Section 1.0
Page 1-20
developing emission factors from optical remote technology applications is beyond the scope of this
document, it is the aim of the handbook to provide the technology background, application
examples, and quality information for optical remote measurements that can assist all data users to
develop emission results comparable to those routinely generated by traditional point source
testing methods.
1.6 General Discussion of the EPA Quality System
The EPA recently issued new guidance on its Quality Program (QP) policy. The document, "EPA
Quality Program Policy"10 states that this policy:
Recognizes existing policies and procedures as the foundation of an Agency-wide Quality
Program,
Establishes an approach for identifying and addressing Agency quality issues,
Provides a structure and procedures to ensure and enhance the effectiveness of the
Quality Program and its application to Agency products and services.
The EPA policy is based on the national consensus standard, ANSI/ASQC E4-1994, Specifications and
Guidelines for Environmental Data Collection and Environmental Technology Programs, developed
by the American National Standards Institute and the American Society for Quality (ANSI/ASQC).11
The ANSI/ASQC E4-1994 specification is consistent with the international standard ISO 17025. The
ANSI document describes the necessary management and technical elements for developing and
implementing a QS by using a tiered approach. The standard recommends documenting: (1) each
organization-wide QS in a Quality Management Plan (QMP) or Quality Manual (to address
requirements of Part A: Management Systems of the standard) and (2) the applicability of the QS to
technical activity- specific efforts in a Quality Assurance Project Plan (QAPP) or similar document (to
address the requirements of Part B: Collection and Evaluation of Environmental Data of the
standard). The EPA has adopted this tiered approach for its mandatory agency-wide QS. This
document addresses Part B requirements of the standard for systematic planning for
environmental data operations.
-------
ORS Handbook
Section 1.0
Page 1-21
In accordance with EPA Order 2106.011, the EPA requires that environmental programs performed
for or by the Agency must be supported by data of the type and quality appropriate to their
expected use. The EPA defines environmental data as information collected directly from
measurements, produced from models, or compiled from other sources such as databases or
literature.
Data Quality Objectives
EPA Order 2106.0 requires that all EPA organizations (and organizations with extramural
agreements with EPA) follow a systematic planning process to develop acceptance or performance
criteria for the collection, evaluation, or use of environmental data. A systematic planning process
is the first component in the planning phase of the project tier (see the bottom tier of Figure 1.5),
while the actual data collection activities take place in the implementation phase.
Systematic planning is a planning process based on the scientific method and includes concepts
such as objectivity of approach and acceptability of results. Systematic planning is a common-
sense, graded approach to ensure that the level of detail in planning is commensurate with the
importance and intended use of the work and available resources. This framework promotes
communication among all organizations and individuals involved in an environmental program.
Through a systematic planning process, a team can develop acceptance or performance criteria
for the quality of the data collected and for the quality of the decision. When these data are being
used in decision-making by selecting between two clear alternative conditions (e.g.,
compliance/non-compliance with a standard), the EPA's recommended systematic planning tool is
called the Data Quality Objective (DQO) Process.
The DQO Process is a seven-step planning approach to develop sampling designs for data collection
activities that support decision-making. This process uses systematic planning and statistical
hypothesis testing to differentiate between two or more clearly defined alternatives.
The DQO Process is iterative and allows the planning team to incorporate new information and
-------
ORS Handbook
Section 1.0
Page 1-22
modify outputs from previous steps as inputs for a subsequent step. Although the principles of
systematic planning and the DQO Process are applicable to all scientific studies, the DQO Process is
particularly designed to address problems that require deciding between two clear alternatives.
The final outcome of the DQO Process is a design for collecting data (e.g., the number of samples to
collect and when, where, and how to collect samples).
The development and implementation of a QS should be based on a "graded approach," that is, the
components and tools of a QS (Figure 1.5) apply according to the scope and nature of an
organization, program, or project and the intended use of its products or services. This approach
recognizes that a "one size fits all" approach to quality management is not appropriate and that the
QS of different organizations and programs should (and will) vary according to the specific needs of
the organization. For example, the quality expectations of a research program are different from
those of a regulatory compliance program because the intended use of the products differs. The
same applies to remote sensing data. Monitoring agencies that use this Handbook are strongly
encouraged to understand their data objectives, perform the DQO Process if needed, and use the
MQOs described in Section 1.4.2 if they are applicable to an agency's program. Additional
explanation and details on the DQO Process can be found in EPA's Guidance on Systematic Planning
Using the Data Quality Objectives Process.12
When an agency or entity is monitoring for non-regulatory purposes (e.g., background
concentrations, modeling applications, or exposure), these MQOs are recommended information.
Meeting MQOs for non-regulatory meteorological monitoring is strongly advised.
Measurement Quality Objectives
Once DQOs are designated for a program or project, measurement indicators must be determined
to understand if the DQOs are being met. Most state/local/tribal agencies that collect data do so to
support programs that are federally mandated or that need to meet federal requirements.
However, other non-regulatory applications exist, such as modeling applications, state
implementation plan development, and forecasting. These programs require different MQOs
-------
ORS Handbook
Section 1.0
Page 1-23
because the application is different (i.e., different DQOs). The following prescribed objectives
should be decided and discussed within the QS.
Measurement. Type of measurements and/or the parameter needed to be collected.
Method. The method is different from the measurement in that a particular instrument
can be utilized in different methods. The method will dictate the precision, bias, and
representativeness of the sampling data.
Reporting Units. Reporting units must be decided before the program begins. If it is a
regulatory program, then ppb or micrograms per cubic meter (|ag/m3) would be the
appropriate units. However, if it is a modeling exercise, then grams per second (g/s) may
be the appropriate unit.
Detection Limits. It is very critical to state the levels of detection (LOD) for a particular
program. The LOD can be very difficult to quantify until the ORS is in the field of
operation. It should also be noted that LODs can be defined in different ways. It is best to
define and state LOD in the quality documents developed for a program.
Minimum Sample Frequency. This objective is required to define how often data must be
collected to meet the end user's requirements for precision and representativeness.
Measurements must be taken often enough to meet model or modeling input criteria.
Completeness. For most programs/projects, there is a minimum amount of data required
to allow the data users to make decisions concerning the environment. A rule of thumb is
75 to 85 percent data completeness.
Precision. Precision is the measure of agreement among repeated measurements of the
same property under identical conditions. This can be very difficult to measure using ORS.
Bias. Bias is the systematic or persistent distortion of a measurement process that causes
errors in one direction. Bias, like precision, can be very difficult to determine with ORS.
-------
ORS Handbook
Section 1.0
Page 1-24
The project or program must be able identify and determine the magnitude of its
measurement bias.
Representativeness. Representativeness is collection of the measurement location,
frequency, duration, and other factors that demonstrate the results correspond to the
emission characterization required by the data users.
1.7 Choosing the Right Tool for the Right Job
This section outlines recommendations on how the different technologies (i.e., instruments) stack
up versus the source type (i.e., area, ducted or fugitive emissions). These table appear here in
pairs; gas and particulate phase. Here is a breakdown on the information in these tables:
Nature of the Source;
Time Resolution;
QA/QC Requirements; and,
Ease of Operation.
Tables 1-1 through 1-8 aren't meant to be the final word on how these technologies and
applications can be utilized, but serve as a reference for those seeking a match between the
technologies and source applications.
What is the nature of the source?
Tables 1-1 illustrates the types of instruments that detect and quantify gases, while table 1-2
covers particulate sources. Three main source types are considered in this handbook: area,
ducted/vented, and fugitive. Identification of the source type is an important first step towards
identifying the proper technology for particular measurement needs.
Area and fugitive sources are similar in that both are often collections of minor sources that are
impractical to consider as separate point sources. More specifically, area sources represent
numerous facilities and activities. Fugitive sources, on the other hand, are often incidental in
nature, such as leaks from valves, tank openings, or carbon sequestration sites, to name a few
examples.
Ducted, or vented sources have also been referred to as point sources. Frequently, these vented
-------
ORS Handbook
Section 1.0
Page 1-25
sources are distillation columns or oxidation vents in a chemical plant, for example. These sources
often lend themselves to measurements under relatively stable conditions.
Table 1-1 Gas Phase Source Type
Source Type
Technology
Area
Ducted or Vented
Fugitive
FTIR
X
X
X
TDL
X
X
X
UV-DOAS
X
X
X
CRDS
X
X
X
Passive Tubes
X
X
DIAL
X
X
Table 1-2. Particle Phase Source Type
Source Type
Technology
Area
Ducted or Vented
Fugitive
Passive Sampler
X
X
X
SMPS
X
X
X
APS
X
X
X
SPAMS
X
X
X
DIAL
X
X
X
Sensors
X
X
X
CRDS
X
X
X
PAS
X
X
X
AMS
X
X
X
ELPI
X
X
X
What is the required time resolution?
Tables 1-3 and 1-4 illustrate which technologies perform in different time resolutions. In addition
to identifying the type of source at which measurements will be taken, a second important
question must do with the required time resolution of the data. Time resolution needs are often
dependent upon several factors such as: available budget, expected stability of the pollutant
emissions being studied, and data reporting requirements, among others. Time resolution can
range from essentially continuous measurements to integrated measurements taken over a period
of weeks or months. Some technologies can be employed at different time resolutions depending
on current needs, and allowing the same technology to be utilized in different scenarios.
-------
ORS Handbook
Section 1.0
Page 1-26
Table 1-3. Gas Phase Time Resolution Matrix
Technology
Time Resolution
real-time seconds minutes hours or longer
FTIR
TDL
UV-DOAS
CRDS
Passive Tubes
DIAL
X
X X
X X X X X
XX X X
Table 1-4. Particle Phase Time Resolution Matrix
Technology
Time Resolution
real-time seconds minutes hours or longer
Passive Sampler
SMPS
APS
SPAMS
DIAL
Sensors
CRDS
PAS
AMS
ELPI
X
XXX XX X
xxxxxxxx
XX X XX
What are the QA/QC requirements for the measurement?
Tables 1-5 and 1-6 illustrate the different levels of QA and QC that are needed to obtain data that
is of known quality. An important consideration when choosing a measurement technology is the
level of QA/QC required of the data. The level of accuracy and precision required of a
measurement will have a significant impact on the technology selected. For example, while a PID
sensor and UV-DOAS instrument can both detect benzene, if the goal is to merely detect benzene,
the PID sensor would be more cost effective and less labor intensive.
Table 1-5. Gas Phase QA/QC Matrix
Technology
Data Type
Qualitative Semi-Quantitative Quantitative
FTIR
TDL
UV-DOAS
CRDS
Passive Tubes
DIAL
Thermal IR Camera
X X X X X X
X
X
-------
ORS Handbook
Section 1.0
Page 1-27
Table 1-6. Particle Phase QA/QC Matrix
Technology
Qualitative
Data Type
Semi-Quantitative
Quantitative
Passive Sampler
X
SMPS
X
X
APS
X
X
SPAMS
X
X
DIAL
X
X
Sensors
X
X
CRDS
X
PAS
X
X
AMS
X
X
ELPI
X
How easy are the data to process?
Tables 1-7 and 1-8 illustrate the ease of operation for each of these technologies. The experience
required to process acquired data and prepare those data for presentation and interpretation is
another important factor to consider prior to selecting a measurement technology. This question
is crucial to recognizing the personnel required for collecting, processing, and analyzing data of the
highest possible quality. In cases where much of the processing is done automatically via
instrument software or settings, experience isn't as important as it is in instances where
processing requires more complex calculations or a greater knowledge of the theory behind the
measurement technique.
Table 1-7. Gas Phase Ease of Operation Matrix
Technology
Years of experience
<1 1-3 >3
FTIR
TDL
UV-DOAS
CRDS
Passive Tubes
DIAL
Solar Occultation Flux
x x
X X X X X
X
-------
ORS Handbook
Section 1.0
Page 1-28
Table 1-8. Particle Phase Ease of Operation Matrix
Technology
Years of experience
<1 1-3 >3
SMPS
X
APS
X
SPAMS
X
DIAL
X
Sensors
X
CRDS
X
PAS
X
AMS
X
ELPI
X
1.8 Future Evolution and Updates of this Handbook
The EPA will periodically update and correct this handbook. Updates will include the addition of
new information as well as feedback from stakeholders. This document will be updated, at the
discretion of the EPA, depending on the availability of resources.
This document does not contain EPA policy information; it is strictly an information document. It is
envisioned that in later editions, new ORS technologies and techniques will be described in this
document.
1.9 References
1. U.S. EPA. 2010 Sources of Pollutants in the Ambient Air - Stationary Sources, 15 July 2010
2. Earthworks. Sources of Oil and Gas Air Pollution.
http://www.earthworksaction.org/airpollutionsources.cfm>.
3. U.S. EPA. Criteria Pollutants, https://www.epa.gov/criteria-air-pollutants
4. U.S. EPA. Hazardous Air Pollutants, https://www.epa.gov/haps
5. U.S. EPA. Greenhouse Gas Emissions, https://www.epa.gov/ghgemissions
-------
ORS Handbook
Section 1.0
Page 1-29
6. U.S. EPA. Ozone Layer Protection, https://www.epa.gov/ozone-layer-protection
7. Other Test Method 10 (OTM 10) - Optical Remote Sensing for Emission Characterization from
Non-Point Sources, FINAL ORS Protocol, https://www3.epa.gov/ttn/emc/prelim/otml0.pdf
8. Ednar, H., P. Ragnarson, and E. Wallinder. 1995. Industrial Emission Control Using Lidar
Techniques. Environ. Sci. Technol. 29:330-337.
9. Mosher, B. W.; Czepiel, P. M.; Harriss, R. C.; Shorter, J. H.; Kolb, C. E.; McManus, J. B.; Allwine,
E.; Lamb, B. K. 1999. Methane emissions at nine landfill sites in the northeastern United States.
Environmental Science & Technology. 33, (12): 2088-2094.
10. US EPA. 2008. US EPA Quality Policy, CIO 2106.0, October 2008.
http://www.epa.gov/quality/qa docs.html
11. Guidance on Systematic Planning Using the Data Quality Objectives Process EPA QA/G-4,
EPA/240/B-06/001 February 2006, https://www.epa.gov/qualitv/guidance-systematic-
planning-using-data-quality-obiectives-process-epa-qag-4
12. US EPA. 2010. Implementation of Quality Assurance Requirements for Organizations Receiving
EPA Financial Assistance, http://www.epa.gov/ogd/grants/assurance.htm.
-------
ORS Handbook
Section 2.0
Page 2-1
2.0 Optical Remote Sensing Technologies
ORS technologies measure the concentration of chemicals in an open-air path or in contained air
samples collected from discreet sampling points. They do this by measuring the interaction of
electromagnetic energy (i.e., different wavelengths of light) with the air's components. Open-path
technologies measure the concentrations of chemicals or particulates across an open path of air.
They do this by emitting a concentrated beam of electromagnetic energy into the air and measuring
its interactions with the air's components. Open-path technologies provide an average
concentration over a line of sight. Point-source applications of these technologies measure the
concentration of a confined sample of air drawn into the apparatus from a point or points in air.
Some technologies, such as Tunable Diode Lasers (TDL), are capable of simultaneously measuring
one or two compounds. Other technologies (e.g., UV-DOAS) can measure several compounds
simultaneously, while others (e.g., FTIR) can measure many compounds simultaneously. ORS
technologies are used to measure the average chemical concentrations over a set distance or at a
stationary point. The path average over a set distance has an advantage over point-source
measurements that may miss high-concentration plumes running between sampling devices. Both
open-path and point-source applications of these technologies have been used to detect hotspots
in area sources and to obtain path integrated averages. Each of the technologies has advantages
and disadvantages for these applications. The technologies described in Chapter 2 can be used
alone or in combination to provide three major types of data: plume characterization, short term
flux measurements and long-term monitoring studies. In Chapter 2, we discuss how the prominent
ORS technologies operates.
-------
ORS Handbook
Section 2.0
Page 2-2
2.1 Fourier Transform Infrared Spectroscopy
FTIR spectroscopy is an optical spectroscopy technology adapted to perform real-time monitoring of
gaseous and volatile organic compounds in air. The technique can detect and quantifying multiple
compounds simultaneously, even in harsh industrial environments, using the characteristic spectral
features of the individual compounds.1
FTIR spectrometers are well-suited to remote sensing applications because they are durable,
portable, and do not require daily routine calibration. The technology, however, is not simple to
operate and requires experienced staff to ensure correct operation and valid results. Both
extractive and open-path (OP-FTIR) environmental applications of the technology have been
demonstrated. The EPA test methods have been written for both open-path, such as other test
method 10, (OTMIO) or toxic organic 16 (TO-16) and extractive (Method 318, Method 320, EPA
performance specific method 15 (PS-15), PS-18 and ASTM D6328-03) measurement by FTIR. In
open-path mode, the IR radiation beam can be directed over distances of up to 400 to 500 meters
to measure selected compounds in emission plumes or dispersed air parcels. Alternatively, in
extractive mode, gas can be drawn into a closed cell with a folded path length of 10 to 100 meters.
Optical remote spectroscopy applications focus on open-path mode. Mobile tracer applications
focus on extractive mode. Compound-specific concentration is determined using standard IR spectra
of known concentration. The onboard computer software and spectra library allow real time
determination of concentration for preset compounds. Post-test processing of IR spectra allow
other compounds in air samples to be determined. Every measurement uses calibrated reference
spectra taken at conditions like the unknown field samples to determine compound concentrations
therein.
Basic Operation
The FTIR instrument sends an IR beam of light through a region (closed-cell or open-path)
containing the compounds of interest and captures the resulting IR spectra from the sample. Figure
2-1 illustrates the basic components of an open-path FTIR spectrometer. Infrared light
-------
ORS Handbook
Section 2.0
Page 2-3
generated by an IR emitting source is guided through an interferometer. The interferometer
consists of an IR source, beam-splitter, mirrors, a laser, and a detector. The IR energy goes from the
source to the beam-splitter, which splits the beam into two parts. One part is transmitted to a
moving mirror and one part is reflected to a fixed mirror. The moving mirror oscillates back and
forth at a constant velocity. This velocity is timed according to the very precise laser wavelength in
the system, which also acts as an internal wavelength calibration. The beam reflected from the
moving mirror and the beam reflected from the fixed mirror have traveled different distances since
being generated by the source and are recombined at the beam-splitter.1'2
Fixed Mirror
Moving Mirror
J
Beam
Spl itter
IR Source
i
T
Laser
Interferometer
1
Reflecting Mirror
I 1
R Detector
Figure 2-1. Diagram Showing Beam Path and Major Components ofFTIR
When the beams are recombined, some of the wavelengths recombine constructively and some
destructively, which creates an interference pattern. This interference pattern is called an
interferogram. The recombined IR beam then passes from the beam-splitter into the open-path
where a portion of the IR energy is absorbed by the gaseous compounds to be measured. The
resulting IR beam reaches the IR detector where the interference pattern is detected, digitized, and
transformed mathematically into a standard single beam infrared frequency spectrum using an
algorithm known as a Fourier transform. A reference or background single beam spectrum is also
collected without a sample and the ratio of the two single beam spectra is computed to produce a
background corrected transmittance spectrum. This transmittance spectrum can be converted to
absorbance by taking the negative base 10 log of the data points.2
The vibrational frequencies of all the infrared absorbing molecules in the IR beam path are
captured in the IR spectrum. When a molecule absorbs light, the energy of the molecule is
-------
ORS Handbook
Section 2.0
Page 2-4
increased arid the molecule is promoted from its lowest energy state (ground state) to an excited
state. Light energy in the infrared region of the electromagnetic spectrum stimulates molecular
vibrations. Molecular species display their own characteristic vibrational structure when stimulated
by IR radiation.3 Figure 2-2 shows the IP, absorption spectra for nitrous oxide, CO.,, CO, NO, NOr
and ammonia. The units of vibrational frequency are wave number. Wave number and vibrational
structure are used to identify a molecule.
036-
034-
0 32-
030-
020-
0.26-
D 24
0 22-
0 30-
5
c
3
018-
1
0.16-
0.14-
0.12-
0 10-
0.00-
006-
0 04-
0 02-
ooo-.
@1075K
Ammonia
(NH3)
Carbon Monoxide
(CO)
Carbon Dioxide
(CO 2)
Nitrous Oxide
(NZO)
Nitrogen Dioxide
(NO?)
Nitrie Oxide
3800 3600 3400 3200 3300 2800 2600 2400 2200 2000
Wavenumber
1800 1600 1400 1200 1000 800 600
Figure 2-2 FTIR Absorption Spectrum recorded at 1075ฐ
Once a compound has been identified, its spectrum can also be used to measure the compound's
concentration because the amount of IR radiation absorbed from the IR beam is proportional to the
concentration of the compound in the sample or open path. According to the Beer-Lambert law,
there is a linear relationship between absorbance and concentration as shown in the following
equation 1,2,3
-------
ORS Handbook
Section 2.0
Page 2-5
A = s*c*l
Where:
A = absorbance intensity
s = absorption coefficient
c = sample concentration
I = sample path length
FTIR systems typically operate in two modes: extractive cell or open-path. Extractive cell
measurements can be conducted either from a single location or from a mobile measurement
platform.4 In the field, OP-FTIR systems can operate with telescopes transmitting and receiving the
IR beam so monitoring of long outdoor paths is possible. The pollutants normally measured in this
process are at ambient temperature and usually in the low ppb concentration range.5 Typical
applications of open-path monitoring include fence-line monitoring of industrial sites, landfill sites,
waste lagoons, urban air monitoring in metropolitan areas, accidental release
detection/identification, and detection of agents or surrogates important to homeland security
monitoring applications.6,7 It should be noted that moisture due to fog or high humidity will cause
spectral interferences, which can limit the use of this technique to pollutants that do not have
overlapping absorption features with gas-phase water.5
Extractive (Closed) Cell Measurement Applications
For extractive or closed-cell FTIR measurements, the beam is sent through a cell that is mounted in
the instrument itself. As shown in Figure 2-1, the IR beam passes through the cell and is focused
onto the detector. Gas-phase samples are pumped into a sealed, constant temperature cell and
analyzed. Sample cell path lengths can vary from 10 cm to 150 m folded-path cells. The longer the
path length the more sensitive the measurement becomes because the IR beam has a greater
chance of interacting with the absorbing compounds. The pollutants measured in this type of
arrangement are usually at higher concentrations than those found in open-path FTIR
-------
ORS Handbook
Section 2.0
Page 2-6
measurements. Typical applications of
extractive FTIR monitoring include stack
testing of flue gases and vehicle exhaust.6'7
gas. This unit is equipped with a 32-rn
Figure 2-3 shows a photo of an FTIR unit
used for extractive monitoring of stack
folded path length cell that extends from
the end of the instrument.
Figure 2-3. FTIR Closed Cell Unit for Monitoring Stack Gases
Open-Path Measurement Applications
There are many instrumental configurations for open-path (OP) instruments. The simplest OP
systems are bistatic configurations. This configuration derives its name from the fact that both the
transmitter and receiver must be fixed in a static position and precisely aimed at each other. The
OP-FTIR equipment projects the IR light beam directly along a path to a detector/receiver. Bistatic
configurations in general have the requirement of supplying power at both the receiver and
transmitter, which can be a disadvantage in some locations. Additionally, there is a requirement for
alignment at both receiver and transmitter, which can be time consuming for mobile systems.8
Monostatic configurations were developed to address issues raised with bistatic designs. In a
monostatic configuration, all the optical components of the transmitter and receiver are in the
same location, and a retro-reflector is used to return the light from the transmitter to the receiver.
This configuration derives its name from the fact that only the transceiver portion of the
instrument needs to be precisely pointed as the retro-reflector returns light to its source regardless
of orientation. A diagram of a typical monostatic configuration is shown in Figure 2-4.
-------
ORS Handbook
Section 2.0
Page 2-7
RctroroHeclof
Light Soiree
Figure 2-4 Basic Setup for Monostotic OP-FTIR
Retro-reflecting mirrors, as they are called, are configured with three perpendicular reflective
surfaces in the shape of a corner. A combination of three mutually perpendicular mirrors reflects
light incident from any direction through 180ฐ as shown in Figure 2-5. Such a combination of
mirrors is called a corner-cube reflector. Corner-cube reflectors beam FTIR light back to its exact
point of origin. This property reduces the divergence of the beam on its return path back to the
detector compared to divergence that would result from a flat mirror. Also, the retro-reflector
array can be very large to capture and return essentially the entire divergent signal from the
telescope.
\ CC_J
CI
1
I ijgfi t- ra^y
1
m i /-ro/
mi
rror
Figure 2-5. The comer cube reflector
(http://farside.ph.utexas.edu/teaching/316/lectures/nodel33.htmn
In the monostatic mode, the IR laser beam is split twice, once leaving the OP-FTIR and once on
turn. This design requires a beam splitter irt the optical path that removes 50 percent of the light
-------
ORS Handbook
Section 2.0
Page 2-8
from the outgoing beam and 50 percent of the light from the return beam for an overall loss of 75
percent of the total light intensity.
The dual-telescope monostatic configuration
has lower detection limits because it does not
utilize a beam splitter in the optical path. A
translating retro-reflector, which is essentially a
portion of a very large cube, is used to return
the light beam offset to align with the
receiving telescope. This single, large retro-
reflector does not have the divergence reversal
properties of the corner- cube array. The
second telescope adds cost and complexity to
the system.8 However, when compared with
monostatic mode, bistatic systems are harder
to align and maintain
Figure 2-6 Typical telescopic FTIR Transmitting and Detection Unit
because any shift in the transmitter or detector can result in system misalignment.7,9 Both
operating modes measure only the compounds that are in the beam path. Emissions outside the
beam path are not measured. In these situations, measurements have been conducted along
multiple beam paths to more accurately characterize the emission plume Figure 2-6 shows a
telescopic FTIR transmitting and detection unit, which would be used for open-path field
monitoring applications. Figure 2-7 shows a typical retro-reflecting mirror. The retro mirrors are
surface coated with a reflective material to reduce extraneous glare from outside stray light. Retro-
reflecting mirrors are often contained in a protective housing, which is closed when the unit is not in
use to protect the sensitive reflecting surfaces from exposure to inclement weather conditions.
-------
ORS Handbook
Section 2.0
Page 2-9
Pollutants and Relative Levels That Can Be Measured
Table 2-1 provides an example list of compounds that have been measured using OP-FTIR
spectroscopy.6,7 This list is not all-inclusive, but shows that many compounds can be measured
via OP-FTIR.
Figure 2-7. Corner Cube Reflector
Another feature of OP-FTIR is that many compounds can be monitored simultaneously as opposed
to other beam technologies that can monitor only single compounds. As with other optical sensing
systems, OP-FTIR produces a path integrated concentration (PIC) in units of parts per million (ppm)
or ppb times length, i.e., ppb (meters).6 Dividing the final ppb (meters) result by the total optical
path length gives the path integrated gas concentration in ppb.
Detection limits can vary widely from compound to compound depending on many factors such as
instrument configuration, the condition of retro-reflecting mirrors, humidity, beam path length and
the absorbance strength of the target compound(s) at the wavelength chosen for analysis.
Detection limits are typically reported in ppm for one meter of path length (ppmm).
-------
ORS Handbook
Section 2.0
Page 2-10
They can be determined empirically using cell based measurements or estimated by solving the
equation on page 2-5 for an absorbance that is three times the mean signal noise if the absorbance
coefficient and noise are known at the wavelength used to measure the compound(s) of interest.
Detection limits for specific sampling episodes are calculated by dividing ppmm by the actual
meters of path length during field sampling.
Table 2-1. Example List of Compounds Measured by FTIR Open-path Systems
Species
Acetaldehyde
1,4-dimethyl piperazine
methyl mercaptan
acetic acid
1,4-dioxane
methyl methacrylate
Acetone
ethane
2-methyl propene
Aceto nit rile
ethanol
morphaline
Acetylene
ethyl acetate
nitric acid
Acrolein
ethylamine
nitric oxide
acrylic acid
ethylbenzene
nitrogen dioxide
Acrylonitrile
ethylene
nitrous acid
Ammonia
ethylene oxide
ozone
Benzene
ethyl mercaptan
pentane
1,3-butadiene
formaldehyde
phosgene
Butane
formic acid
phosphine
Butanol
furan
propane
1-butene
halocarb-11 (CCI3F)
propanol
cis-2-butene
halocarb-12 (CCI2F2)
propionaldehyde
trans-2-butene
halocarb-22 (CHCIF2)
propylene
butyl acetate
halocarb-113 (CFCI2CF2CI)
propylene dichloride
carbon disulfide
hexafluoropropene
propylene oxide
carbon monoxide
hydrocarbon continuum
pyridine
carbon tetrachloride
hydrogen chloride
silane
carbonyl sulfide
hydrogen cyanide
styrene
chlorobenzene
hydrogen sulfide
sulfur dioxide
Chloroethane
isobutene
sulfur hexafluoride
Chloroform
isobutanol
1,1,1,2-tetrachloroethane
m-cresol
isobutyl acetate
1,1,2,2-tetrachloroethane
o-cresol
isobutylene
tetrachloroethylene
p-cresol
isoprene
toluene
Cyclohexane
isopropanol
1,1,1-trichloroethane
1,2-dibromoethane
isopropyl ether
1,1,2-trichloroethane
m-dichlorobenzene
methanol
trichloroethylene
o-dichlorobenzene
methylamine
trimethylamine
p-dichlorobenzene
methyl benzoate
1,2,4-trimethylbenzene
1,1-dichloroethane
methyl chloride
vinyl chloride
1,2-dichloroethane
methylene chloride
m-xylene
1,1-dichloroethylene
methyl ether
o-xylene
dimethylamine
methyl ethyl ketone
p-xylene
dimethyl disulfide
methyl isobutyl ketone
Compounds in bold are EPA Hazardous Air Pollutants (HAPs) CAA -112Title 42, Chapter 85, Subchapter I, Part a U.S. Code 7412 (b).
-------
ORS Handbook
Section 2.0
Page 2-11
Typically, FTIR manufacturers report detection limits for commonly monitored pollutants as part of
the literature for their instrumentation. In general, open-path FTIR detection limits in the single
digit ppb levels can be achieved for many strong IR absorbing compounds.7'10 Extractive FTIR
detection limits for a 10-meter folded path length are typically on the order of 0.1 to 10 ppm. Some
compounds such as benzene have detection limits in the 30-50 ppb range because gas-phase water
interferes with this measurement.7 Other compounds, such as hydrogen sulfide, are weakly IR
absorbing molecules and have detection limits in the 300-800 ppb range.7
Typical QA/QC
To ensure measurement accuracy and data verification, instrumentation response should be
verified annually (detector and IR source) using a known concentration of a standard gas mixture.
Certificates of calibration should be kept on file and available for review. Maintenance records
should be kept in bound notebooks for any equipment adjustments or repairs that could affect
measurement performance. Maintenance notebooks should include the date and description of
maintenance performed. Calibration checks should be performed after major service and regularly
during analysis.11,12 QA and QC procedures for the measurement of gaseous compounds by
extractive FTIR are discussed in great detail in EPA Test Method 320 or ASTM D6328-03. QA and
QC procedures for OP-FTIR are discussed in more detail below.
Calibration Spectra
A gas-phase FTIR reference spectrum is collected at a known temperature and pressure in a fixed
path length enclosed cell for the compound of interest from a sample of known concentration. A
series of measurements can be made at different concentrations and a calibration curve that
relates the measured absorbance and the gas concentration can be developed to confirm a linear
response of signal with concentration. These calibration spectra are stored in a spectral reference
library used by the computer during real-time sample processing. Unknown sample concentrations
can be determined by comparing sample absorption intensities to absorption intensities in the
standard reference spectra. The higher the concentration of compound being measured; the more
-------
ORS Handbook
Section 2.0
Page 2-12
IR radiation characteristic of that compound is absorbed. Complex mixtures of IR sensitive
compounds can be determined from a single spectrum by solving a multiple linear regression
matrix using characteristic wavelengths of compounds and the relative intensities of sample IR
spectral features compared to calibration spectral features.
Tables of absorbance coefficients are available, and standard reference spectra for numerous
compounds can be purchased. Suppliers of reference spectra include Pacific Northwest National
Laboratory, which continues to develop the Northwest Infrared (NWIR) spectral library of
quantitative infrared absorption spectra13 and the high-resolution transmission molecular
absorption database (HITRAN) compiled by Harvard University.14
Figure 2-8 shows an example of a typical single point calibration curve where the sample
absorbance is plotted against concentration. Interpolation of the curve at a given absorbance
measurement gives the concentration of the molecular species being analyzed.
Absorbance vs. Concentration
0.6 r-
0.5
u
ra
a
a
0.2
0.1
0
10
20
30
40
50
60
Concentration (ppbv)
Figure 2-8. Calibration Plot of Absorbance vs. Concentration
-------
ORS Handbook
Section 2.0
Page 2-13
QA/QC for OP-FTIR Instrumentation
Several quality checks should be performed on FTIR instrumentation prior to deployment to the
field and for the duration of the field campaign.15 Prior to field deployment, the spectral baseline is
checked to determine the amount of signal intensity, instrument noise, and baseline drift. Baseline
drift is due to detector signal fluctuations that cause the signal to increase gradually over time.
Typically, instruments are powered on and allowed to warm up for at least one hour prior to data
collection to minimize baseline drift effects. Baseline noise should be checked prior to initial data
collection and on each subsequent day of a field campaign to determine the amount of baseline
signal due to the instrument's electronics and detector noise. All checks must be in accordance
with the method or test protocol being performed.
On the first day of a field study, a stray light instrument check should be performed. This involves
collecting, measuring, and identifying stray light as either background or instrument-related. All QC
checks must be conducted prior to actual data collection and the results must indicate that the
instrument is operating within the acceptable criteria range as specified in the method or protocol
appropriate for the field testing campaign.15 Typical quality control for this technology includes
method quality objectives of 10-25 percent accuracy depending on path length and a precision
target of 10 percent. Spectral quality is verified through the procedures and guidelines set by the
manufacturers and specific EPA method in use.16
In addition to the QC checks performed on the FTIR, the quality of the instrument signal
(interferogram) should be checked regularly during the field campaign. This is done by ensuring
that the intensity of the signal is at least five times the intensity of the stray light signal and
instrument noise. In addition to checking the strength of the signal, checks should be done
regularly in the field to ensure that the data are being collected and stored to the data collection
computer.15
-------
ORS Handbook
Section 2.0
Page 2-14
Data Quality Indicators for Precision and Accuracy for OP-FTIR
Instrument baseline noise and signal intensity are key data quality indicators for OP-FTIR
measurements. Some investigators evaluate the precision and accuracy of the PIC measurements
collected with an FTIR instrument by analyzing nitrous oxide concentrations in the atmosphere. A
typical background atmospheric marker concentration for nitrous oxide is about 315 ppb.17
However, this value may fluctuate due to seasonal variations in nitrous oxide concentrations or the
topographical elevation of the site.17
The precision of the OP-FTIR measurements should be evaluated by calculating the relative
standard deviation of ubiquitous IR active compounds (e.g., nitrous oxide) in each data subset.
A subset is defined as the data collected along one path length during one survey or sampling
episode.15 The number of data points in a data subset depends on the number of sample events
conducted in a survey. For a stable air parcel, the standard data quality indicator (DQI) criterion set
forth for precision is ฑ10 percent.15 The accuracy of the analyte PIC measurements can be evaluated
by comparing the calculated nitrous oxide concentrations from the data subsets to the typical
background concentration of 315 ppb.11 The standard DQI criterion for accuracy is ฑ25 percent.15
Example Applications and Vendors
Details on the OP-FTIR application of open path technologies are provided in Section 3 of this
Handbook. The OP-FTIR has been used for a wide variety of source emission measurements in the
field including applications such as line of sight optical remote, bLS modeling and RPM. Table 2-2
summarizes optical technologies and the typical applications of each of the technologies.
-------
ORS Handbook
Section 2.0
Page 2-15
Table 2-2. Typical Applications for OP-FTIR.
Technology
Applications
OP-FTIR
bLS, RPM, SOF, Tracer Gas Correlation,
TO-16
We are aware of multiple vendors that currently manufacture OP-FTIR units; two of these vendors
have verified their instrumentation through the EPA's ETV program.18,19 The cost of an OP-FTIR field
ready system can range from $75,000 to $120,000 in 2010 U.S. dollars, depending on configuration
and application. Gas standards used in fixed path length enclosed cells to confirm instrument
calibration can range between $300 and $500. Table 2-3 lists several of these vendors and indicates
which have verified their OP-FTIRs. The table also lists potential vendors for FTIR gas standards.
Table 2-3. FTIR Supply Vendors
OP-FTIR Instruments
KASSAY FSI
*Ail Systems Inc
www.kassay.com
*Spectrex, Inc.
http://www.spectrex-inc.com
IMACC Instruments
http://www.ftirs.com/
MIDAC Corporation
http://www.midac.com/
Bruker Optics
http://www.brukeroptics.com/opag.html
ABB/Bomem
http://www.abb.com/analytical
Gas Standard Suppliers**
Air Gas
http://www.airgas.com/
Linde
http://www.linde.com/
Matheson Gas
http://www.mathesongas.com/index.aspx
Spectra Gas
http://www.spectragases.com
Praxair
http://www.praxair.com/
*ETV Verified Technologies
** Requires gas regulator in addition to gas cylinder
In addition to instrumentation and gas standards, tables of absorbance coefficients are available
and standard reference spectra for numerous compounds can be purchased. Suppliers of reference
spectra include Pacific Northwest National Laboratory, which continues to develop the NWIR
spectral library of quantitative infrared absorption spectra13 and Harvard University, which
compiled the HITRAN database.14 These spectra have been measured under tightly controlled
conditions using state-of-the-art instrumentation.
-------
ORS Handbook
Section 2.0
Page 2-16
Strengths and Limitations
FTIR can be used as a qualitative tool to provide specific information about volatile IR energy
absorbing molecules. It can also be used as a quantitative tool to provide the concentration of
many gas-phase molecules. A summary of strengths and limitations is shown in Table 2-4 and Table
2-5. One of its limitations is that gas-phase water and CO^ are a very strong IR absorbing species.
In addition, water has strong absorption features in the 3200^4000 wave number range.5,17,20
Molecules that have coincident vibrational frequencies with water cannot be reliably analyzed
using frequencies in this range. FTIR is also limited to measuring gaseous compounds that absorb
IR radiation. Homonuclear diatomic gases such as nitrogen, oxygen, and halogen gases cannot be
measured by FTIR.
FTIR's major strength is that it can provide real-time, simultaneous analysis of multiple gaseous
contaminants 6 Additionally, the FTIR is a robust field instrument that allows for unattended
sampling for as long as a week period. Not only can the FTIR be used for open path concentration
measurement of a variety of contaminants, but it can also be used for leak and hotspot detection.
-------
ORS Handbook
Section 2.0
Page 2-17
Table 2-4. Summary Table of the OP-FTIR's Strengths
Feature
Strength
Economical
Relatively low instrument cost (about $80,000 - $125,000)
Low-cost long-term deployment
Compact Instrumentation
FTIR equipment is rugged and easily portable
Multiple Wavelength Operation
There are many compounds that are infrared
active (absorb IR light)
Large number of compounds can be analyzed simultaneously.
Spectra can be saved and post analyzed.
Ease of Calibration
No gas calibration standards necessary for field testing
(uses standard reference spectral library). Gas standards
are only needed for laboratory confirmation of instrument
performance and calibration.
Multiple Applications
FTIR can be used to locate discrete emissions hotspots
at a facility/area source
Multi-compound coverage makes FTIR ideal for leak detection or
source location where the facility being monitored has multiple
compounds present (e.g., chemical plants)
Automated Real-time
Measurements
Equipment can be allowed to run with minimal attention for
months at a time with remote access to check instrument
operation, schedule cryogen replenishment and recover data.
No sample collection, handling, or preparation is
necessary.
-------
ORS Handbook
Section 2.0
Page 2-18
Table 2-5. Summary Table o
: the OP-FTIR's Limitations
Feature
Limitation
Spectral Interferences
Gas-phase water spectral interference as well as CO
and CO2 interference5'16'17
Diatomoic Molecules
Not applicable to homonuclear diatomic gases
12 3
such as chlorine, oxygen, and nitrogen ' '
IR Wavelength Range
Weak IR absorption features for many inorganic
molecules such as sulfur dioxide and nitrogen
oxides6
Infrared beam has a limited range and may not
be sensitive enough to meet ambient data
quality objectives.
Path Length Range
Maximum path length is on the order of 400-500
meters
Field Implementation Requirements
Typical infrared detectors require cryogenic
cooling to operate. Liquid nitrogen used for
detector cooling must be refilled and maintained
regularly (weekly).
Field implementation and data collection requires
highly experienced personnel
Setup Time Consuming and Costly
Typical set-up time usually requires about 5 to 8
hours and a minimum of two people
Multiple vertical or horizontal path measurements
necessary to calculate plume flux, can require
significant time and cost to set up and implement
Measurement Limitations
Single beam open-path method measures
concentration along a path. The path must
capture most if not all an analyte plume to
provide accurate measure of emissions.
-------
ORS Handbook
Section 2.0
Page 2-19
References
1. Bell, R. J. 1972. Introductory Fourier Transform Spectroscopy. New York, NY: Academic
Press.
2. Hollas, M. J. 1992. Modern Spectroscopy. 2nd ed. New York, NY: John Wiley & Sons.
3. Laidler, J. K. 1995. Physical Chemistry. 2nd ed. Toronto, Canada: Houghton Mifflin
Company.
4. Galle, B., J. Samuelsson, B. Svensson, and G. Borjesson. 2000. Measurements of Methane
Emissions from Landfills Using a Time Correlation Tracer Method Based on FTIR Absorption
Spectroscopy. Environmental Science and Technology. 201, 35 (1): 21- 25.
5. U.S. EPA. 2009. Measurement and Monitoring Technologies for the 21st Century, CLU- IN.
http://www.clu-in.org/programs/21m2/openpath/op-ftir/
6. Liptak, B. 1995. Instrument Engineers' Handbook: Process Measurement and Analysis. Boca
Raton, FL: CRC Press.
7. Spellicy, R. 2003. IMACC Industrial Monitoring Corporation, Round Rock Texas.
http://www.ftirs.com/
8. Liptak, B. 2003. Instrument Engineer's Handbook: Process Measurement and Analysis.
Radnor, PA: Chilton Book Company.
9. Liptak, B. 1995. Instrument Engineers' Handbook: Process Measurement and Analysis. Boca
Raton, FL: CRC Press.
10. Minnich, T. and R. Scotto. 1999. Use of Open-Path FTIR Spectroscopy to Address Air
Monitoring Needs During Site Remediations. Invited Article Published in Remediation.
Summer 1999.
-------
ORS Handbook
Section 2.0
Page 2-20
11. U.S. EPA. 2007. Evaluation of Fugitive Emissions Using Ground-Based Optical Remote
Sensing Technology. EPA/600/R-07/032.
12. U.S. EPA. 2005. Evaluation of Fugitive Emissions at a Former Landfill Site in Colorado Springs,
Colorado Using Ground-Based Optical Remote Sensing Technology. EPA- 600/R-05/041.
13. U.S. DOE Pacific Northwest National Laboratory. 2009. Optics and Infrared Sensing.
Richland WA. http://infrared.pnl.gov/
14. Harvard University. 2009. The HITRAN Database. http://www.cfa.harvard.edu/HITRAN/
15. U.S. EPA. 2004. Optical Remote Sensing Facility Manual, U.S. EPA National Risk
Management Research Laboratory, Air Pollution Prevention and Control Division, Emissions
Characterization and Prevention Branch. Contract No. EP-C-04-023, Work Assignment 0-33.
16. U.S. EPA. 2006. Optical Remote Sensing for Emission Characterization from Non-Point
Sources. http://www.epa.gov/ttn/emc/prelim/otmlO.pdf
17. Godish, T. 2004. Air Quality, Fourth Edition. Boca Raton, FL: Lewis Publishers - CRC Press.
18. U.S. EPA and Battelle. 2000. Environmental Technology Verification Report: RAM 2000
Fourier Transform Open-Path Monitor. Report prepared by AIL Systems, Inc.
htt p://www.epa .gov/etv/pu bs/0l_vr_a iI. pdf.
19. U.S. EPA and Battelle. 2001. Environmental Technology Verification Report: SafEye 227
Infrared Open-Path Monitor. Report prepared bySpectrex, Inc.
http://www.epa.gov/etv/pubs/01_vr_safeyetwo.pdf.
20. Jaakkola, P., T. Vahlman, A. Roos, P. Saarinen, J. Kauppinen. 1998. On-line Analysis of Stack
Gas Composition by a Low Resolution FT-IR Gas Analyzer. Water, Air, & Soil Pollution. 101
(1-4): 79-92.
-------
ORS Handbook
Section 2.0
Page 2-21
2.2 Tunable Diode Laser
Light amplification by stimulated emission of radiation (LASER) is a technique to generate a narrow
wavelength of light with a small cross-sectional area. Diode lasers generate this beam of light using
a semiconductor material that emits light when electrical current is "injected" into the
semiconductor junction. When the TDL was first introduced, its measurement applications were
limited to laboratory functions because of the instrument functionality and cost. The rise of the
fiber-optics communication industry in the 1980s led to the development of open-path TDL (OP-
TDL) instrumentation that is compact and affordable.1 Since that time, the OP-TDL has become
recognized as a reliable technology for use in the field for in situ measurement of a variety of
gaseous pollutants. New laser development demonstrated in 1994 using a repeated stack of thin
semiconductor layers (Quantum cascade lasers) offers the possibility to produce laser beams at
additional wavelengths and add to the list of compounds that can be measured.2
Laser-based gas detectors are now being used in a wide variety of applications for process and
quality monitoring, and safety and environmental compliance. Laser detectors combine
semiconductor TDLs and optical fibers developed by the telecommunications industry with
detection techniques based on frequency or wavelength modulation (similar to radio). Laser
detectors measure gas concentrations by shining a laser beam through a sample of gas and
measuring the amount of laser light absorbed. Lasers emit light at a single wavelength. In TDLs,
the wavelength can be "tuned" over a small range to match the exact absorption wavelength of a
target compound by adjusting temperature and bias current. The wavelength of the laser is tuned
over a selected absorption feature of the target species. The measured absorption spectra are
recorded and, combined with measured gas temperature and pressure, effective path length, and
known line strength, used to determine a quantitative measurement of concentration. These
properties give laser detectors a combination of selectivity, sensitivity, dynamic range and rapid
response time. The OP-TDL can make quantitative measurements of select gases based on the
principals of Beer-Lambert law. Gas molecules absorb energy at specific wavelengths based on
rotational and vibrational motion within the molecule. By measuring the energy absorbed for a
-------
ORS Handbook
Section 2.0
Page 2-22
compound-specific wavelength over a laser's path, the OP-TDL can determine the concentration
present of a specific gaseous compound. This technology can be used for several open-path and
point monitoring applications.
The OP-TDL is a relatively inexpensive technology that emits very narrow wavelengths in the near IR
ranges. While mid-IR wavelength lasers are available, they are much more difficult to operate or
are currently cost prohibitive for general use. Because the wavelength emitted is very narrow and
can be chosen specific to a vibration or rotation of a specific compound, the OP-TDL eliminates
most interference. Lack of interference and high intensity of the laser beam allows longer open
path lengths, up to 1 to 2 km and therefore, higher sensitivity for the compounds TDL can measure.
The near-IR OP-TDL units currently in use are limited by the small number of compound specific
wavelengths available from commonly available TDLs and the necessity to use a different TDL for
each compound of interest.
Basic Operation
TDL Absorption Spectroscopy instruments rely on spectroscopic principles and sensitive detection
techniques, coupled with advanced diode lasers and optical fibers developed by the
telecommunications industry. Gas molecules absorb energy at specific wavelengths in the
electromagnetic spectrum. At wavelengths slightly different than these absorption lines, there is
essentially no absorption. Measurement of the relative strengths of off-line to on-line transmission
yields a precise and highly sensitive measure of the target gas concentration along the path
transited by the laser beam. Measurements are made by (1) transmitting a beam of light through a
gas mixture sample containing a quantity of the target gas, (2) tuning the beam's wavelength to
one of the target gas's absorption lines, and (3) accurately measuring the absorption of that beam.
The concentration of target gas molecules can then be integrated over the beam's path length.
While results generated by traditional optical instrumentation are generally in concentration units
such as ppb, the output generated by the OP-TDL, like all open-path technologies, represents units
of concentration over distance, such as ppb(m). This is also known as a PIC.
-------
ORS Handbook
Section 2.0
Page 2-23
Each gaseous compound absorbs energy at different wavelengths, usually more than one,
depending on vibrational and rotational excitement within the molecule. Therefore, each
compound has its own "signature" of bands from which energy may be absorbed. Each band is
highly selective, with virtually no absorption occurring outside of a specific wavelength. Because the
OP-TDL emits a laser at a very narrowly tuned wavelength range, it is an ideal instrument for single
compound measurement. The OP-TDL's laser is selected for an overtone band specific to the
compound of interest. The absorption of energy over the laser's path length is measured by the
instrument's detector. The absorption is used to determine the concentration of the target gaseous
compound using the principals of Beer-Lambert Law as described in the equation below.
A =ฃ*c*l
Where:
A = absorbance intensity
s = absorption coefficient
c = sample concentration
I = sample path length
There are many instrumental configurations for OP instruments. The simplest OP-systems are
bistatic configurations. The arrangement of the components of this design for OP-TDL is shown in
Figure 2-9. This configuration derives its name from the fact that both the transmitter and receiver
must be fixed in a static position and precisely aimed at each other. The OP-TDL equipment projects
the laser beam directly along a path to a detector/receiver. Bistatic configurations, in general, have
the requirement of supplying power at both the receiver and transmitter, which can be a
disadvantage in some locations. Additionally, there is a requirement for alignment at both receiver
and transmitter, which can be time-consuming for mobile systems.3
-------
ORS Handbook
Section 2.0
Page 2-24
Receiving,
Q$
-------
ORS Handbook
Section 2.0
Page 2-25
ซs.
Transmitting Laser1 Path
---
" ( _s)
^Return Laser Patlj....-^
Translating
Retroreftector
s '
Transmitting
Optics
Tunable
Owd* I
Source
Receiving
Optics
r o
Detector
Absorbing Medium
I
C
<
Retroroftoctor *
Laser Path
i
r
"
Additional
Beam Splitter
OxxJe Laser
OpOcป
Source
~ t
Detector
Figure 2-10. TDL Monostotic Configuration
cx
I t - ray
Figure 2-11. The corner reflector cube
In the mono-static mode, the IR laser beam is split twice, once leaving the OP-TDL and once on its
return. This design requires a beam splitter in the optical path that removes 50 percent of the light
from the outgoing beam and 50 percent of the light from the return beam for an overall loss of 75
percent of the total light intensity.
The dual-telescope monostatic configuration has lower detection limits because it does not utilize
a beam splitter in the optical path. A translating retro-reflector, which is essentially a portion of a
-------
ORS Handbook
Section 2.0
Page 2-26
very large cube, is used to return the light beam offset to align with the receiving telescope.
This single, large retro-reflector does not have the divergence reversal properties of the corner-
cube array. The second telescope adds cost and complexity to the system.3 However, when
compared with mono-static mode, bi-static systems are harder to align and maintain because any
shift in the transmitter or detector can result in system misalignment.4'5 Both operating modes
measure only the compounds that are in the beam path. Emissions outside the beam path are not
measured. In these situations, measurements have been conducted along multiple beam paths to
more accurately characterize the emission plume.
OP-TDL units are designed to operate under computer control, where the interfacing software
controls the function of the OP-TDL, controls the tuning of the laser, and collects resulting data
from the detector. Commercially available OP-TDL units can be equipped with multiple lasers,
allowing the measurement of several compounds at one time. Field units typically include a
hardware controller, a laptop, a telescope receiver, and a reflector. Instruments can run
unattended via computer control for extended periods of time.6
Pollutants and Relative Levels That Can Be Measured
Near-IR TDLs have been used to measure approximately 20 compounds that have absorbencies in
the 1.4 - 1.8 micrometer (|am) wavelength range. Using an open-path setup, concentrations into
the low ppm range can be detected over a path length of approximately 1000 m to 2000 m. Table
2-6 lists airborne compounds that can be measured by OP-TDL systems and their approximate
wavelengths. The compounds measured by TDLs are limited by the wavelength range commonly
available using electrical current driven semiconductor lasers. Quantum Cascade Laser (QC-Laser)
development offers the possibility of expanding the list by extending available laser wavelengths
into the mid-infrared range, where many compounds of interest strongly absorb these
wavelengths. However, one issue is that the measurements are also limited by the ability of fiber
optic cables to transmit the raw LASER energy in those instruments using remote modules. Current
TDL light sources cost $2,000 to $3,000. Experimental QC-Lasers are available at a cost up to
-------
ORS Handbook
Section 2.0
Page 2-27
$100,000.
Table 2-6. Example List of Gaseous Compounds Measured by Near IR OP-TDL Systems
Species
Approximate near-IR
\ (nm)
Reported Detection
Limit (ppm-m)
ammonia
760,1500
0.5-5.0
carbon monoxide
1570
40-1,000
carbon dioxide
1570
40-1,000
hydrogen chloride
1790
0.15-1
hydrogen cyanide
1540
1.0
hydrogen fluoride
1310
0.1-0.2
hydrogen sulfide
1570
20
methane
1650
0.5-1
nitric oxide
1800
30
nitrogen dioxide
680
0.2
oxygen
760
50
water
970, 1200, 1450
0.2-1.0
acetylene
1520
These compounds are not
commonly measured;
therefore, detection limits
are not readily available.
ethylene
1693
formaldehyde
1930
hydrogen bromide
1960
hydrogen iodide
1540
nitrous oxide
2260
phosphine
2150
propane
1400, 1500, 1700
Typical QA/QC
Three major QA requirements are necessary when using a TDL system: (1) selection of the
appropriate laser and absorption line for the compound of interest, (2) establishment and use of
appropriate calibration procedure, and (3) establishment of QC procedures that ensure the
instrument's performance as measurements are made.6
Selection of the Laser and Absorption Line
A TDL optical system is typically built to generate one wavelength at a time. The range of
wavelengths from each type of laser limits measurement to one compound at a time. Therefore,
laser and instrument selection must be carefully considered. A few lasers can be configured for
one of a limited range of wavelengths, while others provide a wider selection of wavelengths. It
also important to note that many compounds have multiple absorption bands in both the near-
-------
ORS Handbook
Section 2.0
Page 2-28
and mid-IR regions. However, the availability of mid-IR lasers is limited and may not be available
for open-path monitoring or measurement programs. Table 2-7 lists commercially available lasers
producing wavelengths in the near-IR range. Table 2-8 lists other laser types that have been
developed for mid-IR applications. While this list covers most of the lasers available, TDLs
represent a limited set from a larger array of laser types.
Ta
ble 2-7. Near-IR Laser Types Available for OP-TDL Systems
Laser Type
Tunable \ Range
(nm)
Target Compounds
InGaAsP
1200-2000
CO, CO2, NO, CH4, C2H2, HF, HCI, HBr, HI, HCN,
NH3, H2CO, PH3, H20
Antimonide*
2000-4000
CO, C02, NO, N20, CH4, HCI, HBr, H2CO
*Laser emits wavelengths in both the near-IR and mid-IR spectrums.
Table 2-8. Potentially Usable Mid-IR Lasers
Laser Type
Tunable \ Range
(nm)
Target Compounds
AIGalnP
630-690
N02
AIGaAs
750-1000
02, NH3
Vertical Cavity
650-1680
H20, C2H2, HF, H2S, 02, H20, NH3
Antimonide*
2000-4000
CO, CO2, NO, N20, CH4, HCI, HBr, H2CO
Quantum Cascade**
4000-12000
H20, CO, CO2, NO, N02, N20, S02, C2H2,
HCN, NH3, PH3, 03
Lead-salt**
3000-30000
H20, CO, C02, NO, N02, N20, S02, CH4, C2H2,
HCI, HBr, HCN, NH3, H2CO, PH3, 03
*Laser emits wavelengths in both the near-IR and mid-IR spectrums.
** Laser emits wavelengths in the mid-IR spectrum
Because compounds often have multiple absorption bands that can be detected by a TDL system, it
is also important to consider which band is best for quantitative purposes.7 Higher intensity
absorption bands provide the best sensitivity. However, interference from other compounds may
eliminate the use of the most sensitive wavelengths. It also may be worthwhile to measure the
concentrations from a second absorption band to verify the nonexistence of interferences. It is
-------
ORS Handbook
Section 2.0
Page 2-29
highly unlikely that the same interference would exist for both absorption bands.6
Calibration
In a closed-gas cell TDL instrument, known concentrations of the compound of interest are
introduced into the white cell used for sample analysis. Calibration gas is added through the same
line used to collect the sample. Varying concentrations of one compound can be introduced by
adjusting the inlet flow of the calibration gas relative to the dilution gas. For each concentration
step in the calibration curve, the absorption trend should be recorded and the mean and standard
variation calculated.8
The calibration factors are typically determined in the laboratory with short path length gas cells.
One instrument vendor provides an insertion slot that can contain a gas cell of known
concentration into the path of the optical beam during measurement. Other OP-FTIR instruments
can also be calibrated with gas cells of known concentration by introducing the cell into the laser's
path for measurement. Field calibration checks can be accomplished using the absorption signal
provided by the calibration gas cell added to the open path field absorption signal. The signal
increases above the open path signal, proportionally to the gas concentration and path length of
the gas cell. The instrument response is checked using the difference of the measurements with
and without the gas cell. An example of a calibration curve for methane is provided as Figure 2-12.
-------
ORS Handbook
Section 2.0
Page 2-30
SOOO
SOOO
' 4-GOO
_ a-
3000
a_
Q>
y = 1.12x - 219.6
R2 - 0-995
1000
0
lOOO
Ejected PIC Addition (ppm-m)
Figure 2-12. Calibration Data for an OP-TDL System
Calibration frequency depends on the duration of the measurement period as well as the
concentrations of compounds that are measured. Shorter term measurements projects need
calibration verification at the beginning of a measurement episode. Also note that regulatory
requirements may also dictate calibration frequencies. Low concentrations in ambient conditions
may require background and calibration determinations on a weekly or monthly basis because a
small drift in instrument response is more significant at lower measured concentrations.
Quality Control Procedures
Each OP-TDL manufacturer recommends its own QC procedures; however, it is necessary to verify
the accuracy of the calibration throughout a set of field measurements. This can be done by
reinserting a calibration standard cell periodically during a measurement episode to ensure correct
measurement. Recalibration during field measurements may be necessary due to instrument drift
and is typically performed using the instrument's system software.
Example Applications and Vendors Applications
Details on the near-IR TDL application of open path technologies are provided in Section 3 of this
-------
ORS Handbook
Section 2.0
Page 2-31
Handbook. The OP-TDL has been used for a wide variety of source emission measurements in the
field including applications such as line of sight optical remote, bLS modeling, RPM, and mobile
tracer release correlation. Table 2-9 summarizes optical technologies and the typical applications
of each of the technologies.
Table 2-9. Typical Applications for OP-TDL.
Technology
Applications
OP-TDL
bLS, RPM, Tracer Gas Correlation, TO-16
Vendors
While there are many sources for TDL instrumentation that is suitable for laboratory applications,
there are only a few vendors currently offering field ready OP-TDL instrumentation. Vendors
offering instrumentation exclusive to fire detection and monitoring have not been included. The
cost of a TDL field ready system can range from $35,000 to $75,000 in 2010 U.S. dollars depending
on the configuration and application. Table 2-10 lists example vendors and their internet contact
address.
Table 2-10. Near-IR OP-TDL Vendors
Vendors
Boreal Laser
www.boreal-laser.com
OPSIS AB
www.opsis.se
Leister Process Technologies, Axetris
Division
www.ir-microsystems.com
Norsk Elektro Optikk (NEO, Norway)
www.neo.no
PKL Technologies, Inc.
www.pktechnologies.com
PSI Physical Sciences, Inc.
www.tdlas.comwww.psicorp.com
Senscient
www.senscient.com
Simtronics group
www.simtronics.eu
Unisearch Associates, Inc. (Concord, Canada)
www.unisearch- associates.com
Strengths and Limitations
The TDL has an array of strengths and limitations that must be considered for each OP-TDL
application. A summary of strengths and limitations is shown in Table 2-11 and Table 2-12,
-------
ORS Handbook
Section 2.0
Page 2-32
respectively. Perhaps the most striking limitation is the fact that each TDL laser can detect only one
compound at a time and each laser can scan only a limited range of wavelengths. It is also true that
only compounds with overtone absorbencies in the near- and mid-IR ranges can be detected and
quantified, of which there are approximately twenty.9 The instrument's sensitivity is limited
because of noise created by the laser10, though this can be improved by either of the modulations
described above. However, because the laser emits such a narrow bandwidth, interferences from
other gaseous compounds are unlikely and limited to compounds with absorbance at that wave
number.
Table 2-11. Summary Table of the TDL's Strengths
Feature
Strength
Automated Real-time Measurements
24/7 remote monitoring
Can provide near real time data
Unattended measurements collection
Compact instrumentation
Field units are lightweight, typically under 75
Kg, and relatively easy to transport and setup
Economical
0.5 to .01 the cost of alternative technologies
High intensity light source
Wide linear response over a wide dynamic
range resulting in measurements from 0.1
to 1000 ppm
The ability to measure longer path lengths (1
km compared to other ORS technologies)
Solid state technology
Robust field use with low maintenance,
minimal consumables to operate
Low response times
Rapid response - typically 1 second
High spectral resolution
Minimizes interference from other gases
resulting in high compound specificity
Uses fiber optics for signal processing
Lower equipment cost per measurement,
ability to multiplex signals
Vendor-specific calibration cells
Self-calibration, zero and span drift correction
The TDL's strengths in field application are numerous. Technological developments originating
-------
ORS Handbook
Section 2.0
Page 2-33
from the fiber optics communication field have allowed the TDLto become compact, robust, and
economical compared to other technologies. The TDL can be used for several applications,
including open-path, RPM, and cavity ring-down spectroscopy (CRDS) measurements. The high-
powered laser source also promotes fast instrument response times (as low as one measurement
per second) and longer path lengths up to 1,000 meters.
Table 2-12. Summary Table of the TDL's Limitations
Feature
Limitation
Single wavelength operation
Detects only one compound per laser, fewer
measurable compounds, and limited
sensitivity
Mid-IR wavelength range
Quantitation limited to compounds with
overtone absorbencies in the near- and mid-
IR range
Dust and objects block the laser beam
With all open path optical measurements,
blocked beams result in no measurements
References
1. Harris, Shores and Thoma. 2007. Using Tunable Diode Lasers to Measure Emissions from
Animal Housing and Waste Lagoons, 16th Annual International Emission Inventory
Conference Emission Inventories: "Integration, Analysis, and Communications"
Raleigh.. May 14-
17.
2. Faist, Jerome, Federico Capasso, Deborah L. Sivco, Carlo Sirtori, Albert L. Hutchinson,
and Alfred Y. Cho. 1994. "Quantum Cascade Laser" Science 264 (5158): 553-556.
3. Liptak, B. G. 2003. Instrument Engineer's Handbook: Process Measurement and Analysis.
Radnor, PA: Chilton Book Company
4. Liptak, B. G. 1995. Instrument Engineers' Handbook: Process Measurement and Analysis.
CRC Press.
5. Spellicy, R. 2003. IMACC Industrial Monitoring Corporation, Round Rock Texas.
-------
ORS Handbook
Section 2.0
Page 2-34
6. Schiff, H. I. 1987. Measurement of Atmospheric Gases byTunable Diode Laser
Absorption Spectrometry. Monitoring of Gaseous Pollutants byTunable Diode Lasers:
Proceedings of the International Symposium held in Freiburg, F.R.G, 13-14 November
1986. D. Reidel Publishing Company.
7. Ku, R. T., E. D. Hinkley, and J.O. Sample. 1975. Long-Path Monitoring of Atmospheric
Carbon Monoxide with a Tunable Diode Laser System. Applied Optics. 14 (4).
8. Mock, A., C. Roller, J. Jeffers, N. Khosrow. P. McCann, and J. Grego. 2001. Real-time
ground level atmospheric nitric oxide measured by calibrated TDLAS system. Optical
Society of America.
9. Thoma, E. D., Shores, R. C.; Thompson Jr, E. L.; Harris, D. B.; Thorneloe, S. A.; Varma, R.
M.; Hashmonay, R. A.; Modrak, M. T.; Natschke, D. F. and Gamble, H. A. 2005. Open-Path
Tunable Diode Laser Absorption Spectroscopy for Acquisition of Fugitive Emission Flux
Data. J. Air & Waste Manage. Assoc. 55: 658-668.
10. Cappellani, F., G. Melandrone, and G. Restelli. 1987. Post Detection Data Handling
Techniques for Application in Derivative Monitoring. Monitoring of Gaseous Pollutants
byTunable Diode Lasers: Proceedings of the International Symposium held in Freiburg,
F.R.G, 13-14 November 1986. D. Reidel Publishing Company.
-------
ORS Handbook
Section 2.0
Page 2-35
2.3 Ultraviolet Differential Optical Absorption Spectroscopy
The UV-DOAS is an optical remote sensing technology that quantifies concentrations of gaseous
compounds by measuring the absorption of UV light by chemical compounds in the air and applying
the Beer-Lambert law.1
A significant strength of the UV-DOAS is its extremely long path-length capability-typically 500
meters with some research applications up to 10 kilometers.2 UV-DOAS has been deployed in a
wide variety of environmental measurement applications. It is most frequently used to measure or
monitor criteria and smog-related air pollutants. It is also able to accurately monitor several
pollutants that do not produce ideal IR absorption bands. However, because the absorption bands
for UV-DOAS are very wide, there are many compounds that cannot be accurately quantified by UV-
DOAS. Nitrogen and oxygen molecules in the air cause broad spectral scattering and interfere with
many of the compounds that can be measured. The UV- DOAS is reported to have detection limits
in the low (ppb) range and can reach parts per trillion in some research applications when used with
optimum measurement path lengths.2
Basic Operation
In general, UV, visible, and near-IR light is that radiation within the 180-780 nanometer wavelength
range that causes changes in energy between the bonding electrons in molecules that absorb the
light. While wavelength ranges produced by UV-DOAS instrumentation include the rotational and
vibrational transitions caused by near-IR light, the typical application of UV- DOAS restricts the UV
light to a wavelength range of 245 to 380 nanometers. Due to the range of excitations measured,
molecular absorption bands tend to be far broader than that of IR instrumentation. Compounds that
can be accurately detected and measured with the UV-DOAS possess specific chemical structure
characteristics that allow for unique absorption bands, which limits the number of compounds that
can be monitored.3
DOAS is based on the principal that the Beer-Lambert Law (Equation below):
-------
ORS Handbook
Section 2.0
Page 2-36
A = ฃ*C*I
Where:
A = absorbance intensity
s = absorption coefficient
c = sample concentration
I = sample path length
Interferences in the atmosphere cause absorption to occur at all points in the measurement
spectrum. In the atmosphere, the light of the beam undergoes extinction processes by air
molecules and aerosols, turbulence, and absorption by many trace gases. DOAS overcomes the
effects of the beam extinction by mathematically separating and removing the nonspecific beam
extinction from the target gas absorption.4 To address this issue, DOAS measures the difference
between the absorption peak caused by the compound of interest and absorption peaks at
wavelengths on either side of that targeted peak.3 The concentration is determined by the light
intensity in the absence of a structured absorption band, rather than the light intensity in the
absence of all absorption.
A typical UV-DOAS system consists of a light source, optics, a spectrometer, and depending on the
system configuration, a retro-reflector. Most systems employ a tungsten halogen or xenon arc
lamp, though some use deuterium lamps.2 From the source, the light is focused and directed into
the atmosphere by means of a transmitting telescope. A receiving telescope retrieves and focuses
the attenuated light beam and the spectrometer measures the change in absorbance caused of the
UV light. Data collected by the UV-DOAS can be stored in the analyzer and can be transferred off-
site via external storage or Internet connection.5 The digital signal from the spectrometer is
-------
ORS Handbook
Section 2.0
Page 2-37
collected by a computer system arid compared to laboratory-developed reference spectra to ensure
a match between all absorption bands associated with a targeted compound are present to confirm
its identification and quantification.7 Some technologies use specific gas calibrations to fine tune
the library reference spectra and improve instrument performance. Figure 2-13 shows an Opsis
DOAS6 source unit.
Figure 2-13. OPSIS DOAS Instrument
UV-DOAS instruments can practically measure path lengths up to 500 meters. Optimum light path
length depends on the compound of interest, the desired detection limit, the clear line of sight
available, and the expected interferences (e.g., dust and fog). Measurement noise increases and
beam intensity decreases as path length increases.1
Certain chemical species can also pose interference issues at particular wavelengths. For example,
when trying to measure nitrous oxide (N^O) in the presence of other nitrogen oxides (NO, NO^),
absorption from NO and NO,, can cause interference.8 Special considerations must also be made
when measuring concentrations of aliphatic hydrocarbons in ambient air since oxygen is a major
interferent for these compounds.5
Additionally, there are several operational concerns that must be considered when operating a UV-
DOAS in the field. These instruments are approved for use in temperatures of 5 - 305C with
humidity ranging from 0-80 percent. High humidity can cause fog to build up on the receiver
-------
ORS Handbook
Section 2.0
Page 2-38
mirrors and windows, which can substantially decrease the detected light intensity and deteriorate
the condition of the instrument optics and mirrors. This can be corrected by installing heaters on
the mirrors and windows or by changing the site where the UV-DOAS is installed.5
UV-DOAS Field Implementation
There are many instrumental configurations for open-path UV-DOAS instruments. UV-DOAS
instrumentation can be deployed in both bistatic and monostatic open-path configurations.
The simplest OP-systems are bistatic configurations. The arrangement of the components of this
design for UV-DOAS is shown in Figure 2-14. This configuration derives its name from the fact that
both the transmitter and receiver must be fixed in a static position and precisely aimed at each
other. The UV-DOAS equipment projects the light beam directly along a path to a
detector/receiver. Bistatic configurations, in general, have the requirement of supplying power at
both the receiver and transmitter, which can be a disadvantage in some locations. The receiver
and transmitter must be accurately aligned to optimize signal intensity.9
RooEr/rig
Cptb
Figure 2-14. Bistatic Configuration of UV-DOAS
Monostatic configurations were developed to address issues raised with bistatic designs. In a
monostatic configuration, all the optical components of the transmitter and receiver are in the
same location and a retro-reflector is used to return the light from the transmitter to the receiver.
A noted disadvantage of a monostatic system is that the physical path is only half the distance of a
-------
ORS Handbook
Section 2.0
Page 2-39
bistatic system.
"Retro-reflecting" mirrors are configured with three perpendicular reflective surfaces in the shape
of a corner. A combination of three mutually perpendicular mirrors reflects light incident from any
direction through 180ฐ as shown in Figure 2-15. Such a combination of mirrors is called a "corner-
cube reflector." Corner-cube reflectors beam light back to its exact point of origin. This property
reduces the divergence of the beam on its return path back to the detector compared to
divergence that would result from a flat mirror. Also, the retro-reflector array can be very large to
capture and return essentially the entire divergent signal from the telescope.
CL
a
m.ir-r-t->r-
Figure 2-15. The Corner-cube Reflector
The monostatic configuration derives its name from the fact that only the transceiver portion of
the instrument needs to be precisely aimed because the retro-reflector returns light to its source.
-------
ORS Handbook
Section 2.0
Page 2-40
RelfOfcVcior
ฉ
Figure 2-16. Basic Setup Used to Make Monostotic Open-path UV-DOAS Measurements
When compared with monostatic mode, bi-static systems are harder to align and maintain because
any shift in the transmitter or detector can result in system misalignment.10'11 Both operating
modes measure only the compounds that are in the beam path. Emissions outside the beam path
are not measured. Siting and additional QA procedures for ambient measurements found in 40 CFR
Part 58 provide basic guidance for criteria pollutants using open-path measurements. In addition,
EPA QA Handbook Volume 1112 has siting requirements and other useful information on using
UVDOAS in ambient/background monitoring situations.
Passive UV-DOAS
A third configuration is known as passive UV-DOAS. Passive UV-DOAS uses ambient lighting, such
as sunlight, as its light source and does not require a transmitting telescope. Passive UV- DOAS
instruments can be fitted into balloons and used to measure concentrations of pollutant gases at
differing heights in the atmosphere.12
Pollutants and Relative Levels That Can Be Measured
Table 2-13 lists compounds that have been measured with UV-DOAS systems. The list is not
exhaustive and includes only compounds reported in recent literature.2,12 UV-DOAS systems have
the most widespread environmental use in the detection and measurement of inorganic gases and
vapors, monoaromatics (i.e., benzene), and aldehydes.
-------
ORS Handbook
Section 2.0
Page 2-41
Table 2-13. S
pedes Measured with UV-DOAS Systems
Species
1,3-Butadiene
Formaldehyde
Ozone
Acrolein
Hydrogen Fluoride
Sulfur Dioxide
Ammonia
Isoprene
Styrene
Benzene
Mercury
Toluene
Carbon Disulfide
Nitric Oxide
m, p-Xylene
Chlorine
Nitrogen Dioxide
o-Xylene
Ethyl Benzene
Nitrous Acid
Detection limits have been reported in the ppb with at least one research application reporting
cases of detection down to parts per trillion ranges.2 Detection limits vary based on factors such as
the deployment configuration, light path length, measurement noise, and meteorological
conditions.1,2 Table 2-14 gives example detection limits found in the literature.
Table 2-14. Approximate Detection Limits for UV-DOAS
Pollutant
Lower Detection Limit
(ppb)
Path Length (m)
Ammonia
800
200
Benzene
single digit ppb
500
Carbon Disulfide
500
5000
Formaldehyde
single digit ppb
500
Nitrous Acid
single digit ppb
500
Nitrogen Dioxide
single digit ppb
1000
Nitrogen Oxide
240
200
Ozone
single digit ppb
1000
Sulfur Dioxide
single digit ppb
1000
Toluene
single digit ppb
200
m, p-Xylene
10
500
o-Xylene
single digit ppb
500
Typical QA/QC
QA/QC ensure the validity of data and calculations performed by UV-DOAS systems. Each
instrument manufacturer establishes its own quality assurance procedures based on the
specifications of the individual instrumentation, but there are several procedures that should be
-------
ORS Handbook
Section 2.0
Page 2-42
followed universally.
Record Keeping
As with all environmental measurements, it is necessary to keep accurate records during
measurement periods to ensure accurate data collection. For UV-DOAS, information such as
meteorological conditions, path lengths, UV filter numbers, lamp type, light intensities and
measurement times should be recorded.5 Light intensities must be recorded anytime the UV
emitter or receiver is adjusted and compared to the intensities measured when the UV-DOAS was
installed. The measured recoveries of standard gas cells with known concentrations should also be
documented.
Instrument Performance
Before measurements are begun and throughout the measurement process, several instrument
performance checks are required to make sure the instrument is accurately collecting data.
Individual vendors recommend specific instrument performance checks such as correcting for slight
variances in the reference spectrum (i.e., the lamp spectrum with no concentration bands) caused
by changes in the spectrometer and instrument electronics. This is performed by periodically re-
recording the lamp spectrum and comparing it to the initial reference spectrum for agreement. The
reference spectrum is critical to the analysis of collected data and performing regular reference
checks also minimizes noise collected by the instrument.5
Calibration checks are also very important for the collection of accurate data. The analyzer is
checked by measuring gas standards of known concentrations for accuracy. Calibration cells are
filled with the gas standard, allowed to stabilize, and the absorption is measured. A valid
calibration curve should contain six equally-spaced calibration points, including zero, and cover at
least 80 percent of the perceived measurement range. Because UV-DOAS measurements are
based on absorption and, therefore, the number of target compound molecules in a specific path
length, the calibration points can be obtained either by decreasing the measurement path or
-------
ORS Handbook
Section 2.0
Page 2-43
diluting the gas standard.5
A function check is also required to periodically validate the instrument's performance. During a
function test, a cell with a known concentration of gas is placed in front of the receiver. The
instrument measures the concentration of the compound in the cell plus the concentration of the
compound in ambient air. This check serves two purposes: (1) to ensure that the analyzer is
producing accurate measurements and (2) to ensure that no cross-sensitivities occur between the
test gas in the cell and other gases. Function tests must be performed in stable ambient pollution
conditions because spikes in pollutants may cause the results of the function test to be difficult to
interpret.5
Accuracy and precision tests are defined by the EPA regulations at 40 CFR part 58 (Ambient Air
Quality Surveillance). Often these values are determined by performing calibration checks against
known gas standards and verifying the MDL provided by the instrument manufacturer. Instrument
manufacturers provide their own instructions on how to perform accuracy and precision tests in
accordance with EPA regulations.
Example Applications and Vendors
UV-DOAS has been deployed in a wide variety of environmental measurement applications which
are discussed for specific applications in Chapter 3 of this Handbook. UV-DOAS is most frequently
used for monitoring smog-related air pollutants, where its long range is used to verify Eulerian
models that are used in air quality management. Multiple pathways have been used to create 2
dimensional (2-D) and 3 dimensional (3-D) tomographic depictions of pollutants around a large area
source or urban area. UV-DOAS is used for fenceline monitoring of air pollutant emissions. Benzene
has been measured in residential areas downwind of chemical manufacturing plants and ammonia
has been monitored in areas around large-capacity swine feeding operations. UV-DOAS was also
used to measure mercury emissions from a chloro-alkali plant.7 These types of ambient and
fugitive or area source measurements have been useful to government agencies to identify places
where harmful levels of pollution exist and determine the level of injunctive relief necessary.2
-------
ORS Handbook
Section 2.0
Page 2-44
Table 2-15 summarizes those applications that utilize UV-DOAS technology.
Table 2-15. Typical Applications for UV-DOAS.
Technology
Application
UV-DOAS
OTM 10, Tracer Gas, bLS
One developing application for UV-DOAS is known as Multi-Axis Differential Optical Absorption
Spectroscopy (MAX-DOAS). The application provides the ability to derive a vertical profile of
pollutants by completing multiple scans simultaneously using either passive or active techniques.2
Vendors
There are currently five vendors of field-ready UV-DOAS instruments, as summarized by Table 2-16.
Opsis, Inc., has had two separate instruments verified through the EPA's ETV program.13 The Opsis
System has also been designated as an "Equivalent Method" for the measurement of SO^, NO^
and Og in ambient air.
Table 2-16. UV-DOAS Vendors
Vendors
Websites
Argos Scientific
www.argos-sci.com
Environnement S.A. Sanoa UV/Visable DOAS
www.environnement-sa.com
ETG Risorse eTecnologia
www.etgrisorse.com
IMACC
www.ftirs.com
Opsis, Inc.
www.opsis.se
Spectrex
www.spectrex-inc.com
Cerex Monitoring Solutions
www.cerexms.com
Strengths and Limitations
Tables 2-17 and 2-18 summarize the strengths and limitations associated with the use of the UV-
DOAS. Some UV-DOAS can provide concentration data for up to three compounds simultaneously.
In the field, the instrument is portablel5 and can be deployed long-term for continuous remote in
situ monitoring.13 UV-DOAS quantifies compounds more successfully that have strong UV light and
weak infrared absorption characteristics. Since NO^ measurement by UV-DOAS does not require
conversion to NO and measurement by difference, as conventional chemiluminescent monitors
operate, this ORS technology provides a direct rather than indirect measurements.
-------
ORS Handbook
Section 2.0
Page 2-45
Table 2-17. Summary of UV-DOAS Strengths
Feature
Strengths
Automated Real-time Measurements
24/7 remote monitoring
Can provide near real time data
Unattended measurements collection
Economical
Relatively low instrument cost (about
$60,000 - $200,000)
Low-cost long-term deployment
Multiple Wavelength Operation
Broad spectrum instruments allow monitoring of
three criteria pollutants and trace species
simultaneously. Spectra can be saved and post
analyzed
Range of Measurement
Long measurement path length - up to 500 m.
Detectability
Many compounds are detectable in the low ppb
range
Compact Instrumentation
Portable
However, the number of species that can be analyzed with UV-DOAS is limited due to the lack of
appropriate absorption characteristics in the UV-visible wavelength range of many compounds.
Ta
ble 2-18. Summary of UV-DOAS Limitations
Feature
Limitations
Difficulty in Deployment
Alignment of remote receiving optics or reflectors can be difficult at
long path length.
Meteorological Limitations
Fixed observation area. Long term deployment depends on
constant wind direction.
Affected by poor visibility conditions
Limited Compounds
Many species do not have appropriate UV-visible absorption
structures making them undetectable by UV-DOAS
Application Limitations
Some bistatic systems are more difficult to use for radial plume
mapping due to difficulty aligning optics from multiple paths
-------
ORS Handbook
Section 2.0
Page 2-46
References
1. Sigrist, M. W. (Eds.). 1994. Air Monitoring by Spectroscopic Techniques. New York, NY:
John Wiley & Sons, Inc.
2. U.S. EPA. 2009. Measurement and Monitoring Technologies for the 21st Century (21M^).
Open Path Technologies: Measurement at a Distance UV-DOAS.
3. http://www.clu-in.org/download/misc/21M2flier.pdf. Washington, DC.
4. Finlaysson-Pitts, B.J., J.N. Pitts. 1986. Differential Optical Absorption Spectroscopy, from
Atmospheric Chemistry; Fundamentals and Experimental Techniques. John Wiley& Sons.
5. Stutz, J., The Stutz Research Group,_
http://www.atmos.ucla.edu/~iochen/research/doas/DOAS.html
6. Opsis AB. 1996. Quality Assurance and Quality Control using Opsis Analyzers for Air Quality
Monitoring. Version 1.1. Furulund.
7. Thoma, Eben, Measurement of Total Site Mercury Emissions from a Chlor-alkali Plant
Using Open-Path UV-DOAS, EPA/R-07/077, July 2007,
http://www.epa.gov/ord/NRMRL/pubs/600r07077/600r07077.pdf
8. Pundt, I. 2006. DOAS tomography for the localization and quantification of anthropogenic
air pollution. Analytical and Bioanalytical Chemistry, 385 (1) 18-21. doi:10.1007/s00216-
005-0205-4
9. US DOE/ORNL. 1999. NARSTO Measurement Methods Compendium.
10. Liptak, B. 2003. Instrument Engineer's Handbook: Process Measurement and Analysis.
Chilton Book Company. Radnor, PA.
11. Liptak, B. 1995. Instrument Engineers' Handbook: Process Measurement and Analysis.
Boca Raton, FL: CRC Press.
12. Spellicy, R. 2003. IMACC Industrial Monitoring Corporation. Round Rock, TX
http://www.ftirs.com.
13. ETG Resource Tecnologia. 2007. The UV DOAS SENTINEL System Represents the Most Cost
Effective.
-------
ORS Handbook
Section 2.0
Page 2-47
2.4. Differential Absorption Light Detection and Ranging Systems
LIDAR is a technology used to measure area source, fugitive or ambient air pollutants without the
requirement for line of sight or retro-reflector measurement paths. LIDAR operates on the same
principles as radio detection and ranging (RADAR) except light is used rather than radio waves. In
early applications of LIDAR, investigators used search lights, a telescope, and a photoelectric light
detector to collect information about Earth's atmosphere. The technology was used to determine
atmospheric density by studying the backscattered light intensity along the path of the searchlight
beam. Since the 1930s, the use of LIDAR has expanded to applications in aerial surveying, three-
dimensional imaging, chemical warfare agent detection, and forestry. LIDAR is also used to study
atmospheric parameters such as aerosol and cloud properties, temperature, wind velocity, and
species concentration.1'2
There are three generic LIDAR applications:
range finders
Differential absorption LIDAR (DIAL)
Doppler LIDARs
In 1964, a new LIDAR application called DIAL was proposed to locate and measure trace chemical
concentrations in the atmosphere.1 The goal of LIDAR-based DIAL technique was to precisely
measure constituents of ambient air using a simple remote sensing technique that lacked the
complexity of traditional optical techniques such as FTIR. Since then, DIAL has developed into a
commercially available technology capable of mapping concentrations of multiple atmospheric
pollutants.
DIAL uses lasers directed into the atmosphere to measure species concentrations of target
aerosols, dust, and gases in the lower few kilometers of the atmosphere. Using the DIAL approach,
spatial concentrations are obtained from the reflected or backscattered light from two wavelengths
of light: one that is strongly absorbed by the species of interest and the other just outside of the
-------
ORS Handbook
Section 2.0
Page 2-48
absorption range of the target compound, which is used to measure background light scattering.
As these wavelengths of light are emitted from the laser source, the ratio of the backscattered light
intensity between the two wavelengths is measured and coupled with the time delay of the return
signal. The target compounds absorb or reflect and scatter the light back to a telescope or
receiving optics, where intensity of the backscattered light is detected and evaluated.
Concentration is determined based on the amount of light absorbed and the location of the
observed compounds is based on the time delay of the backscattered light at the detector.3 By
measuring the backscatter at different angles from the source, the data can be processed to show
the two-dimensional plume shape of the target compound emission profile.
The main advantage of DIAL over other ORS technologies is the ability to spatially resolve the
concentration of a single compound, or class of compounds, based on the radiation absorption
characteristics of the pollutants being measured.3'4 The ability to spatially resolve concentration
data is a unique advantage when compared to alternate remote sensing methods, which yield
average concentration data over a predetermined path length. The main limitation in using DIAL is
the limited number of wavelengths that can be generated by current laser technology at the precise
wavelength for compounds of environmental interest to be monitored. Additionally, the use of
DIAL in the United States has been limited due to limited availability and the high cost of the
associated equipment.5
Basic Operation
LIDAR technology operates by transmitting a laser into a medium containing gaseous compound.
The laser light can be transmitted in the UV, visible, or IR wave range.6 For the DIAL application of
LIDAR, an appropriate wavelength is chosen for the species to be measured along with a nearby
wavelength that will not be absorbed by the target compound.7
During operation, light generated by a laser is directed through a wavelength switching unit. As the
laser pulses, the switching unit alternates the laser between the two different wavelengths
designated as "on" (the wavelength that is absorbed by the target compound) and "off" (the
-------
ORS Handbook
Section 2.0
Page 2-49
wavelength just outside of the target compound's absorption). Some DIAL systems use separate
lasers to produce both wavelengths. A continuous laser is used when the measurement range is
short or for long measurement time periods. A pulsed laser is used when higher energy is needed
for long measurement ranges or short time intervals.6
A small portion of the laser output is directed to a calibration cell. The cell is filled with a known
concentration calibration gas and the absorption for a given path length is measured. The
remaining narrow beam of laser light is expanded or widened to make it "eye safe" and then
directed into the atmosphere by a series of mirrors and optics.7 The expanded laser beam
interacts with molecules and particles, and is scattered by them. As the light travels through the
atmosphere, the "off" wavelength is scattered elastically by atmospheric particles. The "on"
wavelength is absorbed by the target gaseous compound(s) and scattered at a reduced intensity.3
Light is scattered in various directions and a small portion is reflected back towards its source. This
backscattered light is collected, focused, and a detector converts the light information into a digital
signal for use in determination of pollutant location and concentration.7
DIAL laser systems are chosen based on the absorption characteristics of the compounds under
study. Requirements for lasers include low beam divergence, adequately low pulse repetition
frequency (PRF), and appropriate wavelength specificity.1 Capture of the entire beam by the
receiving optics is preferable to reduce background noise. A low beam divergence (narrow beam) is
necessary to retain the beam in the receiving optics field of view. If a pulsed laser is used, a low
PRF ensures measurement cycles are separated by a sufficient length of time to avoid inference
from one measurement to the next.
Wavelength specificity defines which compounds can be measured by DIAL systems. Although
other applications of LIDAR, such as measuring aerosols, are operated across a range of
wavelengths, DIAL wavelengths are specific to the absorption characteristics of the pollutants being
measured.1 Advances in tunable laser technology have allowed simultaneous multi-wavelength,
hence multi-component, measurements to be made.8
-------
ORS Handbook
Section 2.0
Page 2-50
Target Pollutants
|DIAL UNIT
Figure 2-17. Beam Path and Major Components of DIAL Unit.
The difference in the backscattered intensity between the "on" and "off" wavelengths allows for a
concentration of the compound to be calculated. The "off" wavelength is reflected predictably and
decreases in intensity (P) by the inverse square of the distance the light has traveled. The two
intensities are divided by one another and transformed into concentration data using the Beer-
Lambert law.9 When the time delay of backscattering is added into the calculation, the distance
from the laser to the compound can be determined.3'10'11
DIAL measurements collect pollutant concentration data over a relatively long path length. Beam
path lengths range from a few hundred meters to 3,000 meters.5 Since DIAL systems do not
require a remote detector/reflector, 2-D scans can be completed in approximately 10 minutes4'5'8
Pollution emission flux is calculated by collecting wind speed data and plume concentration during
DIAL testing. Wind speed is multiplied by the pollutant concentration across the emission plane to
obtain a flux value.3
Most of atmospheric sensing DIAL systems operate in a monostatic mode where light is emitted
and received at the same location. A monostatic mode may be deployed in two sub-
configurations; monostatic coaxial mode, where light is received by the same optics through which
-------
ORS Handbook
Section 2.0
Page 2-51
it was emitted, and monostatic biaxial mode, where light is received by optics located adjacent to
the originating optics. The monostatic system allows multiple measurements to be completed
quickly without the need of retro-reflectors or line of site detection systems. Figure 2-18 illustrates
the monostatic coaxial and biaxial modes.
Receiver field
of vision
I
Monostatic Coaxial
Laser Path
T ra rvs mrtto ng f
Receiving
Filter
Detector
Optics
** Laser
Monostatic Biaxial
ฆ--IR Rath
Transmitting Optics
Laser
Sourcซ
Receiving Optics
Filter
Detector
Receiver field
of vision
Figure 2-18. Monostatic Coaxial and Biaxial Configuration for DIAL
The use of a bistatic DIAL configuration is less common. In bistatic mode light is produced and
detected at separate locations. The bistatic configuration requires frequent repositioning of the
detector to obtain an emissions concentration profile.1,5 Figure 2-19 illustrates a bistatic DIAL
system.
-------
ORS Handbook
Section 2.0
Page 2-52
Absorbing
Medium
Reflected
Light
Transmitting
Optics
Receiving
Optics and
Filters
Laser
Source
Figure 2-19. Bistatic Configuration for DIAL
The signal receiving and detection system consists of primary optics, a spectral filtering unit, a
detector, and a photon counter. Receiving optics collect the backscattered light for analysis.
Primary optics vary by application and can include Cassegrain or Newtonian telescopes, multiple
small mirrors, and liquid mirror telescopes among others.1,6 Spectral filtering systems remove
background light and reduce signal noise. Spectral filtering can be accomplished by using simple
systems such as narrow bandwidth interference filters or more complicated systems such as
depolarization techniques.1 A detector converts the incoming spectral signal into a digital signal
for photon counting. Typical detectors include traditional Photomultiplier Tubes (PMTs), Charge
Coupled Devices (CCDs), mercury-cadmium-telluride (MCT) detectors, or avalanche photodiodes.1,6
The photon counter performs two steps. The first step removes dark counts, which are a type of
signal noise created by the detector. The second step counts the number of photons based on the
time they were received. By counting photons on a sequential basis, range resolved measurements
are realized.1
-------
ORS Handbook
Section 2.0
Page 2-53
The DIAL system yields a spatially resolved concentration measurement along the specified path
length. Multiple closely spaced scans are often performed over a cross section of an emission
plume to produce concentration maps as illustrated in Figure 2-20. This type of output is unique
compared to that of measurements such as OP-FTIR spectroscopy, for which the output is a PIC
without spatial resolution along the path length.
DIAL - Mode of Operation
Height
(m)
Concentration
(mg/m*)
Scan
plane
Emissions
Wind
Remote met
stations
& sorption
tubes
Figure 2-20. Illustration of DIAL unit mapping an emission plume
Wind characteristics play a major role in how the DIAL system should be positioned to take
measurements, as well as the validity of data obtained from those measurements. Equipment set
up is recommended to be at least 50 meters from the plume cross section and measurements
should be taken perpendicular to the wind direction. During the measurement period, wind
direction may change, which effectively skews the measurement plane.12 Therefore, wind
direction is typically analyzed throughout the measurement process to accurately adjust the
measurement plane for skew.12,13 Changes in wind direction and speed may cause variation in the
emission plume over time and affect the results of a scan along a measurement plane, which takes
about 10 to 15 minutes to complete. Averaging multiple scans of the same cross section helps to
suppress the error associated with a dynamic emission plume.12 Wind speed analyzed using dual
wind monitors typically do not vary more than 20 percent between the independent wind
measurements. Wind data from three elevations ranging from 15 to 25 meters, provide sufficient
information to determine plume flux through the plane.14
-------
ORS Handbook
Section 2.0
Page 2-54
Pollutants and Relative Levels That Can Be Measured
A variety of atmospheric parameters can be measured with DIAL techniques including:
temperature, pressure, water-vapor concentration, and selected atmospheric gases. Additionally,
back scatter and light absorption of cloud particles and aerosols can be investigated.
DIAL has historically been used to measure criteria pollutants in the upper atmosphere.
Approximately 15 species in the spectral range of ultraviolet to infrared can be detected by DIAL
systems. Table 2-19 lists compounds that can be measured with DIAL systems. The list is not all-
inclusive but displays compounds reported in literature.
Table 2-19: Reported Species Measured with DIAL Systems
Compounds
Acetylene
Hydrogen Chloride
Nitrous Oxide
Alkanes
Mercury
Ozone
Benzene
Methane
Sulfur Dioxide
Ethane
Methanol
Toluene
Ethyl Benzene
Nitric Oxide
Xylenes
Ethylene
Nitrogen Dioxide
Detection limits in the ppb range have been reported at distances of 500 to 3000 meters.6,15
Detection limits vary based on many factors. Atmospheric effects, such as laser beam wander from
atmospheric turbulence, influence the accuracy of DIAL measurements. Laser type and internal
instrument noise can also have a negative influence on detection limits.16,17 The path length also
affects the resolution of the system. Reported minimum detection limits range from 0.001 ppb for
mercury to 90 ppb for hydrogen chloride gas at a 200-meter absorption path length.9 Detection
limits are given as estimates in Table 2-20 and are based on various absorption path lengths.
-------
ORS Handbook
Section 2.0
Page 2-55
Table 2-20. Approximate Detection Limits for DIAL^
Compound
Reported Minimum
Detection Limit (ppb)
Detection Range (m)
Acetylene
26
800
Alkanes
10
800
Benzene
3
800
Ethane
16
800
Ethylene
9
800
Hydrogen chloride
13
1,000
Mercury
0.06
3,000
Methane
76
1,000
Methanol
153
500
Nitric Oxide
4
500
Nitrogen Dioxide
5
500
Nitrous Oxide
56
800
Ozone
3
2,000
Sulfur Dioxide
4
3,000
Toluene
3
800
Xylenes
5
500
The specific wavelength used for detection depends upon the absorption characteristics of the
target compounds. The number of identifiable pollutants is further limited by the number of
absorbing wavelengths that are unique to a specific compound without additional interferences, as
well as the laser technology that is currently available. This technology is improving with
expectation that the range of detectable pollutants will expand.6 DIAL systems can also be used in a
mode like a fugitive source monitor. In this mode, an entire class of chemicals can be measured
using a single laser wavelength that the entire chemical class absorbs. DIAL results from such a
study must be interpreted as an average.
Typical QA/QC
While the EPA provides information by general reference in Other Test Method 10 (OTMIO)13, the
verification of DIAL measurements is challenging due to its unique ability to produce spatially
resolved concentration data. Limited QA/QC guidelines exist that verify such data in the literature.
Information specific to LIDAR technology has been published by the Association of German
Engineers (VDI) in VDI 4210 Part 1 (1999) Remote sensing, Atmospheric measurements with LIDAR,
Measuring gaseous air pollution with DAS LIDAR.18
-------
ORS Handbook
Section 2.0
Page 2-56
Record Keeping
As with all environmental measurements, it is necessary to keep accurate records during
measurement periods to ensure accurate data collection including records of calibrations,
meteorological conditions, path lengths, and measurement times.
Instrument Performance
Initial measurements should be conducted over a period so that emission source fluctuations may
be considered during data analysis; one half to an entire day is recommended for an initial
measurement.14 Scanning the same plume over different days and varying conditions is also
recommended to assess the impact of varying conditions on measurement results.12 An initial site
assessment should also be conducted to determine any interference that may disrupt data
acquisition. Interferences include geographic constraints and off-site upwind emissions sources.13
Some DIAL systems are programmed with verification systems to ensure that quality
measurements are taken. One reported DIAL system employs an internal "wavelength" verification
system, known as a wavemeter, to identify and dispose of any data produced from the emission of
an inconsistent wavelength.8 This prevents erroneous data from being produced if the light source
emits radiation outside of the specified wavelength.
Example Applications and Vendors Applications
With the ability to develop spatial concentration information of air pollutants, DIAL systems have
been implemented in a variety of applications including fenceline monitoring, fugitive emissions
measurement, and plume fate analysis. DIAL may also be used to measure flare efficiency from
industrial processes.5,11 For each application, the strengths and limitations of a DIAL system must
be considered to produce results that meet users' DQOs. For example, the use ofCO^ laser
assumes sufficient aerosol concentration in the atmosphere to provide sufficient backscatter. High
wavelength visible and UV light sources rely on molecular backscatter.
-------
ORS Handbook
Section 2.0
Page 2-57
Table 2-21. Typical Applications for LIDAR.
TECHNOLOGY
APPLICATIONS
LIDAR
DIAL
DIAL systems are either in a fixed or mobile arrangement. Fixed DIAL units used in laboratories are
typically less complicated than mobile systems used for field operation.9 Many mobile systems
have been constructed based on an enclosed truck platform. The truck is driven to the testing site
and positioned accordingly to obtain emissions data.3'7<19 DIAL systems have also been
implemented onboard ocean vessels and in airborne systems, collecting emissions information
while flying over a target area.11'20 Figure 2-21 illustrates a truck-based mobile DIAL platform.
Figure 2-21. Mobile DIAL Unit
Specific field implementation examples include studies completed on gas processing plants and oil
refineries. These studies used DIAL systems and plume mapping techniques in Canada and
European nations. Measured fugitive emissions were four to 20 times greater than factor
estimated fugitive emissions.5'6'14'21 While these studies have concluded emissions factors may
underestimate actual fugitive emissions, objection to using annual emissions figures calculated by
DIAL measurements is apparent. Industry objects to using DIAL-based calculated emissions due to
the short time-period of measurement relative to the long time-period of annual operation.6
-------
ORS Handbook
Section 2.0
Page 2-58
In other studies, DIAL systems have been implemented to measure emissions from mobile sources
such as air and highway traffic. Additional DIAL systems have been developed to detect chemical
warfare agents as well as natural gas pipeline leaks from airplane mounted platforms 6
Vendors
Many vendors manufacture laboratory-scale DIAL applications; however, field-ready measurement
instruments are only offered by a small number of vendors. Table 2-22 lists vendors that
manufacture or provide field-ready DIAL instruments.
Table 2-22: DIAL Vendors
Vendors
Spectrasyne
http://www.spectrasyne.ltd.uk/
LASEN
http://www.lasen.com/
National Physical Laboratory
http://www.npl.co.uk/
ITT
http://www.itt.com/
Strengths and Limitations
A significant limitation of DIAL technology is the cost and limited availability of the measurement
service. Multiple measurements in North America have relied on importing the instrumentation
from the United Kingdom12'15'21 Additionally, the number of chemical species measurable by DIAL is
restricted to the unique absorption characteristics of those species and the availability of laser
technologies able to produce the absorption wavelengths.
The most notable strength of a DIAL system is the ability to spatially resolve pollutant concentration
information in three dimensions in a short period of time. Concentration gradient data obtained in
short periods of time enables DIAL to be deployed in many applications and many configurations.
DIAL has been used in scenarios from ground-based emissions plume monitoring to helicopter and
fixed-wing aircraft aerial surveys. Table 2-23 and Table 2- 24 summarize the strengths and
limitations of DIAL systems.
-------
ORS Handbook
Section 2.0
Page 2-59
Table 2-23: Summary of the DIAL Strengths
Feature
Strength
Concentration data is spatially
resolved
In contrast to PIC measurements taken by other instrumentation
system, DIAL relay concentration data as a function of distance along
the beam path
Beam Path
Reported path lengths up to 3000 meters
Receiving Optics
Collects backscattered light without the use retro- reflectors
Measurement Time
Scan along measurement plane requires 10 to 15 minutes
Mobile Platform
Instrumentation is moved around measurement site obtaining multiple
plume scans from various locations increasing accuracy of plume
characterization
Instrumentation can also be mounted on airborne
platform for increased mobility and expanded applications
Near Real Time Data
Real time data allows for leak identification and inputs to process
change decisions
Multi-Wavelength Operation
Allows for simultaneous concentration measurements of multiple
species and classes of species
Table 2-24: Summary of the DIAL Limitations
Feature
Limitation
Unique Chemical Absorption
Bands
Measurable species are limited to those with unique absorption
bands. Chemical species with common absorption characteristics can
only be measured as classes of compounds
Chemical Species Absorption
Dependencies
The absorption wavelengths of species are temperature and pressure
dependent. It is necessary to check the applicability of wavelengths
selected for measurement based on temperature and pressure
variation in target absorption.
Backscatter Requirements
Sufficient aerosol or molecular material must be in the atmosphere to
create sufficient backscatter
Wind speed and direction
Rapidly changing wind speed or direction may cause measurements to
change rapidly and may affect mixing ratios of measured surrogates to
compounds of interest.
Vendors
Small number of vendors providing DIAL systems and services
Expense
High system cost has limited to amount of commercial DIAL studies in
the United States
-------
ORS Handbook
Section 2.0
Page 2-60
References
1. Argall, P. and R. Sica. 2002. LIDAR in the Encyclopedia of Imaging Science and Technology.
Ed. J.P. Hornak. John Wiley & Sons Inc. New York, NY.
2. Moskal, L., T. Erdody, A. Kato, J. Richardson, G. Zheng, and D. Briggs. 2009. Lidar
Applications in Precision Forestry.
.
3. Edner, H., K. Fredriksson, A. Sunesson, S. Svanberg, L. Uneus, and W. Wendt, W. 1987.
Mobile remote sensing system for atmospheric monitoring. Applied Optics. 26(19): 4330-
4338. doi:10.1364/AO.26.004330.
4. SIRA, Ltd. 2004. Recommendations for best practice in the use of open-path
instrumentation. .
5. U.S. EPA. 2006. VOC Fugitive Losses: New Monitors, Emission Losses, and Potential Policy
Gaps - 2006 International Workshop.
6.
7. U.S. EPA. 2009. Measurement and Monitoring Technologies for the 21st Century (21M^).
Open Path Technologies: Measurement at a Distance UV-DOAS.
8. .
9. Fredriksson, K., B. Galle, K. Nystrom, and S. Svanberg. 1981. Mobile lidar system for
environmental probing. Applied Optics. 20(24): 4181-4189.
10. Weibring, P., H. Edner, and S. Svanberg. 2003. Versatile mobile lidar system for
environmental monitoring. Applied Optics. 42(18): 3583-3594. doi: 10.1364/A0.42.003583.
11. Sigrist, M. W. (Eds.). 1994. Air Monitoring by Spectroscopic Techniques. New York, NY:
John Wiley & Sons, Inc.
12. Spectrasyne, Ltd. 2007. The Technique.
13. .
14. Weibring, P., H. Edner, S. Svanberg, G. Cecchi, L. Pantani, R. Ferrara, and T. Caltabiano.
1998. Monitoring of volcanic sulfur dioxide emissions using differential absorption lidar
(DIAL), differential optical absorption spectroscopy (DOAS), and correlation spectroscopy
(COSPEC). Applied Physics B: Lasers and Optics. 67(4): 419- 426. doi:
-------
ORS Handbook
Section 2.0
Page 2-61
10.1007/s003400050525.
15. Chambers, A. 2003. Well Test Flare Plume Monitoring Phase II: DIAL Testing in Alberta.
Alberta Research Council Inc. Prepared for Canadian Association of Petroleum Producers.
.
16. U. S. EPA. 2006. Other Test Method (OTM) 10 Optical Remote Sensing for Emission
Characterization for Non-Point Sources, .
17. Frisch, L. 2003. Fugitive VQC-Emissions Measured at Oil Refineries in the Province of Vastra
Gotaland in South West Sweden - A Success Story: Development and Results 1986 - 2001.
County Administration of Vastra Gotaland, Report No. 2003:56.
18. Chambers, A., M. Strosher, T. Wootton, J. Moncrieff. and P. McCready. 2008. Direct
measurement of fugitive emissions of hydrocarbons from a refinery. Journal of the Air and
Waste Management Association. 58: 1047-1056. doi: 10.3155/1047- 3289.58.8.1047.
19. Fredriksson, K. A. 1985. DIAL technique for pollution monitoring: improvements and
complementary systems. Applied Optics. 24(19): 3297-3304. doi: 10.1364/AO.24.003297.
20. Ahmad, S. R. 1997. Application of lidar to atmosphere pollutant mapping: a review.
Proceedings of the SPIE. The International Society for Optical Engineering, doi:
10.1117/12.283904.
21. VDI 4210 Part 1. 1999. Remote sensing, Atmospheric measurements with LIDAR,
Measuring gaseous air pollution with DAS LIDAR.
-------
ORS Handbook
Section 2.0
Page 2-62
2.5 Thermal Infrared Cameras
Thermal infrared (IR) cameras that are used for environmental measurements come from a
class of camera known as thermographic or forward-looking IR, which use IR radiation to form
an image in a manner like the way photographic cameras use visible light. The original thermal
IR camera development was funded largely by the astronomy and defense communities for the
purposes of "night-vision" for aircraft and other vehicles and development of heat-seeking
missiles.1 Private companies have adapted military IR technology that does not require
sophisticated cooling or optics to produce commercial IR cameras for environmental
applications.
IR cameras are useful for a wide range of applications, including to monitor watershed
temperature in game habitats, to detect energy loss or insulation defects in buildings, to detect
biomedical abnormalities, to track and aid in target acquisition by the military, to improve
piloting of aircraft in low visibility conditions, to pinpoint ignitions sources during firefighting,
and to aid in search and rescue operations of missing persons.
Environmental applications of IR cameras include the detection of industrial gas leaks that are
invisible to the naked eye. By filtering incoming light to permit only regions of IR radiance
characteristic of hydrocarbon or volatile organic compound (VOC) gases to reach the camera's
detector, the IR video camera allows the user to see images of hydrocarbon gases on the
camera screen in real time.2 These cameras can identify the source and flow path of escaping
gases in a wide variety of applications,3 such as tank vents and gas line leaks.
The major advantages to using the thermal IR camera for leak detection are the technology's
portability and its qualitative ability to display a wide field of view that allows major leaks to be
detected more efficiently than classical leak detection and repair procedures that require each
equipment component to be tested individually. Additionally, IR camera technology allows leak
detection in parts of facilities that may be difficult or hazardous for personnel to access. The
thermal IR camera's major drawback is its inability to measure the quantity or concentration of
gas present in a gas plume. A second limitation is the technology's inability to identify
individual chemicals in a complex gas leak mixture due to the simple optics employed in
portable IR cameras. A third limitation is the technology's inability to detect leaks when the
background temperature is the same as the gas temperature. A fourth limitation is the
-------
ORS Handbook
Section 2.0
Page 2-63
technology's inability to detect leaks when ambient wind conditions are stronger than a
moderate breeze.
Basic Operation
All objects that have a temperature above absolute zero (or 0 Kelvin) have a thermal profile by
emitting IR radiation. In the case of IR cameras designed for the visualization of hydrocarbon
gases, a special band-pass optical filter is placed between the outer lens optic (made of
optically-transmissible germanium) and a focal plane array (FPA) detector that allows only IR
radiation in the range of about 3.2 to 3.4 micrometers (|im) to pass to the detector (Figure
2.22).4
Everything emits
IR radiation
Spectral Filter
IR waves
Germanium
Lens
Signal
Processor
Video
Output
DH
HC gas absorbs/emits IR radiation
in the 3.2 to 3.4 nm region
Concentrated
Infrared Light
Electronic
Signals
Figure 2.22. Overview of Thermal IR Camera Technology Basics.
The detector passes along the information as an electrical signal that is then processed by the
camera software to produce a live video image of the thermogram. The thermogram is an
image of the thermal radiation in the field of view and will display normally invisible gas
emissions on the camera's screen as a clouded area or smoke in real time, as shown in
Figure 2.23.10 If the IR camera is set to display hotter areas on a thermogram as whiter than
the cooler areas in black and white, then a gas plume that is colder than the surrounding
background will appear like a dark cloud or smoke. If the opposite is true and the gas is hotter
than the background, then the gas plume will appear lighter like a white cloud or steam. The
presence of hydrocarbon gas in the thermogram are represented as a change in heat, similar to
how a shadow with a normal camera represents a change in visible light.4"9
-------
ORS Handbook
Section 2.0
Page 2-64
Figure 2.23. Image of a Controlled Gas Release where the Gas is Warmer than the Background.
The IR gas sensing camera creates images based on the IR absorption/emission characteristics
of chemical species within the camera's field of view. IR detection typically occurs in the 3-5
|am wavelength range for hydrocarbon gases,11 but the special lens and filter arrangements
represented by the spectral filter in Figure 2.22 are used to narrow the IR spectrum of
wavelengths detected to about 3.2 to 3.4 urn, thereby allowing the camera to image specific
compounds or compound classes that have electromagnetic signatures in that region.12 Other
wavebands are available in different camera models ranging from 1 to 14 jam. For example, 3.2
to 3.4 jam is used for hydrocarbon detection, 4.52 to 4.67 |im is used for carbon monoxide
detection, 8,0 to 8.6 |im is used for the detection of refrigerant gases, and 10.3 to 10.7 |am is
used for sulfur hexafluoride and anhydrous ammonia detection.13
Each chemical compound has a unique response to radiation from the electromagnetic
spectrum based on the rotational and vibrational energy transition characteristics of the bonds
in each molecule to generate rotating and vibrational "ro-vibrational" spectra. For example,
many hydrocarbon molecules are electromagnetically active in the 3.2 to 3.4 urn range due to
the structure of the carbon-hydrogen bonds. One such chemical, propane, is a short chain of
three carbon atoms with single bonds to each other and to hydrogen atoms (as shown in Figure
2.24). Figure 2.24 also shows the areas where the propane molecule has the greatest amount of
IR absorbance by the peaks in the blue line. Over the IR spectrum (1 to 14 (am), propane has its
-------
ORS Handbook
Section 2.0
Page 2-65
highest peak in the 3.2 to 3.4 |am region with a secondary peak around 6.5 to 7.5 |am. This
indicates that targeting one of these two bandwidths (but especially the primary peak) with the
optics of the IR camera should result in the detection of propane, assuming there are no gases
that may cause interference.
Propane IR Spectrum
1.20E-03
ฆSi 9.9SE-04
3
OJ
7.9SE-04
55 S.9SE-04
c
OJ
qj
o
c
ra
-O
-Q
<
3.9SE-04
1.9SE-04
-S.OOE-06
h
|
,-
1.0
2.0
3.0 4.0 5.0
6.0 7.0 8.0 9.0
Wavelength (nm)
10.0
11.0
12.0
13.0
14.0
Figure 2.24. IR Spectrum for Propane with the molecule Bond Structure (not to Scale).
An IR camera that targets the detection of hydrocarbon gases and has optics that focus the
imaging bandwidth to the 3.2 to 3.4 |am region will have an optical window of transmission like
that presented in the top left panel of Figure 2.24.14The top middle panel of Figure 2.2514,15
illustrates the IR camera window of transmission overlaid with the propane spectrum and the
top right panel is the same but with the spectrum for methane. All three curves are
represented in the bottom panel of Figure 2.25 to illustrate that, although each curve will be
different, if the compound absorbs IR radiation in the window of transmission for the camera,
then the IR camera should theoretically be able to detect and make visible the gas emission of
that compound.
-------
ORS Handbook
Section 2.0
Page 2-66
ฆ
o*
07
0.
0
OJ
0.
JO
ป>ป
2.00E-03
1.80E-03
0.9
1.60E-03 25
0.8
"D
i
=5 0-7
1.20E-03
0.6
ง
on
0.5
c
ra
H 0.4
2
6.00E-04
o
E
oc
4.00E-04 *fi
0.2
2.00E-04
0.1
3.10
3.15
3.20
3.25
3.30
Wavelength (^m)
3.35
3.40
3.50
Figure 2.25. Spectral curves for (top Left) an IR Camera Window of Transmission, (Top Middle) a Propane
Spectral Curve, (Top Right) a Methane Spectral Curve, and (Bottom) All Three Curves put together with
the Dashed Line Indicating the 3.2 3.4 Region.
It is expected that, if the IR camera window of transmission and a major peak in the IR
spectrum of a compound overlap, then the IR camera will be able to image a gaseous fugitive
emission of that compound. However, several factors affect the IR camera's imaging and,
therefore, the sensitivity of the technology. These are:
Ambient thermal energy plays an important role in the sharpness or resolution of the
IR camera's image.
Variations in the thermal profile of the image (called a thermogram) can require that
any of a number of settings be adjusted, such as focusing the lens, changing the
viewing angle, adjusting the temperature range setting, and switching between
automatic and high sensitivity (or enhanced) camera modes to ensure no leaks were
missed.12
-------
ORS Handbook
Section 2.0
Page 2-67
The IR camera's leak detection sensitivity is affected by the temperature of the
target gas and the equipment surface and/or surrounding background in the field of
view.
Reflectivity (reflection of IR light) and emissivity (emission of thermal energy due to
the absorbance of IR radiation) of surrounding materials also play a role in the
camera's sensitivity.
Gas concentration, distance from the leak source, leak pressure, play a role in the
ability to measure using this technology.
Meteorology, such as cloud cover, and wind speed and direction.16
It is possible under certain conditions that the thermal radiance of the leaking gas and the
background are equal. Because the camera uses the temperature differentials to image a gas
plume, the leak will be invisible to the IR camera and, therefore, the operator under these
conditions. The temperature differential between the target gas and the background is
commonly called AT ("delta-T") and is a major influence on the sensitivity of the IR camera.4'13'15
Proper IR camera operator training is required to ensure each of these factors are considered
during leak detection surveys to make sure all possible leaks or vapor clouds are detected.10
The operation of an IR camera is straightforward. At start up, the IR camera must be allowed to
reach operating temperature since the camera's detector is cooled by a Stirling engine to
reduce analytical noise. This start-up process takes an average of about 8 to 10 minutes to
complete, depending on the conditions of the surrounding environment. An additional wait of
10 to 15 minutes after the camera has reached the cooling set point is suggested to allow for
thermal stabilization. At this point, it is recommended that the operator perform a non-
uniformity correction (NUC) to spatially homogenize the detector response to thermal
differences.
IR camera models' basic operation modes can include additional contrast adjustments allowing
easier visualization of leaking gaseous compound against the stationary backgrounds. There
are three potential modes on any given IR camera: Auto or Normal mode, Manual mode, and
High Sensitivity (HSM) or Enhanced (ENH) mode. As the name implies, Auto or Normal mode is
like the Auto mode on a digital camera where software algorithms optimize the image display;
-------
ORS Handbook
Section 2.0
Page 2-68
in the case of IR cameras, this is based on a histogram evaluation of the thermogram in the
field of view. Manual mode indicates the ability to adjust the brightness and contrast (called
level and span, respectively) of the image output. This mode is useful when the temperature of
the target gas is in the bulk of the histogram and not easily distinguishable from the thermal
profiles of other objects in the surrounding scene. An example is the presence of a very hot
object that skews the histogram to one extreme and effectively "washes out" the objects with
a thermal profile on the opposite end. Instead of using Manual mode, it is often the preference
of IR camera operators to use the HSM or ENH mode almost exclusively.
The HSM/ENH mode is executed differently by each individual manufacturer, but, basically,
algorithms in the camera software can perform a type of scene-subtraction whereby the
camera display results in only those objects that are time-dependent. For example, if the only
object moving in a scene is the escaping gas, then the HSM/ENH mode highlights only those
gas pixels that have changed over a series of frames. The camera operator can control how
many frames are included in the analysis, thereby increasing the camera sensitivity by
increasing the number of frames. The highest settings for sensitivity come at the cost of image
object resolution, potential interference, and larger file size. Because this mode can greatly
enhance the sensitivity of the IR camera, it is common that heat convection and the various
phase changes and air mass transportations related to heat are imaged. Frequent examples of
this interference are water evaporation, heat convection from hot or cold objects relative to
the ambient temperature, and water sublimation. There is also a training or warm-up period
associated with using HSM/ENH mode where the operator's brain needs some time to become
accustomed to the features of the HSM/ENH images. Just like an athlete must practice a
certain skill to learn the skill and then warm-up before using the skill during a sporting event,
so too should the IR camera operator practice and warm-up before surveying with an IR
camera.10
Operators are trained to scan each piece of relevant equipment or area of potential leaks from
one end to another, pausing frequently on each new scene of equipment to detect time-
dependent changes, and to perform the scan from a minimum of two different viewing angles
or locations relative to air flow or wind direction to ensure all leaks are detected.2 The first
field ofview is often from a widerangle with a larger viewpoint, while the second field of view
-------
ORS Handbook
Section 2.0
Page 2-69
is at a closer viewing angle. The same areas can be scanned repeatedly to improve the
likelihood that all leaks are detected.12
The images collected by modern IR cameras are digital. Even older IR cameras generated a
video feed that was recordable and allowed archiving for remote viewing and review. The
cameras can operate using battery power for up to eight hours of continuous use, or
connected to AC power for 24-hour monitoring purposes.8 Additionally, IR cameras used for
vehicle inspections have been adapted to use a 12-volt power source. Some IR cameras are
available with global positioning systems (GPS) to automatically record the location of the
camera's use.9
If quantification of a leaking gas is required, it is also possible to couple the IR camera with
additional technology, such as a passive FTIR system, different optical elements, modeling
software, or more traditional leak detection and repair instrumentation such as a portable
instrument meeting EPA Method 21 requirements or mass flow measurement using a bagging
technique. Coupling the IR camera with another technology not only provides means for
quantification, but can verify a detected leak as well as determine its chemical composition.5
Pollutants and Relative Levels That Can Be Detected
Thermal IR cameras can be designed to detect chemical compounds that have absorptions
anywhere in the 1-12 |am wavelength IR absorption range. Typically, the 3-5 |am wavelength
range is used for organic VOC. Depending on a variety of factors, including lens focal point,17
distance from the source, and meteorological conditions, gaseous compound concentrations in
the hundreds of ppm range (> 500 ppm) or leaks above an emission rate of about 12 grams per
hour (g/hr) are detectable by the camera.4'5 Table 2-25 provides a list of example compounds
that have been detected using IR cameras with different wavebands. This list is not all-
inclusive, but shows that many compounds can detected with the technology.
-------
ORS Handbook
Section 2.0
Page 2-70
Table 2-25. List of Example Gaseous Compounds that can be Detected by Thermal IR Cameras at
Different Wavebands
3.2-3.4 urn
4.52-4.67 urn
10.3 -10.7 nm
l-Pentene
Carbon Monoxide
SF6 (Sulfur Hexafluoride)
Benzene
Nitrous Oxide
Acetic Acid
Butane
Ketene
Anhydrous Ammonia
Ethane
Ethenone
Chlorine Dioxide
Ethanol
Butyl Isocyanide
Dichlorodifluoromethane "FREON-12"
Ethyl benzene
Hexyl Isocyanide
Ethyl Cyanoacrylate "Superglue"
Ethylene
Cyanogen Bromide
Ethylene
Heptane
Acetonitrile
Hexane
Acetyl Cyanide
8.0-8.6 urn
Isoprene
Chlorine Isocyanate
MEK
Bromine Isocyanate
R404A
Methane
Methyl Thiocyanate
R407C
Methanol
Ethyl Thiocyanate
R410A
MIBK
Chlorodimethylsilane
R134A
Octane
Dichloromethylsilane
R417A
Pentane
Silane
R422A
Propane
Germane
R507A
Propylene
Arsine
R143A
Toluene
R125
Xylene
R245fa
Typical QA/QC
Maintenance records should be kept for any equipment adjustments or repairs that could
affect measurement performance. Records should include the date and description of
maintenance performed. When the instrument is turned on, it must be allowed to warm up to
the manufacturer's recommended operating temperature and the duration of this initial
period should be regularly noted in the instrument logbook. Once at the appropriate
temperature, the camera can be used to scan a known concentration of a detectable gaseous
compound or the butane from a standard BICฎ lighter to demonstrate that the IR camera is
producing a visible image.4'12
When a leak is detected, a video record should be taken from an angle and distance that
promotes optimum leak visibility. The video should be at least 10 seconds long and stored with
a unique video tag.2 Information about the leak (such as component type, model or style of
component, service, size, process unit, process stream, pressure, vent location and ambient or
-------
ORS Handbook
Section 2.0
Page 2-71
process temperature18,19) should also be entered into a log sheet to further document the
leak.12 For leak detection and repair (LDAR) applications, once the leak has been identified, the
leaking component should be marked with a leak detection ID tag so that it can be easily
identified by maintenance and then repaired.2 For vents, tanks or other major gas emissions
detected by the IR camera, the GPS location and a visible light photograph should be used to
document the observation.
Although no prescribed method exists, a couple theories have proposed the unbiased
evaluation of thermal camera performance using pixel intensity values. The first theoretical
method described is based on a white paper from Dr. Yousheng Zeng,20 and the second is
based on the noise equivalent temperature difference (NETD) method used to quantify the
thermal sensitivity of a thermal imager called the noise equivalent concentration length
(NECL)21 for a gas leak imager. Detailed discussion of these theories is available in reference 13,
however, a summary of these methods is provided below.
An experimental configuration where the filling optical gas cells in front of a controlled
background with a known concentration of test gas (Figure 2.264) yields a pixel intensity
response is the keystone of Dr. Zeng's white paper method. A daily operations quality control
chart (Figure 2.2720) is developed by measuring the change in pixel intensity of the camera
response to the gas concentration in the test cells over different AT set points. Similarly, by
repeating the intensity measurements over different AT set points with different
concentrations can help to define a pixel intensity change over various concentration-path
lengths for a specific gas and camera model.
-------
ORS Handbook
Section 2.0
Page 2-72
Figure 2.26. Calibration/Verification Example Configuration, with IR Camera, Test Cells, Temperature-
controlled Background, and a Gas Delivery System.
Quality Control Chart
< 6.00E*01
Cl
_o 4.0GE+01
w
~ 2.00E+01
a O.OOE+OO
-2.00E+01
c 4.00E>OL
S.O
-20.0 -15.0 -10.0 -5.0
AT (=Tg'Tb) iridEg, F
15.0 20.0
0.0
10.0
Figure 2.27. Example Quality Control Daily Operations Check Chart with Performance Boundaries.
-------
ORS Handbook
Section 2.0
Page 2-73
Al vs, Concentration-Path Length
at Fixed AT (=9.0 deg, F)
2,000 4,000 6000
Cone.-Path Length (ppm-mj
8,000
10,000
Figure 2.28. Anticipated Change in Pixel Intensity for various concentration-path lengths at a Specific AT.
The objective of the NECL method is to allow for the unbiased comparison of various OGI
camera sensitivities from different suppliers. By developing absorption curves that describe an
OGI camera response to set conditions (like the Zeng method), the NECL method uses these
absorption curves to calculate a single number that describes the minimum concentration-path
length that would be detectable above the baseline noise level, The proposed standardized
conditions for developing the absorption curve of a camera for comparison are:
AT = 10ฐC.
The OGI camera to be tested is set up 1.0 m from the gas cell.
After the line of best fit is optimized through the data, the NECL is evaluated at a
concentration-path length (CL) = 0 pprrvm,21
The line of best fit through the experimental absorption curve data is extrapolated to CL = 0
ppm-m to yield the minimum concentration-path length (NECL) that is multiplied by the optical
thickness of the gas plume (path length) to result in the minimum concentration that is
theoretically detectable for that OGI camera. For example, the authors of the study21
-------
ORS Handbook
Section 2.0
Page 2-74
determined that 13 ppm-m is the NECL for methane using a FLIR GF320 OGI camera. Dividing
the concentration-path length by the path length results in the concentration. Therefore,
conducting an OGI survey in the field with a gas plume that is 10 cm in optical depth translates
to the OGI camera being technically capable of detecting the plume if it has a concentration
greater than 130 ppm (13 ppm-m / 0.1 m). This limit of detection will increase, however, in a
manner commiserate with field conditions at the time of detection (e.g., wind speed, leak exit
velocity, background temperature and uniformity, distance from targeted equipment).21,13
Example Applications and Vendors
Thermal IR gas imaging cameras have a wide range of applications, though they are most
commonly used to detect large leaks from process equipment and storage tanks at refineries
and chemical plants.5 The technology is now allowed as a replacement in the current LDAR
requirements for federal rules.4,12'19 The cameras have also been used to detect leaks in natural
gas pipelines through aerial viewing on helicopters.2'3,5 Thermal IR cameras can also be used to
monitor other plant activities that could potentially create fugitive emissions such as truck and
barge loading and unloading and incinerator activities. The cameras can also identify flares that
would otherwise be unnoticed by the naked eye.22 An image of such a flare is included as
Figure 2.29.22 Table 2-26 provides a general description of the applications for thermal IR
camera.
Figure 2.29. Flare Detection by Thermal IR Camera.
-------
ORS Handbook
Section 2.0
Page 2-75
Table 2-26. Typical Applications for Thermal IR Camera-
TECHNOLOGY
APPLICATIONS
IR Camera
Leak Detection
Vendors
Although there are several vendors for standard thermal imaging IR cameras, only a few
companies promote their products primarily as optical remote sensing IR cameras for
pollutants. A standard thermal imaging IR camera used for pollutant detection costs
approximately $80,000. Table 2-27 summarizes these vendors and their website information.
Table 2-27. Thermal IR Camera Vendors
VENDORS
FLIR, Inc.
www.flir.com
Opgal Optronic Industries Ltd.
www.opgal.com
IR Cameras, Inc.
www.ircameras.com
Infrared Cameras, Inc.
www.infraredcamerasinc.com
Strengths and Limitations
The thermal IR camera has a variety of strengths and limitations that should be considered for
each application. A summary of strengths and limitations is shown in Table 2-28 and Table 2-
29, respectively. Utilizing a thermal IR camera is typically a more economical approach to leak
detection than traditional methods. The camera can identify the exact source of a leak from
safe distances within a plant. However, training is required for operating personnel, and
quantitative results cannot be obtained without introducing additional measurement
technology. Additionally, for outdoor use, gas detection becomes more challenging on overcast
days and the IR camera is not waterproof and therefore has limited use on rainy days.
-------
ORS Handbook
Section 2.0
Page 2-76
Table 2-28. Summary Table of the IR Camera's Strengths
Feature
Strength
Economical
Fast screening speed compared to conventional leak
detection methods18
Leak assessment can be done without interruption to
plant operations2,3
Cost-effective compared to traditional leak detection
methods.2,3
Qualitative Results
Accurately assess the size of each leak18
Leak Identification
Better able to isolate the exact source of a leak,
despite proximity to other leaking sources, in real
time and record in a video format.2,3,18
Leak Detection from a Distance
Wide field of view: More likely to identify leaking
components in unconventional places.3,18
Exposure risk minor because leaking components can
be viewed at a distance.2,3,18
Table 2-29. Summary Table of the IR Camera's Limitations
Feature
Limitation
Qualitative Results
Cannot quantify the concentration of a leak without
additional technology3
Training Requirements
Camera use requires individuals with specific
training. Some models are easier to use than
others.3,18
Meteorological Limitations
Cannot be used during rain or fog and is not as
effective during overcast skies.18
The camera has a specified nominal operating range
for ambient temperature.18
Safety Requirements
Operation is not intrinsically safe and use is limited in
hazardous areas.18
References
1. Miseo, E. V. and N. A. Wright. 2003. Developing a Chemical-Imaging Camera. The
Industrial Physicist. 29-32.
2. Trefiak, T. 2006. Pilot Study: Optical Leak Detection & Measurement. Prepared by
ConocoPhillips (October 16).
3. U.S. EPA. 2005. Natural Gas STAR Partner Update.
4. Footer, T.L., J.M. DeWees, E.D. Thoma, B.C. Squier, C.D. Secrest, and A.P. Eisele. 2015.
Performance Evaluations and Quality Validation System for Optical Gas Imaging Cameras
-------
ORS Handbook
Section 2.0
Page 2-77
that Visualize Fugitive Hydrocarbon Gas Emissions. In Proceedings of the 108th Annual
Conference of the Air & Waste Management Association. Raleigh, NC, June 25, 2015.
5. U.S. EPA. 2009. Infrared Spectroscopy and Imaging, http://www.clu-
in.org/characterization/technologies/infrared.cfm. December 21.
6. FLIR Systems. 2008. Data sheet: ThermaCam GasFindIR CO Infrared Camera for Leak
Detection and Repair.
7. FLIR Systems. 2007. Data sheet: ThermaCam GasFindIR LW Infrared Camera for Leak
Detection and Repair.
8. FLIR Systems. 2008. Data sheet: ThermaCam GasFindIR HSX Infrared Camera for Leak
Detection and Repair.
9. FLIR Systems. 2009. Data sheet: ThermaCam GasFindIR GF300 Infrared Camera for Leak
Detection and Repair.
10. Footer, T.L. 2016. "Recent Observations from Optical Gas Imaging Camera Evaluations
for Hydrocarbon Leak Detection." Presented at the ITC InfraMation Conference,
September 27-29, 2016, Las Vegas, NV.
rrl
11. Coffey, Tom. 2009. Alternative Applications for GasFindIR. Presented at the 3 Annual
PetroTherm Conference in League City, Texas (February).
12. Reese, D., C. Melvin, and W. Sadik. 2007. Smart LDAR: Pipe Dream or Potential Reality?
Exxon Mobil Corporation.
13. Footer, TL. 2015. DRAFT Technical Support Document: Optical Gas Imaging Protocol (40
CFR Part 60, Appendix K). Prepared for the U.S. EPA and available as docket document
EPA-HQ-OAR-2010-0505-4949.
14. Rothman, L.S., I.E. Gordon, Y. Babikov, et al. 2013. "The HITRAN 2012 Molecular
Spectroscopic Database." Journal of Quantitative Spectroscopy and Radiative Transfer,
130, 4-50.
15. Footer, T.L., and J. DeWees. 2014. "Recent Observations on the Performance of Optical
Gas Imaging Cameras for Visualizing Fugitive Hydrocarbon Gas Emissions." Presented at
the 14th ISA LDAR Fugitive Emissions Symposium, May 19-22, New Orleans, LA.
16. Spectral Remote Sensing & Detection. 2009. Sensitivity Reliability of Gas Imaging Leak
Detection of SF6. Presented to the 2009 Workshop on SF6 Emission Reduction Strategies
(February).
17. Benson, R., R. Madding, R. Lucier, J. Lyons, and P. Czeripuszko. 2006. "Standoff Passive
Optical Leak Detection of Volatile Organic Compounds using a Cooled InSb Based
Infrared Imager." Proceedings of the 99th A&WMA Annual Conference and Exhibition,
June 21-23, New Orleans, LA.
-------
ORS Handbook
Section 2.0
Page 2-78
18. Picard, D., J. Panek, and D. Fashimpaur. 2006. Directed Inspection and Maintenance Leak
Survey at a Gas Fractionation Plant Using Traditional Methods and Optical Gas Imaging.
Presentation for AWMA Annual Conference (June 22).
19. Federal Register Volume 73, No. 246 (73 FR 78199 78199-78219 [E8-30196]), QA/QC
Requirements Alternative Work Practice to Detect Leaks from Equipment. 12/22/2008.
20. Zeng, Y. 2012. White Paper on A Calibration/Verification Device for Gas Imaging Infrared
Cameras. Providence Photonics, Inc., June 25, 2012.
21. Sandsten, J., U. Wallgren, M. Barenthin Syberg, and H. Hagman. 2015. Optical Gas
Imaging Standard for Sensitivity and Detection of Gases. Proceedings of the Air & Waste
Management Association 108th Annual Conference and Exhibition, Raleigh, NC, June.
22. Bullock, Adam. 2009. TCEQ and Thermal Imaging Technology. Presentation FLIR
Petrotherm 2009 Conference (February 24-25).
-------
ORS Handbook
Section 2.0
Page 2-79
2.6 Cavity Ring-Down Spectroscopy
There are multiple variations on cavity enhanced absorption techniques based on the property of
the time required to reduce the signal due to the absorption or scatter of laser light. CRDS and
Integrated Cavity Output Spectroscopy (ICOS) are examples of laser absorption spectrometry that
measures optical extinction of compounds that scatter and absorb light in a closed sample path.
This chapter describes first generation Cavity Ring-Down Laser Absorption Spectroscopy (CRLAS) as
an example of this widely used optical technology to measure in situ concentrations of gaseous
samples that absorb light at specific optical wavelengths down to the part-per-trillion level.
Although noteworthy for a broad range of applications, this technology is most often used for
measurements of weakly-absorbing or highly-dilute atmospheric samples.
Traditional absorption spectroscopy techniques measure the absolute change in light intensity
after passing through a sample relative to the original intensity of the light. CRDS techniques
improve on these methods by measuring the rate of decay of light intensity exiting from a high-
finesse optical cavity. Using the rate of decay rather than the change in light intensity makes the
CRDS technique less sensitive to fluctuations in the source laser intensity or variations in ambient
conditions (such as humidity). Moreover, the reflectivity of the closed optical (or ring- down) cavity
yields much longer effective sample path lengths for greater detection sensitivity.
In CRDS applications, the measured rate of decay of light intensity over time is a function of the
cavity length, the ability of the optical mirrors to achieve perfect reflectance, and the absorptivity
(s) of the sample. Because the cavity length and mirror reflectance are constant between
successive analyses, the amount of time required for the light intensity to decay to 1/e of is initial
intensity (herein referred to as the 'rate of decay') within an empty ring-down cavity and one where
the target sample is present is entirely the result of the sample absorbance.1
Basic Operation
Each gaseous compound absorbs energy at different wavelengths, usually more than one,
depending on vibrational and rotational excitement within the molecule. Therefore, each
-------
ORS Handbook
Section 2.0
Page 2-80
compound has its own "signature" of bands from which energy may be absorbed. Each band is
highly selective, with virtually no absorption occurring outside of a specific wavelength. Once a
compound has been identified, its spectrum can also be used to measure the compound's
concentration because the amount of infrared radiation absorbed from the IR beam is proportional
to the concentration of the compound in the sample or open path.
the absorption path length may or may not be equal to the cavity length, depending on the
experimental design; the total reflection pathway may be used instead, as with prism cavities.2
The CRDS technique enhances sensitivity to target analytes by significantly increasing the
pathlength using an optical resonator (or ring-down cavity). Increased sensitivity of the CRDS
technique relative to that of conventional absorption spectroscopy was demonstrated in the
inaugural experiments published by O'Keefe and Deacon in the visible region of the electromagnetic
spectrum3.
The absorption spectrum (or collection of spectral features for a single species over a range of
wavelengths) of a gaseous molecule is the spectroscopic equivalent to a fingerprint as each
compound will absorb energy at different wavelengths depending on the quantum properties of
that compound. Given the simplicity of CRDS systems, there are no intrinsic limitations to the
spectral region of which CRDS can be applied.4 Indeed, studies have proven that successful
measurements from the far-IR (12 |-im)5 to UV (197 nm)6 are possible with current technologies.
Theoretically, any spectral region can be probed with CRDS if the following three conditions are
satisfied:
1. Mirrors of sufficiently high reflectivity are available in the spectral region of interest. High-
speed detectors are employed that can confidently measure very small differences in
duration on the order of microseconds (|as) or less.
2. Tunable pulsed lasers or optical wavelength modulation for continuous-wave CRDS
applications are available.
-------
ORS Handbook
Section 2.0
Page 2-81
General Experimental Design
A general schematic diagram of the essential components of a CRDS apparatus is shown in Figure 2-
30. The heart of this technique is the sample cell which is bounded by highly reflective, dielectric-
coated, concave mirrors. Traditionally, the technique evolved using two of these mirrors as
depicted in Figure 2-30, but many mirror configurations that employ three or more mirrors have
been attempted (some of which are available through commercial vendors) and are shown to
improve the measurement quality from CRDS applications. Regardless of how many mirrors are
utilized, the ability of the highly reflective mirrors to achieve maximum reflectivity over the full
wavelength range of interest largely depends on the nature of the dielectric coating selected.
Concave High
Reflectivity Mirrors
Laser n^/saraleV
Photodetector
source Li v. J L
I
Decay Rate
Analysis
Figure 2-30. The essential components of any CRDS experimental set-up
As shown in Figure 2-30, basic CRDS measurements are acquired by optically coupling laser light
through the input mirror of a closed sample chamber bounded by two (input and output) non-
cofocal, highly-reflective optical mirrors and measuring the rate of light intensity decay over time.
To achieve proper optical coupling and cavity reflectance, the bandwidth of the laser radiation
needs to be sufficiently narrow enough to excite only a single optical mode of the cavity while also
being sufficiently narrower than that of the spectral features of the sample to obtain well-resolved
results (this is illustrated later in Figure 2-31).
When the laser light enters the closed optical cavity, it reflects off the bounding mirrors with a
-------
ORS Handbook
Section 2.0
Page 2-82
known amount of light exiting the cavity on each reflection (defined by the mirrors' reflectance). If
the optical cavity is empty, this rate of light intensity decay is characterized by a steady, exponential
decrease to zero (like the single-exponential function plotted in Figure 2-25). If a gaseous species
that absorbs the laser light is introduced into the cavity, then the intensity decay rate will be faster
depending on the concentration of the absorbing species.
The time it takes for the light intensity in an optical cavity to decay to 1/e of its original value is
called the cavity ring-down time (RDT or r) and is illustrated in Figure 2-25. This illustration depicts
the light lost from the cavity with each pass of the reflecting light as measured by a PMT detector.
The smoothed exponential curve above the oscillating data in the figure was derived from an
algorithm applied to the data by instrumental software. The difference between the RDT curve of
an empty cavity and the RDT curve of a cavity that contains sample is directly proportional to the
concentration of the absorbing gas species in the sample. If the empty cavity ring-down time, x, is
known, measurements of the decay rate of light intensity obtained as each laser wavelength is
scanned yields a complete absorption spectrum for each analyte.
Figure 2-31. Schematic representation of the expected rate of decay.
The laser light source can be pulsed or continuous wave (CW), the differences in the experimental
design between the two techniques is mostly in the number and arrangement of optical
components. Romanini et aI7 demonstrated the first application of CRDS with a continuous laser
time
-------
ORS Handbook
Section 2.0
Page 2-83
source (or CW-CRDS) such as a TDL. In their study, Romanini et oi. found that this change in the
laser light source lead to gains in spectral resolution, signal intensity, and data acquisition rate.
Instead of receiving a signal in the shape illustrated in Figure 2-32, the instantaneous power of a
continuous laser is lower, but usually concentrated into a narrower bandwidth.2,8 Figure 2-26
illustrates the difference between the incident pulsed laser light and that of the CW laser light.
Cavity Modes
1 1
cw Laser Line
* Pulsed Laser
- Absorber
J.
' ~
z
fWV I I i I r i ] i I i i I i i : i I I i i I I i i i i ; i i i ฆ I i i 1 i I i i i I I l VVV|
-2
1 0 1
Frequency (GHz)
Figure 2-32. Comparison of Pulsed and Continuous Wave Laser Light Illustrating the overlap between the
band widths of the absorbing species (Absorber), the pulsed laser source, the CW laser source, and the
optical cavity resonance frequency modes.
The CW laser bandwidth must be matched to the narrow transmission limits of the optical cavity to
allow injection of the light into the cavity; this can be done by adjusting the length of the optical
cavity, modulating the laser properties, or a combination of both.2 Looking at Figure 2-26, it is easy
to observe the more passive mode matching of the pulsed laser source with a wider bandwidth
capturing a complete cavity mode versus the more difficult mode matching of the thinner CW laser
source to superimpose on a narrow cavity mode.9 While the CW-CRDS method is more optically
complicated than the pulsed CRDS method, the low-cost, high-performance, and practical energy
requirements of the TDL allows more flexibility for field applications.2
Once the light is transmitted into the cavity, the light source must be turned off to observe the
-------
ORS Handbook
Section 2.0
Page 2-84
decay in light intensity in one of two ways: the operator can either configure the optics to allow
build-up of the light intensity through constructive interference to a predetermined threshold
before extinguishing the light source, or he can turn the light source off immediately once a signal is
produced from the detector. There are many ways to turn off the light source, but the most
common is using an acousto-optic modulator (AOM) as a laser 'shutter.'
Figure 2-33 provides a more detailed illustration of the experimental design for CW-CRDS
applications.8 An optical isolator can be installed immediately after the laser source to reduce on-
axis back reflections and increase the signal-to-noise ratio.10 A piezoelectric transducer (PZT) is
added to modulate the cavity length with a triangular signal to achieve greater cavity resonance,
increasing sensitivity, and further improving the signal-to-noise ratio.11 Optical filters and lenses
can be added immediately before the PZT to augment the laser spectral line selectivity and improve
cavity mode-matching if necessary/desired. Additional optics may be incorporated into the
experimental design after the ring-down cavity for similar purposes, depending on the
experimental application and the selection of the optical detector type. It is important to note that
the PZT is included for CW applications only.
Air
Sample
Dry
purge
PZT gas i
(cw only) .
Dry
purge
gas
Sample
Volume
Volume,
Mirrors
Detector
Mode Matching
Optics
(Optional)
cw Only
(If Needed)
Exhaust
AOM or
other switch
Laser
Isolator
Data Acquisition
Figure 2-33. Optical components schematic of a cavity ring-down spectrometer
When the light is coupled out of the laser, a detector generates a signal that is ultimately relayed to
a personal computer for processing and storage. The type of detector integrated into the
experimental design largely depends on the application and the required signal format. Most
commonly, researchers will use a PMT to detect the CRDS signal because it has good quantum
-------
ORS Handbook
Section 2.0
Page 2-85
efficiency, good spectral range, and a high gain in addition to being lower in cost. Alternatively, a
CCD has comparable, if not better, spectroscopic properties while providing a two-dimensional
read-out and having multiple channel capabilities. PMT and CCD detectors are the two most
frequently cited detectors used in CRDS studies. Figure 2-34 illustrates the difference between the
signals received by either detector.
Fปfri o to m li 11 i pi i
-------
ORS Handbook
Section 2.0
Page 2-86
The most extensive application of CRDS technology has been atmospheric studies. Pulsed laser
CRDS has evolved through many design permutations and is currently manufactured using off- axis
cavity-enhanced absorption (OA-CEA) techniques with advancements to the construction of the
reflective cavity components. Although appropriate for both pulsed and continuous wave light
source applications, CEA or ICOS and off-axis integrated cavity output spectroscopy (OAICOS or OA-
CEA) are more frequently used with the CW approach.8
Most CRDS experiments are performed on gaseous samples due to the simplicity with which gases
can be introduced into the sampling cell. However, attempts to broaden the applicability of the
technology have led to a few studies that sample surfaces, thin films, liquids, and solids, although
no sample medium has been as extensively studied with CRDS than air.
Pollutants and Relative Levels That Can Be Measured
The applications for pulsed CRDS and its numerous variants (such as CW-CRDS, FT-CRDS, CEA/ICOS,
etc.) and the studies performed on these techniques are limitless. Therefore, the list of detectable
pollutants provided in Table 2-30 is only a cursory list and does not represent all possibilities.8
-------
ORS Handbook
Section 2.0
Page 2-87
Table 2-30. Example list of detectable pollutants by CRDS. Wavelength (\), sensitivity, and minimum
detectable mixing ratio at la noise level for common gaseous species.
Species
Method
Approximate A
(nm)
Sensitivity (cm-1)
Mixing Ratio (ppbv)
CH4
CW CRDS
1.65
1.5 x 10-8
52
C2H2
CW CRDS a
1.5
~ 4 x 10-9
4
TNT
Pulsed CRDS b
00
1
to
9 x 10-9
0.075
Chlorobenzenes
Pulsed CRDS
0.266
-
ppmv levels
C02
CW CRDS
1.57
~ 4 x 10-9
2500
CO
CW CRDS c
1.57
~ 4 x 10-9
2000
NH3
CW CRDS c
1.5
~ 4 x 10-9
19
NO
CW CRDS d
5.2
5 x 10-8
0.7
N02
CW CRDS c
0.41
7 x 10-9
0.4
N03
CW CRDS c
0.662
1 x 10-9
0.002
N205
CW CRDS
0.662
1 x 10-9
0.0012
HONO
Pulsed CRDS c
0.354
2 x 10-8
1.7
OH
Pulsed CRDS
0.309
-
ppmv levels
Hg
Pulsed CRDS
0.254
-
0.001
H 180 2
Pulsed CRDS e
0.95
-
7 %o
Aerosol
Pulsed CRDS
0.532
1 x 10-10
-
a h
Samples were analyzed ambient or lab air unless otherwise noted. Sampled in a flame matrix.
Synthetically prepared sample. c Sampled from a lab source, pure gas or carrier gas mixture matrix.
d p
Sampled in human breath. Sampled from prepared standard.
The sensitivity of the CRDS method is determined by the fractional loss of light intensity per round-
trip in the cavity. The absorption of laser light intensity for a single pass through the optical cavity
is given by the Beer-Lambert law, which is used to approximate the minimum detectable absorption
in CRDS. The equation below shows that the minimum detectable absorbance per pass is
dependent upon the reflectivity (R) of the mirrors and the accuracy in the determination of x and
the precision of Ax (or the precision of the number of round trips in the cavity (N).13
-------
ORS Handbook
Section 2.0
Page 2-88
SIm * (l-R)^L = (1 -
r N
Where:
5lm = the minimum detectable change in absorbance
R = mirror reflectivity
Aimin = minimum detectable change in the ring-down time (i.e., precision of Ax)
N = the number of round trips in the cavity
ANmin = the accuracy to which N can be measured
Typical QA/QC
There are three main requirements in the experimental design when CRDS is used. First, the laser
source must emit the wavelength absorbed by the target analyte or a range of wavelengths that
includes the absorption wavelength. Also, the cavity mirrors must be able to reflect the light in the
wavelength region of interest. Finally, the detector must be fast enough to detect changes that
occur in very short time intervals (|as). The rest of the instrumental configuration involves
adjustments (such as cavity length modulation or cavity resonance mode matching) to achieve an
absorbance signal. The optical modes of the entire system must then be harmonized to acquire a
signal from the detector.
Accuracy, precision, linearity, zero/calibration drift, and response time can be evaluated using
known concentrations of gas standards and/or zero air. Simultaneous measurements based on a
reference method such as EPA CTM-027 could provide the reference data for assessing CRDS
comparability. The data completeness is simply a measure of the amount of valid data points
achieved versus the total amount of data points expected for a given period. Other factors that
become important during operation such as maintenance requirements, consumables used, ease-
-------
ORS Handbook
Section 2.0
Page 2-89
of-operation, and frequency of repairs are being assessed through EPA verification tests.15
Example Applications and Vendors Applications
Although the technology is new to regulated environmental applications, it is not foreign to
experiments involving the direct monitoring of environmental contaminants. A work by Berden et
al. [2000] lists a comprehensive overview of published findings with CRDS technology from the
official date of inception (with O'Keefe and Deacon in 1988) to the date of publication (2000).3,4
This list contains well over 200 spectral features detected by CRDS and includes absorption
wavelengths, almost continuously, from 205 nmto 10,617 nm. Moreover, Atkinson [2003] and
Brown [2003] individually identified chemical divisions for the most potential applications of CRDS
in direct environmental contaminant monitoring into the following classes:2'8
Nitrogen oxides and Nitrous Acid (HONO)
Ammonia
Elemental mercury and volatile mercury compounds
Carbon monoxide and carbon dioxide
Methane, hydrocarbons, and formaldehyde
Atmospheric aerosol particulates
The limitations to the application of CRDS technology depend on the development of individual
optical system components. The overall applicability and usefulness of the method has only begun
to be explored and various transfigurations of CRDS will enjoy extensive application in
environmental analytical chemistry as the merits of these methods are proven in time. Table 2-31
summarizes the common applications for CRDS technology.
-------
ORS Handbook
Section 2.0
Page 2-90
Table 2-31. Typica
Applications for OP-TDL.
Technology
Applications
CRDS
Tracer Gas Correlations
Vendors
There are currently three known proprietors of commercial CRDS systems: (1) Picarro, Inc., (2) Los
Gatos Research, Inc., and (3) Tiger Optics, Inc. Each manufacturer has developed a system different
from the others. The main technological differences between each manufacturer's designs have
primarily to do with the ring-down cavity configuration and construction materials. The cost of a
CRDS system ready for field use ranges between $40K to $150K, depending on specific application
and configuration. Table 2-32 lists CRDS vendors and their internet contact information.
Table 2-32. CRDS Vendors
Vendors
Picarro, Inc. (CRDS)
www.picarro.com
Tiger Optics
www.tigeroptics.com
Los Gatos Research (ICOS)
www.lgrinc.com
Strengths and Limitations
CRDS can be used as a qualitative tool to provide specific information about volatile IR energy-
absorbing molecules. It can also be used as a quantitative tool to provide the concentration of
many gas-phase molecules. A summary of strengths and limitations is shown in Tables 2-33 and
2-34. One of the main strengths of CRDS is that it measures time and not absorbance, making the
technique immune to environmental variations and laser intensity fluctuations while concurrently
increasing the linear dynamic range. Moreover, using a high-finesse optical cell, CRDS has greatly
increased the technology sensitivity to target compounds without adding complicated sample pre-
conditioning steps. The use of the optical cell further enhances the technological design to
withstand vibration making field applications of the technology simpler.
The CRDS application is limited mostly by the properties of the high-reflectivity optical mirrors. The
mirrors used for CRDS have a high amount of wavelength specificity but lack the flexibility
-------
ORS Handbook
Section 2.0
Page 2-91
necessary to allow simultaneous multiple species detection and/or a broad species application
range. High-reflectivity optical mirrors currently are only able to reflect about 15 percent of the
target wavelength on either side.
Table 2-33. Summary Table of CRDS Strengths
Feature
Strength
Simple design
Minimal maintenance required and no consumables
are needed. Turnkey operation with the potential for
remote access and control. "User friendly."
Fast detector
Ability to measure very small changes in short time
frames. Can rapidly scan spectra continuously for high
temporal resolution and real- time results.
Multi-pass, high-finesse,
stable optical cell
Greatly increases sensitivity with much longer effective
pathlengths. Insensitive to vibrations during
measurements.
Broad-band source capable
Allows for extended wavelength range scanning,
increasing sensitivity by probing multiple absorption
lines while also eliminating other interferences.
Internal temperature and pressure
controls
Minimal-to-no drift making frequent calibration
unnecessary. Enhanced accuracy and system
stability.
Measures time, not absorbance
Renders the method immune to ambient changes
(such as relative humidity and temperature) and
laser intensity fluctuations.
Direct sampling
Little-to-no sample pre-conditioning or treatment
required before analysis.
-------
ORS Handbook
Section 2.0
Page 2-92
Table 2-33. Summary Tab
e of CRDS Strengths (continued)
Feature
Strength
Compact system
Easy field deployment and installation. Quick
sample exchange in a smaller volume cavity with
moderate flow rates. Advances in components
allow for a rugged portable system
Can use low power optical sources
Logistically simpler for field use to eliminate the
need for a large power source.
Table 2-34. Summary Table of CRDS Limitations
Feature
Limitation
Measures only total extinction
May need to apply sample filtering
components to avoid interferences.
Laser light source
Limits the method to the laser spectral
ranges available.
High-reflectivity mirrors
Are only able to reflect over a small wavelength
range (about ฑ 15%) relative to the center
wavelength.
Multiple species detection difficult.
High quality lasers and mirrors
Key components that typically drive up the cost
of the instrumentation, depending on
application.
References
1. Wheeler, M.D. S.M. Newman, A.J. Orr-Ewing, and M.N.R. Ashfold; (1998). "Cavity ring-
down spectroscopy." Journal of the Chemical Society, Faraday Transactions, 94(3): 337.
2. Atkinson, D.B.; (2003). "Solving chemical problems of environmental importance using cavity
ring-down spectroscopy." Analyst, 128:117.
3. O'Keefe, A., and D.A.G. Deacon; (1988). "Cavity ring-down optical spectrometer for
absorption measurements using pulsed laser sources." Review of Scientific Instruments,
59(12): 2544.
4. Berden, G., R. Peeters, and G. Meijer; (2000). "Cavity ring-down spectroscopy:
Experimental schemes and applications". International Reviews in Physical Chemistry,
19(4): 565.
-------
ORS Handbook
Section 2.0
Page 2-93
5. Bisson, S.E., T.J. Kulp, 0. Levi, J.S. Harris and M.M. Fejer; (2006). "Long-wave IR chemical
sensing based on difference frequency generation in orientation-patterned GaAs." Appl.
Phys. B - Lasers and Optics, 85(2-3): 199.
6. Sneep, M., S. Hannemann, E.J. van Duijn, and W. Ubachs; (2004). "Deep-ultraviolet cavity
ringdown spectroscopy." Optics Letters,29(12): 1378.
7. Romanini, D., A.A. Kachanov, N. Sadeghi, and F. Stoeckel; (1997). "CW cavity ring down
spectroscopy." Chemical Physics Letters, 264: 316.
8. Brown, S.S.; (2003). "Absorption Spectroscopy in High-Finesse Cavities for Atmospheric
Studies." Chemical Reviews, 103(12): 5219.
9. Paldus, B.A, and A.A. Kachanov; (2005). "An historical overview of cavity-enhanced
methods." Canadian Journal of Physics, 83: 975.
10. Hippler, M., and M. Quack; (1999). "Cw cavity ring-down infrared absorption spectroscopy in
pulsed supersonic jets: nitrous oxide and methane." Chemical Physics Letters, 314: 273.
11. Thiebaud, J., and C. Fittschen; (2006). "Near infrared cw-CRDS coupled to laser photolysis:
Spectroscopy and kinetics of the HO^ radical." Applied Physics B, 85: 383.
12. Pemberton, J.E., and R.L. Sobocinski; (1989). "Raman Spectroscopy with Helium-Neon
Laser Excitation and Charge-Coupled Device Detection." Journal of the American Chemistry
Society, 111(2): 432.
13. Crosson, E.R.; (2008). "A cavity ring-down analyzer for measuring atmospheric levels of
methane, carbon dioxide, and water vapor." Applied Physics B, 92: 403.
14. Chen, H., J. Winderlich, C. Gerbig, A. Hoefer, C.W. Rella, E.R. Crosson, A.D. van Pelt, J.
Steinbach, O. Kolle, V. Beck, B.C. Daube, E.W. Gottlieb, V.Y. Chow, G.W. Santoni, and S.C.
Wofsy; (2010). "High-accuracy continuous airborne measurements of greenhouse gases (CO^
and CH^) using the cavity ring-down spectroscopy (CRDS) technique." Atmospheric
Measurement Techniques, 3: 375.
-------
ORS Handbook
Section 2.0
Page 2-94
2.7 Particulate Matter LIDAR
Light detection and ranging (LIDAR) technology is based on measuring the speed and wavelength
of a laser signal that has reflected off compounds of interest. LIDAR operates on the same
principles as radio detection and ranging (RADAR) except laser light is used as the energy source
instead of radio waves. The laser light is aimed toward a material of interest and the properties of
the backscattered (or reflected) light correspond to the physical characteristics of the encountered
material (e.g., gas concentration, density, temperature, humidity, and wind). LIDAR can operate
across many wavelengths, from ultraviolet (UV) to far-infrared (IR), and is therefore utilized in a
variety of applications. From high-resolution mapping to atmospheric content measurements and
from range-finding to autonomous vehicle navigation, the flexibility of LIDAR technology and the
simplicity of its application relative to other remote sensing methods result in a wide distribution
of LIDAR to a multitude of disciplines. This section discusses LIDAR technology as its application to
the measurement of particulate matter (PM) from ground-based sources, with emphasis on
tropospheric applications. LIDAR is well suited to measure tropospheric PM concentrations due to
the high spatial resolution and ability to monitor temporal variations.
The use of LIDAR technology to detect PM in the atmosphere was first explored in the 1960s (e.g.,
Collis and Ligda, 1966).1 The detection of this PM led to some of the first ground-based
observations of the stratospheric structure, but did not have good sensitivity for heights less than
about 20 kilometers (km). Since then, LIDAR technology has advanced with the development of
laser technology to include the following remote sensing techniques:
Elastic-backscatter LIDAR
Coherent, Raman, or Doppler LIDAR
Differential absorption LIDAR (DIAL).
Elastic LIDAR simply measures the changes in magnitude of the reflected light, while coherent and
Raman LIDAR provide information on changes in the wavelength of the reflected light. DIAL,
discussed in Section 2.4, was developed primarily for the spatial measurement of trace chemicals
-------
ORS Handbook
Section 2.0
Page 2-95
in the atmosphere and achieves results by calculating a ratio between two different wavelengths
of laser light. One wavelength is strongly absorbed by the species of interest and is used to probe
for concentration, while the other is just outside of the absorption range of the species of interest
and is used to collect background light scattering. The time-adjusted ratio of these two
wavelengths indicates the location and concentration of the species of interest.2 The increase in
instrumental complexity for the DIAL systems limits the use of this technique due to limited
availability and the prohibitive cost of associated equipment.3
Basic Operation
LIDAR-based systems consist of two basic components: the transmitter and the receiver. The
transmitter consists of the power supply, laser light source (typically a neodymium-doped yttrium
aluminum garnet; Nd:Y3AlsOi2 (Nd:YAG)), and any modulating optics required to direct the laser
light into the material to be sampled. The receiver consists of a sensor, processor, and any optics
required to detect the reflected light. In the simplified illustration of a LIDAR system in Figure
2.354, the laser light source transmitted to the atmosphere is passed through optics that expand
and collimate the light to deliver even light distribution across the laser beam path diameter and
to disperse the laser power over a larger area for eye-safe applications.4 Eye-safe configurations
also include micro-pulsed lasers, wherein the energy transmitted with the laser pulse is below the
ANSI 2136.1-1986 laser exposure safety standard in which the maximum permissible exposure is 5
xlO"7 J/cm2 in the 520-530 nanometer (nm) wavelength range.5 The transmitted light is
backscattered by the dust, gases, and aerosols present in the target atmosphere and collected by
return optics, such as a Newtonian or Cassegrain telescope and spectrally separating optical
components, and directed into the detector (typically, photomultiplier tube or avalanche
photodiode) for signal detection. The electronic signal is then processed and converted to a digital
output for analysis.4
-------
ORS Handbook
Section 2.0
Page 2-96
ATMOSPHERE
Optics
5
Optics
Laser
Power supply
\
Detector
Processor, data
acquisition
Transmitter
Receiver
L J
la
Figure 2-35. Simplified Schematic of LIDAR System.
Currently, the calculation of emission concentrations from LIDAR signals represents the largest
amount of uncertainty in the LIDAR system processing equations.4 The retrieval algorithms must
consider the geometry of the instrumental optics in addition to the geometry of the target
material. Typically, general assumptions on particle "sphericity" and/or a priori information
related to the material extinction coefficient and backscatter coefficient may be required to
calculate the desired measurement results, depending on the LIDAR system design.4
The conventional LIDAR method, elastic-backscatter LIDAR, was the first technology design and
served as the foundation from which other configurations were developed. The simplified
equation for detecting the LIDAR signal from the elastic-backscatter LIDAR system is:
PR = KGRf3RrR
Where:
Pr = power P received at distance R from the object.
-------
ORS Handbook
Section 2.0
Page 2-97
K = measure of the LIDAR system performance.
Gr = range-dependent measurement geometry.
Br = backscatter coefficient at distance R.
Tr = amount of transmitted laser light that is lost when traveling distance R and back.
The geometry term (Gr) in the LIDAR equation is described as Or/R2, where Or is the overlap
function to account for the area where the laser transmission beam and the receiver field of view
overlap. This term refers to the geometry of the LIDAR system (Figure 2.XX2) and, along with the K
term, is adjustable by the LIDAR developer/operator. At the point of transmission, the overlap
function is zero and approaches unity with distance away from the LIDAR. Therefore, a section
called the "blind zone" exists in every LIDAR system where the overlap function is insufficient for
providing robust measurements and the amount of calculation error is too great to report a result
with adequate certainty. This would occur if the distance to R in Figure 2.364 was short enough
such that the diameter of backscatter area reflected from one particle (and illustrated with the
return perception angle in Figure 2.XX2) did not exceed the diameter of the laser beam at the
point of transmission. Where this area occurs relative to the position of the LIDAR depends on the
design optics and can range from <50m to 20 km.6
-------
ORS Handbook
Section 2.0
Page 2-98
receiver field of view
volume
laser
pulse
t
Ri T
fR: I
effective
pulse length/
telescope area
scattering
perception angle
Figure 2-36. Illustration of LIDAR System Geometry.
The variable K describes the performance of a LIDAR system and can be described in more detail
with the following equation:
K= PojAV
Where:
K = measure of the LIDAR system performance
Po = average power of a single laser pulse
c = speed of light
r = duration of the laser pulse
A = area of the primary return optics responsible for the collection of the backscattered
light
q = overall system efficiency.
-------
ORS Handbook
Section 2.0
Page 2-99
The term cr/2 describes the amount of atmospheric volume that is illuminated by the laser light,
known as the effective laser pulse length. The telescope area A and the average laser power
(energy * pulse repetition frequency) are typical LIDAR system design parameters, with the
optimization of q for achieving the best possible LIDAR signal.4
There are two general designs for LIDAR systems, monostatic coaxial and monostatic biaxial, as
shown in Figure 2.37.7 In coaxial systems, the path for the transmitted laser light and the path for
the returning backscatter received are the same and are separated by optics within the
instrument. Biaxial systems have separate pathways for the transmitted laser light and received
backscatter signal.
Receiver field
of vision
4
Monostatic Coaxial
Laser Path
Detect cm
Transrritb ncj.'
Receiving
Optics
Filter
*
Laser
Source
Monostatic Biaxial
-IR Path ~ ~
f
Transmitting Optics
Laser
Source
Receiving Optics
Filler
Detector
Receiver field
of vision
Figure 2-37. Illustrations of Monostatic Coaxial and Monostatic Biaxial LIDAR Configurations.
Researchers at the Utah State University's Space Dynamics Laboratory investigated systems of
both monostatic types. The coaxial system (the AGLITE, Figure 2.3S11) has a three-wavelength
design derived from a Nd:YAG 1064 nm laser that is frequency-doubled and tripled to produce
transmissions that cover UV, visible, and IR wavelength regions for agricultural aerosol monitoring
applications. Light that is elastically backscattered from the quick (~ 10 nanosecond) laser
-------
ORS Handbook
Section 2.0
Page 2-100
transmission pulses is collected into a Newtonian telescope at a repetition rate of 10 kilohertz
(kHz). The operating range for the AGLITE system to produce confident results is about 500 meters
(m) to about 15 km with a resolution of about 5 m given the prototype specifications.
Measurements characterizing the density of probed aerosols as a function of distance from the
LIDAR system are derived from the temporal properties of each laser transmission "return."11
Transmission
Lidar
Return,
Lasers
Beam
Direction
Power
Control
E
U^4-=0=i
Filters/
Detectors
t
DDD
Data
Process
Data
Storage
uzzzzzzzzzzzzzzzzzzzzzzzzzzzzz
Figure 2-38. Illustration of the USU AGLITE LIDAR System.
The biaxial system (the Compact Eyesafe Lidar System, or CELiS) was developed for PM emissions
fenceline monitoring (Figure 2.39).12 Some of the optical components and performance
specifications are comparable to the AGLITE (i.e., Nd: YAG laser source, about 7 nanosecond pulse
width, range resolution of about 5 m), but with a single wavelength transmission at 1,574 nm and
a laser pulse repetition rate of 20 Hz.12
-------
ORS Handbook
Section 2.0
Page 2-101
20a Beam Expander
0.5mi ad divergence
3.3" exit beam
:
Wind
ov aaซl Baffles
1.57jim Laser
25mJ palse energ)
7ns pulse width
20H* PRF
\l
Off-axis telescope
InGaAs APD Detector, aft optics assembly, Barrow band
filter and alignment mount
Computer
Digitizer
Laser Cooling &
Electronics
Pover
Figure 2.39. Illustration ofUSU CELiS LIDAR System.
Pollutants and Relative Levels That Can Be Detected
LIDAR technology can measure atmospheric concentrations of trace gases, aerosols, and clouds in
addition to properties such as density, temperature, humidity, and wind. The flexibility of LIDAR
technology to probe for compounds in the UV, visible, and IR region of the spectra allows for the
detection of unlimited number of pollutants. LIDAR is also sensitive enough to distinguish between
water droplets and ice crystals in clouds and probe stratospheric air masses from the ground or
tropospheric concentrations from satellite. The European EARLINET network of LIDAR
measurement stations detected the 2013 forest fires occurring in the United States.13
Currently, very few commercial options are available for LIDAR PM monitoring systems, because
most LIDAR-only instruments are still in the research phase of development.4 The current systems
are predominantly airborne or satellite-based technologies. USA-based TSI Inc. offers light-
scattering laser photometers, which, instead of transmitting a laser beam of light into the
atmosphere, draw in a sample of the atmosphere and performs the laser light scattering
measurements internally.8 Components are available for laboratories, researchers, and hobbyists
to build their own systems, but manufacturers specifications for detection levels may not be
-------
ORS Handbook
Section 2.0
Page 2-102
available. US company Micro Pulse LiDAR and French companies Leosphere and Cimel
Electronique offer the most complete packages with their models MPL, EZ LIDAR and CE376
Compact Aerosol Analyzer, respectively. Although, both SES (USA) and Raymetrics (Greece) offer
components and/or services to build a custom modular system.
Typical QA/QC
Some LIDAR instruments (e.g., micro-pulse LIDAR) can achieve a few photons per microsecond
detection or less5 and can spatially resolve measurements down to 2 cm.6
Trace gases and water vapor measurements in the atmosphere can be compared to in situ
measurements from balloon sondes and airborne-based monitors for quality control purposes.
The complexity of aerosol measurements, however, are less straight-forward.10 In these cases, the
instrument performance can be verified either with collocated sun photometer measurements (in
the absence of clouds) or intercomparison with a different LIDAR system.
Example Applications and Vendors
LIDAR systems have an extraordinary wide range of applications, though they are most commonly
used to conduct 3D wind and topography surveys. Table 2-35 provides a general description of the
applications for LIDAR systems.
Table 2-35. Typical Applications for LIDAR Systems-
TECHNOLOGY
APPLICATIONS
LIDAR
DIAL, Elastic-backscatter, Raman-
backscatter, Ranging, Optical Density
Monitoring
Vendors
Vendors offering complete solutions for PM measurements via LIDAR are currently very few. Most
systems discussed in the literature are laboratory research developments that are not
commercially available yet. The vendors in Table 2-36 offer either a commercial system
-------
ORS Handbook
Section 2.0
Page 2-103
(Leosphere, Micro Pulse LiDAR, and Cimel), services to build a custom LIDAR (SES), or modules to
incorporate into a custom system (Raymetrics).
Table 2-36. LIDAR Systems Vendors
VENDORS
Micro Pulse LiDAR
www.micropulselidar.com
Leosphere
www.leosphere.com
Cimel Electronique
www.cimel.fr
SES, Inc.
www.sesincusa.com
Raymetrics
www.raymetrics.com
-------
ORS Handbook
Section 2.0
Page 2-104
Strengths and Limitations
As with every technology, the strengths and limitations depend on application and design.
Table 2-37. Table of LIDAR Strengths
FEATURE
STRENGTH
Reliable
Safe and reliable method for dust concentration
measurements, even for very low levels.
User-friendly
Easy to install and user-friendly operation.
Low maintenance, infrequent maintenance
requirements.
Spatial Concentration Resolution
Able to provide density and/or concentration heat
mapping down to very small spatial increments.
Laser Transmission Range
Can cover a wide range of area with one
configuration.
Receiving Optics
Does not require retroreflectors.
Near Real-time Response
Can provide results within minutes of measurement,
depending on sample averaging time.
Table 2-38. Table of LIDAR Limitations
FEATURE
LIMITATION
Backscatter Requirements
Sufficient PM must be in the laser path to create
sufficient backscatter for detection.
Wind Speed and Direction Variability
Rapidly changing wind speed or direction may cause
measurements to change rapidly.
Vendors
Small number of vendors providing LIDAR systems
and services.
Blind Zone
Unable to make accurate measurements in the near-
field.
Expense
Prohibitive cost for complete system solutions.
References
1. Collis, R.T.H. and M.G.H. Ligda. 1966. Note on Lidar Observations of Particulate Matter in
the Stratosphere. Journal of the Atmospheric Sciences. 23: 255257.
2. Edner, H., K. Fredriksson, A. Sunesson, S. Svanberg, L. Uneus, and W. Wendt. 1987. Mobile
remote sensing system for atmospheric monitoring. Applied Optics. 26(19): 43304338.
3. U.S. EPA. 2006. VOC Fugitive Losses: New Monitors, Emission Losses, and Potential Policy
-------
ORS Handbook
Section 2.0
Page 2-105
Gaps - 2006 International Workshop.
http://www.epa.gov/ttn/chief/efpac/documents/wrkshop fugvocemissions.pdf
4. Weitkamp, C (ed). 2005. Lidar: Range-Resolved Optical Remote Sensing of the
Atmosphere. Springer Series in Optical Sciences. Springer, NY, USA.
5. Spinhirne, J.D. 1993. Micro Pulse Lidar. IEEE Transactions on Geoscience and Remote
Sensing. 31(1): 4855.
6. Cimel Electronique. www.cimel.fr.
7. U.S. EPA. 2013. Measurement and Monitoring Technologies for the 21st Century (21M2) -
Open Path Technologies: Measurement at a Distance, Lidar. Available from https://clu-
in.org/PROGRAMS/21M2/openpath/lidar/.
8. TSI, Inc. www.tsi.com.
9. Sicard, M., Md. Reba, M., Tomas, S., Comeron, A., Batet, 0., Munoz, C., Rodriguez-Gomez,
A., Rocadenbosch, F., Munoz-Tunon, C., J. Fuensalida, J. 2010. Monthly Notices of the
Royal Astronomical Society. 405(1): 129142.
10. Matthais, V., V. Freudenthaler, A. Amodeo, I. Balin, D. Balis, J. Bosenberg, A. Chaikovsky, G.
Chourdakis, A. Comeron, A. Delaval, F. De Tomasi, R. Eixmann, A. Hagard, L. Komguem, S.
Kreipl, R. Matthey, V. Rizi, J.A. Rodrigues, U. Wandinger, and X. Wang. 2004. Aerosol Lidar
Intercomparison in the Framework of the EARLINET Project. 1. Instruments. Applied
Optics. 43(4): 961-976.
11. Wilkerson, T.D., G.E. Bingham, V.V. Zavyalov, J. Swasey, J.J. Hancock, B.G. Crowther, S.S.
Cornelsen, C. Marchant, J.N. Cuttis, D.C. Huish, C.L. Earl, J.M. Andersen, and M.L. Cox.
2006. AGLITE: A Multiwavelength Lidar for Aerosols. Space Dynamics Lab Publications,
Utah State University, Paper 143. http://digitalcommons.usu.edu/sdl_pubs/143.
12. Wojcik, M.D. and A.W. Bird. 2012. CELiS (Compact Eyesafe Lidar System): A Portable 1.5
|am Elastic Lidar System for Rapid Aerosol Concentration Measurement. Space Dynamics
Lab Publications, Utah State University, Paper 144.
http://digitalcommons.usu.edu/sdl_pubs/144.
13. Ancellet, G., J. Pelon, J. Totems, P. Chazette, A. Bazureau, M. Sicard, T. Di lorio, F. Dulac,
and M. Mallet. Long-range transport and mixing of aerosol sources during the 2013 North
American biomass burning episode: analysis of multiple lidar observations in the western
Mediterranean basin. Atmospheric Chemistry and Physics, 16, 47254742.
-------
ORS Handbook
Section 3.0
Page 3-1
3.0 Measurements Applicable to Emissions Flux
Optical remote technologies have been applied to answer a variety of fugitive and area source
emissions questions. The range of applications spans both short-term characterization and
measurement of emission flux (which is defined as measurement of pollutants that are either
constant or changing concentration over time) to long-term monitoring of trends in control strategy
performance. Technologies described in this chapter have been used in mobile applications to
screen pipelines or industrial sites for leaks or major sources and in stationary applications to
measure flux from landfills, waste lagoons, and petrochemical plants. In addition, the PM methods
described in this section can also be used in mobile or stationary settings to understand PM flux in
area sources.
Please note that technologies described in Chapter 2 can be used alone or in combination of the
techniques and methods described within this chapter. The techniques and methods in this chapter
also provide three major types of data: plume characterization, short-term flux measurements and
long-term monitoring studies.
Short-term flux measurement applications (e.g. DIAL, RPM, SOF, Tracer Release Correlation etc.)
are useful to determine the emissions from a complex area sources at one point in time. These
measurements provide an estimate of the emissions plume size and concentration of selected
target compounds or surrogates. Concentration profiling involves using ORS such as line of sight
open-path optical techniques. Profiling periodic changes of emissions in one dimension is often the
precursor of pilot stage of long-term monitoring at an area or fugitive emissions site.
Long-term monitoring is used to determine average trends of area sources emissions and to provide
an indication of seasonal or industrial cycle emissions profiles. Continuous concentration profiling is
also useful to determine process upsets or to diagnose the potential source of emissions from a
complex industrial area using back trajectory calculations. Typically, long-term monitoring needs to
be associated with a short-term measurement characterization of the size and composition of the
emissions plume to relate trends in monitoring data to emissions and emission factors.1
-------
ORS Handbook
Section 3.0
Page 3-2
These three types of data provide essential information on the annual emissions rate of fugitive
and area sources as well as a measure of the effect of emissions reductions efforts.
Chapter 3 of this Handbook introduces the use of technologies by describing applications that
measure short-term flux or mass emission rates from open or un-ducted gas and PM sources.
3.0.1 Reference
1. Hashmonay, Ram, Long Term Monitoring of Greenhouse Gas Emissions from Fugitive
and Area Sources, presented at the AWMA Symposium on Air Quality Measurements
Methods and Technology, Los Angeles, CA November 2010
-------
ORS Handbook
Section 3.0
Page 3-3
3.1 Radial Plume Mapping: Other Test Method 10
RPM is an ORS method used to determine fugitive emissions from non-point emissions sources
including fugitive emissions and area source emissions. Its main goals are to identify emission "hot
spots" over large scanned areas and measure emission fluxes. Fugitive emissions include air
pollutants released into the ambient air from pressurized equipment due to leaks and various other
unintended or irregular releases of gases. Examples of fugitive emissions include 0^ precursors,
benzene, and methane.1 For some source categories, fugitive and/or area source emissions are a
significant portion of the total pollutant emissions; therefore, it is important to be able to locate
and quantify these emissions.
The open-path configuration for ORS technologies was originally used to determine the average
concentration of a compound of interest over a path of known length. The line-of-sight, or one-
path, configuration provides an average of the compound of interest concentration per path length
(i.e., ppmm). Open-path monitoring expanded to include the measurement of the average
concentration over several distances along the open path length. These interval measurements
enabled estimation of the concentration profile of the plume. However, the survey of leaks and
hot spots over a large area is not possible with just one optical path. Directing the ORS path over
different lengths, as well as different horizontal and vertical paths, allows additional
characterization of the horizontal or vertical emissions plume profile. RPM, as discussed in this
section, is the outgrowth of multidirectional, multi-pathlength ORS and is used in combination with
mathematical algorithms to characterize the concentration profile over a horizontal or vertical
optical path plane.
General Description of Approach
Measuring the total amount of a fugitive emission over a large area is not simple. Earlier efforts
used traditional point sampling techniques such as canisters, sorbents methods, flux boxes, PID/FID
instruments and others.2 However, the traditional point methods only provide concentrations from
-------
ORS Handbook
Section 3.0
Page 3-4
a single point and fail to capture the temporal and spatial distribution of fugitive emissions over a
large area. These methods also fail to identify, if any, "hot spots" of fugitive emissions. RPM
provides a more complete survey of a large area. Open-path ORS measurement technology
mounted on a programmable aiming platform or scanner can be configured in a vertical plane to
measure emissions flux. When scanning in the horizontal plane (HRPM), results can be used to
locate hot spots at ground level. Emission fluxes are obtained when scanning in the vertical plane
(VRPM) downwind of the area source along with meteorological measurements data.
One-dimensional RPM, which scans along one line, such as an industrial fence line, is used to profile
pollutant concentrations downwind from a source and coupled with wind direction may also be
useful in locating emissions sources.2 Determining which scanning setup to use depends entirely on
the objectives of the project and data quality indicators.
Horizontal RPM Algorithm
The HRPM approach provides horizontal differentiation to path-integrated measurements by
optical remote sensing. This technique yields information on the two-dimensional distribution of
the concentrations in the form of chemical-concentration contour maps. In this application, the
plume mapping identifies chemical "hot spots," the location of high emissions. Horizontal radial
scanning is usually performed with the ORS beams located close to the ground. The survey area is
divided into a Cartesian grid of rectangular cells. A mirror is located in each of these cells and the
OP-FTIR sensor scans to each of these mirrors, dwelling on each for a set measurement time. The
measurement equipment scans to the mirrors in the order of either increasing or decreasing
azimuth angle. The path-integrated concentrations measured at each mirror are averaged over
several scanning cycles to produce time-averaged concentration maps. Meteorological
measurements are made concurrent with the scanning measurements.
The equation on page 3-5 illustrates how the measurements are made.
-------
ORS Handbook
Section 3.0
Page 3-5
Where:
PIC = path integrated concentration
K = Kernel matrix
k = number of index for the pixels
m = number index for the pixels
c = average concentration in the mth pixel
The kernel matrix includes the specific beam geometry as shown in Figure 3-2 where the diagonal
lines represent the K. Each value in the kernel matrix K is the length of the kth beam within the mth
pixel; therefore, the matrix is specific to the beam geometry. The HRPM procedure solves for the
average concentrations (one for each pixel) by solving the non-negative least squares-best fit for
the data. Then the algorithm multiplies the resulting vertical vector of averaged concentration by
the matrix K to yield the end vector of predicted PIC data. The second stage of the plume
reconstruction involves interpolation among the reconstructed pixel's average concentration,
providing a peak concentration not limited to the center of the pixels. A triangle-based cubic
interpolation procedure (in Cartesian coordinates) is currently used in the HRPM procedure.2 The
ORS instrument is typically placed at the origin (in the first quadrant of the Cartesian convention) of
the rectangular area to be measured. Once the HRPM measurement area and the number of path-
determining components (PDCs) have been determined, the area is divided into smaller
rectangular areas called pixels. The total number of pixels required is smaller or equal to the total
number of beam paths.6
-------
ORS Handbook
Section 3.0
Page 3-6
100
ฆ
ro
50
ฆ
PI-ORS
Instrument
x-axis
-50
50
100
Figure 3-1. Horizontal RPM setup
Vertical RPM algorithm
The VRPM algorithm uses multiple beam paths to survey the vertical pollutant concentration
profile as a function of distance from the measurement equipment. Two different beam
configurations of the VRPM methodology have been used: the five-beam (or more) and the three-
beam VRPM configuration. Figure 3-1 shows a VRPM configuration using six beams.3 In the five-
beam (or more) configuration, the ORS instrument sequentially scans the paths of five PDCs. Three
PDCs are along the ground-level crosswind direction, and the other two are elevated on a vertical
structure. Additional beam configurations provide better spatial definition of the plume in the
crosswind direction. In the three-beam configuration, the ORS instrument sequentially scans over
three PDCs. Only one beam is focused at ground level, while the other two are elevated on a
vertical structure. Pollutant data are collected over time as the measurement equipment cycles
between each PDC.2
-------
ORS Handbook
Section 3.0
Page 3-7
PDCs
d
Mean Wi nd Direction
Fugitive Source /
Area of Interest
PI-ORS Instrument
X
Figure 3-2. Vertical RPM setup
Figure 3-2 illustrates a vertical mapping configuration.4 Once the PIC for all beam paths are averaged
for the gas species of interest, the VRPM calculations reconstruct a plume map in the vertical
downwind plane. A two-phase smooth basis function minimization (SBFM) approach is applied
when there are three or more beams focused along the ground level (5-beam or more
configurations). In the two-phase SBFM approach, a one-dimensional SBFM (1D-SBFM)
reconstruction procedure is applied to reconstruct the smoothed ground level and crosswind
concentration profile. The reconstruction is applied to the ground level segmented beam paths of
the same beam geometry to find the crosswind concentration profile. A univariate Gaussian
function is fitted to measured PIC ground level values.2 The 1D-SBFM is also the sum of the squared
errors (SSE), which is also the error function for the minimization procedure. The equation for
calculating SSE is:
-------
ORS Handbook
Section 3.0
Page 3-8
2
Where:
B = area under the one-dimensional Gaussian distribution
n = pathlength of the ith beam
my = mean (peak location)
oy = standard deviation of the jth Gaussian function
PIQ = measured PIC value of the ith path
The SSE function is minimized using the simplex minimization procedure to solve for the unknown
parameters. When there are more than three beams at the ground level, two Gaussian functions
are fitted to retrieve skewed and sometimes bi-modal concentration profiles. This is the reason for
the index j in equation above.2
Once the 1D-SBFM phase is completed, the 2-D phase is applied. The bivariate Gaussian function
used in the second phase is:
I1C
exp
ฆ
CF '
Where:
0^ ^ is the standard deviation along the crosswind direction found in the 1D-SBFM;
rriy ^ is the peak location along the crosswind direction found in the 1D-SBFM procedure, and
oz are the unknown parameters to be retrieved in the second phase of the fitting procedure.
-------
ORS Handbook
Section 3.0
Page 3-9
To solve for the unknowns an error function (SSE) is used which is minimized using the simplex
method to solve for the two unknowns. If measurement equipment uses the three-beam setup (one
at the ground level and the other two elevated), the one-dimensional phase calculation can be
skipped, assuming a wide plume. The standard deviation in the crosswind direction is assumed to
be about 10 times that of the ground-level beam path (length of vertical plane). Thus, if r^ is the
length of the vertical plane, to determine the vertical gradient in concentration we use:
L-.i.
i , s
1 ,
> ttlr1"
i v i
When the parameters of the function are found for a specific run, the VRPM algorithm calculates
the concentration values for every square unit in a vertical plane (pixels). Then, the algorithm
integrates the values incorporating wind speed data at each height to calculate the flux.
One-dimensional RPM algorithm
For 1 dimensional (1-D) plume mapping, the scanning ORS instrument and three or more PDC are
placed in a crosswind direction along a line, such as an industrial site fence line, and PIC
measurements are made. A minimum of three PDCs are needed, but four to six are recommended,
as shown in Figure 3-3, to provide a more detailed concentration profile. PDCs should be placed on
the line-of-sight with an equal distance between each subsequent PDC, if possible. The 1D-RPM
configuration uses the same equations for the VRPM, 1D-SBFM, which reconstructs a mass-
equivalent plume concentration profile along a line-of-sight measurement.
-------
ORS Handbook
Section 3.0
Page 3-10
200 m
Concentration Profile
Instrument
300 m
500 m
PI-
Instrument
Figure 3-3. One-dimensional RPM setup
The RPM model and ORS technique are coupled together by a series of steps. First, the ORS
pollutant concentration data along with wind vector information are processed with the VRPM
algorithm to yield a mass emission flux for the source.2'5'6'7 In a similar way, HRPM and 1D-RPM
algorithms are processed with the concentration data to provide hot spots info or concentration
downwind for the source. The output of the concentration data and algorithm process looks like a
contour map. Figure 3-5 displays VRPM and HRPM contour map outputs where the concentration
patterns are evidence of the distribution of the fugitive emission.
-------
ORS Handbook
Section 3.0
Page 3-11
a.
22
20
18
Concentrations are in ppm
b.
40
20
30
50
70
60
10
50 100 150 200 250
Crosswind Distance (m)
0 10 20 30 40 50 60
Concentrations are in ppm
Figure 3-4. Examples of RPM algorithm outputs. Panel a. corresponds to the VRPM output and panel b.
is HRPM output.
RPM-ORS Technologies
Technologies appropriate for characterizing ground-level area sources and non-point emission
sources such as landfills, lagoons, and industrial complexes3 using RPM methodologies are: OP-
FTIR, open-path Tunable Diode Laser Absorption Spectroscopy (OP-TDLAS), UV-DOAS, and DIAL.
Each technology has its own strengths and limitations and, depending on the objective of the
project, some are more effective than others. The following is a discussion of the conditions and
requirements to deploy each technology.
OP-FTIR has an optical range of 100 - 500 m; it can detect multiple compounds simultaneously at
high temporal resolution with detection limits in the ppb range. The instrument setup is time-
consuming and it requires liquid nitrogen to cool the instrument, so OP-FTIR is best for campaigns
that do not require constant relocation and multiple setups. Another consideration is that CO^ and
water are interfering species in FTIR measurements. OP-FTIR data must be processed to quantify
path-integrated concentrations, so if real-time is needed, OP-FTIR is not the best choice. For more
details about OP-FTIR technology see section 2.1.
OP-TDL has an optical range of up to 1 km. Depending on the topography and location of physical
barriers at the survey area, the distance between the control box and the telescopes may require a
-------
ORS Handbook
Section 3.0
Page 3-12
large amount of fiber optic cable, which can be difficult to deploy2. OP-TDL can detect CO, CO^,
NO , ammonia, methane (CH ) and hydrogen sulfide with detection limits in the ppb range, but can
x 4
only detect one compound of interest at a time. The instrument can produce multiple beam paths,
is lightweight, and is easily deployed. The OP-TDL generates real-time path-averaged concentration
data in the field. When only a single gas is of interest, OP-TDL offers a cost-effective choice
compared to OP-FTIR. OP-TDLAS has been used to monitor the exhaust from natural and
mechanical ventilation systems used in houses, farms and other facilities. The technique has also
been used to measure the flux from a traveling gun sprayer applying swine lagoon liquid to the farm
field.4 For more details about OP-TDLAS technology, see section 2.2.
UV-DOAS detects unstable species like radicals, nitrous acid, aromatic species, and BTX at low
concentrations in the ppb levels. The UV-DOAS can be setup to scan multiple or single beam paths.
For more details about UV-DOAS technology, see section 2.3.
Verification/Validation Studies
The RPM algorithm's capacity to locate/identify sources of fugitive emissions and provide accurate
measurement of emissions flux of fugitive emissions and area sources has been assessed in two
different ways: (1) measurement of known concentration tracer gas releases and (2) comparison
with the measurement results of selected instruments in collocated systems. RPM data is verified
by assessing if the collected data satisfies the objectives of the study.1 The following are several
examples of studies performed to test RPM and the selected ORS instruments to survey fugitive
emissions.
OP-FTIR and OP-TDLAS comparison studies
During the Fort Collins measurement campaigns, methane measurements from two OP-FTIR
instruments were compared.7 Both OP-FTIR instruments contain a Nicolet bench, 12-inch
telescope, and collected data at resolutions of 0.125 cm"1, 0.25 cm"1, 0.5 cm"1, 1 cm-1, 2 cm"1, 4 cm1,
and 8 cm1. The data comparison showed that the instruments were extremely stable and reliable
-------
ORS Handbook
Section 3.0
Page 3-13
for the duration of the campaigns. In separate studies, investigators used OP-TDLAS andOP-FTIRto
compare methane measurements obtained by both instruments and found similar results.6,7 During
the experiments, the two instruments were deployed side-by-side and aligned to an identical
mirror. Methane concentration data were collected with each instrument for a period of 30
minutes.7 The results of the experiment found that methane concentrations measured with the
OP-TDLAS were slightly higher (3 percent) than concentrations measured with the OP-FTIR
instrument. These results are significant because they show that methane concentration data
collected by the two instruments are comparable, and that both can be used interchangeably in
RPM configurations-
OP-FTIR and UV-DOAS comparison: Colorado Springs field studv3
The Colorado Springs field campaign occurred in September 2003 at a former landfill site as part of
an effort to rehabilitate the site for recreational use. The current owners of the landfill and the
State of Colorado requested assistance from the EPA to perform a site assessment to search for the
presence of any fugitive gas emissions from the site. The study used OP-FTIR, OP-TDLAS, and UV-
DOAS instruments. The UV-DOAS instrument was deployed at the site to collect data concurrently
with the OP-FTIR instrument. The UV-DOAS detected the presence of BTX. The concentrations of
toluene measured with the UV-DOAS instrument correlated well with gasoline concentrations
measured with the OP-FTIR instrument during the same period.
VRPM plume capture validation study
During June and July 2006 at Orange County Municipal Landfill, the EPA and ARCADIS performed a
VRPM plume capture validation study. The objective was to capture the emissions from hot spots
located a large distance upwind of the measurement configuration. The experimental design used
OP-FTIR for VRPM measurements and a known concentration of tracer gas released to determine
plume capture. The effectiveness of the RPM configuration in capturing plumes (horizontal and
vertical planes) was evaluated by comparing the actual release rate of the tracer gas to the
-------
ORS Handbook
Section 3.0
Page 3-14
calculated flux values as tracer gases were released at different distances upwind of the
configuration. Releases were made at different distances; 20, 60,100, and 140 m from the VRPM
measurement plane. The study found that if there is no statistical significant difference between
the averaged concentrations along each beam (i.e., no vertical concentration gradient), the VRPM
configuration is not vertically capturing the plume. If the difference between the average
concentrations along each beam is less than 10 percent (i.e., a slight vertical concentration gradient
exists), the VRPM configuration is sufficiently capturing the plume from the upwind release point.
If the difference between the average concentrations along each beam is greater than 10 percent
(i.e. a substantial vertical concentration exists), the VRPM configuration is vertically capturing the
plume from the upwind release point and the releasing location is not close enough to the
maximum upwind location for complete plume capture.
Typical QA/QC
This section describes the QA/QC activities that pertain to the RPM as described above. QA/QC
activities normally depend on pre-determined data quality indicators that address the project
unique objectives. The technologies used for RPM have their own specific QA/QC associated with
the instruments. If interested in the technology QA/QC for OP-FTIR, refer to section 2.1; for OP-
TDLAS, refer to section 2.2; and for UV-DOAS, refer to section 2.3.
The general QA/QC steps are: (1) equipment calibration, (2) assessment of DQI goals and (3) DQI
check for analyte path-integrated concentration measurement. Each ORS instrument has its own
calibration procedure as discussed in Chapter 2 of this Handbook, thus is important to follow the
instrument's manufacturer instructions. DQI goals depend on the compound of interest and the
expected concentration ranges, thus detection limits, accuracy, and precision need to be
determined using appropriate traceable standards. On-site verification using a known
concentration tracer gas release provides a sampling episode specific QC confirmation of test
results.
-------
ORS Handbook
Section 3.0
Page 3-15
Because meteorological data is part of RPM calculations, instruments used to measure ambient
conditions need to be calibrated and their accuracy and precision tested regularly. Table 3-1 shows
the recommended DQIs for the different aspects associated with the measurement of the path-
integrated concentration and RPM.2
Table 3-1. Data qua
ity indicators for the QA/QC process, taken from EPA-OTMIO
Measurement parameter
Analysis Method
DQI
PI-ORS Instrument
Instrument specific
Instrument specific
Wind speed
Side-by-side comparison of two wind
monitors
Within 20%
Wind direction
Comparison to magnetic north
Within 10%
Optical path-length
Measure and compare to known
path
Within 2%
Beam angle
Measure and compare known angle
Within 25
HRPM, VRPM and 1D-RPM
CCF*
>0.8
VRPM flux measurement
Wind direction
-10- to +255 from
perpendicular
1D-RPM
Peak location variability
Reconstructed peak
locations
Concordance correlation factor (CCF) indicates the goodness of fit between measured and predicted path-
integrated concentration, CCF = rA.
Siting Concerns
Certain weather conditions such as rain, fog or snow can obscure the optical beam of the utilized
instrument and affect its ability to continuously measure gaseous concentrations. Transient, but
significant, obscuration can occur during heavy precipitation events, particularly with longer path
measurements. This limits the sensitivity of the PIC measurements or the instrument's ability to
collect data.2
Wind conditions can greatly affect the results of field measurements and should be taken into
account when interpreting data. Calm wind conditions do not affect the HRPM methodology
algorithm for hot spot source location. However, very low wind speeds are not ideal for the VRPM
methodology for emission rate estimation, as the source plume may not be carried through the
vertical plane in the absence of measurable wind. Very high wind speed conditions are not ideal
-------
ORS Handbook
Section 3.0
Page 3-16
for any of the RPM methodologies. High winds may displace or vibrate the optical alignment of the
components of the ORS system used in the setup, and affect the quality of the PIC values acquired
in multiple beam paths. They may also cause displacement of any hot spot identified by HRPM.
Based on controlled studies performed in the past, the following wind speed ranges are
recommended for optimal results:
HRPM methodology: Near 0 to 5 m/s
VRPM methodology: 1 to 8 m/s
1D-RPM methodology: 1 to 8 m/s
In optimal conditions, the prevailing wind direction should be as close as possible to perpendicular
to the VRPM measurement plane. The wind direction needs to be determined for each field study
measurement configuration. These requirements present a challenge when determining sites and
setup locations. HRPM data should be collected for at least one hour in ideal conditions, which can
be difficult when considering locations with highly variable conditions.
Strengths and Limitations
Strengths of RPM-ORS are the ability to measure high time resolution and spatially distributed
emission data, directly calculate emission rates, capture the distribution of all major emissions in an
area and isolate emissions from specific measurement areas. Depending on the ORS instrument, it
can provide real-time PIC data for multiple compounds simultaneously. RPM is limited because it
relies on good wind conditions, it has difficulties characterizing emissions from complex terrain, and
has larger uncertainty when capturing emissions for sources located a large distance upwind of the
VRPM configuration. Depending on the RPM-ORS instrument used, some ambient compounds
cause interferences. No ORS instrument can measure all the possible compounds of interest, which
may create the need for two systems depending on the experimental design. A summary of these
strengths and limitations is presented in Tables 3-2 and 3-3.
-------
ORS Handbook
Section 3.0
Page 3-17
Table 3-2. Summary Table of the VRPM's Strengths
Feature
Strength
Measurement Capabilities
Measures high time resolution and spatially distributed
emission data.
Directly calculates emission rates.
Characterizes the distribution of all major emissions in a
large area and isolates emissions from specific areas.
May provide real-time PIC data depending on the
technology used.
Table 3-3. Summary Table of the VRPM's Limitations
Feature
Limitation
Meteorological Challenges
Characterization is reliant on optimal wind conditions.
Interferences
Each OP technology used has its own interferences
that must be considered.
Topographical Concerns
Difficulties associated with characterizing a plume
from complex terrain (e.g., a side slope)
Large uncertainty when capturing emissions from
sources a large distance upwind of the VRPM Setup.
References
1. Thoma, E. D., Green, R. B., Hater, G. R., Goldsmith, C. D., Swan, N. D., Chase, M. J.
Hashmonay, R. A. 2009. Development of EPA OTM 10 for Landfill Applications. Journal of
Environmental Engineering, 10.1061/(ASCE)EE.1943-7870.0000157.
2. U.S. EPA. 2006. ORS Protocol, Optical Remote Sensing for Emission Characterization from
Non-Point Sources. June 14.
3. Hashmonay, R. A., R. M. Varma, M. T. Modrak, R. H. Kagann, R. R. Segall, and P. D. Sullivan.
2008. Radial Plume Mapping: A US EPA Test Method for Area and Fugitive Source Emission
Monitoring Using Optical Remote Sensing, Advanced Environmental Monitoring. 2:21 - 36.
4. U.S. EPA. Using Tunable Diode Lasers to Measure Emissions from Animal Housing and Waste
Lagoons, .
-------
ORS Handbook
Section 3.0
Page 3-18
Hashmonay, R.A., and M.G. Yost. 1999. Innovative approach for estimating fugitive gaseous
fluxes using computed tomography and remote optical sensing techniques. J. Air Waste
Manage. Assoc., 49: 966- 972.
Thoma, E.D., R.C. Shores, E.L. Thompson, D.B. Harris, S.A. Thorneloe, R.M. Varma, R.A.
Hashmonay, M.T. Modrak, D.F. Natschke, and H.A. Gamble. 2005. Open-Path Tunable Diode
Laser Absorption Spectroscopy for Acquisition of Fugitive Emission Flux Data; Journal of Air
and Waste Management Association. 55: 658-668.
Modrak, M. T.; Hashmonay, R. A.; Varma R.; Kagann, R. 2005. Evaluation of Fugitive
Emissions at a Brownfield Landfill in Ft. Collins, Colorado Using Ground-Based Optical
Remote Sensing Technology. EPA-600/R- 05/042. U.S. Environmental Protection Agency,
Research and Development, Work Assignment No. 0-025. March.
-------
ORS Handbook
Section 3.0
Page 3-19
3.2 Range Resolved Measurements using Differential Absorption LIDAR
The term "range resolved," refers to vertical and horizontal profiles of concentrations for
compounds of interest coupled with meteorological parameters. Generally, range resolved
measurements are performed to study emission rates and fate and transport of the compounds of
interest. LIDAR technology is often paired with the DIAL application to measure range resolved
concentrations of trace species in the atmosphere. DIAL has been used to monitor pollution
species in the lower atmosphere such as water vapor, NO, 0^, SO^ and CH^.1 DIAL can also be
used to measure 0^ concentrations in the middle and high troposphere.1 Atmospheric
temperature measurements are possible by the DIAL technique if the absorption line selected is
temperature-dependent.1
General Description of Approach
Range resolved measurements of plume flux typically employs some technology to measure a
surrogate gas in the plume to estimate the concentration of compounds of interest. DIAL is a dual-
wavelength, elastic (the atom absorbs the photon and instantly emits another photon at the same
frequency), backscatter LIDAR that transmits one wavelength at the absorption line of the target
compound (X ) and one wavelength slightly off-line of the target compound to measure
on
backscatter (X ). The on-line wavelength is absorbed by the gas of interest, while the off-line
off
wavelength is not absorbed as shown in Figure 3-5.2
DIAL
Figure 3-5. Conceptual picture on the operation of DIAL.
-------
ORS Handbook
Section 3.0
Page 3-20
The differential absorption between the two wavelengths is a measure of the concentration of the
gas as a function of range.2 DIAL can provide a 2-D measure or "contour" of concentrations across a
scanning plane. By combining this concentration contour with separately obtained wind speeds, a
contaminant flux can be calculated for the measured compounds, see Figure 3-6.3
Range from DIAL Facility (m)
Figure 3-6. Contour profile ofS02 concentration measured 2.1 km downwind of source, at Cement Works,
by Environmental Measurements Group National Physical Laboratory, UK.
Basic DIAL algorithm to calculate backscatter
The number of photons backscattered correlates to the compound concentration. The basic
equation to calculate number of photons backscattered per unit solid angle due to scattering of
type / (forms of radiation like light, sound) is:
-------
ORS Handbook
Section 3.0
Page 3-21
fR 2
h) t a(r,Ai)oln(Ai)Nl(r)dr
Where:
Xi = wavelength
it = transmission coefficient of the LIDAR transmitter optics
r = range interval
t0= optical transmission of the atmosphere
o1 n = backscatter cross section at the laser wavelength
N1 (r) = number density of scattering centers at range rl.
However, to calculate the number of photons incident on the collecting optic of the LIDAR due to
scattering of type /', one must consider the area of the collecting optic (A):
Where:
Xs = wavelength of the scattered light
^(r) = overlap factor
There are two typical types of detectors: photomultiplier and analog. When using photomultiplier
detectors, the number of photons detected is:
-------
ORS Handbook
Section 3.0
Page 3-22
Where:
it = transmission coefficient of the reception optics
Q = quantum efficiency of the photomultiplier
Quantum efficiency refers to the percentage of photons incident (hitting) the receiver (photo
reactive surface) and it measures the LIDAR electrical sensitivity to light. When using analog
detectors, the equations replace the quantum efficiency of the photomultiplier by the gain of the
photomultiplier (G(A )) combined with the gain of any amplifiers used. After some approximations,
S
the analog detector version of equation is:
PTt{Xl)ATt{Xs)Q{Xs)Ta{R,Xl)Ta{R,Xs)^{R)aln{Xl)Ni{R)SR
Where:
R = range of the center of the scattering volume
Verification/Validation Studies
This section presents studies designed to validate the DIAL technique under various conditions.
Some studies aim to verify that the applications can provide accurate results, thus most of these
studies will have other technologies to compare measured concentrations and emission rates.
Verification of DIAL for Gas Species Measurements
DIAL has been validated in European studies5'6 for hydrocarbon emissions with calculated results
ranging from ฑ3 to ฑ12 percent of the actual value.3 Two validation studies were performed in
Alberta with measured fluxes agreeing within +1 to -10 percent of the known source.3 Over a four-
week period during May-June 2003 in Alberta, Canada, DIAL surveys were performed at four gas
processing plants, one gas well test site located in the foothills, and two solution gas flare sites.
The objective of this project was to field test DIAL technology as a means to:
monitor ambient SO^ concentrations in the vicinity of sour gas well test flares and track
-------
ORS Handbook
Section 3.0
Page 3-23
the SO^ plume position,
measure the combustion efficiency of well test and solution gas flares,
measure fugitive emissions of methane and other hydrocarbons from gas processing
facilities.6
The DIAL measurements were performed by Spectrasyne Ltd., UK. When measuring plume
concentration profiles, Spectrasyne generally located the DIAL equipment at least 50 meters from
the area of interest and relative to the plume source and wind direction so scans could be taken
roughly at right angles to the direction of plume travel.6 Often meteorological changes make this
impossible and the measurements were taken at an oblique angle, which results in profiles that
appear stretched in the horizontal direction.4
The DIAL system included two DIAL lasers, one emitting in the IR range and one in the UV range,
and a self-contained weather station for measuring wind speed and temperature. Meteorological
parameter data is used in mass rate calculations to reprocess the data to account for the angle
relative to the plume direction.4 With this system, total contaminant flux can be calculated and
portions of the plume assigned to specific sources 4
The accuracy of the DIAL system was checked by comparing SO^ mass emissions calculated from
the DIAL measurements of the SO^ plume with SO^ mass emissions calculated from gas plant CEM
instrumentation installed in the incinerator stack 4 Direct comparison between DIAL and fenceline
point source concentration measurements is not possible, since the DIAL measures gas
concentration in a relatively large volume in the atmosphere against a point sampling type
instrument such as a gas chromatograph 4 However, the combination of point source
measurements for a target compound, combined with meteorological data and dispersion
modeling, can provide comparisons useful to verify the DIAL results. Table 3-4 shows some of the
results, the scan time, and calculated fluxes for the DIAL and point source. The DIAL and point
measurements showed a difference of 11 percent, which is within the range of-18 to + 5 percent
-------
ORS Handbook
Section 3.0
Page 3-24
from other calibration studies performed by Spectrasyne Ltd.6 According to this Spectrasyne study
and others, DIAL plume measurements generally underestimate the total mass because some
areas of the plume contain compound concentration below DIAL'S detection limit and are not
included in the plume mass. Additional variability among DIAL measurements is believed to
originate from variation in wind speed and direction, combined with the 4-20 minutes required to
do a full DIAL scan across the plume.
Table 3-4. Results from the comparison of DIAL and plant measurements of SQ2 mass emissions
SCAN NUMBER
SCAN TIME
WIND SPEED
WIND DIRECTION
S02 FLUX
(M/S)
(DEC)
(KG/HR)
1
12:37-12:48
5.2
331
372
2
13:00-13:21
2.2
350
223
3
13:31-13:51
4.3
346
394
4
14:00-14:04
5.6
359
196
5
14:07-14:19
3.9
359
145
6
14:42-14:51
4.0
350
333
7
15:10-15:21
3.9
0
394
Time
Weighted
304
Mean of
DIAL
Plant Data
340
Note: collected on May 26 about 190 m downwind of incinerator stack. Adapted from Chambers,
2003.
Typical QA/QC
This section describes the QA/QC steps that pertain to the applications described above. QA/QC
steps for applications normally depend on pre-determined data quality indicators that address the
project unique objectives. The technology used for the applications presented above have their
own QA/QC associated to specifics of the instruments. If interested in the technology QA/QC for
DIAL, refer to section 2.4.
DIAL measurements are typically verified by running two collocated DIAL systems or one DIAL
system along with another ORS instrument, like FTIR or CRDS. Arcadis has verified DIAL
measurements using OTMIO and developed QA/QC information for conducting DIAL measurement
projects.7 However, as noted in OTMIO, the unique setup of DIAL requires project specific QA/QC
steps with data quality indicators that meet study objectives.
-------
ORS Handbook
Section 3.0
Page 3-25
Siting Concerns
The DIAL equipment, optical housing, electronics, computer equipment and other components
require a climate control enclosure such as a trailer or an aircraft. Therefore, the operating
temperature is controlled to human comfort level ~ 22ฐC. Trailers require relatively flat surfaces
and road access. Changing weather conditions, physical interferences like buildings, trees, traffic
and changing terrain, and interfering chemical species at certain wavelengths will increase
variability in the measurement and possibly result in less accurate results.
Strengths and Limitations
The most significant limitation to DIAL application is the cost and limited availability of
measurement systems. Multiple measurements in North America have relied on importing the
instrumentation from the United Kingdom.4'5'6 Additionally, the number of chemical species
measurable by DIAL is restricted to those that are detectable by the Lidar technology. The most
notable strength of a DIAL system is the ability to quickly resolve pollutant concentrations in two
dimensions. Concentration gradient data obtained in short periods of time enables DIAL to be
deployed in many applications and a number of configurations. Tables 3-3 and 3-4 summarize
these strengths and limitations.
-------
ORS Handbook
Section 3.0
Page 3-26
Table 3-5. DIAL Strengths
Feature
DIAL Strengths
Measurement Capabilities
DIAL provides spatially resolved pollutant
concentration in two dimensions
Measurements are provided in a relatively short
period of time.
Flexibility
DIAL is deployable in many different applications and
configurations.
High intensity light source
The ability to measure longer path lengths (1 to 3 km)
Table 3-6. DIAL Limitations
Feature
Dial Limitations
Limited Availability and Expense
Due to limited availability, DIAL systems used in North
America are typically imported, which increases the expense
of using DIAL for measurements.
Range of Measurement
Chemical species that can be characterized and limited to
those compounds with the unique chemical properties
required to be detected by the LIDAR instrument. Only a
few wavelengths are measured; spectral artifacts cannot be
fixed or investigated
References
1. Argall, P. S. and R. J. Sica. 2002. LIDAR in the Encyclopedia of Imaging Science and
Technology. J.P. Hornak John Wiley & Sons Inc., New York. January.
2. Argall, P. S. and R. J. Sica. 2003. LIDAR in The Optics Encyclopedia. Wiley-VCH, New York,
NY.
3. Robinson, R. 2006. The Application of DIAL for industrial emissions monitoring. Presentation
at the VOC Fugitive Losses: New Monitors, Emission Losses, and Potential Policy Gaps, 2006
International Workshop. Office of Air Quality Planning and Standards, Research Triangle
Park and Office of Solid Waste and Emergency Response, Washington, DC. October 25 - 27.
4. Chambers, A. 2003. Well Test Flare Plume Monitoring Phase II: DIAL Testing in Alberta.
Alberta Research Council Inc. Prepared for Canadian Association of Petroleum Producers.
-------
ORS Handbook
Section 3.0
Page 3-27
.
5. Chambers, A., M. Strosher, T. Wootton, J. Moncrieff. and P. McCready. 2008. Direct
measurement of fugitive emissions of hydrocarbons from a refinery. Journal of the Air and
Waste Management Association. 58:1047-1056. doi: 10.3155/1047-3289.58.8.1047.
6. Chambers, A. K., M. Strosher, T. Wootton, J. Moncrieff, and P. McCready. 2006. DIAL
Measurements of Fugitive Emissions from Natural Gas Plants and the Comparison with
Emission Factor Estimates.
7. .
-------
ORS Handbook
Section 3.0
Page 3-28
3.3 Solar Occupation Flux Measurement
Characterizing and quantitatively measuring fugitive VOC emissions from non-point sources are
challenging. The ability to accurately calculate the flux rate of VOC emissions from large area
sources such as landfills, refineries, waste retention ponds, process areas, and product holding
tanks is highly pursued by government and industry alike. The development and advancement of
optical measurement technologies have increased environmental VOC monitoring capabilities and
are applied in many new monitoring methods. SOF is a method where optical spectroscopic
technologies are used to directly speciate and quantify the chemicals present in a gaseous emission
plume using the sun as a light source.
Because the SOF method uses the sun as a broadband light source, the target compound
possibilities are limited only by the detection equipment and interferences. Depending on the
spectrometer used, the SOF can detect many different gaseous species, even at the same time,
including: ammonia, formaldehyde, VOCs, terpenes, vinyl chloride, CO, ethylene, ethylene oxide,
hydrofluoric acid, (HF) hydrochloric acid, HCI, CH4, SO2,propane, propylene, and hydrocarbons up to
C15.1'2 Due to the ease of mobility with the SOF method, the technique can be used in applications
such as total CO columns of megacities, petrochemical industries, agriculture, refineries, ships, and
volcanoes.3
Volcanic emissions research is a rich source of information on monitoring techniques to determine
emission rates of fugitive gases in atmospheric plumes including SOF. Several measurement
techniques have been employed to measure volcanic gases such as SO^ (i.e., correlation
spectrometer (COSPEC), DOAS, and DIAL). However, the SOF method was developed to improve
volcanic activity forecasting capabilities because multiple gas species can be detected using passive
FTIR, and direct flux measurements can be made based on movement of the instrument view
through the emissions plume.4
Environmental applications of the SOF technique have previously focused on measuring fugitive
VOCs from oil refineries and industry processes.5 VOC gases emitted from these source types
-------
ORS Handbook
Section 3.0
Page 3-29
mostly consist of alkanes, alkenes, and some aromatic compounds.5 These fugitive VOCs have
historically been measured using the DIAL method.6 The DIAL method, described in Chapter 3.2 of
this Handbook, employs short laser pulses directed through the gas plume at different wavelengths
to calculate mass flux measurements by multiplying the resulting concentration integrated over the
plume cross section at different angles by the wind speed.7 However, DIAL is rather complex and
expensive relative to SOF and is not ideal for periodic regulatory monitoring.7 In comparison to
DIAL, the SOF technique uses solar broadband IR or UV/visible spectral radiation as the light source
instead of a laser source, making the SOF method potentially more cost-effective, faster than DIAL,
and easier to automate 7 From the solar spectra, it is possible to retrieve the PIC (molecules/cm2)
of VOCs between the sun and the spectrometer.8 Multiplying this PIC by the local wind speed
results in the mass-flux of target VOCs through the area.
Agricultural application studies have established that the SOF method has a detection limit for
measured hydrocarbons of 0.3 mg/ meter2'3. Similarly, in applications pertaining to refineries and
leak detection, studies conclude that a point source emission of measured hydrocarbons at 0.5 kg/
hour can be measured 50 meters downwind with a precision of three percent and an accuracy of
30 percent.1 For simple sources and several traverses ,it has been found that the accuracy can be
better than 10 percent. Under more complicated conditions, with emissions occurring from a
complex structure with an unknown plume lift, larger systematic errors will occur, primarily due to
uncertainties in assessing the plume lift and the associated wind field. Uncertainty in the wind
speed in more complex terrain and wind stability can be 15 - 30 percent. Consequently, the
accuracy in the SOF method depends on the amount of error associated with the ancillary
meteorological measurements which can dominate the uncertainty in the emissions results.1,3
General Description of Approach
There are three main components to the SOF system: an FTIR spectrometer to capture solar
radiation spectra, a sun tracker to maintain instrumental orientation to the solar zenith, and a GPS
for accurate measurement location relative to the gas plume.5 Two different sun tracker
configurations are shown in Figure 3-7, and a rough schematic of the entire SOF system is provided
in Figure 3-81.
-------
ORS Handbook
Section 3.0
Page 3-30
Figure 3-7. Solar tracker configurations
Note: The left panel shows the sun tracker and mirrors extending out the top of the vehicle to maintain
orientation9 while the right panel shows the path of the sunlight (in yellow) first striking the mirror of the
solar tracker before being directed into the spectrometer5
Sun
Meteorology mast
Solar tracker
GPS
Spectrometer
Radio
Radio
Computer
Figure 3-8. Rough overview of a mobile SO F system.
Flux measurements are difficult to measure in an atmospheric plume using stationary
instrumentation. Figure 3-9 illustrates the capture of a complete plume transect for better
constrained source flux calculations using the SOF technique. In the illustration of Figure 3-99, the
solar tracker maintains the solar zenith, reflecting the solar light into the spectrometer regardless
of the vehicle position as it traverses the width of the plume column1,9 As previous methods
-------
ORS Handbook
Section 3.0
Page 3-31
combined stationary plume measurements with dispersion modeling or tracer gas ratios, the
mobile aspect of the SOF technique allows for direct measurements of gaseous flux emissions if the
full plume column is captured.3
Figure 3-9. Solar Occultation Flux method. The instrument is placed in a vehicle which travels across the
gas plume to capture the plume cross-section (Illustration Karin Sjoberg).
Each gaseous compound absorbs energy at different wavelengths, usually more than one,
depending on vibrational and rotational excitement within the molecule. Fundamentally, this
measurement is a passive form of IR or UV spectroscopy. Therefore, each compound has its own
"signature" of bands from which energy may be absorbed. Each band is highly selective, with
virtually no absorption occurring outside of a specific wavelength. When molecules intercept the
solar radiation before it reaches the detector, the molecular absorption is an extinction (or
occultation) of the solar radiation intensity at the signature wavelengths. Once a compound has
been identified, its spectrum can also be used to measure the compound's concentration because
the amount of radiation absorbed from the solar ray is proportional to the concentration of the
compound in the sample or open path. According to the Beer-Lambert law, there is a linear
relationship between absorbance and concentration as shown in Equation below1:
A = s*c*l
Where:
-------
ORS Handbook
Section 3.0
Page 3-32
s = absorption coefficient
c = sample concentration
I = sample path length
SOF measurements acquired from within the plume column are divided by a reference
(background) spectrum recorded outside the plume. This prevents any background sources such
as the atmosphere, inherent structures of the sun, and instrument functionalityfrom interfering
with accurate measurements.5 Multiple species are simultaneously evaluated for each spectral
acquisition using non-linear, least-squares fit routines with published reference spectra from
HITRAN 2000, Pacific Northwest National Laboratory, National Institute of Standards and
Technology, and Hanst library databases.7 The resulting spectra are evaluated for gas species
absorption intensity to determine concentration using Equation below:
Im(y) = IL(y) * exp | -v4 * aR * LR aMie * LMie - ^
-------
ORS Handbook
Section 3.0
Page 3-33
The PIC is determined by evaluating spectrum measurements individually along a path of
continuous analysis to derive the line-integrated concentration represented by each spectrum. GPS
measurements taken at the beginning and end of each measured spectrum determine the surface
length represented by the spectrum. This value is then multiplied by the line-integrated
concentration and summed over the total plume transect to calculate the PIC. An example of this
process is shown in Figure 3-11.8 Multiplying this PIC by the local wind speed results in the mass-
flux of target VOCs through the measurement plane.
1/ nzittirr es
Ryahainne
Figure 3-10. An example of path-integrated calculations determined from SOF measurements.
Note; The red lines indicate individual spectra, the area between the red lines correspond to the surface-
8
integrated concentration, and the green lines illustrate the wind direction vector-
As shown in Figures 3-9 and 3-10, SOF mobile measurements are made crosswind and near
downwind (about 0.5 to 3 km) from the target source. The total cross-sectional mass of key species
is obtained by summation of ail the species measurements over the plume traverse.10 The process
of determining the emissions flux is summarized mathematically by the Equation below. The
assumption that the wind speed is equivalent to the gas plume velocity is necessary for both the
DIAL and SOF techniques as this value is required to calculate the mass flux by multiplying the
plume velocity by the cross-sectional integrated gas concentration5 Therefore, both methods will
-------
ORS Handbook
Section 3.0
Page 3-34
be susceptible to the same measurement error associated with the wind speed parameter.
Emission ~ j*cohntm\ \ > it dx
3t,
Where:
Column(x) = the total column at distance x across the plume
if
= average wind speed for plume at plume height
Verification/Validation Studies
Because DIAL measurements are a standard method of monitoring VOC emissions at refineries in
Europe, performed a comparison between the results of the two different methods during a 2001
SOF field study. 5 SOF measurements by this group were made at the Preem refinery in Goteborg,
Sweden for four days starting August 1, 2001. DIAL emissions measurements from 1995 and 1999
were recalculated with the annual average wind speed to compare with the 2001 SOF results; this
comparison is shown in Table 3-7.1
Table 3-7. SOF technique VOC emissions compared to DIAL
'""Sea SOT^oofKauij DXAFI999 DIAL1995
2001
Crude oil tanks 56+16 (8) 56 (wind normalised 4 rn/s) 62 (wind normalised 4 m/s)
Process plant E 54 ฑ19 (7) 52 56
Process plant 20 ฑ5 (5) 35 7,5
V
Water 19 ฑ3 (3) 25 11
treatment
Although comparing environmental data that is up to six years apart is not recommended, the data
in Table 3-7 shows that the emissions measured using SOF are generally consistent with previous
DIAL results.1 Regardless of the amount of measurement error imposed on the SOF result by wind
speed approximations, all SOF measurements either match a previous DIAL result exactly or are
very near to the average of the 1995 and 1999 DIAL results.
Annual VOC emissions from the Port of Goteborg were reported to be 1100 tons in 1999 and 2300
tons in 1995 by Shell Global Solutions and Spectrasyne, respectively. Researchers used an SOF
system mounted on a ship to calculate an annual flux of 1770 tons per year based on two
-------
ORS Handbook
Section 3.0
Page 3-35
measurementsthat were not temperature correctedperformed on the same day in August
2001.5 Again, more than two measurements on a variety of days distributed throughout the year
would yield a more accurate snapshot of annual emissions. Nonetheless, a result of 1770 tons per
year is almost the exact equivalent to the average of the reported 1999 and 1995 values and is a
satisfactory result, validating the application of SOF to measure oil refinery gas emissions flux.
The results from two tracer gas correlation studies plus other validation studies were discussed by
researchers3'8 During these studies, sulfur hexafluoride (SF ) was released in an open field and from
6
the roof of a crude oil tank at the previously mentioned Goteborg refinery in May and June of 2002.
Each experiment released SF at a rate of about 2.0 kg/h. Tables 3-8 and 3-9 display the results
6
from the open field and tank roof tracer experiments, respectively.8 As shown in Table 3-8, tracer
correlation measurements over four days in an open field validated that the SOF method can
accurately retrieve emissions flux measurements within ฑ 20 percent error; whereas crude oil tank
measurements done on a single day shown in Table 3-9 yields emissions flux measurements with
up to 50 percent error.3,8 It is worth noting that the crude oil tank measurements were conducted
in the near field as opposed to a medium or far distance from the emission source in addition to
being conducted on only one day. Figure 3-12 illustrates a combination of both scenarios where a
tracer experiment in an open field resulted in 72 percent agreement (within 20 percent) of SOF
measurements to the actual tracer gas emission, while other measurements can vary up to 50
percent from the actual emission.1 When averaged together, however, the resulting value is within
three percent of the actual emission rate.
-------
ORS Handbook
Section 3.0
Page 3-36
Table 3-8. Summary for measurements on the Aby field, 2002.
Day
Emitted
Calculated
Number
Ave. Wind
Ave. Wind
Error
SF6
Average
of
Speed
Direction
(kg/h)
Accepted
(m/s)
Traverses
May 22
1.92
2.3ฑ1.3
4
4.9-8.6
152ฐ-169ฐ
20%
May 23
1.97
2.2ฑ0.6
15
3.9-5.6
120ฐ-142ฐ
10%
June 03
1.97
1.6ฑ0.9
16
2.7-5.3
235ฐ-273ฐ
-20%
June 04
1.89
2.Oil.4
9
5.9-7.8
152ฐ-191ฐ
5%
Table 3-9. SOF traverse done on day 24-June 2002. True emitted SF6 was determined to be 2.0 kg/hr.
Time
Emission SF6
Ave. Wind Speed (m/s)
Ave. Wind Direction
12:45
3.1
6.5
252ฐ
12:54
1.8
7.2
252ฐ
13:05
1.3
6.0
259ฐ
13:17
2.7
7.5
253ฐ
13:29
3.1
5.4
255ฐ
13:56
5.2
7.4
264ฐ
14:05
3.7
7.4
251ฐ
14:24
2.6
7.3
262ฐ
14:31
34
6.5
260
Average
3.0 ฑ 1.1
The results from these studies indicate that multiple measurements in the mid to far field over
multiple days are more representative of overall flux rates and reduce the amount of total
measurement error. Having enough data points to calculate a statistical average is ideal for
eliminating stochastic variances caused my micrometeorological disturbances.
CD
C
o
in
ts>
*E
o
to
LI-
CO
5 t
4.5
4
3.5
3
2.5
2
1 .5
1
0.5
O
850
Uppmatt emission med SOF
-Verklig emission (vagning av tuber)
ft
900
950
1000 1050
Matt id
1 1 OO
1 1 50
1 200
Figure 3-11. Tracer gas/SOF experiment measuring SF6 emissions in an open field over time of day. SOF
measurements are depicted as red circles while the actual emission release rate is drawn as a grey line.
-------
ORS Handbook
Section 3.0
Page 3-37
Typical QA/QC
To make emissions measurements with the SOF method, the operator needs to have
meteorological information (distributed in height, surface, and time); a road to travel along that is
relatively smooth, downwind of the emissions source and near perpendicular to the wind
direction; stable wind conditions; and an unobstructed view of the sun. In addition to meeting
these requirements, measurements can be further validated by calculating the expected and
observed amounts of measurement error. The total amount of measurement error associated
with SOF results is comprised of statistical and systematic error. Statistical error is described with
a normal distribution as this type of error accounts for the natural, stochastic behavior of
atmospheric variability. Researchers qualified this amount, which estimated the oS/ N (or
instrumental precision) from the standard deviation in the baseline of the plume column
measurements during their VOC emissions study at oil refineries.5 This value was about 0.3 to 0.6
mg/ m2, which corresponded to an uncertainty of 0.3 percent to 6 percent for a total column
measurement of 100 to 10 mg/ m2. They also estimated the relative uncertainty in the total
column due to wind direction variability (o0) to be about 1 percent to 7 percent for eight scans.
Researchers state that there is an 8 percent effect on column measurements when a plume is
traversed at a 90ฐ angle and that this error amount is sensitive to the ability to make
measurements perpendicular to the wind direction such that angles of 80ฐ and 70ฐ contribute
errors of 16 percent and 25 percent respectively.5 Relative uncertainty in the wind speed due to
plume height estimate uncertainties (ouH) are discussed in more detail below.
Researchers calculated the statistical error assuming various typical scenarios encountered during
measurements. The results displayed in Table 3-10 show that even between best and worst-case
scenarios, the statistical error amount varies from 14 percent to 19 percent.5 To determine the
total error for a measurement, these statistical error values need to be added to the amount of
systematic error. Output data recovered from SOF measurements do not contain information
related to the height of the gas plume in the atmosphere. Plume height estimates are inferred
from the corresponding temperature and pressure broadening of the spectral absorption lines if
measured from high-resolution (< 0.125 cm-1) spectra.8 This is the largest identified source of
error in the SOF method, if the entire plume cross-section is captured and that the total amount of
absorption equals the total concentration. As mentioned previously, the integrated concentration
of the target species is multiplied by the mass average wind speed of the plume at plume height to
-------
ORS Handbook
Section 3.0
Page 3-38
determine the emissions flux; therefore, accurate wind speed measurements are crucial to
minimizing total flux measurement error. Increasingly complicated site conditions, such as a
complex emissions source structure with unknown plume lift, will have larger amounts of
systematic error.
One researcher states that measuring emission plumes with heights above 20-30 meters at distances
greater than 500 meters downwind are ideal conditions to ensure minimal systematic errors associated
with the wind field.1 This is because of the increased predictability of the wind height profile that
corresponds to the plume height while making measurements under these criteria. Figure 3-13 below
shows the wind speed profile with height over developed land if a 5 m/s wind velocity was measured at 10
meters. It is clear from this figure that wind speeds above 20 meters from the surface during typical SOF
conditions (i.e. sunny, which implies an unstable atmosphere) are past the major inflection point in the
profile and have a less prominent height gradient with increasing height. Fransson and Mellqvist1 made an
estimation from data acquired at 17 meters that ฑ 5 meters is a realistic amount of error associated with
the plume height estimation which translates to about 12 percent error in the wind speed measurement.5
Further studies into the amount of error in the wind field by height concluded that about 14 percent
systematic error in retrieved flux.
ฆ ; ฆ
ฆGG
Velocity it.'e-|
Figure 3-12. Wind velocity profiles by height.
-------
ORS Handbook
Section 3.0
Page 3-39
Siting Concerns
Since the instrumentation for the SOF method is mobile during analysis, issues originating from the
site location are few. Most notably, the path of solar light through the plume to the
instrumentation needs to be unobstructed, and excessive vibration during the mobile operations
can cause noise originating from the interferometer. Therefore, measurement capabilities are
constrained to weather conditions. In addition to the surrounding area of the mobile path, the line
of sight needs to be clear of trees, shrubbery, and buildings that might impede the view of the sun.
Moreover, the mobile path needs to be smooth and near perpendicular to the wind direction to
minimize the amount of method error. SOF measurements are more representative in the mid and
far field downwind from the emissions source where the plume is well-mixed and the solar transect
of the plume will be higher in the wind profile. Plume traverses must be able to capture the entire
plume cross- section plus open atmosphere on either side of the plume such that a representative
background can be collected.
Strengths and Limitations
One major advantage to the SOF method is also a disadvantage: the solar broadband light source.
Using FTIR, the SOF method can precisely and accurately speciate and quantitate multiple gaseous
emissions simultaneously using one instrument. The caveat to this advantage is that the method
can only do so during specific climatic conditions (high sun and steady winds).4 Other strengths
and limitations associated with the SOF method are presented in Tables 3-11 and 3-12,
respectively.
-------
ORS Handbook
Section 3.0
Page 3-40
Table 3-11. Feature strengths of using the SOF method.
Feature
Strength
Direct Measurement
Increases measurement accuracy by reducing uncertainty.
Passive Light Source
Decreases instrumental complexity for field operations and
reduces amount of scattering errors in the UV.
Broadband Light Source
Multiple species detection over a wide range of
wavelengths.
Better Mobility
More suitable for frequent field application.
Lower technical complexity
Decreased cost and easier field application.
FTIR Detection
Higher specificity and better signal-to-noise (relative to DIAL).
Measurements during Sunny conditions
Corresponds to unstable meteorological conditions where wind
gradients due to convection are smoothed out.
Table 3-12. Feature limitations of using the SOF method.
Feature
Limitation
Interferogram Vibration Sensitivity
System requires vibration reduction platform and a
smooth mobile path.
Wind Speed Error
Calculations based on wind speed measurements inherently
add uncertainty due to the stochastic, uncontrollable, and
highly variable nature of wind speed.
No Plume Height Measurement
Uncertainty of plume height increases measurement error
from wind speed term.
Solar Light Source
Inappropriate to make measurements in the presence of
clouds.
"Open Eye" Detection and Roadway
Path Restriction
Difficulty in separating emissions sources that are close
together.
References
1. Mellqvist, J., J. Samuelsson, C. Rivera, B. Lefer, and M. Patel, (2007), Measurements of
industrial emissions of VOCs, NH3, N02 and S02 in Texas using the Solar Occultation Flux
method and mobile DOA S, Final Report: HA RC Project H-53, retrieved off the World Wide
Web on June 21, 2010 from www.FluxSense.com.
2. Frisch, L., (2006), VOC Fugitive Losses: New Monitors, Emission Losses, and Potential Policy
Gaps, 2006 International Workshop. Office of Air Quality Planning and Standards, Research
-------
ORS Handbook
Section 3.0
Page 3-41
Triangle Park. Office of Solid Waste and Emergency Response, Washington, DC. October 25-
27, 2006
3. Mellqvist, J., (2004), Quantifying fugitive emission of VOCs using the Solar Occultation Flux
technique (SOF), Compliment to Fransson and Mellqvist, 2002, Chalmers University of
Technology, Goteborg, Sweden.
4. Duffell, H., C. Oppenheimer, and M. Burton, (2001), Volcanic gas emission rates measured by
solar occultation spectroscopy, Geophysical Research Letters, 28, 16, 3131- 3134.
5. Fransson, K., and J. Mellqvist, (2002), Measurements of VOCs at Refineries Using the Solar
Occultation Flux Technique, Chalmers University of Technology, Goteborg, Sweden.
6. Kihlman, M., J. Mellqvist, J. Samuelsson, L. Tang, and D. Chen, (2005), Monitoring of VOC
emissions from refineries in Sweden using the Solar Occultation Flux method, retrieved off of
the World Wide Web on October 19, 2010 from www.FluxSense.com.
7. Mellqvist, J., M. Kihlman, J. Samuelsson, and B. Galle, (2005), The Solar Occultation Flux (SOF)
Method, a new technique for the quantification of fugitive emissions of VOCs, paper #1377 in
proceeding of A&WMA's 98th Annual Conference & Exhibition, Minneapolis USA.
8. Kihlman, M, (2005) Application of solar FTIR spectroscopy for quantifying gas emissions,
Licentiate Thesis, Chalmers University of Technology, Goteborg, Sweden.
9. Mellqvist, J., M. Kihlman, B. Galle, K. Fransson, and J. Samuelsson, (2005), The Solar
Occultation Flux (SOF) method: a nouvelle technique for the quantification of fugitive gas
emissions, retrieved off the World Wide Web on March 10, 2010 from Fluxsense.com
10. Mellqvist, J., J. Samuelsson, J. Johansson, C. Rivera, B. Lefer, S. Alvarez, and J. Jolly, (2010),
Measurements of industrial emissions of alkenes in Texas using the solar occultation flux
method, Journal of Geophysical Research, 115.
-------
ORS Handbook
Section 3.0
Page 3-42
3.4 Tracer Gas Correlation
Challenges measuring emissions flux from a fugitive or area source such as a landfill, agricultural
waste, industrial fugitive, waste water or oil and gas production source include spatial, temporal
variability of the emission sources and the uncertainty of the measurement technology. Emissions
source variability includes defined and undefined sources like unknown emissions points,
delocalized emissions sources, the timing of periodic or episodic emissions, and atmospheric,
diurnal, seasonal and process variations in emission flux. Defined fugitive sources cover smaller
areas (i.e., less than 1 square kilometer down to a square meter) allowing emissions points to be
identified for direct measurement. Undefined area sources typically originate from large areas
(i.e., greater than 1 square kilometer). For either defined or undefined emissions sources, the
sources of uncertainty cause emissions flux to be difficult to measure and model. Tracer gas
correlation provides a ground based technique that can be applied to both well-defined area
emissions sources and undefined fugitive sources.
Tracer correlation involves a common practice of measuring pollutant emission rates while
releasing a known concentration of a tracer gas. The subsequent simultaneous measurement of
this tracer gas and the pollutant of interest downwind from the release provide sufficient
information to determine or validate the emissions flux measurements. The release of a known
concentration of a tracer gas assists in tracking the plume and sources or sinks of the pollutant of
interest into the plume. The majority of tracer gas studies use cell-based technologies, like CRDS
and FTIR to measure the tracer gas and the compound of interest. However, some studies are
expanding the use of tracer gas release to open-path techniques to evaluate the distribution over a
large area.
General Description of Approach
The use of a tracer gas is common in projects that aim to study emission fluxes. Normally, emission
fluxes are obtained by calculating fluctuations in the vertical and horizontal component of the wind
(w'), fluctuations in the tracer gas concentration (n') over either time (T) or space (S) and surface
-------
ORS Handbook
Section 3.0
Page 3-43
roughness. The following equations provide a method for determining experimentally the surface
fluxes of tracer gases by measuring meteorological parameters near the surface as well as the
vertical gradient in the concentration of these calculate horizontal fluctuations, one must consider
wind velocity (u(z)) near the surface, friction velocity (uj, which is a measure of the drag exerted
by the wind on the surface, and the displacement height (d). The displacement height results from
the canopy acting as a displaced lower boundary layer and whose value is typically 70-80 percent of
the canopy height.
I IterS)
F = fir'jf7/f
T(orS) J0
One must consider wind velocity (u(z)) near the surface, friction velocity (u*), which is a measure of
the drag exerted by the wind on the surface, and the displacement height (d). The displacement
height results from the canopy acting as a displaced lower boundary layer and whose value is
typically 70-80 percent of the canopy height.
u{z ) = 111
Where:
k is the von Karman constant and is equal to 0.4 and zo is the surface roughness.1
Using the Tracer Correlation Approach to measure total emissions over a large subject area is an
alternative to standard dispersion modeling when attempting surface boundary layer methods as
weather conditions are too difficult to measure or estimate accurately. Thus, releasing a tracer gas
with known concentration and rate of release assists in calculating emission rates. If meteorological
conditions affect both the tracer gas and the analyte in the same way, the analyte emission rate is
calculated from simultaneous measurements of the tracer and analyte gas far downwind from the
-------
plume emission source.
ORS Handbook
Section 3.0
Page 3-44
O.AC
0 _ m
~ n " AC,
Where:
Qm = analyte emission rate,
Qt is the tracer gas release rate,
A Cm is the concentration above the background of the analyte observed in the plume (plume -
background concentration)
ACt is the concentration above background in tracer concentration in the plume, relative to the
background.
A tracer gas is released from a canister to transport in the plume along with the analyte. Usually a
tracer gas is chemically stable with no significant sources or sinks while in transport and is expected
to fully mix in the plume. Typical tracer gas field measurements are performed with cell-based
instruments that utilize specific spectroscopic properties to characterize chemical species like CRDS
and FTIR. These instruments can be setup as stationary or mobile to obtain one-point or multiple-
point samples. Figure 3-13 shows a cartoon of tracer gas release setup. Normally, the tracer gas is
released upwind from the source and the cell-based instrument is downwind from the source to
measure a well-mixed plume with the analyte and tracer gas.
-------
ORS Handbook
Section 3.0
Page 3-45
~~~~~~~
~~~~000
c.
Mixed plume
Detected
gases
^spectra
Cell based instrument
vwnw
Figure 3-13 Tracer gas release setup cartoon.
Note: Panel a shows the source with the analyte gas been released to the ambient air. Panel b shows
the tracer gas been released and pushed by the wind into the analyte gas area to mix in the plume.
Panel c shows the cell-base instrument measuring gases, analyte and tracer.
Verification/Validation Studies
Because the tracer gas release concentration is known and there are no significant sources or sinks,
the mixing ratio should remain constant while in transport. Thus, verification of correct
measurements by the cell-based instrument is done by measuring the known tracer gas
concentration with accuracy. Validation is performed depending on the DQOs of the study.
Normally, the study will say how accurate the retrieval of the known concentration of tracer gas
must be to obtain emission fluxes and the quality of the meteorological parameters to measure.
Following are some examples of studies performed to verify and validate the use of tracer gas to
calculate emission fluxes, utilizing CRDS and FTIR.
Mobile plume tracer dilution measurements were taken from May 18 - 21, 2009, driving on
Interstate 36 and East Main Street in Danville, IN. Acetylene (C^H^) tracer gas was released from
four sites located on or near to the center of each landfill facility. The stationary measurements for
gas tracer correlation measurements, the analyzer was located in the plume sufficiently downwind
of the CH^ source and tracer gas release locations for plumes to be well-mixed and appear as
-------
ORS Handbook
Section 3.0
Page 3-46
a single point emission source. During the measurements, the analyzer is stationary and
continuously measures CH^ and C-,H concentrations. The CRDS Picarro Model G1203 methane
ethylene (C2H2) analyzer used in this study is a self-contained, stand-alone unit that provides
continuous measurements of C,H , and CH, concentrations, ambient temperature, and analyzer
location (via high resolution GPS). The C,H,, tracer gas was released from bottles through a mass
flow controller (Alicat Scientific MC Series 16-Bit Mass Gas Flow Controller, model MCP-50SLPM-
DIO-SG-30PSIA/5m) which was attached to the gas cylinder line to ensure a constant release rate
throughout the study. The range of tracer gas release for the test was 20 L/min to 40 L/min. The
release rates were automatically recorded. The following parameters were measured: horizontal
wind speed, horizontal wind direction, temperature, vertical and lateral turbulence, and net solar
radiation. Results from the study shown in Figure 3-14 demonstrated that the Tracer Correlation
Approach can measure highly correlated, (correlation coefficient of 0.84) tracer gas to CH
emissions over a wide range of atmospheric and dispersion conditions.4
O qO^O
>
O o Ocp'o3
03
s:
o 0
Acetylene (ppbv)
Figure 3-14 Methane concentration verses acetylene concentration Tracer gas characterization using
FTIR: Samuelsson et a!., 2001
FTIR was used to obtain time resolved concentration measurements of methane in the downwind
plume of a landfill and N,,0 was selected as the tracer gas.4 FTIR spectroscopy is an optical
-------
ORS Handbook
Section 3.0
Page 3-47
technique allowing a wide spectral region to be recorded simultaneously, thus the detection of CH^
obtained using a long optical path. In this study, a medium resolution (1 cm"1) FTIR spectrometer
was connected to an optical multiple-reflection gas cell with an adjustable pathlength, ranging from
9 to 107 meters. Normally a pathlength of 96 m was used, selected to optimize optical throughput
and absorption levels. The system was built into a well-tempered and mechanically stabilized
optical bench and was in a normal transport van. The recorded spectra were analyzed by multiple-
regression techniques, fitting synthetically derived calibration spectra of all present compounds.
CH^ was analyzed in the wave number region at ~2950 cm4, and N^O around 2200 cm"1. 4
The methodology used to couple the concentration measurements to an actual emission is the Time
Correlation Tracer method. N^O was released in a controlled way from the methane emitting area
by use of several point sources distributed over the landfill. N^O mixed with the emitted CH^ in
the landfill plume, and the emission rate was derived by time resolved analysis of the CH^ and
tracer concentrations collected far enough downwind the landfill. The part of the time series
where the concentrations correlate, can be assumed to have its origin in the area where the tracer
is released, and can be quantified using the known tracer flux according this Equation.
Where:
C = the concentration in the mixing ratio and M to the molecular weight.
A correlation plot between tracer concentration and analyte helps to identify if sampling is within
the plume or outside. If the slopes of the concentration curves coincided as the plume swept in and
out of the location, it could be assumed that the tracer release simulated the entire methane
release. An estimate of the total landfill emission was obtained using the slope of the regression line
and N^O could be done at the same time. Low detection limits and sensitivity of ppb were
Tracer
Tmcer
CfrfT
C-ijf ฃ| C./Z 4
-------
ORS Handbook
Section 3.0
Page 3-48
of methane. Depending on the meteorological conditions, it is estimated that an accuracy of 15-30
percent in the emission estimate is achievable.4
Typical QA/QC
This section describes the QA/QC steps that pertain to the applications described above. QA/QC for
applications steps normally depend on pre-determined data quality indicators that address the
project unique objectives. The technology used for the applications presented above have their
own QA/QC associated to specifics of the instruments.
When using the tracer gas approach, it is important to consider a gas that is stable and has low
reactivity; thus, no significant sources and sinks that will alter the released concentration or, at
least, good knowledge of the background concentrations. Spurious releases of tracer gases that
reach 20 percent of the known concentration are easily identified CRDS, but anything below is
probably not significant. Background levels of the analyte gas must be known to track the
boundaries of the plume and to determine whether the measurements are in or out of the plume.
The time delay between release and arrival at measurement site needs to be carefully determined
before total methane emission results are considered acceptable. Flow rate of tracer gas released
from all bottles be carefully monitored and recorded if total methane emissions from a landfill are
to be accurately determined. A comparison (correlation plot) of analyte and tracer gas
measurements taken close and far away from the source provide a correlation coefficient and the
percentage difference or the total emission rate at close and far locations. Large percentage
differences indicate insufficient overlap of the analyte plume and the tracer gas plume during
stationary tracer-dilution measurements.
Siting Concerns
In general, concerns regarding the use of tracer gases to obtain emission fluxes are associated with
possible loss or gain of the tracer gas while in transport, not fully mixing within the plume, and
changing weather conditions (wind speed and direction for the most part). Other concerns are
-------
ORS Handbook
Section 3.0
Page 3-49
associated with calculations of the emission fluxes when estimating surface roughness or assessing
vertical and horizontal fluctuations.
Strength and Limitations
A key strength of using a tracer gas correlation technique is the ability to determine if varying
weather conditions affect the calculation of emission rates, which is possible by knowing release
rates and concentration. An additional strength is that emission rates are calculated within 15-30
percent precision. However, stationary and mobile setups have their challenges in terms of
logistics, location and whether available roads are near perpendicular to the flow of the plume.
Other limitations are cost of tracer gases cylinders and transportation of these, as well as changing
weather conditions affecting the calculation of emission rates. Tables 3-12 and 3-13 summarize
these strengths and limitations.
Table 3-12. Tracer Gas Correlation Strengths
Feature
Tracer Gas Correlation Strengths
Addresses Meteorology
Can determine if varying weather conditions
affect the calculation of emission rates.
Relatively precise Method
Emission rates are calculated within 15 - 30
percent precision.
Portable instrumentation
Field units are lightweight, rugged, and
relatively easy to transport and operate.
Table 3-13. Tracer Gas Correlation Limitations
Feature
Tracer Gas Correlation Limitations
Meteorological Concerns
Changing weather conditions affect the
calculation of emission rates.
Logistical Concerns
Location and the availability of roads
perpendicular to the plume create
difficulties.
Related Expenses
Tracer Gas cylinders can be expensive to
purchase and ship.
-------
ORS Handbook
Section 3.0
Page 3-50
References
1. Brasseur, G. P., J. J. Orlando and G. S. Tyndall. 1999. Atmospheric Chemistry and Global Change.
New York, NY: Oxford University Press.
2. Eastern Research Group, Inc. 2010. Evaluation of Large Area Methane Emission Source
Methods: Mobile and Stationary Plume Measurements Using the Tracer Correlation Approach.
Final Report for U.S. EPA. May 30.
3. Crosson, E. and S. Tan. 2009. Report on the work undertaken at the Twin Bridges Recycling and
Disposal Facility under an agreement with Eastern Research Group, Inc. August 1.
4. Samuelsson, J., G. Borjesson, B. Galle and B. Svensson. 2001. The Swedish landfill methane
emission project.
-------
ORS Handbook
Section 3.0
Page 3-51
3.5 Backward Lagrangian Stochastic Inverse-Dispersion model
Identifying and quantifying gaseous emission rates from a fugitive or area source to air (e.g.,
emissions from an open-air waste lagoon, confined animal feeding operations, biofuel production
facilities, landfills, etc.) is difficult. Several meteorological techniques are available (e.g., eddy
covariance and flux gradient), but they involve complex instrumentation (e.g., concentration
measurements at many heights and fast-response concentration sensors). They also require the
measurement site to be on a flat location. There is a separate technique that can be used called the
Integrated Horizontal flux that can be used for non-flat locations, but can only be used for small
source areas because it requires vertical and horizontal concentration measurements for the entire
plume. Because many emission sources do not meet these criteria, other techniques must be
implemented to estimate emission rates.1
General Description of Approach
The limitations of traditional meteorological techniques can be addressed by using an atmospheric
dispersion model to calculate the emission rate indirectly. The "inverse-dispersion" technique
provides an accurate and economical alternative for measuring emissions. The technique uses a
mathematical model of the dispersion of target gas from an emission source to a downwind
location, so that a downwind concentration measurement can establish the emission rate.1'2 This
approach has the advantage of requiring only a single concentration measurement and basic wind
information, with substantial freedom to choose convenient measurement locations. Theoretically,
the technique assumes idealized wind conditions; however, with careful selection of measurement
locations, inverse-dispersion modeling can provide a simple means of calculating emissions even in
non-ideal conditions.1,2
Figure 3-15 illustrates the bLS model for estimating emission rates. An area source of known
configuration emits gas at a uniform, but unknown, rate Q in kilograms per meter2 per second
(kg/nr/s). A time-average gas concentration C is measured at point M within the plume. The gas
concentration C can be determined by ORS measurement methods such as open-path FTIR, TDLAS,
-------
ORS Handbook
Section 3.0
Page 3-52
UV-DOAS, or point measurements such as CRDS. With a model prediction of the ratio of
concentration at M to the emission rate (C/Q) , the emission rate can be inferred as indicated in
sim
the Equation below, where C is the background pollutant concentration.1^3'4'5
s=(
-------
ORS Handbook
Section 3.0
Page 3-53
Depending on the trajectory, the particle may or may not sample the target source (i.e.,
"touchdown" within the source). The term Lagrangian indicates that the model releases individual
particles and follows them along their paths through the air, rather than performing calculations at
fixed locations in space. Stochastic indicates that the model mimics the random, turbulent motion
of each particle. As these particles travel through the air, they move through different regions of
interest. Some particles will touch down in the source region and move to the concentration
sensor, contributing to a measured concentration increase.1'2'6
For surface area emission sources, all that is needed to invoke the bLS model to calculate (C/Q are
sim
wind statistics, which can be determined from a few key meteorological observations.12 In general,
the bLS inverse-dispersion method of modeling emissions is cost- effective, requires only a single
field measurement of C, and, under the conditions described in this section, generates emission
rates with accuracy adequate for many applications.
Backward LS Dispersion Model for Calculating (C/Q
sim
In a bLS dispersion model, the upwind trajectories of model particles are calculated from location M
(Figure 3-15). The important information from the backward trajectories is the set of "touchdown"
locations (xq, y ) where particles impact the ground, and vertical "touchdown velocities" at impact
w . From the set of trajectories, the equation below is used to calculate (C/Q) by summing the
0 sim
reciprocal of for touchdowns occurring within the source boundary.3
if CM = -,
\ jf*
Where:
n = the total number of computational particles released from M and the summation covers only
touchdowns within the source area. In the bLS model, thousands of trajectories are calculated
upwind of the prevailing wind conditions.
-------
ORS Handbook
Section 3.0
Page 3-54
Commercially-produced software developed and distributed by WindTrax (Thunder Beach
Scientific, Nanaimo, Canada) is available to solve bLS equations. WindTrax combines the bLS model
with a graphical interface, allowing sources and sensors to be conveniently mapped (see Figure 3-
16).6 To calculate unknown source emission rates and/or concentrations, WindTrax requires the
following information:
upwind and downwind gas concentrations (C and C),
b
wind statistics (e.g. wind speed and wind direction),
roughness of the surface (z ), and stability of the atmosphere near the ground (Monin-
0
Obukhov stability parameter, L).
The latter wind statistics may be obtained from sonic anemometry or estimated within Windtrax. If
concentration is measured in units of ppm or ppb, air temperature and pressure are also needed
(pressure is often estimated from elevation). The particle models used in WindTrax are time-
independent, so the input data must be values averaged over a period, typically 15-30 minutes.
This averaging eliminates the unpredictable variability due to turbulence in the atmosphere on
short time-scales. Conversely, if the averaging time is too long, the more gradual diurnal variation
typical of the surface layer will not be resolved.6
-------
ORS Handbook
Section 3.0
Page 3-55
View | Edit | Actions f Platforms | Sensors | Shaoes ฃiata | Miscellaneous
' WindTrax - C:\Program Fiies\WindTrax\ExaunpIes\Example3.wpf (Modified]
Pointer
I
Mop Surface (Bare Soil) X: 2.9 | Y: 27.3
i
0le Edit J3ew Qraw Simulation 0roject Xools Help
Figure 3-16. Illustration of the WindTrax bLS Modeling Software Graphical Interface
bLS Model Output Units
Open-path ORS (OP-ORS) measurement techniques such as open-path FTIR or TOLAS can be used to
make the concentration measurement in a bLS application. Also, point-measurement sensors such
as closed-cell sensing techniques including CRDS or cell-based FTIR could be used to measure
plume concentration. These techniques provide concentration in units of ppm or ppb over the
optical path used in the measurement (ppm*m or ppb*m). Specific temperature and pressure data
are needed to convert to absolute concentration (i.e. g/ms), which is used to determine the mass
emission rate. The bLS calculation itself produces an emission rate (flux), Q , with units of kg/rru/s
bLS
(for example), but sometimes the emission rate is multiplied by the source area and Q becomes
bLS
an area integrated emission rate with units of kg/s or kg/h.1,2,8,9,
Verification/Validation Study
One of the foremost leaders in the development and application of bLS modeling as it applies to
gaseous emissions for large area sources is Thomas Flesch at the University of Alberta, Canada. He
conducted numerous studies using bLS. In this section, a bLS field trial entitled Deducing Ground-
-------
ORS Handbook
Section 3.0
Page 3-56
to-Air Emissions from Observed Trace Gas Concentrations: A Field Trial is summarized as an
example of the bLS application.1 The summary briefly illustrates how bLS can be used and describes
the quality of data that can be generated by the model.
In 2004, Flesch etal. reported a bLS field trial experiment in which the inverse-dispersion technique
was used to determine Q in an ideal surface-layer setting. A small area source from which
bLS
methane was released at known rates over a wide range in meteorological conditions was
constructed. An open-path laser measured the methane concentration Cat positions located up to
100 meters downwind of the source as shown in Figure 3-17. A corresponding (C/Q was
sim
calculated using a bLS dispersion model, and the resulting estimate of the emission rate Q was
bLS
compared with the known Q. The study objectives were the following: 1) to quantify the accuracy
and uncertainty in Q in an ideal setting; 2) to probe the conditions under which a dispersion
bLS
model based on the Monin-Obukhov similarity theory (MOST) performs poorly; and 3) to validate
an experimental system (i.e., source, sensors, bLS model) for examining the robustness of a bLS
estimate in non-ideal conditions.
Flesch et al. Conditions and Setup
The experiment took place over a 6-day period in May and June of 2001, near Ellerslie and
Edmonton, Alberta, Canada, in a large clover field. From a meteorological perspective, the site was
nearly idealwind conditions were uniform, temperatures were not high enough to cause thermal
convection of the plume out of the measurement window, and the nearest significant change in
land cover was more than 500 m from the source.
-------
ORS Handbook
Section 3.0
Page 3-57
Laser path
Tracer Source
eteo ro log ical
Figure 3-17. Map of the laser paths used in the 2004 Flesch et a I experiment.
Note: The shaded square on the map represents the methane tracer source and the large
"+" symbol indicates the location of the meteorological tower with the 3-D sonic
anemometer.
A synthetic source was created to approximate a 6 m x 6 m square area source. A manifold was
constructed out of polyvinyl chloride pipe and 36 0.5 mm outlet holes were drilled into the pipe. A
gas cylinder was coupled to the manifold through a regulator and rotometer (flow meter). The
methane tracer gas was released from high pressure cylinders (99.1 percent purity) at flow rates
between 15 and 50 L/min. Each release lasted from 1 to 3 hours. The cylinder valve was manually
adjusted to maintain a nominally constant flow rate, with adjustment occurring every minute or
two as necessary. The study estimates a 10 percent uncertainty in Q due to flow-rate fluctuations,
observer error in reading the rotometer scale, and gas temperature variability inside the rotometer
(which affects calibration).
Methane concentration measurements were made using two open-path lasers. A focused beam
from a tunable IR laser was aimed at a distant retro-reflector where it was reflected back to the
receiver optics and a detector. The returning signal strength was proportional to the methane
concentration C between the laser and the retro-reflector. Background methane C was
b
periodically measured at 1.95 ppm. This measured value closely correlated with the average
-------
ORS Handbook
Section 3.0
Page 3-58
methane background for the Edmonton region during the experiment, as routinely measured by
the provincial government of Alberta. The laser units recorded C every minute. These readings
were averaged into 15, 30, or 120-minute values. To convert concentrations from ppm to absolute
concentration in g/m3, Flesch et al used the measured air temperature and atmospheric pressure
for each observation period. A 3-D sonic anemometer was placed on a tower approximately 2 m
above ground and was used to determine values of u*, L, z , and 3 for the bLS simulations. Figure
3-17 shows a map of the various laser paths (dotted lines with arrows) used in the experiment. The
code at the tip of each arrow head is associated with the experimental conditions for that trial
measurement. The wind direction on the map is from west to east and all laser measurements
were made downwind.
Flesch et al Results and Conclusions
When periods of extreme stratification or MOST failure were excluded, the bLS inverse-dispersion
technique diagnosed the strength of a small ground-level source with small bias (mean value of
Q /Q within 2 percent of unity). Poor results were excluded when using a laser path over the
bLS
source because this study dealt with a very small area source, meaning the laser path was not
always sufficiently inside the plume. In situations involving a large source area, a measurement
location above the source is acceptable. The period-to-period variability in Q was acceptably
bLS
small (standard deviation of Q /Q is approximately 0.2). Using path-integrated values of C
bLS
enhanced the accuracy with which Q was diagnosed and rendered the experimental procedure
bLS
very forgiving (the laser could be positioned without being overly concerned about changing wind
direction). Based on their field experiments, Flesch et al made several recommendations for using
a bLS model to infer Q from an area source in an ideal surface layer problem:
bLS
PIC measurement is preferable to a point measurement because PIC gives results that are
integrated over the entire beam path length and are, therefore, more representative than
a single point source measurement of the actual plume concentration.
-------
ORS Handbook
Section 3.0
Page 3-59
Distance of the detector from the source should be small enough that the concentration
rise over background is accurately measured.
Meteorological averaging times of 10-30 minutes are ideal for calculating concentration
and meteorological statistics. Shorter averaging times may not capture an equilibrium
state of the atmosphere, a requirement for the application of MOST.
Periods of extreme atmospheric stability should not be used in assessing Q . An example
bLS
of an acceptable limit is | L| > 10m.4
Disregard periods of low u* (e.g., u* < 0.15 m/s).
Because all testing sites are different, the bLS modeling system has been used with varying site
locations such as ponds, pastures, and other scenarios. More detailed information pertaining to
different types of site locations can be researched in the literature.
Typical QA/QC
For the bLS dispersion model to accurately calculate the emission rate of a source, it is important to
verify that the instrumentation used to collect concentration data for target analytes is appropriate
for bLS calculations. QA/QC guidelines identified in this protocol or other EPA literature for the
technology used should be followed for optimal performance.
Because meteorological measurements are required for the bLS model, it is important to ensure
that accurate measurements are used. Meteorological data collected on site should be
collected with appropriate instrumentation, and applicable EPA guidelines should be followed.
More information on the technology used to collect meteorological data can be found in the
Quality Assurance Handbook for Air Pollution Measurement Systems, Volume IV:
7
Meteorological Measurements, Version 2.0. If meteorological data are not collected on site, it
is necessary to ensure that the data used in the bLS model are taken from a trusted source and
the location of the measurements is near the test site. Wind stability determination requires
more sophisticated instrumentation (e.g., 3-D sonic anemometer, or temperature
measurements at two or more heights).
-------
ORS Handbook
Section 3.0
Page 3-60
WindTrax and other modeling software are available for use to perform the bLS calculations. A
simplistic data set with known results should be used to test the modeling software before use for
field data calculations to verify the software's performance. In special cases, it may be preferable
for the user to develop its own software program to perform the necessary calculations, though it is
not recommended for accurate modeling. It is necessary to test all bLS software on a simplistic data
set for which the results are known to verify the software. In all cases, input data should be
reviewed for accuracy and possible transcription errors.
Siting Concerns
The bLS dispersion model utilizes the average wind and turbulence statistics of the atmosphere to
calculate (C/Q) . MOST states that the statistical properties of the wind in the surface layer are
sim
determined by a few key parameters: the friction velocity u*, which is determined by the vertical
transport of horizontal momentum near the surface; the Obukhov stability length L, which
quantifies the stability of the atmospheric surface layer; the surface roughness z , which is related
to the height of the plants, soil, or other elements covering the ground; and the wind direction |3. In
the field, these parameters are typically determined with the use of sensors such as a 3-D sonic
anemometer. The placement of sensors relative to sources can have a major effect on the quality of
the predictions generated by the models. If no concentrations are measured downwind of a
source, then emission rates cannot be determined.6 Upwind ambient gas concentrations must also
be measured that may be coming onto the site. The variability of wind direction might cause some
simulations to fail while others succeed. When multiple unknown sources are present, the
calculations are very sensitive to sensor placement and measurement error. Where flow obstacles
such as buildings or fences are present, measurements are often better made further downwind of
the source, well away from the obstacles; at least 10 obstacle heights downwind are a useful rule
of thumb for measurement location (M) placement.1'2'5'6
If the source and the detection point M lie within a horizontally homogeneous surface layer, the
application of the bLS dispersion model technique is reasonably straightforward. Many agricultural
-------
ORS Handbook
Section 3.0
Page 3-61
and environmental source estimation problems potentially fit this category. These problems may
include emissions from small soil research plots, feedlots, ponds, industrial grounds, and so on, that
often occur in circumstances for which it is reasonable to assume that the local wind flow is
uniform (i.e., wind statistics that do not deviate more than 10 to 20 percent from their spatial
average over the region from source to detector).1
An important advantage of bLS models is the ease with which complex source shapes can be
handled. One of the most important factors affecting model error is the size of the regions of
interest through which the analytes travel. As more analytes travel through a given region (i.e., the
concentration is higher), more samples of the region are taken and the model error is reduced. In
practice, this means that the larger the source target, the smaller the error, and conversely, the
smaller the source target, the greater the potential for error.1'2'3 However, the use of bLS within
large source areas has been very successful in many studies.
Strengths and Limitations
bLS models have several advantages over Gaussian and Eulerian models. For example, bLS models
are more physically valid than Gaussian models, which do not incorporate wind shear or other
meteorological information, and they do not require artificial diffusivity, as do the Eulerian models
for convective transport. A traditional disadvantage of bLS models is their computation time
requirements, which can be several orders of magnitude larger than those required to solve
algebraically reduced Gaussian models or even Eulerian models. This is because of the need to
calculate thousands of unique atmospheric trajectories. However, modern computing power has
rendered this problem to be of limited concern for most users. Tables 3-15 and 3-16 summarize the
bLS model's strengths and limitations in more detail.
-------
ORS Handbook
Section 3.0
Page 3-62
Table 3-15. bLS Model Strengths
Feature
bLS Strengths
Simplicity of
Measurements
Requires only a single concentration measurement - as opposed to
many concentration measurements made in the vertical or horizontal
plane of the plume.
Flexible input requirements: various wind statistics can be entered
into software and needed conversions are done internally; different
types of concentration observations are possible (point or line
average).
Siting Concerns
Substantial freedom to choose convenient measurement locations
Handles complex source shapes and sizes with relative ease
Can be used in locations with wind disturbances if sensor locations are
chosen with care
Economical
An economical alternative for determining emissions
Free downloadable software available online
Table 3-16. bLS Model Limitations
Feature
bLS Limitations
Meteorological Concerns
Assumes idealized atmospheric conditions unless care is taken
with sensor placement.
Rapid atmospheric changes or extreme stability invalidate MOST
and cause QbLS estimates to be inaccurate
Training Requirements
Some judgment required to identify poor measurement locations
(for both winds and concentration). Poor measurements locations
(e.g., close to building) can significantly impact the quality of
emission calculation.
Some basic experience or training in micrometeorology or
atmospheric gas transport is required to generate high-quality
data from the bLS model
Siting Concerns
When multiple unknown sources are present, the calculations are
very sensitive to sensor placement
Where flow obstacles such as buildings or fences are present,
measurements are often better made further downwind of the
source, well away from the obstacles.
Time Limitations
Can require lengthy computational time.
-------
ORS Handbook
Section 3.0
Page 3-63
References
1. Flesch, T.K., J.D. Wilson, L.A. Harper, B.P. Crenna, and R.R. Sharpe. 2004. Deducing
ground-air emissions from observed trace gas concentrations: A field trial. J. of Applied
Meteorology. 43:487-502.
2. Flesch, T.K., J.D. Wilson, and L. A. Harper. 2005a. Deducing ground air emissions from
observed trace gas concentrations: A field trial with wind disturbance. J. of Applied
Meteorology. 44:475-484.
3. Flesch, T.K., J.D. Wilson, and E. Yee. 1995. Backward-time Lagrangian stochastic dispersion
models, and their application to estimate gaseous emissions. J. of Applied Meteorology.
34:1320-1332.
4. Wilson, J.D., T.K. Flesch, and L.A. Harper. 2001. Micro-meteorological methods for
estimating surface exchange with a disturbed wind flow. J. of Agricultural and Forest
Meteorology. 107:207-225.
5. McBain, M.C. and R.L. Desjardins. 2005. The evaluation of a backward Lagrangian
stochastic (bLS) model to estimate greenhouse gas emissions from agricultural sources
using a synthetic tracer source. J. of Agricultural and Forest Meteorology. 135:61-72.
6. Thunder Beach Scientific, Nanaimo, British Columbia, Canada, 2009.
http://www.thunderbeachscientific.com.
7. U.S. EPA, Quality Assurance Handbook for Air Pollution Measurement Systems, Volume IV:
Meteorological Measurements, Version 2.0, EPA-454/B-08-002, March 2008.
8. Harper, L.A., T.K. Flesch, J.M. Powell, W.K. Coblentz, W.E. Jokela, and N.P. Martin. 2009
Ammonia emissions from dairy production in Wisconsin, J. of Dairy Science. 92:2326-
2337.
9. Flesch, T.K., L.A. Harper, J.M. Powell, and J.D. Wilson. 2009. Use of dispersion analysis for
measurement of ammonia emissions from Wisconsin dairy farms. Agric. Eng. 52:253-265.
10. Harper, L.A., T.K. Flesch, K.H. Weaver, and J.D. Wilson. 2010. The effect of biofuel
production on swine farm ammonia and methane emissions. J. Environ. Qual. (accepted
for publication, August 5, 2010)
-------
ORS Handbook
Section 3.0
Page 3-64
3.6 Geospatial Measurement of Air Pollution, Remote Emission Quantification -
Other Test Method 33
Identifying and quantifying gaseous emission rates from a fugitive or area source to air (e.g.,
emissions from an open-air waste lagoon, confined animal feeding operations, biofuel production
facilities, landfills, etc.) is difficult. The challenges encountered trying to measure area source
emissions flux include spatial and temporal variability of the emission sources and the uncertainty
of the measurement technology. As discussed in previous sections, emissions source variability
includes defined and undefined sources such as unknown emissions points, delocalized emissions
sources, the timing of periodic or episodic emissions, and atmospheric, diurnal, seasonal, and
process variations in emission flux. Defined fugitive sources cover smaller areas (i.e., less than 1
square kilometer down to a square meter) allowing emissions points to be identified for direct
measurement. Undefined area sources typically originate from large areas (i.e., greater than 1
square kilometer) and attempts at measuring the air quality have included ambient air monitoring.
Traditional ambient air quality measurements are primarily collected from fixed-placement
monitoring stations and provide information on long-term trends of air pollutants over the larger air
shed region. Direct (on-site) measurements of air pollutants are conducted at the immediate source
of the emission. By contrast, the methods of geospatial measurement of air pollution utilize source
assessment schemes that provide data for air quality conditions in the spatial and temporal ranges
in between for both defined and undefined areas. Figure 3.18 illustrates the operation regime of the
geospatial methods.1
-------
ORS Handbook
Section 3.0
Page 3-65
Distance From Source (increasing)
r
Direct Source
Measurement
GMAP
Ambient
Measurement
Required Detection Sensitivity (increasing)
Figure 3-. 18. GMAP Operational Regime.
Since 2006, EPA has investigated the use of mobile measurement systems under its Geospatia!
Measurement of Air Pollution (GMAP) program for a variety of air quality assessment applications,
including those involving complex, undefined, or large area sources.1 The GMAP vehicle
instrumentation system combines fast response analytical instruments with precise global
positioning systems (GPS) to characterize pollution emissions. Other Test Method 33 (OTM 33) is a
mobile measurement method series that resulted from the GMAP efforts using approaches that are
designed to quantify source emissions on scales ranging from near-field inspections of small fugitive
releases to whole facility mass emission rate measurements.2
OTM 33 systems typically have two possible modes of operation: (1) downwind mapping surveys to
detect and locate emission sources, and (2) quantification procedures to characterize concentrations
and mass emission rates. Because the OTM 33 techniques can be applied to many different
situations with different approaches, sub-methods OTM 33a, 33b, 33c, and 33d are in development
to address more details and specifics regarding the different assessment approach schemes. This
OTM 33 method series allows for the use of many different instrumentation configurations and
vehicle mobility schemes to assess air quality concerns at a variety of spatial scales.
The OTM category of measurement methods contains methods that have not undergone the federal
-------
ORS Handbook
Section 3.0
Page 3-66
rulemaking process, but have been reviewed by the EPA's Emission Monitoring Center (EMC) and
are potentially useful to the emission measurement community. Such methods may be considered
for use in federally enforceable state and local programs and can also be considered as candidates
for alternative methods to meet federal requirements under 40 CFR Parts 60, 61, and 63 through an
approval process.
General Description of Approach
Geospatial measurements of air pollution, or GMAP, is a general term defined in OTM 33 as
"referring to the use of fast-response instruments and precise global positioning systems (GPS) in
mobile formats to spatiotemporally-resolve air pollution patterns in a variety of use scenarios."1 The
geospatial measurement of Air Pollution-Remote Emissions Quantification (GMAP-REQ)
measurement method series (OTM 33) uses ground-based vehicle platforms fitted with
instrumentation to make mobile measurements proximal to a driving route. OTM 33 is a general
description of instruments, approaches, and assessment schemes that are further defined in sub-
methods. The source assessment schemes covered in these sub-methods include:
Concentration Mapping (CM) - Used to find the location of unknown sources and to
evaluate the impact these source emissions may have on local air shed pollution
concentrations.
Source Characterization (SC) - Used to improve the understanding of known or discovered
sources through the collection of secondary measurements (such as remote imaging or
canister grab sample speciation).
Emissions Quantification (EQ) - Used to measure (or estimate) the source emission rate.
Emission information on gas-phase criteria pollutants, particulate matter and ultrafine particles,
volatile organic compounds (VOCs), hazardous air pollutants (HAPs), and greenhouse gases (GHGs) is
collected when a vehicle outfitted with appropriate analytical and auxiliary equipment is driven
around the target facility. This information is spatially and temporally resolved, meaning that the
-------
ORS Handbook
Section 3.0
Page 3-67
data can be analyzed for leak detection and repair programs, periodic fenceline monitoring,
gradient-type local airshed concentration mapping, and source emission rate characterization. As
the instrumented vehicle is driven around the facility, air pollution concentration levels are recorded
both inside the emission plume and outside the plume, capturing background levels and the location
where the plume intersects the vehicle pathway.
Figure 3.19 illustrates a limitation of the method based on wind direction and path of the
instrumented vehicle.1 Also, if the wind conditions are not favorable, or there are other
interferences such as buildings or topographic obstacles to the path of the emission plume, then
representative measurements of pollutants in the plume may be difficult.
(A)
Roadway
Wind
Direction
Emission
Plume
\
a
Emission js
detectable by
GMAP-REQ
\
GMAP
Vehicle
(B)
Roadway
Wind
Direction
Emission
Plume
a
Emission is not
detectable by
GMAP-REQ
\
GMAP
Vehicle
Figure 3-19. GMAP (OTM 33) Limitations
The analytical instruments included in the design of the GMAP vehicle must be robust enough to
withstand mobile applications in remote areas. Unpaved or rough roads will likely be encountered
during the course of a facility survey. The pollutant measurement instrumentation must have near-
ambient level detection capabilities with an appropriate dynamic range. Because OTM 33 sub-
methods typically use a combination of data measurements (such as GPS location, pollutant
concentration, wind information, etc.) to determine source detection, emission source location, and
-------
ORS Handbook
Section 3.0
Page 3-68
mass emission rate, the instruments should also be time-synchronized to the second.
Currently, only OTM 33A is promulgated by EPA while other sub-methods are in development (and
are, therefore, not final). Following are the anticipated sub-methods for OTM 33:110
OTM 33A - Discovery/Characterization of Near-Field Fugitive Sources
See Section 3.7 for a more in-depth discussion.
OTM 33B - Mobile Tracer Correlation
See Section 3.4 for more discussion.
OTM 33C - Solar Source Techniques
See Section 3.3 for more discussion.
OTM 33D - Regional Mobile Sensing Approach
This method will be like OTM 33A, but will be designed for a broader geographical survey.
Details on the method procedures, quality assurance requirements, and performance specifications
are (or will be) available in the sub-methods, however, EPA expects that project-specific quality
assurance project plans will be developed according to EPA requirements for each measurement
event. Table 3.16 presents a brief overview of the performance characteristics of various mobile
measurement approaches that can be used for OTM 33 sub-methods.
-------
ORS Handbook
Section 3.0
Page 3-69
Table 3.16. Summary of mobile measurement approaches.
OTM 33A
OTM 33B
(Near-field)
OTM 33B (Far-
field)
Mobile SOF
Mobile Flux
Plane
Work Truck
CM, SC
(various types)
V
V
V
V
V
V
EQ Approach
Stationary
single point /
inverse model
Tracer
correlation
Tracer
correlation
Extended flux
plane /
integration
Finite flux
plane /
integration
Mobile single
point / inverse
model
Primary
Source Type
Point-like
Point-like
Large area /
facility
Large area /
facility
Point-like
Point / area
Distance to
Source (m)*
20 -150
50 - 300
300 - 4000
500 - 3000
20 -150
20 - 500
Elevated
Source
-
Possible
Possible
V
-
-
EQ
Observation
Mode
20 min.
stationary
Drive-by
Drive-by
Drive-by
Drive-by
Drive-by /
stationary
Analyte
Limitations
CMI-limited
CMI-limited
CMI-limited
Column
background
limited
CMI and
storage tube-
limited
CMI / sensor
cost limited
Site Access
Required
-
V
V
-
-
-
Key Use
Limitation
Open areas /
meteorological
Road access/
meteorological
Road access/
meteorological
Road access/
sunny
conditions
Open areas /
meteorological
Meteorological
Anticipated
Accuracy
Goals*
< 30%
< 15%
< 15%
< 20%
< 20%
< 50%
Application
Cost
Low
Mid
Mid
Low/mid
Low
Very low
Verification/Validation Studies
Verification/validation studies were conducted on the sub-methods to OTM 33. Therefore, see the
referenced sections above to review studies specific to the different OTM 33 techniques.
Typical QA/QC
Each sub-method to OTM 33 will have its own QA/QC parameters that are specific to the
instrumentation used, measurement technique, field site, and purpose of the study. QA/QC for
applications steps normally depend on pre-determined data quality indicators that are unique to
-------
ORS Handbook
Section 3.0
Page 3-70
the project objectives. The technology components used for OTM 33 applications have their own
QA/QC associated to specifics of the instruments. For the technology QA/QC for OP-FTIR, refer to
section 2.1; for OP-TDLAS, refer to section 2.2; for UV-DOAS, refer to section 2.3; and for CRDS,
refer to section 2.6.
Although EPA OTM 33 discusses potential interferences and QA/QC parameters for the various sub-
methods, this section pertains only to the method overview. Readers should refer to the published
sub-method for further discussion of potential interferences and QA/QC parameters.
Siting Concerns
In general, concerns regarding the use of OTM 33 to make representative air pollutant
measurements are associated with roadway access in and around the target facility, topography of
the landscape, buildings around the facility, and meteorological conditions. Referring to Figure 3.20,
it is easy to see how these factors could prevent a successful measurement survey. Having limited or
no roadway access could eliminate a site from being a candidate for OTM 33 measurements
altogether. On the other hand, complex site topography and/or meteorological conditions may
make the acquisition of representative measurements difficult, but not entirely impossible. Each
measurement survey at each individual site must have a project specific quality assurance plan that
evaluates the characteristics of the target facility site.
Other factors may exist that would indicate the need for one OTM 33 sub-method over another. For
example, if the leak source is suspected to be significantly elevated off the ground, then sampling via
inlet may not be able to sample the source plume, and therefore, total column techniques that use
the sun as the optical source (such as OTM 33C) would be more appropriate. Or, if there is the
possibility for multiple overlapping sources (such as that shown in Figure 3.20), a near-field OTM 33
sub-method (such as OTM 33A) may be more appropriate than a farther-field (such as OTM 33D).1
-------
ORS Handbook
Section 3.0
Page 3-71
Roadway
Wind
Direction
[
Emission
Plumes
V
G M AP-REQ may
detect one
combined plume
\
G MAP
Vehicle
Figure3-20. GMAP (OTM33) Overlapping Plume Sources
Strength and Limitations
The OTM 33 method series is advantageous for the flexibility that the method and sub-methods
allow. This flexibility means that one mobile sampling platform could be used in a wide variety of
applications simply by adjusting method parameters.
OTM 33 methods fill an important gap in air quality measurement spatial and temporal scales, but
can operate at the extremes as well (more locally at the source, to more regionally at ambient
levels). The air quality measurement instruments installed in the mobile platform will, by design,
have a broad dynamic range with a detection limit near ambient concentrations. An additional
-------
ORS Handbook
Section 3.0
Page 3-72
strength of the OTM 33 methods is that emission rates are calculated within 15-50 percent
precision.1,2
Although OTM 33 methods have their strengths, stationary and mobile setups have challenges in
terms of logistics, location, and whether available roads are near perpendicular to the flow of the
plume. For example, one major limitation of the OTM 33 method series is the requirement for
roadway access. Not all target facilities will have ideal perimeter survey roadways that completely
encircle the facility. If some roadway access exists on one side of the facility, but not on another,
then measurements should be taken during times when the wind conditions favor plume
transportation to the side with good roadway access. The representativeness of the emission
estimates will heavily rely on a limited range of appropriate atmospheric conditions and an
unobstructed transport pathway for the plume to travel from the source to the sampling location.
Tables 3.17 and 3.18 list the general strengths and limitations of the general approach, respectively.
Table 3.17. Strengths of the General OTM 33 Approach
Feature
OTM 33 Strengths
Flexible application
Is appropriate for a variety of applications with
only simple adjustments.
Good inspection approach
Individual emission rates are calculated within
60% accuracy (OTM 33A), improving with each
replicate measurement.
Portable instrumentation
Field units are rugged with a high temporal
resolution for real-time analysis and fast
deployment.
Site access not required
OTM 33A does not require access to the
surveyed site.
-------
ORS Handbook
Section 3.0
Page 3-73
Table 3.18. Limitations of the General OTM 33 Approach
Feature
OTM 33A Limitations
Meteorological concerns
Sustained and varying wind conditions are
required to transport the emission plume.
Susceptible to atmospheric stability
Atmosphere must be stable, but not overly
stable.
Absence of near-field obstructions
Obstructions to proper plume transportation
(such as trees, fences, etc.) can cause
inaccurate estimates.
Logistical concerns
Location and the availability of roads that are
perpendicular to the plume create difficulties.
Distance from emission source
Measurements must be made between about
10 and 200 m (OTM 33A). Accurate distance
measurements are required for the emission
estimation.
References
1. U.S. Environmental Protection Agency (EPA). 2014. Draft "Other Test Method" OTM 33
Geospatial Measurement of Air Pollution, Remote Emissions Quantification (GMAP-REQ).
Available online at http://www.epa.gov/ttn/emc/prelim.html.
2. Thoma, E.B., H. Brantley, B. Squier, J. DeWees, R. Segall, and R. Merrill. 2015. Development
of Mobile Measurement Method Series OTM 33. Proceedings of the 108th Annual
Conference & Exhibition of the Air & Waste Management Association, Raleigh, NC, June 22-
25, 2015.
3. DeWees, J. 2014. Development of a Mobile Tracer Correlation Method for Assessment of Air
Emissions from Landfills and Other Area Sources. Presented at the 38th SSSAAP Conference,
Point Clear, AL.
4. Foster-Wittig, T.A., E.D. Thoma, R.B. Green, G.R. Hater, N.D. Swan, and J.P. Chanton. 2014.
Development of a Mobile Tracer Correlation Method for Assessment of Air Emissions from
Landfills and Other Area Sources. Atmospheric Environment, 102, 323 - 330.
5. Albertson, J.D., T. Harvey, G. Foderaro, P. Zhu, X. Zhou, S. Ferrari, M.S. Amin, M. Modrak, H.
Brantley, and E.D. Thoma. 2016. A Mobile Sensing Approach for Regional Surveillance of
Fugitive Methane Emissions in Oil and Gas Production. Environmental Science & Technology,
50, 2487 - 2497.
-------
ORS Handbook
Section 3.0
Page 3-74
6. Rella, C.W., T.R. Tsai, C.G. Botkin, E.R. Crosson, and D. Steele. 2015. Measuring Emissions
from Oil and Natural Gas Well Pads Using the Mobile Flux Plane Technique. Environmental
Science & Technology, 49, 4742 - 4748.
7. Thoma, E.D., B.C. Squier, D. Olson, A.P. Eisele, J.M. DeWees, R.R. Segall, M.S. Amin, and M.T.
Modrak. 2012. Assessment of Methane and VOC Emissions from Select Upstream Oil and
Gas Production Operations Using Remote Measurements, Interim Report on Recent Survey
Studies. Proceedings of the 105th Annual Conference of the Air & Waste Management
Association, San Antonio, TX, June 19-22, 2012.
8. Brantley, H.L., E.D. Thoma, W.C. Squier, B.B. Guven, and D. Lyon. 2014. Assessment of
Methane Emissions from Oil and Gas Production Pads using Mobile Measurements.
Environmental Science & Technology, 48,14508 - 14515.
9. Brantley, H.L., E.D. Thoma, and A.P. Eisele. 2015. Assessment of Volatile Organic Compound
and Hazardous Air Pollutant Emissions from Oil and Natural Gas Well Pads using Mobile
Remote and On-site Direct Measurements. Journal of the Air & Waste Management
Association, 65:9,1072 - 1082.
10. Foster-Wittig, T.A., E.D. Thoma, and J.D. Albertson. 2015. Estimation of Point Source Fugitive
Emission Rates from a Single Sensor Time Series: A Conditionally-sampled Gaussian Plume
Reconstruction. Atmospheric Environment, 115.101 - 109.
-------
ORS Handbook
Section 3.0
Page 3-75
3.7 Geospatial Measurement of Air Pollution, Remote Emission Quantification:
Direct Assessment - Other Test Method 33A
Identifying and quantifying gaseous emission rates from a fugitive or area source to the atmosphere
is difficult due to a source's spatial and temporal variability and measurement technology
uncertainty. However, spatiotemporal measurements of air quality around a facility can be made
with mobile measurement systems. Mobile measurements refer to the geospatial measurement of
air pollution (GMAP) series of methods wherein a vehicle is outfitted with fast-response air quality
instruments and precise global positioning systems (GPS) that can be deployed in a variety of
scenarios.1,2 Other test method (OTM) 33, or "Geospatial Measurement of Air Pollution-Remote
Emissions Quantification (GMAP-REQ)," describes the general overview of the GMAP methods and
serves as the background for method subsets. This section discusses the first of these sub-methods,
OTM 33A, a GMAP-REQ approach called "direct assessment" (DA). The DA approach directly
measures the concentration of atmospheric pollutants and then uses wind measurements to
calculate the emissions flux rate. OTM 33B for comparison, requires the known release of a tracer
gas to determine emissions flux as described in Section 3.4.
OTM 33A is a mobile measurement technique that is used to characterize emissions from near-field,
ground-level point sources and is intended for rapid deployment without requiring auxiliary
stationary instrumentation, site-specific modeling, or direct site access. Typically, this method is
applied to sources that are small in spatial extent and are near (about 20 to 200 m) to the driving
route.2'3
Emission sources are categorized as either "defined" or "undefined." Undefined area sources
typically originate from large areas (i.e., greater than 1 square kilometer), whereas defined fugitive
sources cover smaller areas (i.e., less than 1 square kilometer down to a square meter) allowing
emissions points to be identified for direct measurement.1 The types of sources targeted with OTM
33A are mostly categorized as defined sources.
-------
ORS Handbook
Section 3.0
Page 3-76
General Description of Approach
Conducting OTM 33A requires the deployment of a vehicle equipped with fast-response analytical
instruments1 and precise GPS to characterize pollution emissions as the GMAP vehicle is driven
around the facility. Emission information on gas-phase criterial pollutants, particulate matter and
ultrafine particles, volatile organic compounds (VOCs), hazardous air pollutants (HAPs), and
greenhouse gases (GHGs) is collected when a GMAP vehicle outfitted with appropriate analytical
and auxiliary equipment is driven around the target facility. This information is spatially and
temporally resolved, meaning that the data can be analyzed to supplement regular air monitoring
programs such as leak detection and repair programs, periodic fenceline monitoring, gradient-type
local air shed concentration mapping, and source emission rate characterization.1 As the GMAP
vehicle is driven around the facility, air pollution concentration levels are recorded both inside the
emission plume and outside the plume, depending on where the plume intersects the GMAP vehicle
pathway (Figure 3.21).1 If the wind conditions are not favorable, or there are other interferences
such as buildings or topographic obstacles to the path of the emission plume, then representative
measurements of pollutants inside the plume may be difficult. Measurements outside the plume are
considered background levels.
1A suitable instrument example is cavity ring-down spectroscopy (CRDS) described in Section 2.6.
-------
ORS Handbook
Section 3.0
Page 3-77
(A)
Roadway
Wind
Direction
ฆ=>
Emission
Plume
\
a
Emission js
detectable by
GMAP-REQ
it
\
GMAP
Vehicle
(B)
Roadway
Wind
Direction
<=ฆ
Emission
Plume
Emission is not
detectable by
GMAP-REQ
it
\
GMAP
Vehicle
Figure 3.21. GMAP (OTM 33) Limitations.
The flexibility of OTM 33 methods allow for multiple source assessment schemes possible using one
measurement system. The source assessment schemes possible for OTM 33A include:
Concentration Mapping (CM) Used to find the location of unknown sources and to evaluate the
impact these source emissions may have on local air shed pollution concentrations.
Source Characterization (SC) - Used to improve the understanding of known or discovered
sources through the collection of secondary measurements (such as remote imaging or canister
grab sample speciation).
Emissions Quantification (EQ) - Used to measure (or estimate) the source emission rate.
Concentration mapping (CM) involves driving the GMAP vehicle around a target facility to determine
the area of highest emission concentrations. As illustrated in Figure 3.22, the mobile measurements
combined with wind information can identify a "hot spot" area (shown in red as higher
concentration).2
-------
ORS Handbook
Section 3.0
Page 3-78
Wind
Figure 3.22. GMAP-REQ-DA (OTM 33A) Concentration Mapping Survey of an Industrial Facility. Red bars
indicate elevated emission concentrations.
Once a hot spot area has been identified, further investigations (i.e., source characterization, leak
detection and repair (LDAR)) may be conducted to provide additional information regarding the
discovered emissions. As explained in the OTM 33A method, these source characterization activities
can include multiple repeated transects of the GMAP vehicle through the emission plume or
mapping upwind and downwind of the source to help refine the source location and observe any
potential background interferences. Source characterization activities may also involve the
collection of other data forms and samples such as recording infrared camera images or collecting a
canister grab sample while in the emission plume. Auxiliary data can also be collected using the real-
time measurement data to position the GMAP vehicle in the plume at a safe distance downwind of
the emission source. If site access is limited and there is no other way to determine the source of the
emission, then in-plume canister sampling may help inform the actual source by elucidating the
chemical composition, which may indicate one component of the process over another.2
While positioned in the plume, EQ efforts can also be accomplished. Although it is possible to
perform EQ measurements both in mobile and stationary scenarios, EQ measurements are typically
acquired using a stationary approach such as Point Source Gaussian (PSG).2 When conducting PSG
-------
ORS Handbook
Section 3.0
Page 3-79
measurements, the GMAP vehicle is stationary at a location downwind of the emission source while
the concentration measurement and wind measurement instruments of the vehicle collect data
over a period of 15 to 20 minutes for one EQ. measurement. A PSG-based computer program uses an
inverse algorithm to estimate the strength of the source emissions and bin the concentration data
by the wind angle at the time of sample collection (see example in Figure 3.23).2 The combined
information is then used to estimate the mass emission rate of the emission source assuming a point
source release and using Gaussian plume dispersion tables.
Methane Concentration vs Wind Direction
(A)
Emission source
n
Emission plume
Wind direction
variability
Predominant wind
direction
O
Observation point
(B)
pgi 3 pqy 9.77
ti_0 16134 maxc_17.1
wsf_0 angf_60
0.5
100
150
200
250
300
350"
o Data
xQ= 185 97
Rป 0.99714 (lin)
150 200 250
Wind Direction (deg)
Figure3-23. (A) An Illustration of a Stationary EQ Observation, and (B) the Resultant Time-Integrated,
Wind Angle-Resolved Data File and Gaussian Fit.
The analytical instruments included in the design of the GMAP vehicle must be robust enough to
withstand mobile applications in remote areas. Unpaved or rough roads will likely be encountered
during a facility survey. The pollutant measurement instrumentation must have near-ambient level
detection capabilities with an appropriate dynamic range. Because OTM 33 sub-methods typically
use a combination of data measurements (such as GPS location, pollutant concentration, wind
information, etc.) to determine source detection, emission source location, and mass emission rate,
the instruments should also be time-synchronized to the second. A typical GMAP vehicle includes a
cavity ring-down spectroscopy instrument (e.g., 10 Hz G1301-fc or 0.5 Hz G1204 by Picarro Inc.,
Santa Clara, CA, USA; a 1 Hz GG-24-r instrument by Los Gatos Research, Mountain View, CA, USA; or
similar) to provide real-time methane measurements. Common additional equipment includes a
sonic anemometer (e.g., model 81000 3-D by R.M. Young, Traverse City, Ml, USA), a compact auto-
-------
ORS Handbook
Section 3.0
Page 3-80
north weather station (e.g., model AIO by Climatronics Corp., Bohemia, NY, USA), and a GPS system
(e.g., Hemisphere Crescent R100 Series GPS by Calgary, AB, Canada or similar). A custom computer
program (e.g., LabView by National Instruments, Inc., Austin, TX, USA) is also employed to time-align
the wind and concentration data stream, while a custom analysis program (e.g., MATLAB by Math-
Works, Natick, MA, USA) processes the data and calculates source emission rate estimates based on
the PSG calculation approach.2'3'4'5'6'8
The PSG calculation uses the measured source distance from the mobile platform when stationary
(determined using an optical gas imaging (OGI) camera and laser rangefinder) and a representative
atmospheric stability indicator (ASI)derived from the average turbulence intensity from the 3-D
sonic anemometer and the standard deviation of the 2-D wind direction from the compact
meteorological stationto input values for the horizontal (oy) and vertical (oz) plume dispersion as
indicated in the PSG reference tables (EPA OTM 33A Appendix Fl). The PSG emission estimate (q) is
a simple 2-D Gaussian integration (reflection term is omitted) where the plume dispersion is
multiplied by the mean wind speed (u) and the maximum plume concentration (c), calculated as:2'3
q = 2 ฅT x Sy x 6 z. x u x c
Through validation studies discussed in the section below, data quality indicators (DQIs) were
developed to filter out data that were not ideally obtained. This process eliminates the potential for
high or low biases in the concentration measurements due to skewed plume acquisition decreasing
the representativeness of the plume measurement, plume concentrations too low for the method
limits or incomplete plume capture, or improper wind conditions or plume channeling due to
upwind obstructions. Such DQIs include (but are not limited to):
Concentrations corresponding to wind directions that are ฑ 30ฐ of the mean source wind
direction
An average in-plume concentration of 0.1 ppm or greater
A Gaussian fit with an R2 > 0.80.
-------
ORS Handbook
Section 3.0
Page 3-81
For example, if the atmospheric boundary layer increases in height due to unstable atmospheric
conditions, the emission plume height will increase such that the measured concentration is greatly
reduced (Figure 3.24).2 In this example, the change in atmospheric stability is typified by two
emission plume mobile OTM 33A measurements taken about 2 hours apart on the same day. In the
mid-morning (shown in the red trace), when the atmosphere is more stable, the advection (lateral)
movement of the plume is more dominant than the convection (vertical) movement of the plume
and the full representation of the gas emission concentration is captured. Conversely, as the ground
warms towards mid-day (blue trace) and the energy in the atmosphere increases, the number of
rising parcels of air increases and the convection movement strengthens, causing the plume to
disperse more vertically and effectively dilute the gas emission concentration at the sample probe.2
The measurement shown in the blue trace would be eliminated from the final data set after the
application of the < 0.1 ppm DQI, thereby insuring that the data set analyzed by the end-user is of a
known quality and certainty.
2.20
2.15
2.10
Q.
Cl
jl 2.05
CD
^ 9:47 AM Transect
11:52 AM Transect
CD
o 2.00
c
o
O
5 195
1.90
1.85
1.5
0.0
0.5
1 .0
Transect Path (km)
Figure 3-24. Effect of varying atmospheric conditions with red showing stable atmosphere and blue
showing unstable atmosphere.
As mentioned in Section 3.6, the quality assurance project plan (Q.APP) will identify the DQIs used
and EPA's OTM 33A document includes considerations for sampling technique and execution.
-------
ORS Handbook
Section 3.0
Page 3-82
Verification/Validation Studies
Although many groups are actively developing OTM 33A GMAP vehicles and techniques, only a
handful of have published the results of their studies and many of these are from EPA-directed
research. What is now known as OTM 33A was first described by Thoma et al. in 2010 as a mobile
method designed to locate fugitive emissions, estimate the methane emission rate, and use the
methane result to calculate VOC emissions using the ratios of compound abundance relative to
methane in evacuated canister samples. This preliminary study provided proof-of-concept and laid
the groundwork for method development. An overview of the studies performed to date is available
in Table 3.19 and are discussed in chronological order below
Table 3.19. Summary of GMAP-REQ-DA Studies.
Thoma et al.,
20104
Thoma et al.,
20125
Brantley et al.,
2014s
Brantley et al.,
20157
Foster-Wittig et
al., 2015s
Study dates
2009
2010-2011
2010 - 2013
July 2011
2010 - 2013
Chemical
compounds
measured
Methane & VOC
Methane & VOC
Methane
VOC & HAPs
Methane
Study Focus
Proof-of-concept
Interim Report
Data statistics and
comparison with
direct, onsite
measurements
Comparison of
onsite and remote
VOC/HAP
measurements
Comparison of
emissions
calculation
methods
Key Results
> There is
potential for
the approach
> Multiple repeat
measurements
are best.
> Proper plume
transport is
important.
> Accuracy = ฑ
60%
> Methane
emissions
mostly
correlated with
gas production.
> OTM 33A
captures large,
stochastic
emission events
versus smaller
leaks.
> Onsite and
remote
concentration
measurements
are similar.
> OTM 33A can
be used to
identify when
emissions are
not effectively
controlled.
> Although
simplified, PSG
estimates are
good
approximations.
> Repeat
measurements
reduce estimate
error.
As research progressed and the technique was applied in multiple field studies, Thoma et al.5
released an interim report in 2012 that discussed different emissions estimation calculation
methods and the potential use and limitations of the method. The data presented in this study
included measurements from both field campaigns and controlled release experiments using a
-------
ORS Handbook
Section 3.0
Page 3-83
preliminary DQI filter for the wind direction of ฑ 60 degrees from the predominant wind vector. The
field campaign measurements indicated that the methane emission rates and VOC compositions for
each basin are unique to that basin such that there is potential to determine the source of a
measurement just from these two attributes, like a compositional fingerprint.5 The controlled
release experiments investigated the accuracy and precision of two emission calculation methods:
PSG and backwards Lagrangian stochastic (bLs) modeling. Although individual measurements can
exhibit a large amount of variability (see Figure 3.25a), repeat measurements can significantly
reduce the measurement error and yield estimates for 0.6 grams per second (g/s) methane release
of 0.56 g/s using PSG and 0.57 g/s using bLs calculation methods (with a 1 standard deviation (a) =
0.17 g/s and 0.23 g/s, respectively). In addition, the controlled release experiments showed that
unstable, low wind speed conditions (< 1 meters per second (m/s)) do not produce much usable
data and measurement farther away from the emission source than 100 m require more favorable
wind conditions for sufficient plume transport to the sampling location.5
Release Distance (m)
PSG Emission Estimate (g/s)
Figure 3-25. results from controlled release experiments with (a) measurement results for both psg and
bis calculation methods, and (b) a comparison between the two methods.
The data plotted in Figure 3.25a represent the spread of the two calculation methodsPSG and
bLsusing the error bars with the circle indicating the average of the two methods. The first data
point in blue of panel (a) represents the average of all the data with o = 0.18 g/s as the error bars.5
-------
ORS Handbook
Section 3.0
Page 3-84
The largest underestimates in panel (a) occur at the farther distances while the largest overestimate
in the measurement series (1.03 g/s) occur where plume flow obstructions were observed and
therefore indicate the effect of plume "channeling." The overall measurement uncertainty between
the two methods is quite small as evidenced by the data set average in panel (a) being very close to
the actual release rate and the high r2 value of 0.83 in panel (b). Most notably, the data shows that
individual measurements using the OTM 33A approach can have an accuracy of about ฑ 60%, while
ensembles of repeat measurements tend to be much more accurate.5
Field studies using this approach continued to 2013, when the researchers of Brantley et al.6 had
enough data to evaluate the statistics of the data set. The data used for Brantley et al., 2014
resulted from the PSG calculations and were pre-screened using the following 3 DQI criteria:
Peak concentrations that corresponded with wind directions that were within ฑ 30ฐ of the
source direction
An average in-plume concentration greater than 0.1 ppm
A Gaussian fit with an R2 > 0.80.
The filtered data set was then subjected to multivariate linear regression analysis and comparisons
with previous, onsite, direct measurement studies.
The Pearson correlation coefficients shown in Table 3-20 represent the results of the multivariate
regression analysis. Here, the amount of correlation in the data set is shown as a fractional number,
with +1.00 being the highest amount of positive correlation possible and -1.00 indicating the most
negatively correlating variables (where, if one goes up, the other goes down at a rate that is
inversely proportional). As illustrated in Table 3-20, the most correlating variable to methane
emissions is the amount of gas production per day, and it is not a relatively strong correlation with a
coefficient of only 0.29, accounting for only 10% of the emissions data set variation. The negative
correlation of methane emissions with mean facility age indicates that the amount of methane
emissions measured at a facility should decrease as the facility age increases. However, with only a
correlation coefficient of -0.20, the age variable of a facility is not expected to predict methane
emissions.6
-------
ORS Handbook
Section 3.0
Page 3-85
Table 3.20. Pearson Correlation Coefficients of Emission and Production.
Methane Emissions
(Mscf/day)
Gas Production
(Mscf/day)
Hydrocarbon
Liquids Production
(bbl/day)
Water Production
(bbl/day)
Methane Emission
(Mscf/day)
1.00
Gas Production
(Mscf/day)
0.29
1.00
Hydrocarbon
Liquids Production
(bbl/day)
-0.01
0.44
1.00
Water Production
(bbl/day)
0.22
0.77
0.40
1.00
Mean Age (years)
-0.20
-0.59
-0.34
-0.57
In relation to direct, onsite, measurements, the data captured from OTM 33A appears to only
represent the leaks with the highest emission rates. For the study comparison, Brantley et al.6
compared OTM 33A field results with those from Allen et al. (2013)10 and ERG (2011),11 where the
methane emissions were measured using three very different approaches. The results from the
study comparison are shown in Figures 3.26 and 3.27, where the ERG study captured more of the
very low leak rates relative to the other studies, and the OTM 33A study captured more of the
higher leak rates.6
Although the studies collected measurements from different basins during different years (the OTM
33A study used data from 2010 to 2013) and from very different facility populations, some general
observations are still possible. For example, the ERG study data in Figure 3.26 was conducted in the
Barnett Basin and exhibits many more measurements that are well below the whiskers of the box
plot (the boxes represent 1st and 3rd quartiles of the data and the whiskers extend to 1.5 times the
interquartile range; the black circles and bars indicate the data means and 95% confidence intervals,
respectively). The data mean for the ERG study is also much lower, accounting for the more
measurements collected for leaks of smaller emission rates. Whereas, the OTM 33A study was only
able to capture the larger emission rates at the farther sampling location. As discovered by Thoma et
al. (2012),5 each basin should have a distinct emission rate relative to its production type and region;
however, the comparisons between similar facilities in the same basin should be similar.
-------
ORS Handbook
Section 3.0
Page 3-86
100.0000 ฆ
(/?
O)
ซ 1.0000-
V)
V
CD
CL
"g 0.0100-
E
UJ
x 0.0001 -
n = 43
Allen etal. (2013)
0.1<
n = 58 n = 21
0.13*
0A5\
n = 17
0.03f
ERG (2011)
n = 295
0.14*
Appalachian Gulf Coast MidcontinentRocky Mountain Barnett
Basin
This study (OTM 33A)
n = 43 " = 74 n - 107
- I
J.
0.33t
1.14*
0.59'
1 1 1
Barnett DJ Pinedale
Figure 3-26. Comparison of methane measurements by basin.
Allen et al. (2013) ~
ERG (201 1 )
This study (OTM 33A)
CL>
IQ
CL>
~o3
0.6 -
7ฑ=r 0.4 -
en
0.2 -
O.O -
/V
1 -OO
CD
ง 0.75
Q
0.50 -
=5 0.25
O.OO
0.001 0.010 0.100 1.000 10.000
(b)
0.001 0.010 0.100 1.000 10.000
CH4 Emitted Per Site (g/s)
Figure 3-27. Density (a) and cumulative density (b) of methane emission measurements.
Figure 3.27 elucidates the potential reason for the discrepancy between the two Barnett Basin data
sets. In panel (a), the ERG study represents many more measurements that are lower in emission
-------
ORS Handbook
Section 3.0
Page 3-87
rate as seen by the "shoulder" in the left-hand part of the ERG line. The OTM 33A study used a DQI
filter of only measurements at a concentration > 0.1 ppm, which limits the method to about 0.010
g/s compared to the < 0.0001 g/s measurement limit for the onsite sampling approaches. When
looking at the cumulative density, the OTM 33A measurements do not start until a higher emission
rate and do not peak until the very upper end of the measurement distribution. Therefore, Brantley
et al. (2014)6 not only revealed that OTM 33A captures the emissions from higher rate leaks, but also
observed that many of the onsite measurements did not or were unable to represent these larger
leaks in their data sets. Most likely, these leak omissions were due to the presence of stochastic
emission events (such as tank flashing and malfunctioning equipment), which are not easily
measured using onsite techniques.
Brantley et al. (2015)7 continued analysis of OTM 33A measurements by investigating the
concentration and composition of VOCs and HAPs collected using OTM 33A versus those measured
onsite. During stationary measurements, when the GMAP vehicle was in the peak of the emission
plume as indicated by the real-time methane concentration measurements, the GMAP vehicle
operator initiated sample acquisition with a 30-second evacuated canister grab sample. After
applying the same DQI criteria as for the previous study (Brantley et al., 20146), the PSG method was
used to estimate methane emission rates from the associated 20-minute OTM 33A sample and used
the following ratio calculation to estimate the canister compound emission rate:
F= = [ (C= to) /Co] [Mc/Mo]
Where:
Fc = Emission rate estimate of canister compound (g/s)
Cc = Measured concentration of canister compound (ppb)
F0 = Emission rate estimate of methane using PSG (g/s)
Co = Methane concentration measured from the canister sample that is above
-------
ORS Handbook
Section 3.0
Page 3-88
background (ppb)
Mc = Molecular weight of the canister compound (g/mol)
Mo - Molecular weight of methane (g/mol)
Unexpectedly, this study discovered that the high-volume sampler (HVS) used for the onsite
measurements has a now-documented (Howard et al., 2G15)9 malfunction when transitioning from
lower to higher measurement categories. This is evidenced in Figure 3.28 where the measurements
between the HVS (Hi-Flow) and OTM 33A (canister) are well correlated at concentrations of up to
about 10.5% hydrocarbons (HC)/
EPA Onsite (All)
EPA Onsite (<1Q.5%)
60-
10.0-
a Non-tank
o Tank
7.5-
40-
LL
5.0-
20
O o.
2.5-
0.0-
0.0
2.5
5.0
HC % Canister
7.5
10.0
HC % Canister
Figure 3-28. Comparison of onsite measurements and remote (canister) measurements. The solid black
line represents y =x.
The presence of flash emissions during this study seem to correlate with higher methane, VOC, and
HAP emissions and highlights the need for a cost-effective method that can identify significant leaks
that have a larger amount of temporal variability. Brantley et al. (2015)7 also compared the VOC and
HAP emissions estimates from the canister samples to those modeled by the API E&P TANKS
potential-to-emit estimates. By comparing the measured and modeled values, Brantley et al,
discovered that the OTM 33A remote sampling approach can be used to identify situations where
-------
ORS Handbook
Section 3.0
Page 3-89
facility emissions are not being effectively controlled to the 95% control level. The overall results of
this study show that the OTM 33A approach can be used as a remote inspection technique to survey
facilities for large and/or intermittent fugitive emissions.7
Recently, a study by Foster-Wittig et al. (2015)8 took a closer look at the methane emission estimate
calculations by comparing PSG results with those from plume geometry inverse calculation
approaches based on a wind field data model and plume geometry reconstruction from angle-
resolved concentration data. The modeled and reconstructed calculation approaches attempt to
produce more robust measurement results by calculating atmospheric dispersion and plume
geometry in more detail than what is represented using the atmospheric stability look-up tables
associated with the more simplistic PSG method. Controlled release data was filtered using the
following DQIs:
Data were acquired during a mostly neutral atmosphere (the absolute value of stability-
source height divided by source distanceis less than 0.2).
Measurements where the average concentration sorted by corresponding wind direction
measurements were greater than 0.3 ppm for any wind angle category.
The standard deviation of the wind direction over the measurement duration is greater
than 20ฐ to allow full plume capture.
The mean wind direction over the measurement duration is within ฑ 50ฐ of the
measurement location direction relative to the plume source.
After DQI data filtering, 41 of the original 106 measurements were calculated using the modeled and
reconstructed methods and compared to the PSG results.8 Figure 3.29 illustrates these results,
where the shaded bands indicate the controlled methane release amount with a ฑ 5% release rate
accuracy.8 In this figure, the circles represent the estimated emission rate as calculated by the
modeled atmospheric dispersion approach (Sm), the asterisk data points represent the
reconstructed geometry approach (Sr), and the triangles represent the conventional PSG estimates
(Se). Similarly, the authors compared the calculated emission values average by study in Figure 3.30.8
-------
ORS Handbook
Section 3.0
Page 3-90
S0 ฑ 5%
00 0.6
4
File#
Figure 3-29. Source emission strength calculations for each controlled release.
1.4
1.2
ii 0.8
s
m o.6
0.4
0,2
0SM
- SR
ASn
S0 ฑ 5%
o
A - ฃ-
*
5 A
i
* 4 8 *
ฎ A
*
8
ฃ
A * A .
0123456789
Study
10 11 12 13 14 15
Figure 3-30. source strength calculation averages by controlled release study.
Note for Figure 3.30 that only studies 3, 4, 5, 6, 7, and 14 along the x-axis have three or more
measurements in their averages.8 Each study average represents at least one measurement and can
have up to seven measurements. The amount of error represented by the data in Figures 3.30 and
3.31 for each calculation approach is -47 to 27%, -39 to 29%, and -42 to 158% for the modeled,
reconstructed, and PSG results, respectively.8 Overall, the mean percent error for the modeled,
-------
ORS Handbook
Section 3.0
Page 3-91
reconstructed, and PSG results are -5%, -2%, and 6% with a standard deviation of 29%, 25%, and
37%, respectively. The mean percent error results show that the modeled and reconstructed
measurements tend to underestimate the release rate whereas the PSG measurements tend to
overestimate the release rate; however, it was observed that the largest error occurs when there
are less than three measurements available for the study average, indicating that multiple repeated
measurements with help to reduce the overall error in the source emission rate estimate. In
conclusion, although the PSG calculation approach yields higher standard deviations versus more
complicated calculation approaches, it will still provide a good estimation of source emission rates
that become more accurate with repeat measurements.8
Typical QA/QC
The OTM 33A approach involves many potential stages to application and, therefore, can be subject
to multiple stages of QA/QC related to site characteristics, measurement execution, and acquired
data quality. The facility type and expected emitted compounds must be able to be detected by the
GMAP vehicle equipment, depending on the real-time concentration measurement instrument
incorporated in the GMAP vehicle design. Because OTM 33A is a mobile approach, the site must
have road access proximal to the target equipment (within 20 to 200 m), there must be a clear, open
area between the potential source and the GMAP vehicle sample inlet, and the atmospheric
conditions must be such that the plume is transported from the source to the GMAP vehicle with
minimal uplift or dispersion (typically sustained wind speeds should exceed 2.0 m/s and the
atmospheric stability indicator level of 3 or greater).2 Repeated transection during mobile
measurements or samples during stationary measurements are important to determine temporal
variability and to provide confidence in the overall OTM 33A assessment. The application of auxiliary
data, such as infrared camera videos, help to verify or pinpoint the source location and reduce the
overall amount of measurement error. The number of repeated measurements is generally
proportional to the importance of the study data, the temporal characteristics (constancy) of the
emission, and the study measurement accuracy objectives.2
-------
ORS Handbook
Section 3.0
Page 3-92
The GMAP vehicle must be outfitted with batteries such that the OTM 33A equipment can perform
measurements while the vehicle is off, eliminating measurement interference from unintended
sampling of vehicle emissions. The real-time concentration measurement instrument should have a
time resolution of 1 Hz or better (about 10 Hz for the wind-field measurements), be within ฑ 10% of
the actual concentration during an onsite accuracy test, and can resolve a sustained plume six times
the standard deviation of a baseline (background) concentration level.2
The OTM 33 method series includes sub-methods that operate like the general OTM 33 design. As
such, the general OA elements discussed in OTM 33 also apply to applications of OTM 33A and
reader should be aware of these additional requirements. The OTM 33 sub-methods also allow for
some flexibility in the choice and design of the GMAP vehicle equipment and application; therefore,
detailed QAPPs that include discussion on engineering design, field application, procedures for
operation and calibration and site-specific analysis of method applicability, potential interferences,
measurement data quality objectives (DQOs) and DQIs are required.2 An effective QAPP lists the
measurement objectives, the intended use of the data, and the measurement error tolerances for
the project through the definition of DQOs and infield and analysis DQIs. The project measurement
objectives are defined in the DQOs, and the monitoring of field operations and performance against
the DQOs are executed through DQIs. For example, to perform EQ measurements, the
transportation pathway of the plume must be free from obstruction (DQO). If there are any trees or
manmade structures in the near-field between the source and the GMAP vehicle, then the plume
geometry will be affected and enforcing a DQI where the measurements must show a Gaussian
distribution with an R2 > 0.80 when sorted into discrete bins based on wind direction can cause the
measurement to fail the QA requirements.
The flexibility of the OTM 33A approach allows for a variety of GMAP vehicle engineering designs
and equipment installed. As mentioned in the studies reviewed in Section 3.7.2, there are many in-
field DQIs that can be developed to monitor data quality and that may be applicable to all OTM 33A
configurations, such as:
-------
ORS Handbook
Section 3.0
Page 3-93
Peak methane concentration within ฑ 30ฐ of the source direction
Average, in-plume concentration greater than 0.1 ppm
Gaussian fit with an R2 > 0.80 for data binned by wind direction
Standard deviation of the wind direction greater than 20ฐ and less than 60ฐ
Sufficient data acquisition rate as evidenced by the number of data points recorded per
measurement duration.
Whenever deploying a newly developed GMAP vehicle for field use, it is good practice to validate
the system for remote measurements by performing technique performance testing using a
simulated leak source in a realistic environment (i.e., controlled release test). This primary QA tool
helps to evaluate the system performance and develop in-field DQIs before the system is deployed
for field use. With a combination of pre-deployment approach validation, in-field DQIs, and post-
acquisition QA analysis procedures, in addition to proper operation of equipment and application of
techniques, the OTM 33A method can provide an estimation of source emission strength from
remote locations.2
Siting Concerns
Many of the siting concerns associated with the application of the OTM 33A approach are discussed
in the interferences section (Section 4) of the draft OTM 33A method. Because OTM 33A is a remote
measurements method, these interferences are associated with the height of the emission source,
the distance of the GMAP vehicle sample inlet to the emission source, the orientation of the GMAP
vehicle inlet in relation to the emission source and predominant wind direction, and the presence of
any near-field obstructions to the expected transportation of the emission plume to the GMAP
vehicle sample inlet.2
At least three background source factors should be considered when making methane
measurements from near-ground level, proximate sources: (1) potential for interference from
mobile sources, (2) potential for interference from nearby methane sources, and (3) potential for
interference from far away methane sources.2 To mitigate the first potential interference, local
-------
ORS Handbook
Section 3.0
Page 3-94
traffic patterns immediately proximal to the measurement route should be noted, the GMAP vehicle
sampling inlet should be located in the front portion of the vehicle, and the vehicle must be turned
off during stationary measurements. Making repeated mobile measurements around all aspects of
the target facility help to elucidate multiple potential sources and/or far away background
interferences, thus reducing the potential for the second and third categories of potential
interferences. Auxiliary data such as site photos and infrared camera videos can help to verify or
pinpoint the emission source and evaluate the upwind background for any potential interferences.
Far away background sources should be stable enough to be considered a part of the background
signal. However, if a background source is too close to the target equipment, the interference will
show up in the variance of the 5% background determination used for the PSG calculation DQI
analysis.2
When the GMAP vehicle is positioned for stationary measurements, the operator must use the real-
time concentration display to ensure that stationary sampling occurs at the point of the
predominant wind direction relative to the source locationthis means finding the predominant
area of peak concentration along the mobile measurement pathway. Once the location of
predominant peak concentration is determined, the GMAP vehicle must be in proper orientation to
avoid causing channeling interferences. Typically, this involves positioning the GMAP vehicle such
that the sampling inlet is facing upwind towards the emission source and there are no obstructions
from the vehicle between the inlet and the source location.
The site characteristics and expected emission source locations should also be considered during the
pre-field assessment. For example, eddies and potential recirculation of emissions near to a tank
source could overestimate the emission rate if the sampling location is not at least five tank
diameters downwind. Other near-field obstructions may exist that will obscure the general
transportation of the emission plume, such as a tank battery or other man-made structures. The use
of an infrared camera can be helpful to positively identify the emission source so that an evaluation
of any near-field obstructions can be made and mitigated, if possible.2
-------
ORS Handbook
Section 3.0
Page 3-95
Strengths and Limitations
OTM 33A is a fast-deployable inspection approach that can be used to make source emission
estimates without requiring onsite access or complex modeling. However, because the
measurements are taken remotely, the representativeness of the emission estimates will heavily
rely on a limited range of appropriate atmospheric conditions and an unobstructed transport
pathway for the plume to travel from the source to the sampling location. Tables 3.XX3 and 3.XX4
list the general strengths and limitations of the approach, respectively.
Table 3-21. Strengths of the OTM 33A Approach
Feature
OTM 33A Strengths
Flexible application
Is appropriate for a variety of applications with
only simple adjustments.
Good inspection approach
Individual emission rates are calculated within
60% accuracy, improving with each replicate
measurement.
Portable instrumentation
Field units are rugged with a high temporal
resolution for real-time analysis and fast
deployment.
Site access not required
OTM 33A does not require access to the
surveyed site.
-------
ORS Handbook
Section 3.0
Page 3-96
Table 3-22. Limitations of the OTM 33A Approach
Feature
OTM 33A Limitations
Meteorological concerns
Sustained and varying wind conditions are
required to transport the emission plume.
Susceptible to atmospheric stability
Atmosphere must be stable, but not overly
stable.
Absence of near-field obstructions
Obstructions to proper plume transportation
(such as trees, fences, etc.) can cause
inaccurate estimates.
Logistical concerns
Location and the availability of roads that are
perpendicular to the plume create difficulties.
Distance from emission source
Measurements must be made between about
10 and 200 m. Accurate distance
measurements are required for the emission
estimation.
References
1. U.S. Environmental Protection Agency (EPA). 2014a. Draft "Other Test Method" OTM 33
Geospatial Measurement of Air Pollution, Remote Emissions Quantification (GMAP-REQ).
Available online at http://www.epa.gov/ttn/emc/prelim.html.
2. U.S. Environmental Protection Agency (EPA). 2014b. Draft "Other Test Method" OTM 33A
Geospatial Measurement of Air Pollution, Remote Emissions Quantification - Direct Assessment
(GMAP-REQ-DA). Available online at http://www.epa.gov/ttn/emc/prelim.html.
3. Thoma, E.B., H. Brantley, B. Squier, J. DeWees, R. Segall, and R. Merrill. 2015. Development of
-------
ORS Handbook
Section 3.0
Page 3-97
Mobile Measurement Method Series OTM 33. Proceedings of the 108th Annual Conference &
Exhibition of the Air & Waste Management Association, Raleigh, NC, June 22-25, 2015.
4. Thoma, E.B., B.A. Mitchell, B.C. Squier, J.M. DeWees, R.R. Segall, C. Beeler, M.T. Modrak, M.S.
Amin, A.B. Shah, C.W. Rella, and R.L. Apodaca. 2010. Detection and Quantification of Fugitive
Emissions from Colorado Oil and Gas Production Operations Using Remote Monitoring.
Proceedings of the 103rd Annual Conference & Exhibition of the Air & Waste Management
Association, Calgary, Alberta, Canada, June 22-25, 2010.
5. Thoma, E.D., B.C. Squier, D. Olson, A.P. Eisele, J.M. DeWees, R.R. Segall, M.S. Amin, and M.T.
Modrak. 2012. Assessment of Methane and VOC Emissions from Select Upstream Oil and Gas
Production Operations Using Remote Measurements, Interim Report on Recent Survey Studies.
Proceedings of the 105th Annual Conference of the Air & Waste Management Association, San
Antonio, TX, June 19-22, 2012.
6. Brantley, H.L., E.D. Thoma, W.C. Squier, B.B. Guven, and D. Lyon. 2014. Assessment of Methane
Emissions from Oil and Gas Production Pads using Mobile Measurements. Environmental Science
& Technology, 48,14508 - 14515.
7. Brantley, H.L., E.D. Thoma, and A.P. Eisele. 2015. Assessment of Volatile Organic Compound and
Hazardous Air Pollutant Emissions from Oil and Natural Gas Well Pads using Mobile Remote and
On-site Direct Measurements. Journal of the Air & Waste Management Association, 65:9,1072-
1082.
8. Foster-Wittig, T.A., E.D. Thoma, and J.D. Albertson. 2015. Estimation of Point Source Fugitive
Emission Rates from a Single Sensor Time Series: A Conditionally-sampled Gaussian Plume
Reconstruction. Atmospheric Environment, 115,101 - 109.
9. Howard, T., F. Thomas, and A. Townsend-Small. 2015. Sensor transition failure in the high-
volume sampler: Implications for methane emission inventories of natural gas infrastructure.
Journal of the Air & Waste Management Association, 65:7, 856-862.
10. Allen, D. T.; Torres, V. M.; Thomas, J.; Sullivan, D. W.; Harrison, M.; Hendler, A.; Herndon, S. C.;
Kolb, C. E.; Fraser, M. P.; Hill, A. D. 2013. Measurements of methane emissions at natural gas
production sites in the United States. Proceedings of the National Academy of Sciences of the
U.S.A., 110 (44), 17768-17773
11. ERG. 2011. City of Fort Worth Natural Gas Air Quality Study Final Report; Fort Worth, TX.
Available online at http://fortworthtexas.gov/gaswells/?id=87074.
-------
ORS Handbook
Section 3.0
Page 3-98
3.8 Hyperspectral Imaging
Hyperspectral imaging is a powerful measurement technique that produces a visual representation
of a typically invisible fugitive gas plume and provide spectral information of that gas plume for
chemical identification and quantification. Any remote sensing system is a hyperspectral imager if it
collects spectral information for each pixel in an image and enables that information to be
processed to help identify and measure objects in the scene.
Hyperspectral imaging developed alongside aeronautical advances in the defensive and space
exploration programs starting around 1960 with multispectral imaging, and experienced a significant
increase in development and application after the arrival of 2-dimensional (2D) charge-coupled
device (CCD) detector arrays in the 1980s.1'2 The technique is popular with agricultural,
astronomical, biomedical, geological, and geospatial (land-use, land-cover) applications and has
recently been deployed for environmental purposes. This section discusses specifics related to
hyperspectral imaging for the detection and measurement of volatile organic compounds (VOC) and
hydrocarbon gases for environmental monitoring of industrial locations.
Hyperspectral imaging is typically conducted in the infrared (IR) region of the electromagnetic
spectrum using IR sensing principles and can oftentimes include multi- or ultra-spectral imaging.
Multispectral, hyperspectral, and ultra-spectral imaging operate by essentially the same principles,
the main difference being the spectral resolution and the amount of spectral information collected
as shown in Figure 3.31.2
Hyperspectral imaging is slightly different from multispectral imaging by the amount of wavelength
data captured at any given moment. With multispectral imaging, about 3 to 10 discrete wave bands
are analyzed with bandwidths commonly between 50 to 120 or more nanometers (nm), whereas
hyperspectral imaging collects information from hundreds of wave bands that are smaller in width
(typically 1 to 15 nm) and, therefore, produces a higher resolution.1 Figure 3.313 shows that there
can even be ultra-spectral measurements in which thousands of wave bands are evaluated;
however, this is not currently used for gas detection and is therefore beyond the scope of this
-------
ORS Handbook
Section 3.0
Page 3-99
document. IR detection of hydrocarbon gases is typically conducted using the mid-wave IR (MWIR)
region of 3 to 5 micrometers (|am) or the longwave IR (LWIR) region of 8 to 12 |am due to the
presence of an atmospheric transmission window (a region in the electromagnetic spectrum where
common atmospheric gases such as water vapor and carbon dioxide are not very
electromagnetically active). The shortwave IR (SWIR) region is not typically used due to the lack of
electromagnetic activity for VOC and hydrocarbon gases in that region.
Broadband
SWIR
LWIR
Multispectral
Band
Band
Band
Band
Band
Band
1
2
3
4
5
7
ฆ45-.52
52-.60
.63-.69
ฆ79-.90
1.55-1.75
2.08-2.35
Band
6
10.4-12.4
Hyperspectral
Ultra spectra I
100s of Bands
1000s of Bands
Figure 3.31. Example of Differences between MultispectralHyperspectral, and Ultra spectral Data
Resolution.
These different spectral resolutions fall under the banner of hyperspectral imaging because, for as
many applications that exist for the technology, so too do many systems exist that acquire
hyperspectral information. These systems are described and discussed in the following section.
General Description of Approach
Hyperspectral imaging is a passive optical remote sensing technique that both provides a visual
display of the measurement area and spectral information for each pixel of the image. The resulting
data are therefore 3-dimensional (3D) "datacubes," where the (x,y) coordinates correspond to the
2D image values and wavelength (A) is the spectral information in the third dimension (on the z-
axis). Hyperspectral imaging is also a radiometric method, meaning that the measurement method
-------
ORS Handbook
Section 3.0
Page 3-100
is based on the thermal properties of the elements in the field-of-view.
Operating Principles
As a radiometric method, the radiation perceived by the hyperspectral imaging device can be
simplified into the radiative transfer model depicted in Figure 3.32.8 For IR hyperspectral imaging
devices discussed in this section, the spectral radiance reaching the device can be expressed as:
L1 = (1 T1)B1 + T-[ [(1 T2)ฃป2 + r2^3J
Where:
Li = Spectral radiance reaching the hyperspectral imaging device
t
1 = The transmittance of layer 1
T
2 = The transmittance of layer 2
Bi = The Planck function for a blackbody evaluated at layer 1
B2 = The Planck function for a blackbody evaluated at layer 2
Lj3 = Radiance entering the gas cloud from the background
Layer 1
Spectrometer
Layer 2
'$ w
Cloud
1 ^ Atmosphere
ฆ .. . ,
ฆ Background
Figure 3.32. Radiative Transfer Model Illustration for IR Remote Sensing of Airborne Pollutants.
-------
ORS Handbook
Section 3.0
Page 3-101
Many variations of hyperspectral imagers exist, but they all basically operate on the radiative
transfer principle. In most designs, the instrument detects the spectral difference between the
original background radiation and the radiation after it has passed through a gas cloud. Typically, the
magnitude of this difference is converted into chemical species concentration using the Beer-
Lambert Law, which relates the amount of transmittance of a material to its optical depth
(concentration) and absorbance:
A 8 * C * I
Where:
A = Absorbance intensity (transmission = initial radiation - absorbance)
s = Absorption coefficient of the pollutant
c = Pollutant concentration
1 = Measurement path length through the plume
Although a wide variety of system designs exist, most hyperspectral imagers fit into one of two
categories: IR cameras with modified optics and imaging Fourier transform instruments (FTIR).5
Occasionally, these two types of technology are combined. The basic operation principles of these
types of technology (IR cameras and FTIR) can be found in Chapters 2.1 and 2.5 of this handbook,
respectively.
Hyperspectral imagers based on IR camera technology will have additional filtering systems such as
a Fabry-Perot etalon or an internal filter wheel. For example, the "Second Sight MS" system by
Bertin Technologies uses luminance differentials between various optical cut-off filters to acquire
multispectral information using an IR camera.6
-------
ORS Handbook
Section 3.0
Page 3-102
Example Instruments
Bertin's Second Sight MS System
The Second Sight MS system operates by cycling through a specially designed filter wheel that
contains one reference filter and five active filters to limit the spectral range of thermal radiation
reaching the camera's detector. The scene information for each filter is captured sequentially in
time to a 2D focal plane array (FPA) detector and then processed to subtract the background (or
reference filter data) from the target (active filter) results. Figure 3.336 is an example schematic
where the dark blue line represents the incident radiation being transmitted from the imaged scene
to the instrument and the reference filter allows the spectral data from about 9.5 |am and greater to
reach the camera's detector (purple line). Active Filter 2 (orange line) captures the spectral data to
almost 8 |am, and Active Filter 3 (green line) captures data to about 7 |am, which also includes the
area of absorption for methane shown by the dip in the dark blue line. When the spectral luminance
data from the reference filter and/or Active Filter 2 are subtracted from the data captured by Active
Filter 1, pixel by pixel, the image of the methane gas plume is isolated (the 7 to 8.3 |am region) and
the plume composition can be identified. The resulting data from the Second Sight MS system is,
therefore, not highly resolved spectrally but covers band segments and is a multispectral technique.
-------
ORS Handbook
Section 3.0
Page 3-103
Active
Filter 1
Active
Filter 2
Reference
Filter
Methane Gas
6 7 8 9 10 11 12 13 14 15
Wfevelength (micrometer)
Figure 3.33. Example of Multispectral Imaging using Spectral Differentiation of Methane Gas.
Physical Sciences' AIRIS
The Adaptive IrifraRed Imaging Spectroradiometer (AIRIS) developed by Physical Sciences, Inc. is an
IR camera that is fitted with a Fabry-Perot interferometer on the front (Figure 3.34).5 The
interferometer is tunable and can therefore select the wavelength of interest to reach the camera's
detector. The MWIR model operates in the range of 3-5 |im and the LWIR model operates between
8-12 jim. Both models have a quick tuning velocity of 10-20 milliseconds (ms). The selected
wavelength radiation enters through the camera's optics and is detected by a mercury-cadmium-
telluride (HgCdTe) FPA, as seen with other, conventional IR cameras.'
-------
ORS Handbook
Section 3.0
Page 3-104
foterferameter IR
Lenses
Figure 3.34. AIRIS Interferometer Design where the Interferometer Lens (a) is Placed on the Front of the IR
Camera (b).
Rebellion Photonics' Gas Cloud Imager
Rebellion Photonics, Inc. added subdivisional mirrors, an array of spectroscopes with dispersive
elements, and a recombination scheme to develop their datacube in a system they call an image
mapping spectrometer (IMS).9
Once the radiation enters through the lens optics, it encounters a mirror (called the "mapping
mirror" in Figure 3.35) with a multifaceted surface that divides the image into numerous strip
images. The mapping mirror directs each collection of strip images into its respective spectroscope
(prism and lenslet arrays in Figure 3,35),9 which disperses the bandwidths so that each bandwidth
for each strip image is detected by a singular, coupled Sofradir microbolometer detector array. The
2D detector array image of multiple strip images are then recombined into the 3D datacube by a
remapping algorithm. An example of the recombined image is provided in Figure 3.XX6.
-------
ORS Handbook
Section 3.0
Page 3-105
imaging lens
mapping
mirror
Figure 3.35. Illustration of the IMS Optical Layout.
The resulting image in Figure 3.369 panel (a) was first divided into strip images. Those strip images
were reflected into the spectroscope array where they were further divided by wavebands. The strip
images of each waveband were recorded by the detector array and recombined using a remapping
algorithm. Panels (b) through (e) in Figure 3.36 represent each waveband result after the
recombination of the corresponding strip images. In this image, the four wavebands are spaced
apart at (b) 463 nm, (c) 523 nm, (d) 595 nm, and (e) 622 nm. The IMS instrument typically has 8 to
20 spectral channels available, indicating that this is a multispectral technique.
-------
ORS Handbook
Section 3.0
Page 3-106
1
\
W %
Ik x,*
(c)
(e)
\ _ k
^ ^ V
(a)
1J
TV C j "
H \
(b)
(d)
Figure 3.36. Example Datacube (a) with Waveband Elements at (b) 463 nm, (c) 523 nm, (d) 595 rim, and
(e) 622 nm.
Telops' HyperCam and MS-IR
A true imaging Fourier transform spectrometer (i'FTS) was developed by Telops, Inc. over a period of
about 20 years. The current name for this iFTS system is the HyperCam (Figure 3.37, left panel).12,14
The HyperCam uses a Michelson interferometer to modulate the optical radiance allowed to pass to
the FPA detector (see Section 2.1 for more detail on FTiR), The wave form of the radiance exiting the
interferometer plotted over the scan location (optical path difference (OPD) value of the
interferometer range) yields an interferogram. The interferometer conducts a single scan in time at
one OPD setting for all pixels. The interferometer then moves to the next OPD setting and records
for all pixels again. This is repeated for the full range of the interferometer so that a full
interferogram is constructed for each pixel. The pixel interferograms are then mathematically
converted from time-domain information to wavelength-domain information using Fourier
transform algorithms and the spectrum for each pixel is thus developed. False color identification of
plumes is produced by the Telops software highlighting the regions of gas cloud detection over a
scene image that comprises the total radiance broadband information for each pixel from the
datacube (Figure 3.38).10
-------
ORS Handbook
Section 3.0
Page 3-107
Beamsplitter
Control
Electronics
Actuator
Retro-Reflactor
Detector Lens
Focal Plane Array
i
Data Output
Figure 3.37. Telops HyperCam (Left) Hyperspectral Imager and MS-IR Infrared Camera (Right)
Multispectral Imager.
a Telops Reveal D&I
Fie Ytevi Scenario Tools 'A'reiovi rtetp
Q OpenSซfiSno start Standby ฃ Badsgrouid
Delecbon view
Figure 3.38. Example Telops Software Output Where the Detection View is the Total Radiance with the
Gas Cloud Highlighted in False Color.
-------
ORS Handbook
Section 3.0
Page 3-108
Depending on the spectral resolution selected by the operator, the HyperCam can capture up to 320
wavelengths per pixel (the maximum for the size of the FPA), thus making this system a true
hyperspectral imager. Telops has also developed a multispectral imager similar to the Second Sight
system, where an IR camera has an added filter wheel to collect information at from up to 8
different wavelength regions (Figure 3.37, right panel).12
Additional Considerations
How hyperspectral systems develop the resulting datacube is a key characteristic in the instrument's
design. There are two main methods for acquiring a datacube: either by collecting 2D cube slices in
rapid succession over time (scanning), or by collecting all information simultaneously and dividing it
onto 2D elements to be recombined into a cube with post processing (snapshot). The scanning
method is shown in panel (a) of Figure 3.39 as the "pushbroom spectrometer" and the snapshot
method is represented by the "snapshot imaging spectrometer" in panel (b).4
The scanning detection method measures time-sequential 2D slices of the datacube, whereas the
snapshot method takes one instantaneous measurement of the scene and divides the measurement
information into multiple 2D elements on the FPA that can then be recombined into a datacube
(Figure 3.40).2
whiskbroom''
spectrometer
snapshot imaging
spectrometer
filtered
camera
Figure 3.39. Detector Methods for DataCube Acquisition.
-------
ORS Handbook
Section 3.0
Page 3-109
Figure 3.40. Example Snapshot Detection with FPA Divided into smaller Collections.
For the instruments used as examples for system designs, Table 3.23 lists their type of detection.
The "Pushbroom spectrometer" depicted in panel (a) of Figure 3.39 illustrates how a scanning
method would acquire the spectral information for each pixel by evaluating all spectral wavelengths
across the x-axis for one row in the y-axis per scan over time. In Table 3.23, a similar concept is used
to illustrate that the "Infrared Camera" detection scheme in Figure 3.39 is really a pushbroom
spectrometer in the spectral domain (along the A-axis). The designation of "Spectral Pushbroom" in
Table 3.23 therefore indicates that all pixels are evaluated for the same wavelength per instrument
measurement (scan).
Table 3.23. Detection Method by Instrument.
Instrument
Instrument Type
Detection Type
Second Sight TC
Multispectral
Spectral Pushbroom
AIR IS
Hyperspectral
Spectral Pushbroom
Gas Cloud Imager
Multispectral
Snapshot
HyperCam
Hyperspectral
Spectral Pushbroom
MS-IR
Multispectral
Spectral Pushbroom
There are documented advantages to using a snapshot detection method, most notably the higher
light throughput and the simultaneous collection of all datacube information. The snapshot
-------
ORS Handbook
Section 3.0
Page 3-110
technique has had some exposure in the astronomy community, but application has been limited to
coupling with telescopes and very little attention has been given to the technique for other
applications. To understand the advantage of using the snapshot technique, the rest of this section
discusses the number of voxels (the "pixels" of the datacube) illuminated per measurement
integration.4 Simply, the amount of radiation collected through a filtering or scanning system (panel
a in Figure 3.39) is significantly less than that of a full throughput snapshot collection (panel b in
Figure 3.39). The light collection efficiency of a system is related to the number of elements in the
datacube recording illumination for a single measurement integration according to the following
equation:
1
LCE=Ti
Nx,y,X
Where:
LCE = light collection efficiency
Nx,y,\ = number of elements in the scan
Many more snapshot system architectures are reviewed in reference 4. However, limitations in the
manufacturing of large FPAs and precision multi-aperture optical elements have slowed the advance
of this technology for many applications, such as environmental monitoring.4
Verification/Validation Studies
The technology for hyperspectral imaging is still being developed and therefore only a couple of
studies are currently available. The complexity of FTIR instruments requires the regular verification
of proper operation and measurement accuracy.
In June 2015, Rebellion Photonics participated in a validation study hosted by EPA in North
Carolina. Two controlled release platforms were utilized to try and quantify the accuracy of
the Rebellion hyperspectral method while also noting any limitations on its application. The
first release platform had two release positions that varied in height (Figure 3.41) and the
-------
ORS Handbook
Section 3.0
Page 3-111
second was an industrial leak simulator custom-built by Eastern Research Group, Inc. (ERG)
(Figure 3.42).
Figure 3-41. High/Low Controlled Gas Release Platform
-------
ORS Handbook
Section 3.0
Page 3-112
Figure 3-42. ERG Controlled Leak Simulation Platform
The Rebellion Gas Cloud Imager was able to detect the leaks from all platforms; however, the
accuracy from the High/Low EPA platform (left panel of Figure 3-43) was better than that
achieved from the partially obscured ERG platform (right panel of Figure 3-43). The data in
Table 3.XX2 illustrate that the average error for the unobstructed view with the High/Low
platform was about ฑ 10% and about ฑ 31% for the partially obstructed view with the leak
simulation platform.15
-------
ORS Handbook
Section 3.0
Page 3-113
2,500
- 2.000
- 1,500
V
I
s
-1.000
Figure 3-43. Example Releases as Seen from the GCIfor the (Left) High/Low Platform and (Right) Leak
Simulation Platform.
Ta
3le 3.24. Summary of Rebellion GCI EPA Controlled Release Results.
Trial
Test Gas
Viewing
Distance (m)
Release Rate
(g/s)
Measured Rate
(g/s)
Error
Error %
1
Methane
30
0.20
0.193
-0.007
-3.4
2
Methane
30
1.20
1.020
-0.180
-15.0
3
Methane
30
0.02
0.018
-0.002
-9.6
4
Methane
30
0.12
0.123
0,012
10.3
5
Methane
17
0.20
0.207
0.007
3.3
6
Methane
17
1.20
0.983
-0.217
-18.1
7
Methane
17
0.02
0.022
0.002
10.5
8
Methane
17
0,12
0.107
-0.013
-10.5
9
Propane, hidden
16
0.02
0.028
0.008
38.7
10
Propane, hidden
16
0.12
0.088
-0.032
26.7
11
Propane, hidden
16
0.02
0.027
0.007
33.1
12
Propane, hidden
16
0,12
0.084
-0.036
-29.7
13
Methane, hidden
16
0.02
0.028
0.008
39.7
14
Methane, hidden
16
0.12
0.096
-0.024
-20.0
Typical QA/QC
The typical quality assurance/quality control (QA/QC) associated with each hyperspectra!
imager will be dependent on the base technology for that instrument. See Section 2.1.3 for
the typical FTIR QA/QC for the Telops Hypercam or Section 2.5.3 for the typical thermal
r 4,000
I 3,500
- 3,000
2,500
- 2,000 5
3
- 1,500
1.000
- 500
L o
-------
ORS Handbook
Section 3.0
Page 3-114
camera QA/QC for many of the other hyperspectral imagers discussed. As the technology
continues to advance, methods will be developed that contain more specific quality
objectives.
Summary of Thermal IR Camera QA/QC
Thermal IR cameras are less complex and, hence, so are the QA/QC procedures associated with
using these instruments. If the IR camera is used for temperature measurements, then the
instrument should have the temperature calibration settings verified annually by the manufacturer
or other appropriate service. When preparing the instrument at the beginning of the day, the IR
camera must be allowed to cool and reach a sort of thermal equilibrium. The design of an IR camera
will typically not allow the camera to be used until the desired temperature is reached, but it is also
important to allow an additional 10 to 15 minutes after this point to allow for thermal stabilization.
Once the camera indicates it is ready and some additional time has passed to allow for thermal
stabilization, if possible, reset the detector values by performing a non-uniformity correction (NUC)
with the lens cap still in place. When the lens cap is still in position over the lens, the camera should
theoretically see a completely uniform image. The FPA of an IR camera is known to suffer from fixed-
pattern noisewhere there is a spatial non-uniformity in the photo-response of the detectors in the
arrayand so the NUC homogenizes the FPA output by measuring the amount of difference
perceived for each pixel and mathematically correcting the pixel response to appear identical across
the entire array. The fixed-pattern noise phenomena common for FPA responses can drift over time
so it is suggested that a NUC is performed a couple times throughout the imaging duration, if longer
than 1-2 hours.
The IR camera operator may wish to record evidence of proper operation before beginning a survey
if the IR camera results may potentially be required in a legal proceeding. However, this step is not
necessary for the camera preparation and a survey may begin at this time. It was determined that
the gas emission rate of butane from a standard, non-adjustable, BICฎ lighter is constant at 4.00
g/hr, and so this may be used as evidence of proper operation by recording video footage of the
butane release. Similarly, if the IR camera has additional optics that will allow for the measurement
-------
ORS Handbook
Section 3.0
Page 3-115
of a gas release, then the standard BICฎ lighter can be used as a calibration gas check as well.15
Siting Concerns
The siting concerns for hyperspectral imaging techniques are dependent on the technological
design. For example, if the instrument is a modified IR camera, then the siting concerns will be like
those for thermal cameras in Section 2.5 of this document. The same is true for FTIR-based systems;
they will be subject to restrictions like those for the FTIR technology in Section 2.1 of this document.
Regardless of the base technology, performing hyperspectral imaging will absolutely require the
ability to have an unobstructed view of the gas leak emission. As evidenced by the study discussed in
Section 3.8.2 using the Rebellion Photonics Gas Cloud Imager, measurements derived from gas
plumes that are partially hidden from view lead to greater error in the overall measurement.
Therefore, the hyperspectral system must be mobile enough to allow for optimal positioning with
respect to the target plume location and transport direction.
Strengths and Limitations
As mentioned for typical QA/QC and siting concerns, the strengths and limitations for each system
will be influenced by the base technology that the hyperspectral imager is designed around.
However, in general, the hyperspectral imaging technique has specific strengths and limitations as
listed in Table 3.24.
-------
ORS Handbook
Section 3.0
Page 3-116
Table 3.25. Strengths and Limitations of Hyperspectral Imaging.
Strengths
Limitations
Faster and easier to deploy than traditional
active techniques
High cost
Nondestructive acquisition of all spectral data
preserved for post-processing
Can be subject to wind speed and other
meteorological conditions
Results in an image of the pollutant plume for
superior spatial resolution
About one order of magnitude less accurate
than laboratory techniques
Mobile/portable design for better
maneuverability and optimum siting angles
Requires a complete view of the emission
plume
References
1. Hungate, W.S., R. Watkins, and M. Borengasser. 2007. Hyperspectral Remote Sensing: Principles
and Applications. CRC Press, Boca Raton, FL, USA. Print ISBN: 978-1-56670-654-4.
2. Hagen, N., and M.W. Kudenov. 2013. "Review of Snapshot Spectral Imaging Technologies."
Optical Engineering, 52(9). DOI: 10.1117/1.0E.52.9.090901.
3. AltiGator. 2017. Webpage: Multispectral and Hyperspectral Drone Imagery. Last accessed on
February 7, 2017 from http://altigator.com/multispectral-and-hyperspectral-drone-imagerv/.
4. Hagen, N., R.T. Kester, L. Gao, and T.S. Tkaczyk. 2012. "Snapshot Advantage: a Review of the
Light Collection Improvement for Parallel High-Dimensional Measurement Systems." Optical
Engineering, 51(11). D0l:10.1117/1.0E.51.11.111702.
5. Kastek, M., T. Pigtkowski, P. Lagueux, M. Chamberland, and M.A. Gagnon. 2015. "Passive
Optoelectronics Systems For Standoff Gas Detection: Results Of Tests." WIT Transactions on
Ecology and the Environment, 198, 89104. DOI: 10.2495/AIR150081.
6. Naranjo, E., S. Baliga, and P. Bernascolle. 2010. "IR Gas Imaging in an Industrial Setting."
-------
ORS Handbook
Section 3.0
Page 3-117
Proceedings ofSPIE, ThermosenseXXXII, 7661, 76610K. DOI: 10.1117/12.850137.
7. Gittins, C.M., W.J. Marinelli, and J.O. Jensen. 2002. "Remote Sensing and Selective Detection of
Chemical Vapor Plumes by LWIR Imaging Fabry-Perot Spectrometry." Proceedings ofSPIE, 4574,
63-71.
8. Harig, R., M. Grutter, G. Matz, P. Rusch, J. Gerhard. 2007. "Remote Measurement of Emissions
by Scanning Imaging Infrared Spectrometry." CEM2007, 8th International Conference on
Emissions Monitoring, Zurich, 34-39, 2007.
9. Hagen, N., R.T. Kester, and C. Walker. 2012. "Real-time quantitative hydrocarbon gas imaging
with the gas cloud imager (GCI)." Proceedings ofSPIE, 8358. DOI: 10.1117/12.919245.
10. Chamberland, M., and M.-A. Gagnon. 2017. Personal communication.
11. Gagnon, M.-A., J.-P. Gagnon, P. Tremblay, S. Savary, V. Farley, E. Guyot, P. Lagueux, M.
Chamberland, and F. Marcotte. 2016. "Standoff Midwave Infrared Hyperspectral Imaging of Ship
Plumes." Proceedings ofSPIE, 9862. DOI: 10.1117/12.2218643.
12. Telops. 2014. "Application Note: Time-resolved Infrared Multispectral Imaging of Gases." Last
accessed on January, 4, 2017 from http://telops.com/products/multispectral-cameras.
13. Chamberland, M., P. Lagueux, P. Tremblay, S. Savary, M.-A. Gagnon, M. Kastek, T. Pigtkowski,
and R. Dulski. 2014. "Standoff gas detection, identification and quantification with a thermal
hyperspectral imager." WIT Transactions on Ecology and The Environment, 181. DOI:
10.2495/EID140571.
14. Kastek, M., T. Piatkowski, M. Zyczkowski, M. Chamberland, P. Lagueux, and V. Farley. 2013.
"Hyperspectral Imaging Infrared Sensor Used for Environmental Monitoring." Acta Physica
Polonica A, 124(3), 463-467.
15. Footer, T.L. (ERG). 2015. "Draft Technical Support Document: Optical Gas Imaging Protocol (40
CFR 60, Appendix K)." Technical document posted as docket ID: EPA-HQ-OAR-2010-0505-4776.
Last accessed on 1/24/2017 from https://www.regulations.gov/document?D=EPA-HQ-OAR-
2010-0505-4949.
16. Hagen, N., S. Rajaraman, R.P. Mallery, and R.T. Kester. 2016. "Quantifying gas cloud emission
rates with the Gas Cloud Imager." Proceedings of the A&WMA 2016 Air Quality Measurement
Methods and Technology Specialty Conference, March 15, 2016, Chapel Hill, NC.
-------
ORS Handbook
Section 3.0
Page 3-118
3.9 Fenceline Passive Sampling - Method 325 A/B
A variety of industrial facilities, including energy production and refining operations, can emit
greenhouse gases and other pollutants due to equipment leaks and process malfunctions. Of
specific interest is monitoring of volatile organic compounds (VOCs) such as benzene. For example,
in 2015 the Environmental Protection Agency (EPA) amended the petroleum refinery rules (40 CFR
part 63, subpart CC) to add fenceline monitoring provisions for fugitive benzene emissions (80 FR
75178).
One method to monitor fenceline emissions gaining considerable interest is the deployment of
diffusive samplers. The samplers collect gas or vapor pollutants at a controlled rate via physical
process (diffusion) rather than pumped air. Samplers have been developed specifically to collect
emissions of benzene and several other VOCs. Method 325A and 325B (325A/B) were created to
outline proper sampling and analysis, respectively, of passive samplers to help manufacturers
comply with the new ruling.
General Description of Approach
Qualitative versions of diffusive samplers were first introduced in the 1930s. The diffusion and
permeation processes through which the samplers collect pollutants of interest can be described
through derivations of Fick's first law of diffusion:
Where:
J
D
Diffusion flux of amount of substance per unit area per time (mol/m2s)
Diffusion coefficient (m2/s)
Concentration of ideal mixture (mol/m3)
x
Position of the mixture (m)
-------
ORS Handbook
Section 3.0
Page 3-119
The derived expressions from Fick's law relate the sampler's mass uptake to the concentration
gradient, the sample time, and the surface area of the sampler to ambient air.1
Because pollutant concentrations are directly proportional to the sampler's mass uptake, it is
important that the sampling rate is constant and does not change due to the sampler deployment
time or the concentration of pollutant it is exposed to. Concentrations are also positively correlated
to the diffusion coefficient of the pollutant, the time of exposure, and the cross-sectional area of the
diffusion path. Conversely, the pollutant concentration is inversely proportional to the length of the
diffusion path. It is important to note that the sampling rate can be negatively affected as the
sampler's sorbent reaches saturation.
It is generally necessary to expose the sampler over a period of days to collect sufficient
concentrations of the target pollutant for analysis in environmental samples because the sampling
rate is relatively low compared to active (i.e., pumped) methods.
Passive Tube Samplers
The first quantitative application for diffusive samplers occurred in 1973 when E.D Palmes2
developed a tube-type sampler that collected sulfur dioxide.1 Since then, a variety of passive
samplers have been developed for the quantification of a variety of pollutants. Method 325 A/B
uses a tube-type sampler similar to the one described by Palmes. Please see example of a passive
tube design that was proposed for Method 325 A/B in Figure 3-43.3
-------
ORS Handbook
Section 3.0
Page 3-120
Sorb en t Retaining
Gauze
Sampling End
t
Sorbent Bed length DG
(up to 60 mm}
Brass Cap
Figure 3-43. Cross-Section of the PS Tube.
A passive sampler (PS) uses a thermally de-absorbable carbon-based sorbent packed in a stainless-
steel tube (approximately 3.5" long and internal diameter). The cross-sectional area of the tube is
19.6 mm2 with an internal diffusion gap (DG) of 1.5 cm between the diffusion cap and the sampling
end. Each tube is equipped with leak-proof storage caps to prevent contamination during both
storage and transport. Several laboratory equipment manufacturers create tubes appropriate for
Method 325 A/B sampling. Table 3.26 lists the manufacturers that currently sell passive samplers.
Table 3.26. Passive Sampler Manufacturers
Manufacturer
Website Information
Camsco
www.camso.com
Markes
www.markes.com
Sigma-Aldrich
www.sigmaaldrich.com
SKC
www.skcinc.com
Although Method 325 A/B is written specifically for the measurement of fugitive emissions for
-------
ORS Handbook
Section 3.0
Page 3-121
benzene, the PSs are suitable for sampling a variety of pollutants. Each pollutant has a different
diffusion coefficient, therefore the maximum concentration collected over the same time period
may differ significantly between different pollutants. Table 3.27 lists pollutants that can be sampled
using Passive Samplers.
Table 3.27. Pollutants that can be Collected with Passive Diffusive Samplers
Pollutants
1,1-Dichloroethane
Carene
1,1-Dichloroethene
Chlorobenzene
1,1,1-Trichloroethane
Ethylbenzene
l,l,2-Trichloro-l,2,2-
triflouorethane
Labile Sulfur
1,1,2-Trichloroethane
Limonene
1,2-Dichloroethane
p-Dichlorobenzene
1,2-Dichloropropane
Styrene
1,3-Butadiene
Tetrachloroethene
3-Chloroproprene
Tricholoroethene
Benzene
Toluene
bis-Chloromethyl Ether
Xylenes (m,p- and o-)
Carbon Tetrachloride
-------
ORS Handbook
Section 3.0
Page 3-122
Sampler Deployment
Prior to deployment, passive samplers received from the manufacturer must be conditioned as
described in EPA Method 325B, Sections 6.2 and 9.2. This can be accomplished with a dedicated
tube conditioning unit or an analytical thermal desorber (TD) system. The TD system should be used
only if it supports a dedicated tube conditioning mode that allows effluent from the tubes to vent
without passing though the system's sample flow path. Alternatively, if a tube conditioning unit is
used, the unit must be leak-tight and allow for reproducible temperature selection within a 5 ฐC
range that is at least that of the thermal desorber. The unit must also allow inert gas flows up to 100
mL/min.3
Great care must be taken to avoid contaminating the passive tube samplers during transport,
sample deployment, and analysis. Each tube must be capped with brass long-term storage caps that
are fit with polytetrafluoroethylene (PTF) ferrules. While capped, the tubes must be stored in a
clean, air-tight container that is designated for un-sampled passive tubes (un-sampled tubes must
never be stored in the same container as sampled tubes.) Tubes should be handled while wearing
cotton or nitrile gloves to prevent contamination on the surface of the tubes themselves, especially
if the analytical thermal desorption equipment used to extract the tubes does not exclude
contamination from the surface of the samplers.3'4
Upon arrival at the sampling location, the tubes should be allowed to equilibrate with local ambient
conditions for 30 minutes to an hour before removal from their transport container. After inspecting
the tube for any potential damage or contamination sources, the tube should be secured to a pole
or other structure so that the bottom of the sampling cap is approximately 5 to 10 feet above the
ground. The passive tube should be oriented vertically with the sampling end pointing down. To
protect the tube from bad weather, a sorbent tube protection cover, as shown in Figure 3.44/
should be used. The storage cap on the sampling end of the tube must be replaced with a diffusive
-------
ORS Handbook
Section 3.0
Page 3-123
sampling cap arid the start time arid sampling location for the passive sampler must be recorded.
Method 325A recommends a minimum sampling period of 14 days for passive tube samplers.4
Figure 3-44. Sorbent Tube Protection Cover.
Note: Tubes must be labeled for identification purposes; however, many labeling substances may cause
contamination. See EPA Method 325A, Section 8.6.2, for specific information about sampler labeling.
Radio-frequency identification (RFID) labels compatible with many TDs are commercially available and
allow each sample to be programmed with relevant sample information.
Passive Sampler Recovery
After the 14-day sampling period is complete, the diffusion cap must be removed from the PS and
replaced with the storage cap. The storage cap must be tightened sufficiently to prevent further
exposure during transport. The date and time the end cap was replaced must be recorded and the
tube must be placed in an air-tight container that is not used for clean conditioned tubes. If the
tubes are being shipped to the laboratory for analysis, they may be shipped at ambient
Diffusive ^
sorbentTube
Diffusion cap
Weather hood with
tube bracket
-------
ORS Handbook
Section 3.0
Page 3-124
temperature.
Meteorological Data Collection
Time-resolved meteorological data used in conjunction with concentration measurements from the
passive samplers can be used to determine both the source of pollutants and calculating the mass
flux of the plume. Therefore, a meteorological station must be deployed at or near the monitored
facility in locations that are representative of the conditions to which the passive sample is exposed.
These stations must be equipped to monitor ambient temperature, barometric pressure, and wind
speed and direction.4 Specific information about meteorological station siting can be found in
Section 8.3 of EPA Method 325A; more information about meteorological instrumentation can be
found in Meteorological Monitoring Guidance for Regulatory Modeling Applications (EPA-454/R-99-
005).5
Passive Sampler Analysis
Due to the variety of analytical equipment available for this analysis, analysis procedures will vary
between laboratories. For specific procedures, reference EPA Method 325B, Section 11. Presented
here is a general overview of the analysis of PSs. The samples must be analyzed no later than 30
days from the end of sampling to ensure accurate concentration results for benzene. It should be
noted that if quantification is desired for certain compounds that are reactive in nature, analysis
should occur no more than 10 days from the end of sample collection.
The gas chromatographer/mass spectrometer (GC/MS) must be tuned to the manufacturer's
specifications using a 50-ng injection of bromofluorobenzene. Tuning must be completed prior to
the start of each analytical run and every 24 hours thereafter for continuing analysis. Tuning criteria
is summarized in Table 3.28.
-------
ORS Handbook
Section 3.0
Page 3-125
Table 3.28. GC/MS Tuning Criteria.3
Target Mass
Rel. to Mass
Lower Limit %
Upper Limit %
50
95
8
40
75
95
30
66
95
95
100
100
96
95
5
9
173
174
0
2
174
95
50
120
175
174
4
9
176
174
93
101
177
176
5
9
The TD system must also be checked for system integrity and go through additional steps prior to
each analysis. Many commercial TD systems implement these procedures automatically. A leak
check must be performed of the PS tube, the GC carrier gas pressure, and the sample flow path. A
dry purge of the PS may also be performed to remove water vapor and other interferences, and
internal standards must be added to the PS, either automatically or manually. Each sampler is then
pre-purged to remove oxygen to prevent damage to the analytical system.
The TD/GC/MS system must be calibrated using at least five concentrations that represent the
expected monitoring range of the samples for each compound monitored with sampling. PS tubes
prepared as described in Section 10.7 of EPA Method 325B can be used for calibration and may be
stored up to 30 days when refrigerated. The TD/GC/MS system must be calibrated prior to the start
of PS analysis, following any significant instrument maintenance, or if the instrument fails continuing
calibration verification (CCV).
The preferred analytical sequence is detailed in Section 11.1 of EPA Method 325B. If an automated
TD GC/MS, then the end caps of each sample can be removed and loaded into the instrumentation.
However, if a manual system is used, each sample must be uncapped and analyzed individually. A
CCV must be analyzed at the beginning and end of each analysis and every 10 samples throughout.
Step-by-step procedures for operating the TD and the GC-MS are included in Section 11.3 of EPA
Method 325B, although instrument-specific operating manuals should also be used.
-------
ORS Handbook
Section 3.0
Page 3-126
Recordkeeping and Data Analysis
Throughout each step, records must be maintained to both verify a sample's validity and to provide
optimal analytical results for data analysis. The following information should be available for each PS
tube:
The sorbent lot used for each PS, as well as records of tube packing if the tubes are
manually prepared.
Information about the conditioning and blanking of each tube, including the measured
background.
A chain of custody that documents each tube's sampling location and start and end times.
Duplicate copies must be maintained for both field and laboratory staff purposes.
Meteorological data records collected during the sampling period for each sample.
The day the sample was received at the laboratory, as well as the day it was analyzed.
Analytical method data and sample results must be maintained by laboratory personnel.
The calculations required for this method are included in EPA Method 325B, Section 12.2. This
section includes both the calculations necessary to verify calibration performance, as well as the
calculations used to determine the target compound's concentration in micrograms per cubic meter
(|ag/m3). Valid VOC concentrations determined by this method are expected to be between 0.5 and
5.0 |ag/m3, although that is dependent upon such factors as the split ratio used, the dynamic range
and noise provided by the analytical instrumentation, and any interfering background of VOCs on
the PS tube.
Verification/Validation Studies
Flint Hills West Refinery6
At the end of 2008, the EPA was permitted access to Flint Hills West Refinery by Flint Hills Resources.
Production at Flint Hills West Refinery, which refines up to 260,000 barrels per day of crude oil, was
typical during the sampling period described. With the aid of the on-site leak detection and repair
contractor hired by Flint Hills Resources, a total of 26 PS sets were deployed between December 3,
2008 and December 2, 2009. Each PS was exposed for 14 days while attached to the boundary fence
-------
ORS Handbook
Section 3.0
Page 3-127
line at approximately 5 feet above the ground. In addition to the 18 locations along the fence line,
PSs were also deployed at two continuous air monitoring stations (CAMS) south and east of the
facility. The CAMS sites perform automated GC benzene measurements on an hourly interval;
therefore, the PS collection data could be compared to the GC results at both sites. In total, 579
samples were collect at or near Flint Hills West Refinery, including 56 duplicates and 49 field blanks.
Meteorological data were collected at the CAMS site located south of the facility. The wind data
from the site indicates a six-month period of mixed wind directions, as well as another six-month
period where the wind direction is typically from the southeast. These periods of high wind speeds
with a uniform direction can affect comparison analysis results with time-integrated monitors
significantly.
The benzene concentration data collected by the PSs was compared to the 14-day average of the
hourly concentration data collected by the GCs at both CAMS sites. The MDL for the PS was slightly
lower than that of the GC, at 35 parts per trillion by volume (pptv) as opposed to 50 pptv. A linear
regression of PS and GC data correlated well, with an unconstrained R2 value of 0.86 and a slope of
0.90. Unfortunately, due to the distance from the refinery, the range of concentrations collected at
the CAMS sites were not truly representative of the data collected at the fence line sites, and
therefore not optimal for validation of the data collected at those sites.
The average benzene concentration of PSs collected at the facilities fence line was 1,075 pptv, with a
median value of 709 pptv over a range of 122 to 29,280 pptv. Duplicate samples were collected at
two sites for the entire study period and a third duplicate site was added towards the end of the
study. This data also correlated well, with an average relative percent difference (RPD) of 8.5%. The
maximum RPD (33%) occurred for a set of samples with a low concentration reading. Field blanks
were also collected at two sites. Analysis yielded results that averaged 8.0 pptv, which is well below
the 35 pptv MDL calculated for the study.
Further analysis of the data from the study showed that the PS results are very likely a result of
emissions from the facility, rather than an outside emission source. The benzene data also showed
-------
ORS Handbook
Section 3.0
Page 3-128
that the two-week sampling period was able to reflect changes in wind direction around the facility.
Additionally, it was determined that changes in temperature in humidity due to seasonal changes at
the facility had little bearing on the performance of the PSs. Overall, the study verified that benzene
concentrations collected by PS tubes are in fact reflective of emissions from the target facility, have
good precision and accuracy, and yield time resolved data that reflects meteorological conditions at
the site.
South Philadelphia PS and Sensor Study7,8
To continue its monitoring efforts, the EPA performed a PS study in South Philadelphia from June 18,
2013 to March 25, 2015. Carbopack X passive sampling tubes were deployed at 17 locations that
included fenceline locations around a petroleum refinery (Philadelphia Energy Solutions) and sites
near other oil and natural gas operations in the area. Samplers were also installed in community
areas near the refinery. Passive samplers were deployed in duplicate at each location for a two-
week sampling period. EPA Method 325A was used for sample collection, with the exception of
Section 8.0, which describes the procedures for determining sampling locations, and a modified
version of EPA Method 325B was used for analysis. After duplicate averaging, 41 valid sampling
periods yielded 655 total PS results. Each PS was installed between 2 and 4 m from the ground on
stakes, light poles, or billboards. In addition to benzene, the samples were also analyzed for
ethylbenzene, toluene, styrene, and xylene isomers.
Meteorological conditions during the study were variable, though the prevailing wind direction was
from the west. Northerly winds were more common during the winter months. To capture
downwind emissions, most of the samplers were situated to the east of the petroleum refinery.
However, other sources such as traffic and other facilities that emit VOCs did impact the sampling
locations.
In addition to the PSs, AMS also deployed an open-path ultra-violet differential optical absorption
spectrometer (UV-DOAS) at one of the PS locations. The EPA also tested two prototype leak
detection sensors called SPod and Sentinel to further investigate the viability of low-cost leak
-------
ORS Handbook
Section 3.0
Page 3-129
detection options for fenceline monitoring.
Benzene concentrations from each of the 17 monitoring sites ranged from 540 pptv to 5,020 pptv,
with an overall study mean of 700 pptv. Analysis of the data shows that nearly all the minimum
concentrations were collected at the community sites furthest away from the refinery, while the
majority of the maximum concentrations for each study period were collected at or near the
fenceline of the facility. Coupling the PS data with the distance from the facility and the collected
meteorological data allowed the researchers to determine a general concentration gradient, where
the highest concentrations of benzene were seen downwind (generally, east) of the refinery, while
lower concentrations were seen at sampling locations north of the refinery. Further examination of
the data yielded time-resolved concentration data that clearly indicates the source of pollution.
Typical QA/QC
Field Sampling QA/QC3
EPA Method 325A/B requires the collection of both collocated or duplicate samples as well as field
blanks. One collocated or duplicate sample must be collected for every 10 field samples so that
precision measurements can be calculated for the sampling period. The relative percent difference
(RPD) is calculated between sample pairs; if the RPD exceeds 30%, then the sample data must be
qualified.
One field blank must be collected per sampling period (no less than two for an entire study) to verify
PSs were not contaminated during sampler transport to and from the sampling locations or during
the sampling period. Field blanks must be conditioned at the same time as the correlating PSs used
for field samples. They are deployed at the sampling location during the entire two-week period
under a protective hood; however, the long-term sample caps are not removed. Field blank must
have less than one-third the target analyte concentration of the field samples. If a field blank fails to
meet these criteria, all associated criteria for the sampling period must be qualified as such.
-------
ORS Handbook
Section 3.0
Page 3-130
Laboratory QA/QC3
Most of QA/QC for EPA Method 325 is carried out in the laboratory. EPA Method 325B describes the
QA/QC requirements for the method in detail. Table 3.29 below lists the required analytical control
procedures that must be performed frequently throughout a PS study.
Ta
ble 3.29. Analytical QA/QC Procedures for EPA Method 325B3
QC/QA
Parameter
Frequency
Acceptance Criteria
Analytical
Precision
During initial method setup
and once per year thereafter.
ฑ20% RPD between two spiked
tubes of the same concentration.
Desorption
Efficiency and
Compound
Recovery
During initial method setup.
Repeat analysis of the same
standard tube must have > 95%
recovery.
Audit Samples
As available.
ฑ 30% of known spike
concentration.
Tube
Conditioning
Blank
Blank values must be verified
on each new sorbent tube and
10% of each batch of re-
conditioned tubes.
The larger of:
1. < 0.2 ppbv, or
2. < 3x the detection limit, or
3. < 10% of target compound
mass at the regulated limit.
Five Point
Calibration
Every three months, or
following any major repair to
the analytical system or if the
daily CCV analysis does not
meet criteria.
1. Percent deviation of
response factors must be ฑ
30%.
2. Relative retention times
(RRTs) for target peaks
must be ฑ0.06 units from
the average RRT.
Instrument
Tune
Performance
Check
Prior to the analysis of
samples.
Described in Table 3.XX3.
Analytical Bias
(Initial
calibration
verification and
CCV)
The initial calibration
verification (ICV) must be
analyzed immediately
following the calibration
curve.
CCVs must be analyzed prior
to the analysis of samples,
every ten samples thereafter,
and
The response factor must be within
30% of the average response factor
for the calibration curve.
-------
ORS Handbook
Section 3.0
Page 3-131
QC/QA
Parameter
Frequency
Acceptance Criteria
Laboratory
Blank
Beginning of each analysis
following the CCV.
Must meet the same criteria as the
tube conditioning blank.
Method detection limits (MDLs) are determined by calculating as described in Method 301, Section
15. Seven PSs are spiked with a concentration of pollutant within a factor of five of the estimated
detection limit. The standard deviation of the seven measurements is determined and then
multiplied by three to calculate the MDL. The MDL should be around 50 parts per trillion (ppt), or no
more than one-third of the lowest concentration of interest.
Siting Concerns
Section 8 of EPA Method 325A4 contains detailed instructions for determining the number and
location of optimal PS deployment locations around a facility. This section presents a general
overview of considerations that should be made when planning PS deployment around a facility.
Prior to any sampler deployment, a site visit should be made to note the size and shape of the
facility, any potential obstructions that could impede airflow, and any potential source interferences
from off-site locations. Obstructions that impede air flow include large groups of trees, buildings,
and changing topography. Both air flow obstructions and external emissions sources can cause the
PSs to report elevated levels of VOCs not related to the monitored facility.6
During the site visit, potential locations for a meteorological station should be identified as well. The
meteorological station should be placed in a location that has representative air flow and ambient
temperatures for the facility, and thus should be sited in open terrain far removed from buildings
and at least 30 meters from large paved areas. In the case of facilities located within complex
topography, more than one meteorological station may be required to accurately represent
different sampling locations around the facility.4
PSs should be deployed in a perimeter around the source location along the internal boundary of
the facility. Locations may circle the facility's geometric center at different angles, or placed at
-------
ORS Handbook
Section 3.0
Page 3-132
different distances based on the length of the facility's perimeter. Method 325A discusses PS
locations around a facility of both regular and irregular shape. In some cases, monitored facilities
have permanent monitoring stations close by that are run by state or local agencies. If access is
available, it may be beneficial to deploy a PS at this location as the emissions data
Strengths and Limitations
EPA Method 325A and 325B have several strengths and limitations; however, it is the strengths of
the sampling and analysis methods that have made PSs attractive to facilities that need to meet the
EPA's new requirements for fenceline detection of benzene. Primarily, PSs can be deployed for
much cheaper cost compared to other methods currently available for fenceline monitoring of
VOCs. The sampling method is easy to deploy; current facility staff (such as leak detection and repair
contractors) can successfully perform this sampling method without extensive training.5 The PSs
themselves are robust and can be deployed for two-week periods while exposed to a variety of
weather conditions.
Because of the required two-week sampling period, PSs are not suitable for immediate leak
detection and repair purposes. Additionally, the samplers must be transported and analyzed at an
off-site laboratory that is able to do the analysis described in EPA Method 325B. Fenceline
monitoring with PSs may also prove challenging for facilities with complex topographies or facilities
that are closely surrounded by external emissions sources such as busy highways or other oil and gas
facilities. Finally, meteorological data must be collected throughout the sampling period to
accurately assess the sample data. Table 3.30 lists the strengths and limitations of EPA Method
325A/B.
-------
ORS Handbook
Section 3.0
Page 3-133
Table 3.30. Strengths and Limitations of EPA Method 325A/B
Strengths
Limitations
Inexpensive to deploy.
Not suitable for more immediate leak
detection and repair.
Minimal training required for monitoring
staff.
Samplers must be analyzed at an offsite
laboratory.
Samplers are robust and can withstand a wide
variety of weather conditions.
Challenging to use with complex site
topography and close external emissions
sources.
Meteorological data are required for data
analysis.
References
1. Brown, R.H. The Use of Diffusive Samplers for Monitoring of Ambient Air. Pure & Appl.
Chem., Vol. 65, No. 8, pp. 1859-1874, 1993.
2. Palmes, E.D. and A.F. Gunnison. Personal Monitoring Device for Gaseous Contaminants.
American Industrial Hygiene Association Journal. Vol. 34, pp. 78-81. 1973.
3. US EPA. Method 325B-Volatile Organic Compounds from Fugitive and Area Sources:
Sampler Preparation and Analysis.
4. US EPA. Method 325A-Volatile Organic Compounds from Fugitive and Area Sources:
Sampler Deployment and VOC Sample Collection.
5. US EPA. 2000. Meteorological Monitoring Guidance for Regulatory Modeling
Applications. EPA-454/R-99-005. Office of Air Quality Planning and Standards, Research
Triangle Park, NC. February 2000.
6. Thoma, Eben D., Michael C. Miller, Kuenja C. Chung, Nicholas L. Parsons, and Brenda C.
Shite. Faciilty Fence Line Monitoring Using Passive Samplers.
7. Thoma, Eben D., et al. South Philadelphia Passive Sample and Sensor Study. Journal of
the Air & Waste Management Association., Vol. 00, No. 00, 1-12. 2016.
8. Mukergee, Shaibal, et al. Spatial Analysis of Volatile Organic Compounds in South
Philadelphia Using Passive Samplers. Journal of the Air & Waste Management
Association., Vol. 66, No. 5, 492-498. 2016.
-------
ORS Handbook
Section 3.0
Page 3-134
3.10 Method to Quantify Particulate Matter Emissions from Windblown Dust -
Other Test Method 30
Dust emissions from arid playa sources1,2 affect not only human health, but also create
environmental concerns around loss distribution3, mineral cycling, and even cloud formation.1,2
Sources of dust emissions in the United States include Owens and Mono Lakes, both of which are
exposed lake beds in California that were the result of water being diverted from the lakes into
populous areas. Other Test Method (OTM) 30 was developed so that particulate matter (PM)
emissions can be monitored downwind of places susceptible to wind erosion and potentially
remedied through characterization of high emission sources.
General Description of Approach
The basic premise of OTM 30 is that PM emissions can be quantified by comparing saltation flux to
the difference in upwind and downwind ambient PM concentrations.3 Saltation is sand-sized
particles hopping over erodible surfaces caused by wind. The saltating particles bombard the ground
surface, releasing tiny particles of dust. It is this saltation bombardment, not aerodynamic lift, that is
the primary mode by which dust particle movement is initiated.4 When the tiny dust particles are
lofted by bombardment, winds can carry those particles many kilometers downwind. Figure 3.453 is
an illustration from OTM 30 that shows the transport of dust particles through the saltation process.
2 Playa is the flat bottom of a desert basin that can become a shallow lake at times.
3 Loess is an unstratified loamy deposit found in North America, Europe, and Asia and is chiefly deposited by the wind.
-------
ORS Handbook
Section 3.0
Page 3-135
Wind
Suspended Dust
/ ^ (pm10,pm25&pm10.25)
's '
ft' Saltation
//> ~
/
M
Cox Sand Catcher*
/ (particle collection bin)
/ 7
Sensit*
(real-time particle
sensor)
*Typical sampling height is 15 cm for saltating particles
Figure 3-45. Saltation and Dust Production Process for Windblown Dust.
Theoretical and experimental evidence indicates that the vertical flux of windblown dust is
proportional to the horizontal flux of saltating sand-sized particles for soils with the same binding
energy. The following equation shows the two components are proportional by a factor of K:
F = Kf X qls
Where:
F = Vertical PMio emission flux [g/cm2-s]
Kf = Seasonal K-factor; Non-dimensional proportionality constant
qis = Horizontal sand flux 15 cm above the surface [g/cm2-s]
Studies have shown that surface moisture and temperature changes that occur seasonally affect the
binding energy of the soil, thereby affecting Kf, or proportionality constant. Dispersion models such
-------
ORS Handbook
Section 3.0
Page 3-136
as CALPUFF or AERMOD are used to determine the Kf for changing surface characteristics.
Instrumentation
OTM 30 requires the deployment of two instruments to calculate sand flux measurements: (1) a
passive instrument that monitors sand catch over a sampling period, and (2) a real-time particle
impact sensor to time-resolve the sand catch data.
The sand catch instrument recommended by OTM 30 is the Cox Sand Catcher (CSC) manufactured
by the Great Basin Unified Air Pollution Control District (GBUAPCD) in California. The sand catchers
are deployed so that the inlet is 15 cm above the surface of the source as seen in Figures 3.466
through 3.46. Sample tubes that collect the sand inside the CSC are collected monthly to be weighed
in a laboratory. Another sand catch instrument that has been used successfully in the Big Springs
Number Eight (BSNE), manufactured by Custom Products in Big Springs, TX (Figure 3.47).7 The BSNE,
however, has a smaller capacity than the CSC, and daily site trips may be required to prevent
overloading the instrument.
C^ircular Top
E2.6 cm ปl
ImpxieLor Roci |
Removable Cap
r
\ 2 i
Figure 3.46. Schematic of CSC Placement for Sampling.
-------
ORS Handbook
Section 3.0
Page 3-137
Reference #
Feature
Description
1
Roof
1/8" thick by 2 3/4" diameter PVC sheet
2
Roof Support
3/4" schedule 40 PVC pipe 2" in length
3
Sample Inlet Opening
1 cm from bottom of roof to top of PVC coupling. Tolerance is 0.5 mm.
4
Support Pins
1/4" diameter PVC rod glued in place
5
Head
2" schedule 40 PVC coupling, specify long coupling approximately 2 3/4" in length
6
Catch Tube Seal
rubber shank washer cut to fit
7
Catch Tube
2" diameter clear plastic soil sample tube variable length to fit application*
8
Connecting Pipe
2" schedule 40 PVC pipe** 3 1/2" in length
9
Stainless Pipe Clamp
10
Adjustment Coupling
2" diameter rubber plain and flexible pipe coupling 3 1/2" in length
11
Body
2" schedule 40 PVC pipe** variable in length to fit application, 25" for 2' CSC
12
Catch Tube Stopper
rubber stopper or plug
13
Bottom Cap
2" schedule 40 PVC cap with a flat top
14
Bottom Plate
1/8" thick by 3 7/8" diameter PVC sheet
*Note: The Catch Tube shown here is partially filled with sand.
**Note: The inner diameter of PVC pipe varies with manufacture. Make sure the sample catch tube slides freely into the pipe before purchasing.
Figure 3.47. Cut-out of a CSC and Construction Specifications
-------
ORS Handbook
Section 3.0
Page 3-138
Figure 3.48. CSC Placement in the Field with a Height Adjustment Tool.3
Figure 3.49. 8SNE Sampler Shown Sampling in the Field.
-------
ORS Handbook
Section 3.0
Page 3-139
The only real-time particle impact sensor to be successfully used for an OTM 30 study is the Sensit,
manufactured by Sensit Company in Portland, ND. Sensits use a piezoelectric crystal to detect and
measure saltation activity. The resulting particle count and kinetic energy measurements are
proportional to the mass flux of PM. When co-located with CSCs and installed at the same height,
hourly Sensit readings are used to time-resolve sand catch data over a sampling period to determine
hourly sand flux.
OTM 30 also requires that PM is monitored with instruments capable of collecting hourly data;
tapered element oscillating microbalance (TEOM) PMio monitors have proven to have good success
in existing OTM 30 studies. Beta-gauge and beta-attenuation monitors that provide hourly PM data
should also work with this method.
A 5-10 m meteorological tower that measures scalar wind speed, direction, and sigma-theta (the
standard deviation of the wind direction over the period of measurement) is also required to be
near the study area. The meteorological tower must be able to record hourly average data and
should be sited and operated in accordance with the federal monitoring guidelines described in US
EPA Quality Assurance Handbook for Air Pollution Measurement Systems, Volume IV:
Meteorological Measurements.
A complete list of supplies and equipment needed to perform OTM 30 is available in Section 6.0 of
the test method.
Dispersion Modeling and K-factors
Both the AERMOD and CALFPUFF dispersion models are EPA-approved and have been
demonstrated to work well with OTM 30. Both are applied following the EPA modeling guidelines
detailed in 40 CFR, Part 51, Appendix W. Both dispersion models calculate Kf using the PM emissions
calculated from the horizontal flux assuming an initial K-factor, Ki, of 5 x 10"5. The hourly Kf is then
determined post-processing using the following equation:
-------
ORS Handbook
Section 3.0
Page 3-140
Kf = Kt
C0 Cb
-m
Where:
Ki = Initial K-factor (5 x 10~5)
C0= Observed hourly concentration
Cb= Background concentration
Cm = Modeled concentration
OTM 30 offers specific guidance on hourly K-factors that should be screened due to varying
conditions that affect PM emission data. Once K-factors are screened, the seasonal K-factors can be
calculated based on the geometric mean of the hourly K-factors calculated above. It is the seasonal
K-factor that is used in the calculation that determines vertical PMio emission flux from horizontal
sand flux. More specific instructions for data analysis and calculations are discussed in Section 10.0
of OTM 30.
Sample Collection
Prior to the start of monitoring, the meteorological instrumentation and PM monitor should be
calibrated in accordance with EPA monitoring guidelines and all data logger functionalities should be
checked. A tare weight must be determined for the empty sand catcher sampling tubes used in the
CSCs and recorded. After installing the empty sampling tube in the CSC, the height must be adjusted
to 15 cm. The Sensit functionality can be tested by simply tapping on the sensor. Once the height of
the sensor is verified to be 15 cm, sampling and recording can be initiated. OTM 30 recommends
that site visits occur monthly when no erosion event is occurring. It is important to note that
maintenance activities may be required more frequently depending on the conditions the
-------
ORS Handbook
Section 3.0
Page 3-141
instruments are exposed to. Specific procedures about the collection of CSC samples and Sensit data
are listed in OTM 30, Section 7.4. Verification/Validation Studies
Owens (dry) Lake. Inyo County California1
The goal of the study was to characterize the dry lake bed to determine which areas within the area
create the highest PM emissions, therefore needing dust controls. The methodology used in this
study is the basis on which OTM 30 was written.
In all, 135 sand flux sites were installed to monitor sand flux rates on an hourly basis. Each site was
spaced approximately 1 km apart. The sites were separated into four sampling zones based on
different geological characteristics and source activity that seemed to affect PM emissions. Each
measurement site included a CSC, which collects saltating sand-sized particles, and a collocated
Sensit electronic sensor that provided hourly sand flux rates. Kfactors were determined using the
Calpuff dispersion modeling system at six representative monitoring sites. Using this system,
researchers could quantitate PM emissions for each of the square kilometer measurement sites and
determined that approximately 77 square km of the lake bed required dust controls to reduce PM
emissions to within federal standards.
Mono Lake. California6
A second large source of windblown dust is Mono Lake, California. The lake is a large shallow lake in
the Great Basin of Yosemite National park in Mono County, north of Los Angeles. Diversion of water
from the lake to the City of Los Angeles between 1941 and 1989 lowered the lake level, exposing
over 24 square km of lakebed to wind erosion. Though an air quality analysis completed in 1994
determined that a lake level of 6,391 feet above sea level would reduce PM emissions to within
federal standards, the lake level only reached 6385 feet twice, in 1999 and 2006. PM concentrations
continued to exceed federal standards even at the highest lake levels.
The GBUAPCD implemented a refined method for estimating PM emissions from Mono Lake based
on the methodology used to characterize PM emissions from Owens (dry) Lake. The monitoring
effort started in July 2009 and ended in June 2010. The monitoring network, shown in Figure 3.503,
-------
ORS Handbook
Section 3.0
Page 3-142
included 25 CSCs, two Sensits, a meteorological tower, and one TEOM PMio monitor, all of which
were operated on the north shore of Mono Lake. Hourly data were collected from each network
component so that time-resolved PMio concentration data could be calculated. Researchers at
Mono Lake used the AERMOD dispersion modeling system to determine K-factors. The AERMOD
system differs from the CALPLJFF model in that its AREAPOLY area source algorithm is better suited
for irregularly-shaped area sources.
Figure 3-50. Monitoring Network at Mono Lake, CA.
Met Tower
200 Meters
Over the 1-year study period, the 24-hour federal PMio standard was violated 25 times, with the
highest 24-hour average concentration measuring 14,147 |ig/m3, nearly 10 times the federal
standard. Predicted PMio concentrations calculated via dispersion modeling compared well to the
concentrations recorded by the TEOM PMio monitor throughout the study period.
Typical QA/QC
All data logged by the monitoring instrumentation must be examined for missing or anomalous data.
In the case of missing data, which may be caused by a low battery or a data logger malfunction, data
from the nearest operating Sensit can be used to replace the missing data and time-resolve the CSC
sand catch data. Anomalous data, which includes sensor responses not caused by suspended dust
-------
ORS Handbook
Section 3.0
Page 3-143
(such as operators tapping on sensors to check functionality) should be removed from the dataset.
The ratio of the individual Sensit readings to sand catch ratio for each monitoring site and sampling
period should be examined for consistency between sampling events. Large changes in this ratio
may be an indication that a Sensit should be replaced.
Data logged by the meteorological station and PM monitors must also be examined for missing and
anomalous data. Any anomalous data should be investigated and removed as needed. While
operation staff conduct calibration activities at the site on a routine basis, OTM 30 requires that a
third-party quality assurance audit be performed as well. For example, meteorological equipment
must be audited within 30 days of the start of monitoring and every six months thereafter.
All balances used to weigh sample tubes and particulate collected in the CSCs must be checked
before and after each weighing session with certified National Institute of Standards and Technology
(NIST) Class-F weights. All weights must be recorded in a balance log. The balances must be certified
and calibrated annually through a third-party.
Siting Concerns
Section 7.0 of OTM 30 contains specific recommendations about creating a monitoring network
using OTM 30. This section summarizes the major recommendations and specific concerns.
Monitoring networks can range in scope from one sand flux, meteorological, and PM monitoring site
for small source areas to over 100 sand flux sites with a representative number of meteorological
and PM monitoring sites. While measurement accuracy does improve with the number of
monitoring sites, data yielded by few sites are also beneficial.
CSC and Sensit monitoring locations should include all significant windblown dust sources between
the upwind and downwind PM monitors. Dust sources that are not included in the background
concentration measured at an upwind monitoring site could bias hourly K-factor calculations,
including dust sources that do affect the downwind monitoring site. Thus, it is very important to
include all potential sources of PM in the dispersion model. Each CSC and Sensit monitoring site
-------
ORS Handbook
Section 3.0
Page 3-144
should be between 100 and 1,000 m apart and each should have a designated source boundary.
PM monitors should be installed upwind and downwind of the site per the prevailing wind direction
for high wind events. A collocated PM monitor should be positioned at the site of maximum impact,
or a downwind position, to produce the most accurate K-factor, as well as enhance the defensibility
of data collected during the study. The meteorological tower and PM monitor must be sited in an
area that avoids structures and topographical features that interfere with wind flow patterns
between the dust source and the downwind monitors.
Strengths and Limitations
The methodology provided in OTM 30 has several strengths and limitations. This flexible method
allows all windblown dust events at a monitoring site to be sampled, flux calculations can determine
specific sources of PM that require remediation, and the on-site instrumentation is simple to
operate. OTM 30 does, however, require all potential PM sources that affect an area to be
monitored. Extensive data analysis is also required to yield accurate results. Table 3.31 summarizes
the strengths and limitations of OTM 30.
Table 3.31. Strengths and Limitations of the OTM 30 Approach
Strengths
Limitations
All windblown dust events during a sampling
period are sampled.
All PM sources that contribute downwind
concentration must be included in the
dispersion model.
Sand flux measurements indicate specific
locations where emissions originate.
Extensive data analysis is required to achieve
accurate results.
Emissions vary hourly based on changing
conditions that affect erosion rates.
Instrumentation is simple to operate.
Flexibility in monitoring network size.
References
-------
ORS Handbook
Section 3.0
Page 3-145
1. Gillette, D., D. Ono and K. Richmond. 2004. A Combined Modeling and Measurement
Technique for Estimating Windblown Dust Emission at Owens (dry) Lake, California. Journal of
Geophysical Research. 109:F01003.
2. Koehler, K. A., S.M. Kriedenweis, P.J. DeMott, A.J. Prenni, and M.D. Petters. 2007. Potential
Impact of Owens (dry) Lake Dust on Warm and Cold Cloud Formation. Journal of Geophysical
Research. 112:D12210.
3. U.S. Environmental Protection Agency (EPA). 2012. Other Test Method-30: Method to
Quantify Particulate Matter Emissions from Windblown Dust. Available online at
https://www.epa.gov/emc/emc-other-test-methods
4. Shao, Y., M.R. Raupach, and P.A. Findlater. 1993. Effect of Saltation Bombardment on the
Entrainment of Dust by Wind. Journal of Geophysical Research. 98:12719-12726.
5. Office of Air and Radiation, EPA. 2016. Clean Air Excellence Awards Ceremony Booklet.
Available online at https://www.epa.gov/caaac/clean-air-excellence-awards-ceremony-
booklet.
6. Ono, D., P. Kiddoo, C. Howard, G. Davis, and K. Richmond. 2011. Application of a Combined
Measurement and Modeling Method to Quantify Windblown Dust Emissions for the Exposed
Playa at Mono Lake, California. Journal of the Air & Waste Management Association. 61:1036-
1045.
7. Acosta-Marinez, V., S. Van Pelt, J. Moore-Kucera, M.C. Baddock, and T.M. Zobeck. 2015.
Microbiology of Wind-eroded Sediments: Current Knowledge and Future Research Directions.
Aeolian Research, 18, 99-113.
-------
ORS Handbook
Section 3.0
Page 3-146
3.11 Determination of Emissions from Open Sources by Plume Profiling - Other
Test Method 32
General Description of Approach
Other Test Method 32 (OTM 32)1 utilizes plume profiling to characterize particulate matter (PM)
emissions from open sourcesspecifically from roadways. Plume profiling, synonymous with the
terms exposure profiling or plume flux profiling, is an open source emissions test method based on
the exposure profiling concept.2 This concept describes how exposure can be defined as the mass
flux of a pollutant at a sampling point, where mass flux is calculated as the product of pollutant
concentration and wind speed.
OTM 32 uses a "conservation of mass" approach to calculate emissions factors and rates,3 where
the quantity of emissions (concentration) is determined by the spatial integration of mass over the
cross section of the plume. The PM concentration is calculated by the following formula:
Where:
C = PM concentration (Ib/VMT4)
m = Mass collected on the filter or substrate (lb)
Q = Flow rate of the sampler (VMT)
T = Sampling time
Exposure is the net particulate mass that passes through a unit area during the test. Exposure is
calculate using the following formula:
E = (C ~Cb)UT
Where:
E = Exposure (mass/area)
C = Downwind PM concentration (mass/volume)
Cb = Background PM concentration (mass/volume)
U = Approaching wind speed
T = Sampling time
Because exposure values are determined at specific points within the plume, those values will vary
4 VMT = Vehicle-mile traveled. This value is defined as the product of the vehicle count during the sampling period and
the length of the road segment being represented by the profiling data.
-------
ORS Handbook
Section 3.0
Page 3-147
over the vertical length. A one-dimensional integration is used to obtain the total PM emissions
from a line source according to the following equation:
H
A1 = I Edh
o
Where:
A1 = Integrated exposure (mass/length)
E = Exposure (mass/area)
h = height from the ground (length)
H = vertical height of the plume (length)
PM emission factors are then determined by normalizing the integrated exposure against a
measure of source activity, such as dividing the integrated exposure by the number of passing
vehicles.
Sampling Equipment
To determine pollutant (PM) concentrations, an array of samplers is deployed to provide time-
averaged concentration data at points within the plume. In the case of PM emissions, the sampler
is fitted with a removable sample collection substrate, such as a filter, that is submitted to a
laboratory for analysis after sampling is completed. The PM fraction of interest will determine the
specific sampler used in the study, as described in Section 6.1 of OTM 32. Because sampler flow
must be maintained at a sufficient level to ensure adequate sample collection, a flow
measurement device is typically incorporated into the sampler. Figure 3.511 illustrates the
configuration of a traditional PM sampler. All samplers should be maintained and calibrated as
recommended by the manufacturer prior to sampling.
-------
ORS Handbook
Section 3.0
Page 3-148
Lid (removes for cleaning)
Cyclone Body
Outlet Tube
Transition Piece
8 x 10 in filter holder
Figure 3.51. Traditional PM Sampler Configuration.
Filters used for sampling must be prepared in the laboratory before sample collection begins as
described in OTM 32 Section 8.1.2. To summarize, a clean filter with no imperfections must be
equilibrated in a controlled room with an analytical balance for at least 24 hours. Ideal room
conditions for gravimetric analysis are 40% relative humidity with a temperature of 34ฐ C (93ฐ F).
Once the balance's calibration is verified, each filter is weighed, providing an initial tare weight for
the sample. At least three blank filters should be weighed and handled as a blank for each test day
to assess the handling effects of the study. Each filter is then placed in a glassine envelope,
followed by a numbered file folder. The folders are stored for submittal to the field in a heavy-duty
cardboard box.
Because wind speed is an integral part of the mass flux calculation, meteorological equipment is
also needed at the sampling location. The meteorological monitors should provide time-averaged
wind speed data at 5 to 15 minute intervals. Ambient temperature and relative humidity are also
collected at the sampling site. Barometric pressure data can be obtained from a local weather
station to report concentration data in local conditions, or can be approximated based on
elevation to report data in standard conditions. All meteorological equipment should be
maintained and calibrated in accordance with Volume IV of EPA's QA Handbook series.4
-------
ORS Handbook
Section 3.0
Page 3-149
Sample Deployment
Because OTM 32 relies on both pollutant concentration and wind speed data to generate mass
flux concentrations, simultaneous multi-point measurements are made over the emission plume in
a sampling plane perpendicular to the pollutant's (PM) direction of transport. Figure 3.521
illustrates the ideal sampler array for a point source, where W is about 75% of the observed visible
plume width and H is about 75% of the observed visible plume height at a distance D.1 For mobile
(or line) sources, the samplers are oriented in a vertical array at equal points along a support
tower with a maximum height of seven meters (Figures 3.53 and 3.54).1 Wind speed monitors are
typically deployed at a height of 2 to 3 meters from the ground. If the length of the line source is
at least ten times the downwind distance to the sampling array, only a single vertical array is
needed to adequately characterize the emissions plume.
Figure 3-52. Illustration of Fixed Point Source Sampling Array.
-------
ORS Handbook
Section 3.0
Page 3-150
Wind
Diesel exhaust.
15 111 :
Upwind Sampler
t .#<9
Q Downwind
Sampling
Array
\ ฆ
ฆ; ,--s_. .ฆ ฆ ฆv.-'-9. "ฆLi
J.
ssiS' ^
Suspended dust
plume
Direction of
travel
Area swept out
by earthmoving
equipment
Figure 3-53. Illustration of Moving Point Source-Unpaved Road Dust Emissions.
~
LEGEND
E
LCI
0
Wind
E
LD
9 Sampler
| | Anemometer
5 m +
Path of moving equipment
15 m t
Figure 3-54. Example Sampling Array for Moving Point Source.
Sampling must be conducted at specific meteorological conditions. Specifically, if measurable
precipitation is forecast, sampling should not occur. Additionally, average wind speed must be
between 3 and 20 miles per hour (mph) in a direction that creates a 0 to 45-degree angle to the
-------
ORS Handbook
Section 3.0
Page 3-151
perpendicular of the line source for at least two consecutive 5-minute time-averaged periods. If
either condition drifts outside of these requirements for more than two consecutive 5-minute
time-averaged periods, sampling must be suspended until acceptable meteorological conditions
return.
Once the sampling event has concluded, sample substrates may be removed from the samplers.
Filters used for PM collection are inserted into a glassine envelope upon sample collection, then
further protected by a file folder and/or heavy-duty plastic bag. Sampled filters should be stored
separately from clean filters to prevent contamination.
Sample Analysis
Once clean filters are exposed and returned to the laboratory, gravimetric analysis is required to
determine final weights for each filter. Once the filters are equilibrated as described above for the
tare weight, and the balance's calibration is verified, an analyst weighs each filter and the final
weight recorded. Then, a second analyst independently verifies ten percent of exposed filter
weights. If the second measurement does not agree with the first within three times the standard
deviation of the reweighs for the filter blanks, then the entire batch must be reweighed and
verified a second time.
Verification/Validation Studies
United Taconite and U.S. Steel Minntac3
Midwest Research Institute (MRI) conducted a field tests at United Taconite and U.S. Steel
Minntac in the Mesabi Iron Range in northern Minnesota to characterize the PM emissions factors
for fugitive dust emissions that result from haul truck operations at taconite5 mines in January and
July of 2007. The plume profiling method was performed in July to calibrate results from mobile
monitoring efforts. The study objectives included developing PM2.5/PM10 ratios for haul road dust
emissions, and developing water control methods to reduce PM emissions. MRI implemented both
mobile monitoring and the plume profiling method simultaneously at haul road sites for each
mine. Sampling towers as described above were deployed at each reference site in the study
downwind from the near edge of the haul road. Upwind PM10 samplers were deployed upwind to
5 Taconite is a low-grade iron ore.
-------
ORS Handbook
Section 3.0
Page 3-152
determine background PM emissions. Sampling periods were typically between 1 and 2 hours
long. The sampling sites were located at road segments where there was unobstructed wind flow
over the road and there were no interferences from other PM sources. To prevent interference
from truck exhaust emissions, the grade of the road segments was very small.
A mixture of haul trucks, maintenance vehicles, and light-duty vehicles passed through the
sampling locations during the sampling periods. To account specifically for haul truck emissions,
MRI converted the other vehicles into haul truck equivalents and determined the other vehicles
accounted for less than 10 percent of the total haul truck passes through the sampling site. Six
plume profiling tests were conducted at each facility, resulting in twelve total tests. Both PMio and
PM2.5 samplers were deployed to determine PM2.5/PM10 emissions ratios for each of the reference
test sites. Overall, emissions factors for the United Taconite facility were determined to be lower
than those at U.S. Steel Minntac, which can be explained in part by rainy conditions at United
Taconite and watering of the road segment overnight.
The average uncontrolled emission factors (both daytime and nighttime combined) were
determined to 6.0 Ib/VMT for United Taconite, and 8.4 Ib/VMT for U.S. Steel Minntac. Annual
emission factors previously estimated by the Taconite Industry Working Group based on literature
and emissions factors determined at other facilities (6.2 lb PM10/VMT) compared well with the
emission factors determined at the test sites when considering that the Working Group numbers
were derived from daytime values only. Overall, MRI determined the average PM10 emission factor
for taconite mine roads to be 7.2 lb PM10/VMT, while the average PM2.5 emission factor is 0.72 lb
PM2.5/VMT.
-------
ORS Handbook
Section 3.0
Page 3-153
Typical QA/QC
Several types of quality assurance and quality control methods are utilized in OTM 32 to ensure
accurate measurements and sample traceability.
Quality Control Samples
Blanks are collected minimally for 10 percent of samples collected in the field. A blank is collected
in the same manner as a sample, however, no ambient air is passed through the collection
substrate or filter, as is the case for PM. Blank recoveries demonstrate that filters or other
sampling media were not contaminated during sample handling.
Quality Assurance
OTM 32 recommends performance audits be conducted routinely by an independent operator
during the study period. The auditor reviews the flow rate calibration, re-analyzes the exposed
substrate (filter), and reviews the data processing.
It is also necessary to ensure each sample is traceable throughout the sampling process. Each
substrate, or filter, should receive a unique identification number that is recorded along with the
date the filter was obtained. Each sample must also be coded with the sample location and test
series, as well as any other data pertinent to sample collection. All sample transfers, such as from
the field to a laboratory, should also be recorded.
Precision
To determine method precision for mobile or line sources, vertical sampler towers can be
collocated on the road segment of interest or by repeating tests on a single sampler tower under
constant source conditions. An example of a collocated tower for mobile (line) source sampling is
showing in the image of Figure 3.55.1 A minimum of three tests is required to estimate method
precision; however, due to the cost of deploying multiple sampler towers, method precision is
often estimated based on the precision of previous studies with similar source configurations.
-------
ORS Handbook
Section 3.0
Page 3-154
Les pass
Figure 3.55. deployment of Collocated Plume Profiling Towers at Roadside Location.
Siting Concerns
In the case of OTM 32, the contract under which the sampling occurs may dictate a specific
location. If this is not the case, the method outlines recommended site selection criteria in Section
8.2, which are summarized as follows:
The sampling area should be flat, open terrain where;
o The height of the nearest downwind obstruction is less than the distance from the
obstruction to the sampler.
o The height of the nearest upwind obstruction is less than one-third the distance
from the obstruction to the sampler.
The site should be at least 15 meters from the upwind edge of the source and at least 10
meters from the downwind edge of the source.
For mobile sources treated as a line source, the mean wind direction should be a 0 to 45-
degree angle from a line drawn perpendicular to the source at the time of sampling.
-------
ORS Handbook
Section 3.0
Page 3-155
Average wind speed should be greater than 3 mph at the time of sampling.
Adequate source activity, i.e., traffic, should be present at the time of sampling to provide
sufficient pollutant (PM) mass for sampling.
To prevent interference from PM emissions due to vehicle exhaust, the road grade should
be near zero.
OTM 32 recommends the use of a portable weather station to assess wind conditions at the
sampling location at the time of sampling. The guidance provided in Volume IV of EPA's QA
Handbook series4 should be followed when collecting meteorological data for flux emissions
calculations.
Strengths and Limitations
Table 3.32 summarizes the strengths and limitations of OTM 32.
Table 3.32 Strengths and Limitations of the OTM 32 Approach
Strengths
Limitations
Cost effective for determining PM emissions
factors when used in conjunction with other
test methods.
Potential interferences due to upwind pollutant
(PM) concentrations.
Sampling array set up allows for the entire
plume to be characterized.
Specific meteorological conditions are required
for sampling.
OTM 32 isolates a single emission source in
ambient conditions to provide accurate
emissions factors.
Accuracy of testing results is highly dependent
on wind conditions during sample periods.
Does not require interference of traffic
patterns.
Direct measurements do not utilize generalized
atmospheric dispersion models to estimate
results.
-------
ORS Handbook
Section 3.0
Page 3-156
References
1. U.S. EPA. 2013. Other Test Method - 32: Determination of Emissions from Open Sources by
Plume Profiling. June 2013.
2. Cowherd, C., Jr., K. Axetell, Jr., C. M. (Guenther) Maxwell, and G. A. Jutze. 1974.
"Development of Emission Factors for Fugitive Dust Sources," EPA Publication EPA-450/3-
74/037, NTIS Publication PB-238 262, June 1974.
3. Cowherd, C., Jr., Donaldson, J. Kies, R., and Murowchick, P. 2008. "Field Study of Emissions
from Haul Roads." Prepared for the Iron Mining Association of Minnesota, Duluth MN,
October 30, 2008.
4. U.S. EPA. 2008. "Quality Assurance Handbook for Air Pollution Measurement Systems,
Volume IV: Meteorological Measurements Version 2.0 (Final)." EPA-454/B-08-002. March
2008.
-------
ORS Handbook
Section 3.0
Page 3-157
3.12 Method to Quantify Road Dust Particulate Matter from Paved and Unpaved
Roads - Other Test Method 34
General Description of Approach
Other Test Method 34 (OTM 34)1 is a mobile monitoring method used to quantify road dust
particulate matter (PM) emissions from vehicles travelling on paved and unpaved roads. Although
this method can be applied to any particulate size fraction in principal, this method specifically
quantifies emissions for PM2.5 (PM less than 2.5 microns) and PM10 (PM less than 10 microns.) To
accurately quantify emissions, OTM 34 relies on an increase of PM emissions from background
levels caused by the interaction of vehicle tires with the road surface. Measurements are collected
at one or more locations along a stretch of paved or unpaved road at least 100 meters long.
PM emissions are created primarily by the aerodynamic and mechanical suspension of roadway
particles caused by the interaction of vehicle tires with the road surface. Roadway particles include
loose, erodible soil, dust from brake or tire wear, and dust from erosion of the road surface (i.e.,
asphalt or cement.) As road dust PM is suspended, a PM plume disperses behind each tire and in the
wake of the vehicle. The test method quantifies the increase of PM above ambient conditions at a
known travel speed and fixed distance from the PM source. Background or ambient PM emissions
are measured at a location not influenced by the test vehicle's own emission plume, such as the
hood of the vehicle. Application of this test method relies on four key assumptions:
1. The speed of the vehicle determines the degree to which the emission plume disperses
behind the vehicle tires;
2. The degree of plume dispersion is not measurably affected by other vehicles on the road,
the magnitude of PM emissions, or the mechanism of PM emissions;
3. PM emissions are solely the result of the interaction of vehicle tires with the road and the
contribution of PM from turbulence created by the vehicle is negligible; and
4. The additional PM created by a vehicle other than the test vehicle is either small or can be
accounted for through known mathematical relationships.
Beginning in 1999, two concurrent mobile monitoring methods were tested for the measurement of
road dust emissions. One, Fitz2, estimated the profile of the dust plume in the wake of the vehicle
while the other, Kuhns et al3, measured PM concentrations directly behind the vehicle's front tire
and related those concentrations to emission factors determined as prescribed in the US EPA AP-42
-------
ORS Handbook
Section 3.0
Page 3-158
Guidance document (4th edition.)4 The latter study that found that for the same roadway, PM
emissions behind a vehicle tire increase exponentially with vehicle speed. The relationship between
road speed and PM emissions is described in the following equation:
EFps MSCiiVipS X Ki(y)
Where:
EFps = Emission factor of PM road dust (g/km)
MSCi,v,ps = Measured PM concentration
Ki = Constant specific to the mobile monitoring system
v = Travel speed
In addition to determining emission factors for roadways of interest, measurements yielded by OTM
34 can be used to provide insight on what types of road conditions yield the highest PM
measurements, as well as evaluate the effectiveness of dust control techniques currently in use like
chemical treatments.
Instrumentation
The mobile measurement system must provide PM concentrations with a high time resolution, on
the order of seconds or less. Typically, this requires the use of PM monitors that estimate mass
concentrations based on the light scattering properties of aerosols such as PM. OTM 34 specifically
mentions the DustTrak 3563 nephelometer, however this instrument has been discontinued. The
manufacturer, TSI, Inc., has introduced a newer product called the DustTrak Aerosol Monitor
(http://www.tsi.com/environmental-dusttrak-aerosol-monitor/, models 8540 and 8543), both of
which measure PM at various size fractions. However, any similar monitor that can provide sufficient
time-resolved PM concentration data may be used.
PM monitors may also be equipped with a cut device or size-selective inlet (SSI) that allows for the
measurement of a specific fraction of PM, such as PMio. Although manufacturer-specific SSI devices
-------
ORS Handbook
Section 3.0
Page 3-159
do not typically meet National Ambient Air Quality Standard (NAAQS) compliance requirements for
PM monitoring, the widely-used SSI option is considered acceptable for compliance provided the
nominal cut size is within 5 percent of the size fraction being measured and the size cut
characteristics of the SSI are documented.1 In the case that a manufacturer-provided SSI is not
sufficient per project requirements, another option is to subject the entire sample flow within the
sample inlet to an SSI, such as a cyclone for PM2.5. This second option proves more difficult for PM10
measurements, as the SSI for this size tend to be larger and may not fit well behind the test vehicle's
tires.
Other common equipment needed for the mobile monitoring system includes pumps or blowers to
pull sample air through the inlet lines. Flow meters such as pitot tubes are required to measure the
air flow through the sample inlet.
Due to logistical and safety reasons, the PM monitor typically cannot be installed in the ideal
location (behind the tire) on a test vehicle to measure PM emissions, therefore, a sample inlet line is
used to channel sample air into the PM monitor. One common configuration is to install the sample
inlet directly behind the front tire, as shown in Figure 3.56.1 With this configuration, particle
sampling loss and size biases are small because there is little difference between the air speed just
outside the sample collection point and the air speed through the inlet. Typically, one sample inlet
will be installed behind the left front tire and another behind the right front tire with both lines
measured independently.
-------
ORS Handbook
Section 3.0
Page 3-160
Figure 3-56: Sample inlet installation behind the front tire.
Another common sampling configuration is to install the measurement instruments and sample inlet
line on a trailer that is pulled behind the test vehicle, as shown in Figure 3-57.1
Figure 3.57. Sample inlet installation on a trailer pulled behind the test vehicle.
-------
ORS Handbook
Section 3.0
Page 3-161
The assumption is that PM emissions from both the left and right tires will affect the sampler
simultaneously. Regardless of the instrumentation set up, there are several requirements that must
be met to minimize particle loss between the inlet and the measurement instruments. These
requirements are discussed in detail in Section 6.1.1 of OTM 34.
Because vehicle location and speed are critical to OTM 34, a global positioning system (GPS) capable
of providing high time-resolved data should be used in conjunction with the test method. Because
measurement collection is automated, a datalogger such as a laptop is required to collect the data
provided by the various instrumentation. A laptop is ideal because the data can be viewed real-time
so adjustments to the instrumentation can be made as needed. While not required to perform OTM
34, Geographical Information System (GIS) software is a useful tool to map the travel route, select
subsets of data, and examine spatial trends.
Calibration
To quantify the coefficient Ki(v) in the equation above, the mobile monitoring system must be
calibrated against an external standard. The likely optical-based mobile monitoring system is related
to PM horizontal flux measurements determined by mass-based instruments using a method such as
OTM 325, "Determination of Emissions from Open Sources by Plume Profiling." Calibration should
occur under conditions like those expected during the mobile monitoring effort, especially the range
of expected vehicle travel speeds.
The calibration procedure requires that the horizontal flux of PM is measured perpendicular to the
road segment of interest, at points both upwind and downwind of the site. PM is measured at
varying heights above ground level simultaneously with wind speed and direction. The values are
combined to quantify the rate at which PM is added to the atmosphere by vehicular travel along the
road segment. It is assumed for the purposes of this method that the road itself represents a
homogenous emission source, therefore small variations in wind direction are insignificant to
results. For more information about this method, see Section 3.11 of this document.
-------
ORS Handbook
Section 3.0
Page 3-162
It is important to note that if no calibration data are available, mobile monitoring data collected
using OTM 34 are still valuable for determining the ratio of PM emitted by different road segments,
which is useful for determining the effectiveness of road dust mitigation efforts. Calibration is
required, however, if absolute emissions data are required.
Data Collection
Prior to testing, all PM monitors and flow measurement instruments should be calibrated according
to manufacturer recommendations, or shown to have been cleaned, serviced, and recalibrated
within one year (12 months) of the test. Sample lines should be inspected and cleaned as needed.
The flow through the sample lines should be verified with an independent flow measurement device
to ensure any pitot tubes in the sample line are measuring accurately. Field personnel should log all
repair, calibration, and replacement information for the mobile monitoring system in a maintenance
log that is kept in a field notebook that resides in the test vehicle. An example pre-test preparation
checklist is provided in Appendix A of OTM 34. Please follow this Internet link:
https://www3.epa.gov/ttnemc01/prelim/otm34.pdf).
Once the mobile monitoring system is set up and fully functional, the measurement collection
process is automated. All data are uploaded to two separate data storage locations for redundancy.
Data files should be checked to ensure the time stamps for the first and last data points collected
match the duration of the measurement effort. Any discrepancies should be noted in the field
notebook.
Data Analysis
Once data have been collected and stored, data analysis begins. OTM 34 outlines eight steps that
are needed for appropriate data analysis. For more details about each of these steps, refer to
Section 10 of OTM 34.
1. Review One-Second Raw Data Records: Review raw data for missing information and invalidate or
remove data as discussed later in Section 3.12.3.
2. Calculate One-Second Net Raw PM Concentrations: Subtract the background PM concentrations
from the measured PM concentrations collected at the same time.
-------
ORS Handbook
Section 3.0
Page 3-163
3. Apply Time Delay Correction to Raw Data Records: Apply the time delay created by the movement
from the sample inlet to the sampler to the data so that events recorded by the GPS can be
accurately correlated with the correct PM measurements.
4. Apply GPS-based Validity Criteria: Exclude measurement periods where GPS data indicate a quality
assurance issue in the sample data (see Section 3.12.3 for more information.)
5. Apply PM Mass/Optical Correction: If a filter-based method for measuring PM was performed in
parallel with the mobile measurement study, the ratio of the two measurements should be applied
to all raw signals calculated using step 2.
6. Calculate Emission Factors: Calculate the PM emission factors on a one-second basis using the
equation above.
7. Calculate Road Segment-Average Emission Factors: Associate each data point with a road segment
and calculate average emission factors.
8. Tabulate Results: Collect road segment emission factors in a spreadsheet or database for further
review. It is also possible to use GIS software to illustrate this information.
Verification/Validation Studies
Clark County. Nevada
In February 2005, road dust emission factors were measured on a 100-mile stretch of road in Clark
County, NV.6 Over four consecutive days, the same measurement route was completed using the
Testing Re-entrained Aerosol Kinetic Emissions from Roads (TRAKER) mobile monitoring system. At
the time measurements were collected, the TRAKER system was not calibrated. However, a larger
study called the Clark County Stage IV study7 was performed in 2006 that provided a calibration
scheme that was retroactively applied to the data collected in 2005.
During the latter study, the calibration values were determined via the plume profiling method used
during the Clark County Stage IV study. The profiling method used a tower equipped with five
nephelometer-style PMio monitors at five different heights above the ground on the downwind side
of the road. The data collected were then mass-corrected by comparing the values collected by a
nephelometer and filter-based monitor when the same road dust material was suspended in the
laboratory. The horizontal flux was then calculated by numerical integration. After examination of
the data generated during Clark County Stage IV, it was determined that the ratio of the calculated
emission factor to the raw measurement data collected by the mobile monitoring system was not
-------
ORS Handbook
Section 3.0
Page 3-164
dependent upon the speed of the test vehicle. This means the value for Ki, shown in the above
equation, is constant with speed.
The value for Ki determined during the Clark County Stage IV study was then retroactively applied to
the measurement data collected during the Clark County study performed in February 2005. At the
same times, silt loading measurements were performed in accordance with USEPA AP-42, 5th
Edition.8 Although the entire 100-mile measurement route was executed in three to four hours, it
took the same amount of time to perform the silt loading measurements for each site.
Because the same measurement route was monitored four consecutive days, the repeatability of
the TRAKER mobile monitoring system could be determined. Using the assumption that road dust
emissions were constant over the four days of monitoring, the study showed that the average
coefficient of variation (COV, a measurement of uncertainty) decreased as the test vehicle travel
speed increased, as shown in Figure 3-58.1
80%
60%
20%
0%
Test Vehicle Travel Speed (mph)
Figure 3-58 Coefficient of Variation (COV) determined from data collected in Clark County, NV in February
2005.
-------
ORS Handbook
Section 3.0
Page 3-165
Typical QA/QC
Several types of quality assurance and quality control methods are utilized in OTM 34 to ensure
accurate measurements and sample traceability.
Screening Criteria
Due to the variable nature of driving a test route, it may be necessary to screen, or eliminate, data
associated with events not related to goals of the monitoring effort. OTM 34 recommends that data
collected during the following events should be removed from the overall dataset:
Travel speed is less than 11 miles per hour (mph);
Acceleration or deceleration is outside of the prescribed range (ฑ 1.3 mph);
The vehicle is performing a turn and the wheel angle is greater than 3ฐ from the vehicle
body;
Exhaust or road dust plume from another vehicle interferes with measurement equipment;
Corresponding GPS data are invalid;
Measured concentrations are outside of the instrument's measurement range; and
Correlating background concentrations indicate interference from other sources of PM.
Depending on the monitoring program, it may also be necessary to screen data collected during
events such as the proximity to an intersection or location where excess road debris is present, such
as a construction site.
Collocation
The mobile monitoring system used to quantify emissions during testing should be collocated with a
second identical mobile monitoring system, or other system that measures the same quantities with
the same or better precision and bias. When the data are reviewed, PM concentration data changes
should be recorded by both instruments within similar time periods. The data collected from each of
the two systems correlate well when a regression line fit between two instruments of the same PM
fraction yields a slope ranging from 0.8 to 1.2 for a dataset of 100 or more points.
-------
ORS Handbook
Section 3.0
Page 3-166
Likewise, it is beneficial to compare concentration measurements collected from the left and right
sides of the vehicle in mobile monitoring systems where the sample inlets are installed behind the
front wheels. In general, PM concentrations from both sides of the vehicle will rise and fall at the
same time. It is important to note that there may be periods where PM concentrations between the
two sides of the vehicle are very different because the lane of a road is not considered homogenous.
Siting Concerns
The role of the testing operator is to drive a route within the road network of interest, ensuring that
the route driven is consistent with the goal of the monitoring effort. Considerations when designing
a measurement route include types of roads, type of activity studied, and time of year.
Strengths and Limitations
Table 3.33 summarizes the strengths and limitations of OTM 34.
Table 3.32. Strengths and Limitations of the OTM 34 Approach
Strengths
Limitations
Dust emissions can be measured over many
miles of roadway, allowing many
measurements over a roadway network.
Exhaust from nearby vehicles may result in
higher PM concentrations than can be caused
by road dust emissions.
Many miles of roadway can be monitored in the
time it takes to plan, set-up, and take
measurements at a single site using stationary
road dust monitoring methods.
Measurement data must be calibrated against
an external standard.
Meteorological data collection is not necessary,
unlike stationary monitoring methods.
Wet roadway conditions can cause erroneous
estimates of PM emissions.
Extensive maintenance and repair of PM
measurement instruments may be required if
allowed to get significantly wet.
-------
ORS Handbook
Section 3.0
Page 3-167
References
1. U.S. EPA. Other Test Method - 34: Method to Quantify Road Dust Particulate Matter
Emissions (PM10 and/or PM2.5) from VehicularTravel on Paved and Unpaved Roads.
January 2014.
2. Fitz, D.R. (2001) Measurements on PM10 and PM2.5 Emission Factors from Paved Roads in
California. Final Report to the California Air Resources Board under Contract No 98-723,
June 2001.
3. Kuhns, H., Etyemezian, V., Landwehr, D., MacDougall, C., Pitchford, M., and M. Green.
(2001). Testing Reentrained Aerosol Kinetic Emissions from Roads (TRAKER): A New
Approach to Infer Silt Loading on Roadways. Atm. Env. 35: 2815-2825.
4. U.S. EPA, 1993. Emission factor documentation for AP-42 4th edition, Section 13.2.1: Paved
Roads. Report prepared for U.S. EPA, March 8,1993 by Midwest Research Institute, 425
Volker Boulevard Kansas City, Missouri 64110-2299 USA.
5. U.S. EPA. Other Test Method - 32: Determination of Emissions from Open Sources by
Plume Profiling. June 2013.
6. Etyemezian, V., H. Kuhns, and G. Nikolich (2005). The Las Vegas Road Dust Emissions
Technology Assessment, Phase II: Final Report. Prepared for the Clark County Department
of Air Quality and Environmental Management, Las Vegas, NV. July, 2005.
7. Langston, R., R.S. Merle Jr., V. Etyemezian, H. Kuhns, J. Gillies, D. Zhu, D. Fitz, K. Bumiller,
D.E. James, and H. Teng (2008). Clark County (Nevada) Paved Road Dust Emission Studies in
Support of Mobile Monitoring Technologies: Final Report, 122 pages.
8. U.S. EPA (1999). Compilation of Air Pollutant Emission Factors - Vol. I, Stationary Point and
Area Sources. Report No. AP-42, 5th ed. U.S. Environmental Protection Agency, Office of
Air Quality Planning and Standards, Research Triangle Park, NC.
-------
ORS Handbook
Section 4.0
Page 4-1
4.0 Meteorological Measurements
Meteorological conditions at the site of optical remote measurements are an important
component of many of the applications. Please note that in-depth description of meteorological
measurements information can be found in Volume IV of EPA's QA Handbook series.1 Quality of
the meteorological parameters measurements is as important as the quality of the optical remote
sensing results used in emissions flux calculations. The quality of meteorological parameters
measurements includes topics such as:
Meteorological tower siting and setup,
Wind speed (horizontal and vertical) and direction,
Relative humidity and dew point,
Temperature,
Solar radiation and
Atmospheric pressure.
Historical meteorological parameters measurements are important to the applications of
measurements because emission rates from open sources are affected by ambient conditions (e.g.,
high wind speed over the source can increase the emission rate). If the data are intended to
evaluate exposure or determine emissions factors, the ultimate use of the measurements should
include evaluation of how well the meteorological conditions encountered during the test compare
to the annual average meteorological conditions for the sampling site. Ideally, the test report
summary includes data or commentary that addresses the representativeness of the
meteorological conditions. The meteorological conditions recorded during the test can be
compared to the historical trends for a site. If site-specific meteorological measurements are not
available, current local conditions can be compared to the average statistics available from the
nearest National Weather Service (NWS) monitoring station. If there is more than one NWS station
-------
ORS Handbook
Section 4.0
Page 4-2
near the measurement site, then the comparison should be made against the average
measurement of the NWS stations.
Wind stability is often important to measurements and refers to atmospheric turbulence.
Atmospheric stability or turbulence is the vertical and horizontal transportation of an air mass.
Atmospheric turbulence is the collective differences between small-scale air motions driven by
winds that vary in speed and directions for a given parcel of air. Measurements that lead to wind
stability account for convection, diffusion, buoyancy, rapid variation of pressure and wind velocity.
Turbulence is responsible for mixing the atmosphere and is what distributes water vapor,
particulate matter and gases.
However, one does not measure or calculate turbulence directly to assess atmospheric turbulence;
instead we examine the atmospheric stability and the potential for vertical and horizontal
transport, as well as mixing while in transport. Vertical and horizontal transport is normally
associated with atmospheric stability (although it is important to note that wind speed will play a
major role in the horizontal plane).2'3'4
A practical use for "wind stability" or turbulence for ORS measurements is to examine atmospheric
stability and assess the atmosphere's potential to mix and transport (vertically and horizontally).
Another concept that is associated with "wind stability" is surface roughness as it relates to
topographical and landscape variability. Determining if one is under stable or unstable
atmospheric conditions is an important step in evaluating when conditions are appropriate for
sampling, depending on the objectives of the measurement application. For example, the
combination of vertical and horizontal wind speed and direction with solar radiation can be used to
determine the atmospheric stability class of the air parcel by stability categories.5 These stability
categories can be used as a quality indicator for representative ORS measurements.
4.1 Meteorological Station Siting
Typically, the terrain associated with an optical remote monitoring measurement program is
complex only to the extent that the local elevation provides an elevated or depressed land area
-------
ORS Handbook
Section 4.0
Page 4-3
with sloped sides. A properly located meteorological monitoring station should provide
meteorological measurements for an entire facility. For facilities or sites that exceed one square
mile in area, an additional meteorological monitoring station may be required for each additional
square mile of area to obtain accurate local meteorological conditions.
The meteorological monitoring station must be positioned at the center of the highest point near
the measurement location. Winds blowing across the top of elevated areas and winds down the
slopes contribute to transport and dispersion characteristics. By positioning the station at the
center and highest land mass near the measurement site (i.e., as determined by Google earthฎ
map elevations), all upwind wind patterns sweeping the source emissions are represented in the
wind speed and direction measurements.
Obstructions must also be considered during siting of a meteorological monitoring station. There
is no specified distance that the station must be positioned away from the measurement location.
By positioning the station at the center of the highest elevation, effect/contribution of the winds
blowing over elevated areas are normalized with respect to the direction of the wind to the
greatest extent possible.
Meteorological measurement sensors are typically setup or positioned at 2 meters above ground
level. The wind/speed direction sensors must be positioned so that they are located no closer
than 2 meters from the temperature sensor. Wind direction sensors are oriented to true north
using a digital compass. The timers/clocks on the meteorological station and the optical
measurement equipment should be synchronized to allow the concentration data and
meteorological data to be directly compared to the measurement data during post-processing. If
synchronization is not possible, the offsets between the clocks will be recorded at the start and
end of data collection of each day to estimate any differences in the clock times or rates.
Horizontal Wind Speed and Direction
Compact weather stations6 can provide all-in-one measurement of temperature, horizontal wind
speed, and horizontal wind direction at optical remote measurement sites. The compact weather
-------
ORS Handbook
Section 4.0
Page 4-4
station typically uses 2-D ultrasonic measurement technology for wind speed. For wind direction
measurements, the system automatically and continuously self-aligns to magnetic North. Typical
units can provide the following performance measurements:
Wind speed range from 0-50 m/s with an accuracy of ฑ 0.5 m/s
Wind direction range 09 - 3609 with an accuracy of ฑ 59
Vertical Wind Speed and Lateral Turbulence
If wind stability is required as an ancillary measurement for the optical remote measurement
application, then a 3-D Ultrasonic Anemometer may be necessary. Typical accuracy of this
instrument is ฑ 2-degree compass (1 to 30 m/s) or ฑ 5 degrees compass (30 to 40 m/s).
Relative Humidity
Relative humidity (RH) is important to measurements because it provides information on water
interference and predicting cloud or fog formation which interfere with many techniques. Of the
many atmospheric variables describing water vapor content in the atmosphere, RH is the most
common for routine monitoring programs. RH is the ratio (percent) of actual vapor pressure of
moist air to the saturation vapor pressure at the same temperature. A corollary measure, dew-
point temperature (or dew point) is the temperature to which a moist air parcel must be cooled to
achieve saturation over water at constant pressure and water vapor content. RH and dew point
are measured with electrical hygrometer, chilled mirror, or psychrometric instruments. Typical
accuracy of this equipment is ฑ 0.5% RH.
Temperature
Temperature is important because it determines or controls vertical transport of air which changes
portion of the plume measured by the ORS technology. Standard meteorological equipment
includes a thermocouple or thermistor temperature sensor with a temperature range - 509C to
+509C and an accuracy oft 0.29C.
-------
ORS Handbook
Section 4.0
Page 4-5
Net Solar Radiation
Net Solar Radiometer is required for wind stability determination. The typical accuracy of these
units is > 90 percent of the Daily Total Solar Radiation.
Atmospheric Pressure
For air quality and meteorological purposes, atmospheric pressure is generally measured with
mercury, aneroid, or electronic barometers. Most, if not all, of the atmospheric pressure sensors
available provide analog or serial output that is directly interfaced with a data acquisition system.
A mercury barometer measures the height of a column of mercury that is supported by the
atmospheric pressure. It is a standard instrument for many climatological observation stations,
but it does not afford automated data recording. An aneroid barometer consists of two circular
disks bounding an evacuated volume. As the pressure changes, the disks flex, changing their
relative spacing which is sensed by a mechanical or electrical element and transmitted to a
transducer. Most electronic barometers of recent design use transducers which transform the
sensor response into a pressure-related electrical quantity in the form of either analog or digital
signals. Current digital barometer technology employs various levels of redundancy to achieve
long-term stability and accuracy of the measurements. One technique is to use three
independently operating sensors under centralized microprocessor control. Even higher stability
and reliability can be achieved by using three completely independent barometers, incorporating
three sets of pressure transducers and microprocessors. Each configuration has automatic
temperature compensation from internal-mounted temperature sensors. Triple redundancy
ensures excellent long-term stability and measurement accuracy, even in the most demanding
applications.7
Differential Global Positioning for Tracking Monitoring Locations
Global position information using a high-resolution Differential Global Positioning System (DGPS) is
required for mobile tracer correlation optical measurements and temporary stationary
-------
ORS Handbook
Section 4.0
Page 4-6
measurement locations.8 Global position high resolution accuracy of <0.6 meters will meet most
application requirements. The DGPS data collection includes a time stamped data stream acquired
in real time. The DGPS clock should be synchronized with the measurement system clock to
simplify measurement location and concentration information.
In the field coordinates measured by the DGPS, unit results should be compared to known Google
Earthฎ coordinates for known geodetic marks as a quality check of the location system.
Collection of Process Information
Process information is site-specific data on factors or activities that affect the production or release
of pollutants to be measured. Process parameters are measured indicators of process
performance such as duct flow, operation temperature, or fuel use. The measurement of
emissions rate, also called emission flux, may be extended beyond mass per unit air volume to
determining emissions factors when the flux can be related to activity factors of the source.
Emission flux is often converted into an emission factor to estimate air pollutant emissions from a
process or activity (e.g., fuel combustion, chemical production). The simplest form of an emission
factor is an expression of the mass of pollutant emitted per unit of activity generating the pollutant
(e.g., pounds of particulate matter emitted per ton of coal burned). Typically, emission factors for
stationary point sources are developed by dividing the source's emission rate by an appropriate
parameter (e.g., number of widgets made per hour) that represents the activity responsible for
generating the emissions. Therefore, gathering process information related to production,
chemical use, energy use, heat or power generation is important to assess the relative rates of
pollution produced by a stationary source. Once process information is available, ORS flux
measurements of open source activity can yield meaningful emission factors.
In developing emission factors for point sources, identification of the underlying activity that
generates the emissions is typically straightforward. For example, particulate matter emissions
from fuel combustion are a direct function of the type and amount of fuel burned. However, for
open sources (e.g., landfills) the pollutant generation and emission release mechanisms tend to be
-------
ORS Handbook
Section 4.0
Page 4-7
more complex. This complexity can make the selection of appropriate underlying activity,
emission precursor, or process parameter(s) for use as an emissions surrogate more difficult than
for point sources. Consequently, emission factor developers should have a thorough
understanding of the pollutant generation and emission release mechanisms for a given open
source to accurately interpret the results from the ORS test and to properly apply the ORS data for
emission factor development.
One strength of the techniques in this Handbook is measurement of fugitive emissions from open
sources such as landfills, wastewater treatment systems, agriculture operations, and equipment
leaks at petrochemical and industrial facilities. At landfills, wastewater treatment systems, and
animal agriculture operations, fugitive emissions are generated by biological decay of organic
matter present in the waste. The rate of biological activity is affected by ambient conditions (e.g.,
bioactivity increases with increasing temperature) and process parameters such as chemical
conditions (pH, reduction/oxidation potential) in the waste. Pollutant emissions are also affected
by site-specific parameters such as the process information on the configuration of the source and
the process steps involved in handling and disposing of the waste. At petrochemical and industrial
facilities, fugitive emissions from equipment leaks are a function of process information such as
the type of equipment; the number of equipment components; the concentration and vapor
pressure of pollutants in the in-service gas; and process parameters such as temperature and
pressure. The complex transport and diffusion mechanisms, and the chemical/biological reactions
inherent in certain types of open sources, mean that the emissions from the open source may not
be easily related to an industrial process or activity. For example, an industrial process generates a
liquid waste stream that is discharged to an open wastewater treatment system. Although the
pollutant loading to the treatment system may be relatively constant over time, the emissions
from the system may not consistently track the loading rate due to changes in the rate of pollutant
formation caused by increases/decreases in process parameters of the treatment system such as
temperature.
The output ultimately obtained from a test is an emission rate in terms of mass of pollutant
-------
ORS Handbook
Section 4.0
Page 4-8
emitted per unit of time. For point sources, the activity typically selected for emission factor
development has a time component such that the use of the activity in the denominator of the
factor cancels out the time units of the measured emission rate. For example, the use of a boiler's
fuel feed rate (Mg coal/hr) as the activity converts the measured emission rate (g pollutant
emitted/hr) into an emission factor in terms of g pollutant emitted/Mg of coal fired. The activity
data selected for use in development of an emission factor based upon measurements may or may
not have a time component.
Simple emission factors for a specific site can be developed using the emission rate measured by
one of several of the technologies described in Chapter 2 and a characteristic activity factor related
to the emissions factor of interest. However, the source characteristic is not necessarily a simple
time-dependent activity independent of the site. The addition of site-specific or process- specific
information improves emissions factors estimates. For example, a simple emission factor (kg of
pollutant/ft2 of landfill surface area) could be developed using test data and the surface area of the
landfill. However, because landfill fugitive emissions are also dependent on the type of cover and
gas collection system (if applicable), the type of material contained in the landfill, the retention
time of the material, and the size and number of landfill cells, the applicability of the simple
emission factor to other landfills based solely upon surface area would be limited. A more refined
landfill emission factor (e.g., in terms of mass of pollutant emitted/mass of pollutant generated)
could be developed using the measured emission rate and an estimated pollutant generation rate,
such as one calculated from the site specific biological decay model discussed in Section 2.4.4.1
(Municipal Solid Waste Landfills) of EPA's AP-42, as the activity. This refinement (i.e., emission
factor) would an intermediate estimate and more site-specific than the simple factor, yet less
specific than uniquely measuring the emissions flux for every landfill of interest. The additional
value of the process parameters necessary to utilize the decay model (e.g., pollutant generation
potential of the waste, time since initial waste placement in the landfill) is less costly than multiple
field sampling episodes at different landfills.
-------
ORS Handbook
Section 4.0
Page 4-9
Attribution of Emissions to Source of Intent
Results from an ORS sampling episode can be used in a dispersion modeling analysis conducted in
reverse. In dispersion modeling for point sources, the emission rate from the source is known and
the concentrations at receptor points downwind of the source are estimated based on the release
characteristics of the source (e.g., stack height, exit velocity, gas temperature) and the
meteorological data (e.g., wind speed and atmospheric stability parameters) used in the modeling
analysis. In ORS sampling, the sampling path or plane effectively serves as a downwind receptor
and the emission rate from the open source is back-calculated based upon measured downwind
concentrations and the wind speed and atmospheric stability data measured during the test.
Consequently, the open source emission rate is directly dependent upon the ambient conditions
encountered during the test.
The emission rates determined using ORS techniques can also be affected by background pollutant
concentrations in the atmosphere surrounding the open source, and by how well the placement of
the instrumentation (e.g., transmitters, receivers, retro-reflectors) captures the area source
emissions. Consequently, data users should determine whether background concentrations were
accounted for in the measurements. Typically, background emissions are determined by
measuring pollutant concentrations in sampling paths or planes located upwind of the emission
source and subtracting those concentrations from the concentration values measured at the
downwind locations. For assessing the effectiveness of the instruments to capture the source
emissions, the developer should review the placement of the instrumentation, and any
assumptions made regarding the prevalent wind direction, relative to the emission source to be
measured. The technique could be applied properly, but the configuration of the sampling
equipment relative to the emission source and the prevailing wind direction may not adequately
capture the source emission plume under varying meteorological conditions.
4.2 References
1. Quality Assurance Handbook for Air Pollution Measurement Systems, Volume IV:
Meteorological Measurements Version 2.0 (Final), EPA-454/B-08-002 March 2008.
-------
ORS Handbook
Section 4.0
Page 4-10
2. Daniel J. Jacob, Introduction to atmospheric chemistry, Princeton University Press, 1999.
3. John M. Wallace and Peter V. Hobbs, Atmospheric Science an introductory survey,
Academic Press, 1977.
4. David G. Andrews, An introduction to atmospheric physics, Cambridge University Press,
2000.
5. EPA 2000. Meteorological Monitoring Guidance for Regulatory Modeling Applications,
EPA-454/R-99-005.
6. A Climatronics Corporation AIO compact weather station A 2-D Ultrasonic Anemometer
(AIO Compact Weatherstation, Model 102780, Climatronics, Bohemia, NY).
7. World Meteorological Organization. Guide to meteorological instruments and methods
of observation; Draft seventh edition ed.; Secretariat of the World Meteorological
Organization; Geneva, Switzerland, 2006.
8. DGPS, Cressent R100 Series, Hemisphere GPS, Calgary, AB, Canada.
-------
ORS Handbook
Section 5.0
Page 5-1
5.0 Data Validation and Verification
This section provides information on validation and verification of remote optical measurements
starting from field observation through test report review. The emphasis in this section is how an
optical technology report recipient can evaluate and verify the quality of the reported data.
Information is provided on how data quality indicators can be used to assess and verify optical
monitoring data in the most general way.
Data review, verification, and validation are techniques used to accept, reject, or qualify data in an
objective and consistent manner. Verification can be defined as the process of evaluating the
completeness, correctness and conformance of a specific data set against the data quality
requirements.i Verification can be done by examination and objective evidence that the final data
meets specified QC recommendations or requirements and fulfills the data users' requirements.
Validation can be defined as a confirmation by examination and objective evidence that the
recommendations for a specific intended use are fulfilled. The criteria for deciding the degree to
which each data item has met its quality specifications should be described in an organization's site
specific QAPP. The QAPP should clearly indicate the plan to meet the end user's DQO. The DQO
process was described in Chapter 1 Section 1.4.
This data validation and verification section describes the techniques used to make assessments of
the application of remote optical air methods to field measurements. In general, the initial
assessment activities are performed both by persons implementing the environmental data
operations and by personnel "independent" of the operation, such as post test data reviewers and
the organization's QA personnel. The procedures, designated personnel, and frequency of
assessment should be included in an organization's QAPP. These activities should occur prior to
submitting the final data report and before they are used in models or emissions factors
development. Field testers should verify results from a field test program before data users
validate measurement results and use them to make decisions.
-------
ORS Handbook
Section 5.0
Page 5-2
5.1 General Approach
Specific QC specifications for each optical technique are provided in the individual technology or
applications sections in Chapters 2 and 3. The general information in this section can be used for
most optical remote measurement projects. How closely a measurement represents the actual
environment at a given time and location is a complex issue that must be considered during
development of the sampling design. Testers check (verify) each measurement for conformity to
the specifications, including type and location (spatial and temporal). Modelers and other
secondary data users should compare project quality specifications to their data needs and
determine (validate) if the optical remote measurement data is useful for their purpose. By noting
the deviations in sufficient detail, modelers and secondary data users will be able to determine the
data's usability for scenarios different from those included in project planning.
Remote sensing methodology and meteorological data are often linked. Pollution enters the
atmosphere directly, or is formed by chemical reactions in the atmosphere, or it is the result of a
process. Photochemical pollutants, such as 0^ and sulfates, are generally produced over a period
of time. Ozone forms by the interaction of VOCs and NOx under the right meteorological
conditions when low wind speeds, variable wind directions, and relatively high temperatures are
present. Other pollutants are generated by point, mobile, and area sources. Winds, a
meteorological variable, can transport pollutants from their sources to affect populated areas.
5.2 Data Validation Methods
Therefore, if possible, meteorological data should be verified and validated at the same time as
remote sensing data, not separately. Figure 5-1 is an illustration of a typical verification and
validation process. The left column shows the "levels of data review." These levels of data are
described in this Section. The right column illustrates the type of verification or validation that
usually occurs during the process. The numbers in parentheses reference the section numbers in
this document that provide additional details on the data review process at each step in the
hierarchy.
-------
ORS Handbook
Section 5.0
Page 5-3
Levels of Data Quality Review
Generally, there are four "levels" of air quality measurements data review. These levels are similar
to those defined by Mueller and Watson2 and Watson et. al.3 When a data set has undergone
each level of review, it passes on to the next level. The entire process is used to determine the
validity of the data.
Level 0 verification includes raw calibration data and initial setup observation prior to collecting field
data. Testing staff should report results of manufacturer calibration and verification that they
perform prior to a field campaign. These data can include background and noise measurements
made to establish a baseline for sensitivity of the measurements. Level 0 verification also includes
field observation of the equipment setup and function. At this level, the data may be reduced and
possibly reformatted, but are unedited and un-reviewed. These data have not been adjusted for
known biases due to interfering components in the air at the test site or other problems that may
have been identified during field maintenance checks or audits. These observations and data may
be used to monitor instrument operations during the measurement episode but should not be used
for regulatory purposes. Section 5.2 provides more details on this level of validation.
Validation and Quality Review
Verification Level Activity
Level 0
Review
Level 3
Review
Verify Quality checks were performed
during sampling
Verify Jniaal setup and calib'a;ion data
instalment -s ftnc'.jcning as expected
Validate data relative to ancillary
measurements and/or historical data.
Full post tsst Validation of data by an
independeci OA reviewer.
Figure 5-1. Generalized Data Verification and Validation Process Flow
-------
ORS Handbook
Section 5.0
Page 5-4
Level 1 data verification involves quantitative and qualitative reviews for accuracy,
completeness, and internal consistency. Quantitative checks are performed by instrument
software screening programs, and qualitative checks are performed by field staff who
manually review the data for outliers and problems. Quality control flags are assigned, as
necessary, to indicate the data quality. Data are only considered verified at Level 1 after
final QC checks have been completed and any adjustments, changes, or modifications to
the data have been made. Section 5.2.2 provides more details on this level of data
validation.
Level 2 data validation involves comparisons with independent data sets. This function
includes, for example, making comparisons to other simultaneous emissions
measurements or historical data on the source emissions.
Level 3 data validation involves a more detailed analysis and final screening of the data.
The purpose of the final step is to ensure that there are no inconsistencies among the
primary optical data and related data (such as meteorological measurements). The
reviewer examines the overall consistency of the data and the consistency of related data
(i.e., checking emissions patterns against time of day or wind speed and wind direction).
5.3 Data Verification Methods
Data verification is defined as the confirmation by examination and objective evidence that
specified quality requirements have been fulfilled according to the standard procedure for the
method. This verification contributes to the confidence that the data will be valid for the original
decision or purpose of the data collection. These recommendations should be included in the
organization's QAPP as part of the method quality indicators. The data verification process involves
two basic steps: visual inspection and analysis and verification performed by data review. Both
techniques are needed to verify optical measurements data. Each is described in the following
sections.
-------
ORS Handbook
Section 5.0
Page 5-5
Visual Data Verification
Verify Initial setup and calibration data
Instrument is functioning as expected
Figure 5-2. Level 0 Verification Checks the Most Fundamental Quality Requirements
The monitors and equipment used for remote sensing rely upon a radiation source (UV, visible, or
IR) and a detector used together to identify and quantify the levels of certain chemicals in the
atmosphere. These monitors are typically used in a continuous monitoring mode and monitor one
or several compounds simultaneously. Although the overall design requirements for the different
spectral ranges are significantly different, the basic components of these technologies are similar.
In general, these monitors contain at least the following components:
Radiation source
Optics
Detector
Data processing algorithms
The radiation sources for these technologies belong to one of three distinct groups. The monitors
operating in the UV region of the spectrum use a continuous or non-continuous lamp that provides
broad-band radiation in the UV and visible regions. The monitors, using TDL technology, use a laser
to provide radiation over a very narrow spectral range in the near-IR. That spectral range can be
tuned over a small range with a single TDL and is selectable over a wider range using multiple TDLs.
The FTIR monitors use a broadband IR source. Passive technologies such as IR cameras and Passive
FTIR measure natural IR radiance from the compounds being measured. The optical components
of these monitors typically are used to guide the radiation from the source, through the
atmospheric path to be monitored, to the detector. The detectors and configurations for these
monitors vary according to specific applications. They are typically chosen to maximize signal-to-
Level 0
Review
-------
ORS Handbook
Section 5.0
Page 5-6
noise ratio for the spectral region and operating temperature.4
Level 0 verification includes review of calibration of instruments and equipment. Periodic
calibration and/or calibration checks must meet MQOs identified in project QAPPs. Typical MQOs
are listed Chapter 2 for each measurement technology. Calibration data should be reviewed by
field test staff and data validation staff. The following questions should be answered:
Were the calibrations performed within an acceptable time prior to generation of data?
Were they performed in the proper sequence?
Were they performed using standards at the conditions expected during field
measurements?
Were acceptable linearity checks and other checks made to ensure that the measurement
system was stable when the calibration was performed?
Level 0 verification can also include field inspections to visually verify optical measurement
technologies performance during field acquisition of data. Field verification can be technical
systems audits (internal or external) or simple inspections by field operators. For example, optical
equipment often generates visible light as a direct or indirect result of the measurement process.
Field inspection can verify measurement equipment is aimed correctly and operating if reflected
visible light is apparent from the optical path. Optical equipment and associated reflectors can
gather dust or moisture, and observation of these two interferents can be made visually during
field inspection. Several questions might be asked during a visual verification process:
Is the equipment operational? Verification is performed by observing that the optical
equipment is collecting and storing data.
Was the equipment aimed correctly to make measurements? Verification is performed
by observing that the data collected is different from zero or full saturation.
-------
ORS Handbook
Section 5.0
Page 5-7
Part of the verification process is a review of the optical remote data over a period of
time. A quick visual inspection may reveal some anomalies that do not match other
parameters.
Continuous long term optical measurement verification programs should include documentation of
periodic field observations that ensure equipment is operating. Figure 5.3 is an example of a visual
observation records. Many environmental samples can be flagged (qualified) during the periodic
visual inspections.
Weekly Visual Quality Control Check Sheet
Optical Remote Instruments
Site Month/Year Site Number_Technician Date:_
[ ] System is powered, and operating
[ ] Source is- generatingadequateradsatianformea3urements
[ ] Optka! alignment is correct a ijowmg beam detection
[ ] Detectors maintained at the proper temperature and generating spectral signal
[ ] Calibration, span, zero checks are performed as appropriate
ฃ ) Standard Gas mixtures. are available if required
{ ) System operation includes periodic QC
AH comments must be noted in the system log.
Reviewed by Date
Figure 5.3 Example of Optical Remote Measurement Visual Check List
-------
ORS Handbook
Section 5.0
Page 5-8
Data Review and Verification
Level 1
Review
Verify Quality checks were performed
during sampling
Figure 5.4 Level 1 Verification Process
In the late 1990s, optical remote systems were developed with personal computer compatible
data collection routines. Many optical remote instruments offer remote access and download of
data from systems that are in continuous use. Steps preparatory to data validation should include
the daily transfer of raw data (e.g. signal averaged processed data) to a central data processing
facility and the transfer of raw data files to create an edited database. The raw data files should be
stored separately to insure data integrity. Backup copies of the data should be prepared and
maintained on-site and off-site.
For continuous optical remote monitoring systems, data can be processed and QC operations
parameters can be evaluated to determine if equipment maintenance is required. These types of
verification techniques can be extremely useful because the program can "sense" a change in
operating conditions or instrument response and a prediction of the possibility of equipment
failure.
Data reviewers should answer some typical questions during their remote download data review
such as:
Did the signal intensity drift or diminish significantly since the last equipment
maintenance?
Did the regular QC check for noise and or calibration exceed acceptable limits?
Did the optical system trip any electronic limits or indicate data collection failure?
Does the data fall within the measurement range of the instrument or was the data
saturated or zero?
-------
ORS Handbook
Section 5.0
Page 5-9
Does the system have a minimum detectable limit?
Validation of Primary and Ancillary Measurements
Level 2
Review
Validate data relative to ancillary
measurements and/or historical data
Figure 5-5 Level 2 Quality Checks Start the Data Validation Process
Both manual and computer-oriented systems require individual reviews of all data tabulations. As
with all environmental measurements, it is necessary to keep accurate records during
measurement periods to ensure a complete data collection. A site logbook and calibration sheets
should be maintained at the data collection site. The site logbook will include information such as
meteorological conditions, path lengths, UV filter numbers, lamp type, light intensities and
measurement times. Light intensities must be recorded anytime an optical source or receiver is
adjusted and compared to the intensities measured when the equipment was installed.
Calibration sheets include the record and results of system calibration checks or audits performed
with known concentrations of a target or surrogate analyte.
Initial data verification steps should be performed by the station operator and later by data
validation staff. All necessary supporting material, such as audit reports and site logs, should be
available for Level 2 validation. Access to daily ancillary measurements such as wind speed,
direction, should be provided for use in relating suspect data to local and regional conditions. If
measurements are taken down wind of a facility, process information and production schedules
are useful to interpret trends or excursions in optical remote data. Questionable data, such as data
flagged in an audit, manual review should be corrected or invalidated during Level 2 data
validation.
For long-term continuous measurements programs, the data should be reviewed on a regular
schedule and at least monthly. For short-term measurements programs, the data should be
reviewed by the site operators at the end of each day. Optical measurement instruments typically
-------
ORS Handbook
Section 5.0
Page 5-10
include onboard personal computers that allow operators to view and evaluate data visually.
Graphs or plots of data or a summary table of data can be evaluated for outliers or obvious data
collection failures. Graphing data can be a quick method of visualizing the data relative to other
parameters. Graphs can show longer term trends and relationships that are difficult to see when
data validation staff are looking at large amounts of tabular data.
The purpose of manual data inspection is to spot unusually high (or low) values (outliers) that
might indicate a gross error in the data collection and to verify signal intensity. Manual review of
data tabulations also allows detection of uncorrected drift in the zero baseline of an optical sensor.
Zero drift may be indicated when the daily minimum values tend to deviate (increase or decrease)
from the expected minimum value over a period of several days.
In an automated data processing system, procedures for data validation can be incorporated into
the basic software. As noted in Section 5.2.2, the computer can be programmed to scan data for
extreme values, outliers, or ranges. These checks can be further refined to account for time of
day, time of week, and other cyclic conditions. Questionable data values flagged on the data
tabulation may or may not indicate possible errors. The system operator should check all the data
flagged by the acquisition system program and investigate whether the data flagged should remain
flagged. In some cases, extreme conditions can occur rapidly and the data may reflect real values.
For example, if a spill or leak occurs and moves through a measurement area the optical monitor
may record high values or extreme interference in the data. The system operator should note such
excursions and alert the data validation and reporting staff that these data may reflect real
conditions. Data validation in Level 2 evaluates the data completeness and representativeness
against the project DQO requirements to ensure sufficient data is collected for data users to draw
conclusions or make decisions.
A useful data validation method is to compare the difference between successive data values. Logic
dictates that rapid changes in values in a 1 to 15 minutes acquisition period would normally not be
expected. When the difference between two successive values exceeds a predetermined value,
-------
ORS Handbook
Section 5.0
Page 5-11
the data in question can be flagged for further evaluation. Screening is an iterative process in
which range checks and other screening criteria are revised as necessary based on experience. For
example, an initial QA pass of a data set using default criteria may result in flagged values which,
upon further investigation, are determined to be valid for a particular site. In such cases, one or
more follow-up QA passes using revised criteria may be necessary to clearly segregate valid and
invalid data.
Final Validation and Evaluation of Measurements for Data Users
Full posttest Validation of data by an
independent QA reviewer.
Figure 5-6 Level 3 Quality Checks Ensure the Data is Usable for the Purpose Intended
Data validation is a routine process designed to ensure that reported values meet the quality goals
and objectives of environmental data operations. A progressive, systematic approach to data
validation must be used to ensure and assess the quality of data. The purpose of this step in the
process is to detect, compare, and perform a final screening on all data values. Any final data that
may not represent actual conditions at the sampling site will be detected at this stage. Effective
data validation procedures usually are handled independent of the procedures of initial data
verification, that is, by different staff. It is important that data validation staff be independent of
field operators.
If data assessment results clearly indicate a serious response problem with the optical technology,
the agency should review all related information to determine whether the optical remote
assessment data, should be invalidated. Some problems that may escape detection during an
audit are often easily identified during data validation. Data validation should be performed by a
person with appropriate training in the optical technology who has a basic understanding of
instrument operation and typical results from similar measurement projects.
Level 3
Review
Data flagged by the QC screening should be evaluated by personnel with optical measurement
-------
ORS Handbook
Section 5.0
Page 5-12
expertise. Reasons for changes in the data resulting from the validation process should be
documented. If system problems are identified, corrective actions should also be documented.
Edited data should continue to be flagged so that their reliability can be considered in the
interpretation of the results of modeling analyses for which the data are used. Flags can be used in
the field and by the data reviewers to signify data that may be suspect due to calibration or audit
failure, special events, or failed QC limits. When calibration problems are identified, data
produced between the suspect calibration event and subsequent recalibration should be flagged.
Because flag combinations can be overwhelming and cannot always be anticipated, an organization
needs to review these flag combinations to determine whether single values or values from a site
over a period should be invalidated. Procedures for screening data for possible errors or anomalies
should also be implemented. When calibration problems are identified, data produced between
the suspect calibration event and any subsequent recalibration should be flagged to alert data
users.
5.4 References
1. Guidance on Environmental Data Verification and Data Validation, EPA QA/G-8, EPA/240/R-
02/004, November 2002.
2. Mueller, P. K.; Watson, J. G." Eastern regional air-quality measurements. Volume 1, Section
7" ;EPRI-EA -1914-Vol.l, final report prepared for Environmental Research and Technology,
Inc., Concord, MA, by Electric Power Research Institute, Palo A Ito, CA. 1982.
3. Watson, J. G.; Lioy, P. J.; Mueller, P. K." The measurement process: precision, accuracy, and
validity" . In Proceedings, Proceedings, Air Sampling Instruments for Evaluation of
Atmospheric Contaminants; 7th ed.; Hering, S. V., Ed.; American Conference of
Governmental Industrial Hygienists: Cincinnati, OH, 1989, pp 51- 57.
4. Battelle ETV Generic Verification Protocol for Optical Open-Path Monitors 2002.
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-454/B-18-008
Environmental Protection Air Quality Assessment Division August 2018
Agency Research Triangle Park, NC
------- |