United States
Environmental
Protection Agency
EPA Science Advisory
Board (1400F)
Washington, DC
EPA-SAB-CASAC-05-006
May 2005
www.epa.gov/sab
EPA's Final Draft National
Ambient Air Monitoring
Strategy
An Advisory by the Ambient Air Monitoring
and Methods Subcommittee of the EPA
Clean Air Scientific Advisory Committee
-------
\
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON D.C. 20460
April 20, 2005
EPA-SAB-CASAC-05-006
OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
Honorable Stephen L. Johnson
Acting Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, NW
Washington, DC 20460
Subject: Clean Air Scientific Advisory Committee (CASAC) Advisory on
Implementation Aspects of the Agency's Final Draft National Ambient Air
Monitoring Strategy (NAAMS) (December 2004)
Dear Acting Administrator Johnson:
The Clean Air Scientific Advisory Committee's (CASAC) Ambient Air Monitoring and
Methods (AAMM) Subcommittee ("Subcommittee") met on December 15, 2004 to conduct an
advisory meeting on implementation aspects of the Agency's Final Draft National Ambient Air
Monitoring Strategy (NAAMS or "Strategy"). The public meeting was held at the staff offices
of the EPA Science Advisory Board (SAB) in Washington, DC. Members of the Subcommittee
are recognized, national-level experts in one or more of the following disciplines or areas: (1)
atmospheric sciences and air quality simulation modeling; (2) human health effects and exposure
assessment; (3) air quality measurement science; (4) ecological risk assessment; and (5) State,
local agency or Tribal experience. The Subcommittee roster is found in Appendix A of this
report.
In general, the CASAC finds that the Agency's ambient air quality monitoring program is
beginning to implement the changes necessary to bring it in line with the NAAMS strategy
document. There are and will be a number of scientific issues that will arise as the progress is
made in reconfiguring the network and as new knowledge with respect to monitoring, modeling,
and effects becomes available. CASAC AAMM Subcommittee members' individual review
comments are provided in Appendix B.
1. Background
The CASAC, which comprises seven members appointed by the EPA Administrator, was
established under section 109(d)(2) of the Clean Air Act (42 U.S.C. 7409) as an independent
scientific advisory committee, in part to provide advice, information and recommendations on
the scientific and technical aspects of issues related to air quality criteria and national ambient air
-------
quality standards (NAAQS) under sections 108 and 109 of the Act. The CAS AC, which is
administratively located under the SAB Staff Office, is a Federal advisory committee chartered
under the Federal Advisory Committee Act (FACA), as amended, 5 U.S.C., App. The SAB Staff
Office established the CASAC AAMM Subcommittee in July 2004 as a standing subcommittee
to provide the EPA Administrator, through the CASAC, with advice and recommendations, as
necessary, on topical areas related to ambient air monitoring, methods and networks. The
Subcommittee complies with the provisions of FACA and all appropriate SAB Staff Office
procedural policies.
In late 2002, EPA's Office of Air Quality Planning and Standards (OAQPS), located
within the Office of Air and Radiation (OAR), issued a draft National Ambient Air Monitoring
Strategy. OAQPS subsequently requested that the CASAC review the draft NAAMS document
and provide advice and recommendations to the Agency on the technical bases and design
aspects of the Strategy. The SAB Staff Office announced the formation of the NAAMS
Subcommittee of the CASAC on November 5, 2002 (67 FR 67403). The CASAC NAAMS
Subcommittee held a public meeting in Research Triangle Park, North Carolina, on July 8-9,
2003 (68 FR 34945, June 11, 2003) to conduct this review of the draft Strategy document. The
primary recommendations of the CASAC NAAMS Subcommittee, through the chartered
CASAC, included a request for an implementation plan, and added emphasis on rural- and
ecosystem-oriented monitoring, support for the National Core Monitoring Network (NCore)
Level 1 program, and training and quality assurance to enhance data consistency across the
Nation. The CASAC NAAMS Subcommittee's complete report from this review is found on the
SAB Web page at URL: http://www.epa.gov/sab/pdf/casacl04001.pdf. OAQPS updated the
NAAMS document after the CASAC's review of the Strategy. The revision incorporated EPA's
responses to the CASAC NAAMS Subcommittee's recommendations.
Last spring, the SAB Staff Office announced (69 FR 19180, April 12, 2004) the
formation of the CASAC AAMM Subcommittee. This subcommittee replaced the former
CASAC NAAMS Subcommittee. Subsequently, OAQPS asked the CASAC AAMM
Subcommittee to conduct an advisory meeting for the purpose of providing advice and
recommendations on the implementation plan for its updated final draft NAAMS, which is
posted on EPA's Ambient Monitoring Technology Information Center (AMTIC) Web site at the
following URL: http://www.epa.gov/ttn/amtic/files/ambient/monitorstrat/allstrat.pdf.
2. CASAC AAMM Subcommittee Advisory on Implementation Aspects of the Agency's
Final Draft NAAMS
The purpose of this meeting was to review the progress toward implementation of the
Final Draft NAAMS. The OAQPS requested that the Subcommittee provide expert advice and
recommendations on the following charge questions, which focus on key implementation issues:
1. The CASAC has expressed its support for the Agency's proposal to redesign the routine
PM monitoring network to support PM precursor gas measurements (CO, SO2, NO/NOx,
NH3, HNOs) at NCore Level II multiple-pollutant sites, and for air quality management
decisions and to obtain relevant exposure data for research programs. Given limited
budgetary resources, does this represent both an appropriate and adequate balance, as
reflected by the relative resource allocations provided in Section 11, "Draft
-------
Implementation Plan," of the Final Draft NAAMS Document? In addition, are the
relative adjustments in the training and guidance approaches proposed in the draft
implementation plan consistent with the overall objectives of the Strategy?
2. The implementation plan proposes a series of communication actions to advance the
NCore Level 2 network, in order to more directly support long-term health effects
research and provide better support to ecosystem assessments through an increased level
of coordination. Does the CAS AC AAMM Subcommittee have additional suggestions
for addressing this need for integration and communication to the broader community of
"users," including scientific researchers (i.e., human health, atmospheric, ecological) and
State, local and Tribal (SLT) Agency representatives? More specifically, what is the
most effective manner for EPA both to reach-out to this broad user community and,
where appropriate, to incorporate their feedback and design input on such issues as
monitoring site locations and parameters?
3. One of the remaining technical issues relates to harmonizing rural- and urban-based
PM2.5 chemical speciation networks such that both categories of networks utilize
consistent sampling and analysis protocols. For example, EPA is considering converting
all of the Speciation Trends Network (STN) speciation sites to Interagency Monitoring of
Protected Visual Environments (IMPROVE) samplers and IMPROVE laboratory and
sample handling protocols. What are strengths and weaknesses of this approach?
4. As EPA implements the National Ambient Air Monitoring Strategy to address multiple
monitoring objectives, it will be looking to spatially optimize the ambient monitoring
networks. This may mean that some redundant monitors in adjacent, but separate,
geopolitical areas (e.g., neighboring counties) are "divested" from a given network.
Although technically sound, these divestments could result in data gaps which might, in
turn, adversely impact regulatory decision-making. The Agency is willing to adopt
alternative approaches for assessing regulatory issues such as non-attainment
designations, so long as such approaches are scientifically justifiable; hence, the rationale
for initiating discussion of these issues with the CASAC. Is it scientifically acceptable to
generate isopleths of airborne species concentrations through modeled observations
and/or integratedpredictive/observational fields that would be of appropriate uncertainty
for use in the regulatory decision-making process?
Before the discussion of the implementation plan and related issues, the Subcommittee
would like to commend the OAQPS staff for the responsive manner in which they have revised
the original draft monitoring strategy in response to the comments provided by the CASAC
NAAMS Subcommittee (EPA-SAB-CASAC-LTR-04-001) [the URL for this report is found in
the Background section above]. Several members of the present Subcommittee were also
members of this earlier group, and there was strong approval by these individuals to the changes
that have been made in the Final Draft NAAMS Document.
The Subcommittee discussed these issues sequentially and this report will provide the
subcommittee recommendations. Individual comments from the subcommittee members on each
of these issues are provided in Appendix B to this report.
-------
Question 1
The first issue is that of resource allocation as the redeployment of monitoring resources
starts to be implemented. The Subcommittee continues to support the need for Level 1 (LI) sites
as a means for testing of new monitoring technology and moving it to routine use. Substantial
advances in monitoring technology have been obtained through the EPA Particulate Matter (PM)
Supersites. That exercise provided the opportunity to test and refine new instrument concepts. It
is recognized that such an effort cannot be sustained at the level of the Supersites, but some level
of continuing support is critical to continue the improvements in data that are needed to provide
the underpinning of future regulatory decisions. Since one of the main purposes of LI sites
would be to help transition new methods into routine use, it may be useful to develop the LI sites
as cooperative efforts between a research team and state or local regulatory agency personnel.
There was some discussion as to potential sources of funding that could be diverted to the
establishment of LI sites. There were suggestions from several members that Photochemical
Assessment Monitoring Stations (PAMS) monitoring could potentially be reduced beyond the
recommendations of the implementation plan. It may be that there is not a need for as many
standard monitors and that more advanced instrumentation integrated with a few LI sites or
limited duration special studies might be more effective in providing the information needed for
making the needed air quality management decisions. However, in those locations where it is
felt useful to maintain existing data collection, consideration should be given to extending this
operation through the entire year. Although it is unlikely that there will be ozone violations in
the winter, the data on organic constituents can be useful for understanding exposures to and
sources of toxic hydrocarbons and PM2.5 precursors, for tracking effects of changes in gasoline
or diesel fuel content, and for evaluating emissions inventories and air quality models.
An important consideration is the allocation of resources for data analysis as an integral
part of the network. End uses of the data are too often afterthoughts, and planning for an initial
set of data analyses is an essential part of the design process. There are too many examples of
inadequate planning and resources to use the data collected by routine network operations and it
would be useful to better match the data uses with the effort to acquire the data as an integral part
of the overall plan. Although all of the possible uses for data cannot be anticipated, many of the
typical analyses can be anticipated and included in the implementation plans together with the
requisite budget. CASAC strongly recommends that the Agency give greater consideration to
identification of the amounts of financial and human resources that will be allocated for data
analyses and interpretation activities. In this connection we call special attention to the written
comments on "outreach and extension of findings" and "analysis and interpretation of
monitoring results" by CASAC members in Appendix 2.
There were suggestions that there needed to be more effort placed on time-resolved
measurements. Long-term integrated measurements lose critical information. It is generally
better to get more detailed time-resolved information for shorter time intervals than to have long
time interval integrated measurements. With careful design and appropriate statistical methods,
these episodic measurements can still lead to adequate descriptions of annual averages and
trends. There is so much additional benefit to the time-resolved measurements that the effort
needs to be made to move strongly in that direction. As indicated in the report from the earlier
-------
Subcommittee on Fine Particle Monitoring, it is important to retain an appropriate number of
integrated filter samplers to provide a basis for comparisons with the past monitoring record and
appropriate quality assurance checks on the monitoring program.
Although the commercially-available instruments commonly used for measuring NOX
concentrations are appropriate for ensuring compliance with the health-based national ambient
air quality standard for nitrogen oxides, they are not sensitive enough to be useful for making
decisions about NOX vs. VOC sensitivity of ozone non-attainment areas. Such information is also
important for understanding particulate matter formation. Most commercial instruments have a
minimum detection limit for NO of about 1 ppb, but decisions about NOX vs. VOC sensitivity
require instruments that have minimum detection limits of about 0.1 ppb. Similarly there is a
need for improvements in other gas phase species methods to provide the sensitivity required for
all of the other objectives other than compliance testing. There is also a need for more
measurements of key gas-phase species like NH3 and HNOs. At this time it appears there needs
to be additional development efforts to provide semi-continuous measurements of these species
and their particle-phase counterparts.
Finally, there needs to be greater consideration of general HC characterizations in NCore.
The treatment of carbonaceous compounds is currently a weak part of the monitoring strategy.
As a first step, continuous total NMOC measurements should be included in the NCore L2 sites
or at least a subset thereof.
With the promulgation of the Clean Air Mercury Rule on March 15, 2005 that will
require controls on mercury emissions from coal-fired power plants, it is important that
monitoring be put in place to measure the atmospheric concentrations of mercury species. Such
an effort should be put in place as quickly as possible so as to gather baseline data before
controls can actually be put in place and maintained for a sufficiently long period as to ascertain
the changes in atmospheric concentrations that occur as a result of these controls. Given that
there is likely to be regional caps, as part of a cap and trade program, appropriate regional
strategies will be needed to provide the necessary accountability data.
Question 2
There is consensus within the committee that the single most effective way for EPA to
reach out to potential users of its data is to make these data easily accessible via the internet. The
committee was encouraged by the staff presentation on current plans for Web access, and
expressed a desire to move rapidly toward getting this portal into operation with the flexibility to
adjust it as experience with its use is obtained. There is concern that there may be bottlenecks
between network operation and data availability. We support pushing the access to raw data
behind EPA's "AIRNow" Web site as quickly as possible and would like to encourage their
Steering Committee to move things ahead.
Another issue that needs to be addressed is computer-to-computer access. For users who
continuously update graphical displays or model calculations, access for an automatic download
of data is an important aspect of the accessibility of the data. Providing protocols for such access
should be part of the plans for the data access Web site.
-------
It would be helpful to combine ancillary data from other agencies such as the visibility
data from airports into the data base. It may be useful for EPA to discuss the development of
combined data sets with the variety of other Federal agencies that have air-quality data sets such
as NASA, NOAA, and DOE.
In terms of communications, it may be useful to consider organizing regional workshops
to get broader input into network modifications and data use. In particular, this may be the
appropriate way to bring the ecological community into the discussions. The move to site more
monitoring in rural areas needs the active participation of the ecological community to balance
the needs of the air quality modelers with those interested in understanding the ecological effects
of air pollution. Since ecosystems vary widely from region to region across the country, regional
workshops or similar forums can provide a means for meaningful participation by those
interested in ecological systems.
It is not clear that additional efforts are needed to get the health community involved
other than making the data readily-accessible. It is likely that a number of health researchers
would make use of data if they are easily accessible. However, changes to the network affect the
ability of epidemiologists to use the data and there needs to be effective communication with this
community to permit EPA to understand the potential impacts of any changes that are planned
for the network. The network is being redeployed to serve multiple objectives and regulatory
needs will take priority, but good communications can permit maximal exchange of points of
view and an opportunity to make better decisions on the redeployment of monitors.
Question 3
There is strong general support for making substantial changes to the 54 STN Trends
sites to ensure compatibility with IMPROVE data. Currently, it is not possible to fully combine
STN and IMPROVE data in spatial studies or in model evaluations since there are significant
differences in sampling and analysis. However, it needs to be fully understood what is meant by
IMPROVE-protocol sites. At present, the IMPROVE program controls all aspects of the system
from sampling to analysis. STN utilizes state and local agencies to operate the samplers.
The IMPROVE sampler uses a critical orifice rather than active flow control, and thus,
provides a less stable particle-size cutpoint. IMPROVE uses 25 mm filters and higher flow rates
that have the advantage of having higher areal density of material on the filters as compared with
the STN samplers. However, there has been concern regarding the possibility of clogging of the
Teflo filters that could lead to invalid samples during extreme events and/or to changes in the
cut-point for the cyclone inlets. IMPROVE-protocol sampling has been conducted in
Washington, DC; Seattle, WA; and Phoenix, AZ for a number of years without clogging being a
substantial problem. The IMPROVE experience has been that clogging has only been observed
in major fire episodes so it may not be a problem or could be accommodated by changing the
sampling protocol to a half-hour within each hourly time interval or by terminating the sample at
less than 24 hours if flow rates drop significantly. These data would be flagged appropriately to
indicate that they do not represent the full time interval.
-------
There are existing problems of harmonization of data from IMPROVE and STN that need
to be addressed. It has recently been recognized that there are significant differences in the XRF
results from the two networks and quite different approaches to the reporting of errors for the
XRF results. There are now on-going discussions between RTI, ChesterNet Lab and the
University of California at Davis regarding the XRF analyses and reporting of the XRF errors,
and it is anticipated that these differences may be resolved. There may need to be some
additional samples run by all of the laboratories in round robin studies to resolve discrepancies in
the values and reported errors, and OAQPS should support these efforts.
There are substantial differences in the organic and elemental carbon fraction (OC/EC)
methods. These differences have been reported in the literature. However, there is no "right"
answer since the measured species are defined by the analytical method. IMPROVE uses a
dynamic blank to correct the OC data while STN uses only field and laboratory blanks. Blank
sampling and reporting is thus an area requiring harmonization. Another key issue is the type of
filter. Because of potential differences in the type of quartz filters used by the STN and
IMPROVE programs, it is recommended that the STN use the same filter type as used by
IMPROVE (i.e.., Pall Tissuequartz) to ensure consistency. There are a number of other issues
such as differences in the storage, the analysis of cations, shipping conditions, etc. that need to be
harmonized. It should be noted that IMPROVE will soon be changing their method because of
the need to replace the existing analytical instruments. A commercial instrument is now
available to implement the IMPROVE time-temperature protocol. Studies to date suggest that
the OC and EC values by both the old and new instruments are comparable, but the values of the
thermal fractions can have significant variations. There will then be changes in the data being
reported in the IMPROVE network. Therefore, it is an appropriate time to change the STN
network to what will shortly be the new IMPROVE protocol so that there will be comparability
between the measurements.
Thus, the key question is whether or not to change the Trends sites to the IMPROVE
protocol at this time. If these 54 STN sites are fully converted to IMPROVE, it will guarantee
the comparability of the data and thus, permit the comprehensive use of the compositional data
obtained in the future. For other sites in the network, it would be useful to provide this
alternative to the SLT organizations that are control these sites. At this time, it is unclear if there
is sufficient capacity at the University of California at Davis to handle these additional sites. If
the filter preparation and analyses are done by the existing STN contractor, there will still be a
problem of data harmonization that needs to be addressed to ensure comparability of results from
the different laboratories. The Subcommittee recommends that to achieve fully comparable data,
it will require that all of the samples be collected in an identical manner with identical samplers.
The samples would then be analyzed by a single laboratory for any given chemical constituent
with a single approach to error estimation, data validation, etc.
The decision to change the monitoring approach for the STN and/or the IMPROVE
network must also take into account emissions sampling protocols and methods. If the two
databases are incompatible, this will inhibit accurately identifying source impacts. Similar
studies that compare ambient measurements (i.e., between the IMPROVE and STN methods)
need to be done for emissions measurements. If one technique (e.g., the NIOSH vis • «vis
IMPROVE for EC/OC) proves better for conducting emissions measurements (i.e., provides a
-------
more stable measurement), this should affect the choice for use in ambient monitoring (and vice
versa). Whichever techniques are ultimately used, all parties should recognize and account for
these tangential impacts and issues. Thus, there also needs to be the development of source
sampling methods that utilize the same sampling and analytical methods as the ambient network
so that the resulting source characterizations can be used in the interpretation of the ambient
aerosol composition data.
Question 4
The generation of surfaces of air quality parameters needs to be generated by a
combination of measurements and model simulations. Both measurements and model results
have uncertainties associated with them. Measurements are made at specific locations and
represent a limited geographical area. Alternatively, models average results over the minimum
size of the grid cell and cannot fully reproduce the local environment.
Thus, the use of integrated predictive/observational fields is the preferred approach.
While there is much work to be done here, this approach will help tackle multiple issues. First, it
is probably the best way to extend an observation (or sets of observations) both spatially and
temporally, if necessary. Second, it can be used in the process of source apportionment (or vice-
versa: source apportionment can be used in extending the use of observations). Third, it will
help identify uncertainties in the representativeness of the observations at a monitoring location.
Fourth and finally, it will produce the type of information that can be used by groups identified
in Question 2.
Therefore, conceptually the combination approach is the appropriate way to proceed.
However, we still do not have the best approach laid out, and the enabling technologies
developed (e.g., software/hardware environments to provide this information). These issues can
probably be tackled within three or four years. Such a process should utilize the observations
available, both in situ and remote (for example, satellite) as well as PM modeling (with data
assimilation). There should be a feedback loop where the information provided by the integrated
system utilizes additional approaches to assess the quality of the fields developed (e.g., data
withholding, etc.). While this may seem ambitious, EPA should set as a goal, to have a field of
source apportioned daily/monthly/yearly PM for the U.S. by 2010 (e.g., target PM2.5 and coarse
PM for 2008, with the apportioned fields developed by 2010). The work should also include
fields of the uncertainties in the integrated daily PM levels and, at least, in the annual source
apportionments. These results should be updated on an ongoing, annual basis. If the resources
were currently available, it is likely that this integrated approach could be achieved by 2008
(using 2006 data).
Thus, we find that the Agency's ambient air quality monitoring program is beginning to
implement the changes necessary to bring it in line with the NAAMS strategy document. There
are and will continue to be a number of scientific issues that arise both as progress is made in
reconfiguring the network and as new knowledge with respect to monitoring, modeling, and
effects becomes available. Thus, there is the need for a continuing process for the ongoing
review and assessment of the monitoring program. The CAS AC has found it to be very valuable
to review and offer its advice and recommendations to the Agency in the development of the
current monitoring strategy, and further advises that periodic reevaluation be built into the
-------
monitoring network plans to ensure that the network continues to meet the multitude of needs
that are expected to change over time. Therefore, we recommend that the CASAC Ambient Air
Monitoring and Methods Subcommittee continue to serve in this role. As always, the CASAC
wishes the Agency well in this very important endeavor.
Sincerely,
/Signed/
Dr. Rogene Henderson, Chair
Clean Air Scientific Advisory Committee
Appendix A - Roster of the CASAC AAMM Subcommittee
Appendix B - Review Comments from Individual CASAC AAMM Subcommittee Members
cc: Steve Page (MD-10) Jake Summers (MD-12)
Rich Scheffe (MD-14) Anthony Maciorowski (1400F)
Phil Lorang (MD-14) Fred Butterfield (1400F)
Tim Haniey (MD-14)
-------
Appendix A - Roster of the CASAC AAMM Subcommittee
U.S. Environmental Protection Agency
Science Advisory Board (SAB) Staff Office
Clean Air Scientific Advisory Committee
CASAC Ambient Air Monitoring and Methods (AAMM) Subcommittee*
CHAIR
Dr. Philip Hopke, Bayard D. Clarkson Distinguished Professor, Department of Chemical
Engineering, Clarkson University, Potsdam, NY
Also Member: SAB Board
CASAC MEMBERS
Dr. Ellis Cowling, University Distinguished Professor At-Large, North Carolina State
University, Colleges of Natural Resources and Agriculture and Life Sciences, North Carolina
State University, Raleigh, NC
Mr. Richard L. Poirot, Environmental Analyst, Air Pollution Control Division, Department of
Environmental Conservation, Vermont Agency of Natural Resources, Waterbury, VT
SUBCOMMITTEE MEMBERS
Mr. George Allen, Senior Scientist, Northeast States for Coordinated Air Use Management
(NESCAUM), Boston, MA
Dr. Judith Chow, Research Professor, Desert Research Institute, Air Resources Laboratory,
University of Nevada, Reno, NV
Mr. Bart Croes, Chief, Research Division, California Air Resources Board, Sacramento, CA
Dr. Kenneth Demerjian, Professor and Director, Atmospheric Sciences Research Center, State
University of New York, Albany, NY
Dr. Delbert Eatough, Professor of Chemistry, Chemistry and Biochemistry Department,
Brigham Young University, Provo, UT
Mr. Eric Edgerton, President, Atmospheric Research & Analysis, Inc., Gary, NC
Mr. Henry (Dirk) Felton, Research Scientist, Division of Air Resources, Bureau of Air Quality
Surveillance, New York State Department of Environmental Conservation, Albany, NY
A-l
-------
Dr. Rudolf Husar, Professor, Mechanical Engineering, Engineering and Applied Science,
Washington University, St. Louis, MO
Dr. Kazuhiko Ito, Assistant Professor, Environmental Medicine, School of Medicine, New
York University, Tuxedo, NY
Dr. Donna Kenski, Data Analyst, Lake Michigan Air Directors Consortium, Des Plaines, IL
Dr. Thomas Lumley, Associate Professor, Biostatistics, School of Public Health and
Community Medicine, University of Washington, Seattle, WA
Dr. Peter McMurry, Professor and Head, Department of Mechanical Engineering, Institute of
Technology, University of Minnesota, Minneapolis, MN
Dr. Kimberly Prather, Professor, Department of Chemistry and Biochemistry, University of
California, San Diego, La Jolla, CA
Dr. Armistead (Ted) Russell, Georgia Power Distinguished Professor of Environmental
Engineering, Environmental Engineering Group, School of Civil and Environmental
Engineering, Georgia Institute of Technology, Atlanta, GA
Dr. Jay Turner, Associate Professor, Chemical Engineering Department, School of
Engineering, Washington University, St. Louis, MO
Dr. Warren H. White, Visiting Professor, Crocker Nuclear Laboratory, University of California
- Davis, Davis, CA
Dr. Yousheng Zeng, Air Quality Services Director, Providence Engineering & Environmental
Group LLC, Baton Rouge, LA
SCIENCE ADVISORY BOARD STAFF
Mr. Fred Butterfield, CASAC Designated Federal Officer, 1200 Pennsylvania Avenue, N.W.,
Washington, DC, 20460, Phone: 202-343-9994, Fax: 202-233-0643 (butterfield.fred@epa.gov)
(Physical/Courier/FedEx Address: Fred A. Butterfield, III, EPA Science Advisory Board Staff
Office (Mail Code 1400F), Woodies Building, 1025 F Street, N.W., Room 3604, Washington,
DC 20004, Telephone: 202-343-9994)
* Members of this CASAC Subcommittee consist of:
a. CASAC Members: Experts appointed to the statutory Clean Air Scientific Advisory Committee
by the EPA Administrator; and
b. CASAC Subcommittee Members: Experts appointed by the SAB Staff Director to serve on one
of the CASAC's standing subcommittees.
A-2
-------
Appendix B - Review Comments from
Individual CASAC AAMM Subcommittee Members
This appendix contains the preliminary and/or final written review comments of
the individual members of the Clean Air Scientific Advisory Committee (CASAC)
Ambient Air Monitoring and Methods (AAMM) Subcommittee who submitted such
comments electronically. These comments are included here to provide both a full
perspective and a range of individual views expressed by the Subcommittee members
before, during and after the Subcommittee's December 15, 2004 advisory meeting on
implementation aspects of the Agency's Final Draft National Ambient Air Monitoring
Strategy (NAAMS). These comments do not represent the consensus views of the
CASAC AAMM Subcommittee, the CASAC, the EPA Science Advisory Board, or the
Environmental Protection Agency (EPA) itself. The list of Subcommittee members
providing individual comments is provided on the next page, and their review comments
follow.
B-l
-------
Panelist Page#
Dr. Ellis Cowling B-3
Mr. Richard L. Poirot B-l 1
Mr. George Allen B-17
Dr. Judith Chow B-20
Mr. Bart Croes B-29
Dr. Kenneth L. Demerjian B-33
Dr. Delbert Eatough B-36
Mr. Eric Edgerton B-41
Mr. Henry (Dirk) Felton B-44
Dr. Rudolf Husar B-48
Dr. Kazuhiko Ito B-53
Dr. Donna Kenski B-56
Dr. Thomas Lumley B-58
Dr. Peter McMurry B-61
Dr. Armistead (Ted) Russell B-65
Dr. Jay Turner B-70
Dr. Warren H. White B-73
Dr. Yousheng Zeng B-75
B-2
-------
Dr. Ellis Cowling
Comments by Ellis Cowling on
Implementation Aspects of EPA's National Air Monitoring Strategy (NAAMS)
After careful reading of all 12 sections of the Final Draft NAAMS and many of the 164 public
comments and EPA responses in the Addendum to the Final Draft, I offer the following
comments especially with regard to the first question in the Charge to the CASAC Ambient Air
Monitoring Subcommittee.
General Comment 1)
With regard to the Agency's proposal to redesign the routine precursor gas measurements (CO,
SO2, NO/NOy NH3, HNO3) at NCore Level II multiple pollutant sites and for air quality
management decisions and to obtain relevant exposure data for research programs, permit me to
mention that scientists in the Southern Oxidants Study drew the following general conclusion
from our major field measurement campaigns in Atlanta and Nashville in 1990, 1992, 1994, and
1999:
"In comparison with research-grade instruments, the commercially-available instruments
commonly used for measuring NOX concentrations in air: are not sensitive enough to
measure NO reliably, and are not specific enough to measure NO2 reliably."
Although the commercially-available instruments commonly used for measuring NOX
concentrations are appropriate for ensuring compliance with the health-based national ambient
air quality standard for nitrogen oxides, they are not sensitive enough to be useful for making
decisions about NOX vs. VOC sensitivity of ozone non-attainment areas. Most commercial
instruments have a minimum detection limit for NO of about 1 ppb, but decisions about NOX vs.
VOC sensitivity require instruments that have minimum detection limits of about 0.1 ppb.
Perhaps I missed it, but I was unable to find in the Final Draft Strategy document an adequate
discussion of the minimum detection limits for nitrogen oxides and the other precursor gases
listed above.
General Comment 2)
The preamble to Charge Question 1 asks:
"Given limited budgetary resources ... is the information on resource allocations provided in
Section 11 with regard to support for PM precursor gas measurements at NCore Level II multiple
pollutant sites appropriate and well balanced and consistent with the overall objectives of the
Strategy."
My response to this question is: "Yes, in part, with respect to the proposed gas measurements at
NCore Level II multiple pollutant sites."
-------
The specific wording of Charge Question 1 goes on to ask: "Are the relative adjustments in the
training and guidance approaches in the plan consistent with the overall objectives of the
Strategy?"
My response to this specific question is "Yes, by in large." But if the question is rephrased — as
I believe it should be — to read: "Are the relative adjustments in the training, guidance, analysis,
interpretation, and communication approaches consistent with the overall objectives of the
Strategy?" my response would be a resounding "No!"
Sections 1, 2, 9, 11, and 12 of the Strategy all give strong emphasis to analysis, interpretation,
and communication of results from the various monitoring networks described in the NAAM
Strategy.
This is especially true of the statements in Section 1.2 on 'Goals and Objectives,' Section
1.3 on 'Scope of Participants and Key Operating Principles' (most notably) its subsection on
'Data Analysis and Interpretation,' and Section 1.4 on 'Recommendations" (especially the
statement "A strong public communications program is advocated, both at national and local
levels."
This is also true of the statements in Section 2.2 on 'Network Assessment,' and Section
2.6 on 'Communication and Outreach.'
It is especially true of many statements in the early parts of Section 9 on
'Communications and Outreach,' (most notably) in Section 9.1.1 on 'Benefits to State and Local
Agencies, Public Interest Groups and the General Public,' Section 9.1.2 on 'Benefits to the
research and Academic Community,' and Section 9.1.3 'Benefits to Tribal Communities.'
But the description in Section 9.2 about 'How This Information Will be Communicated'
seems to be both 'inconsistent and not appropriately balanced' with the statements in Sections
9.1.1, 9.1.2, and 9.1.3. The 'Fact Sheet,' 'Quarterly Newsletter,' 'Specialized Briefing
Presentations' using 'packaged slide presentations,' and the 'Monitoring Strategy Brochure' all
appear to be focused on descriptions of the 'monitoring program' rather than to 'policy-relevant
scientific findings' from the monitoring measurements.'
Various parts of Section 11, especially Section 11.3 on 'Resource and Funding Strategy,'
and Section 11.3.1 on 'Implementation Using Current Funding Basis,' Tables 11-1 and 11-2 and
especially Table 11-3 on 'Proposed summary of redistributed Federal resources,' and even more
notably Table 11-6 on 'Implementation Schedule' give little or no information on 'Data
Analysis,' and even less information on 'interpretation,' 'communication,' and 'outreach.'
In Section 12 on 'Issues' the distinction between 'Classical' and 'Value' perspectives is
discussed. Please note especially the last part of the section on 'Value:'
"The real success of the Strategy ultimately will require ... cultural modification upgrades
[within both EPA and state, local and tribal communities] that allow for a [more] meaningful
dialogue across data generators and data user communities." In this respect the statements in
Section 12.4 on 'Addressing Data Availability and Data Analysis Needs,' and in Section 12.4.2
on 'Data Archiving, Distribution and Analysis Efforts' are relevant.
So far as I can find in the Strategy document, there is only one monitoring program in
which a specific discussion about needs for data analysis and interpretation is discussed directly
with specific dollars allocated for data analysis and interpretation. This is the PAMS program.
Please note the first paragraph on page 11-5. It reads in part as follows (see also Table 11-3 on
page 11-9):
B-4
-------
c. PAMS. PAMS requirements have been scaled down to allow for more specific special
studies of interest by local area/regions. The current $14 M Federal 105 STAG
contribution to PAMS should be reduced to $12M, an amount sufficient to cover the
revised, minimum PAMS monitoring requirements. There has been a wealth of data
collected from the PAMS program, but very limited and often sporadic analysis and
interpretation of the data. To address this gap and yield value from the PAMS data bases,
$0.5M will be set aside for analysis of the PAMS data.
[This amounts to an allocation of only about 4% of the current annual cost of the PAMS
program and does not consider cumulative funding for PAMS during years in which
analysis and interpretation was 'very limited and often sporadic.']
My recommendation is that our CASAC AAMM Sub-Committee offers the following
recommendations regarding Implementation Aspects of the NAAM Strategy:
1) Financial and human-resource allocations for analysis, interpretation, and communication
of monitoring results should be increased substantially for each of the several monitoring
programs described in the National Ambient Air Monitoring Strategy including the
programs for SLAMS, NAMS, PAMS, PM2.5, Toxics, CASNET, IMPROVE, and the
NCore network.
2) A summary table should be developed for inclusion in a revised Section 11 that shows
the approximate relative allocations within each monitoring program — for the
monitoring measurements themselves, and separately for analysis, interpretation, and
communication of scientific findings from each of the monitoring networks described in
the NAAM Strategy.
3) CASAC should initiate an endeavor together with OAQPS, selected universities, and
various state, local, regional and tribal organizations of air quality managers. The
objective would be to: a) Identify individuals and organizations that have been unusually
successful in analysis, interpretation, and communication of scientific findings from
monitoring networks; and b) Use these lists to foster and encourage more frequent
analysis, interpretation, and communication of results from the several monitoring
networks described in the Final Draft National Ambient Air Monitoring Strategy. In this
connection, please see the attached documents on the Environmental Statistics Program at
NC State University
Further Background Information for General Comment 2:
On July 22, 2004, CASAC held a "Consultation on Methods for Measuring Coarse-
Fraction Particulate Matter (PMc) in Ambient Air, Based upon Performance Evaluation Studies
Conducted by EPA." At that time I made two points that may be worth repeating in connection
with the current CASAC discussions on "Implementation Aspects of the Final Draft of the
National Ambient Air Quality Monitoring Strategy."
Point 1) EPA and many other federal research and monitoring organizations need to guard
against the tendency to allocate so much of the funds used in monitoring programs and field
measurement campaigns to "making careful measurements" so that inadequate funds are
B-5
-------
available for "scientific analysis and interpretation" to determine and communicate to users what
the measurements really mean.
These cautionary remarks about problems in field monitoring programs were suggested
originally by the late Glenn Cass, formerly of Cal Tech and later of Georgia Tech, on the basis of
his career-long experience in various environmental monitoring programs — programs in which
too much funding was allocated to "measurements" and too little to "analysis and interpretation"
of the data and "communication of results" from the field measurements. In this connection,
please note the attached excerpts from a paper published in Water, Air and Soil Pollution in
1995.
Please also note especially the suggestion in the second item 9 about a "50:50
distribution" of funding allocations between "measurements" and "analysis and interpretation" of
monitoring data rather than the (90:10 or 80:20 distribution) that is typical of many monitoring
programs in EPA and other agencies. But please also note that an even better suggestion was
made by Mary Barber, former executive leader with the Ecological Society of America who
spoke in opposition to the "50:50 distribution" idea at a recent Whitehouse Conference on
monitoring. Mary Barber insisted, and I agree with her, that it would be even more appropriate
to distribute the funding into three rather than two categories of investments — with equal shares
going to "measurements," "analysis and interpretation," and "outreach and extension of findings"
to interested clientele and "customers" for the results of routine field monitoring programs.
This problem is so commonplace — not only in EPA but in many other agencies in this
country and around the world — that I commend these "lessons that are available to be learned"
(and perhaps even the "15 reasons why this happens" and the "13 things to do about it") for
inclusion among the comments from individual participants in the CASAC meeting on
"Implementation Aspects of the Final Draft National Ambient Air Monitoring Strategy."
Point 2) EPA should also guard against the tendency to give undue emphasis to "Data Quality
Objectives" in the selection and evaluation of instruments and subsequent implementation of
field monitoring programs to the exclusion of concern about "Science Quality Objectives" and
"Policy Relevancy Objectives."
Experience within the Southern Oxidants Study and other large-scale field measurement
and monitoring campaigns have demonstrated repeatedly that undue emphasis on "Data Quality
Objectives" often leads to:
1) Serious lack of attention to the scientific hypotheses and assumptions that are inherent in
any choice of scientific instruments, the appropriateness of the ground-based sites at
which the instruments are located, the skills of the instrument operators, the data
processing and data-display programs used, and especially the scientific quality of the
conclusions and statements of findings that are drawn from analysis and interpretation of
the measurements that are made; and
2) Equally serious lack of attention to the policy relevancy of the measurements being made
— relevancy to the general or specific enhancements of environmental protection that are
the real reason behind the public health or public welfare concerns that led to the decision
to establish a monitoring program in the first place.
B-6
-------
In this latter connection, permit me also to call attention to the attached "Guidelines for
the Formulation of Scientific Findings to be Used for Policy Purposes." These guidelines
were developed originally by the NAPAP Oversight Review Board led by Milton Russell,
former Assistant Administrator for EPA. Please find attached below, an electronic version of
these Guidelines which we have adopted and very slightly adapted for use in formulating
policy relevant scientific findings in the Southern Oxidants Study.
The original version of these Guidelines was published as Appendix III of the April 1999
Report titled "The Experience and Legacy of NAPAP." This was a Report to the Joint Chairs
Council of the Interagency Task Force on Acidic Deposition of the Oversight Review Board
(ORB) of the National Acid Precipitation Assessment Program. As indicated in Appendix
III:
"The following guidelines in the form of checklist questions were developed by the
ORB to assist scientists in formulating presentations of research results to be used in
policy decision processes. These guidelines may have broader utility in other programs at
the interface of science and public policy and are presented here with that potential use in
mind."
Excerpts from:
Cowling, E., and J. Nilsson. Acidification Research: Lessons from History and Visions of
Environmental Futures. Water, Air and Soil Pollution 85:279-292. 1995.
LESSONS LEARNED ABOUT THE SOCIAL DYNAMICS OF MONITORING PROGRAMS
AND FIELD RESEARCH CAMPAIGNS
We also observed that there is a social tendency among research scientists, to spend more time and
energy developing new monitoring networks than in analysis and interpretation of already existing data,
especially long time-series measurements. Glen Cass of the California Institute of Technology was one of
the first to call this tendency to our attention. But we also have discussed this matter with other scientists
and research leaders in Sweden, Norway, Germany, The Netherlands, Canada, and the United States. As
a result we have come to believe that there are three general reasons and at least 15 specific reasons why
this happens and four general things and at least 13 specific things that can be done about it.
Three General and Specific Reasons Why this happens
Personal/psychological reasons:
1) Some scientists think it is not appropriate to analyze data collected by other people.
2) Some scientists underestimate the time and creative energy it takes to do thorough analysis and
interpretations of field data.
3) Some scientists prefer to concentrate on field measurements rather than the sometimes more demanding
intellectual work of data analysis and interpretation.
4) Some instrumentalists and data analysts exaggerate small deficiencies in measurement methods so that
they too readily agree to "get out and do it right" rather than "milk the data for all it is worth" before going
to the field once again.
5) The tendency to believe that "there is nothing more tragic than a beautiful hypothesis slain by an ugly
fact in your own data".
6) The tendency to "move on to other things" rattier than recognize that "it is a sign of maturity if you
finish things and a sign of immaturity if you just start new ones".
7) Some scientists think analysis and interpretation of long term trends is boring.
B-7
-------
Social/psychological reasons:
8) The contagious enthusiasm that field campaigns seem to engender in groups of scientists, and which
occurs less frequently in the usually more private intellectual work of data analysis and interpretation.
9) The tendency of some scientists to believe that those who collect the data "own" the data and thus can
choose (regardless of what organization paid the bills for the program!) when, where, and under what
circumstances of authorship, priority, etc., the data should be validated, archived, analyzed, interpreted,
and made available to others.
10) The tendency of research leaders to avoid conflicts about the policy implications of research findings.
For example, atmospheric monitoring data usually are not threatening, but data analyses sometimes lead
to situations that have policy consequences that make for uncomfortable disputes. Since non-
controversial portions of research programs proceed to completion more readily than parts of research
programs facing opposition, it is not surprising that the atmosphere is over-measured and under-
interpreted.
Budgetary/logistical reasons:
11) The tendency for unreasonable budget optimism. Exaggerated promises often are made to make a
project appealing to sponsors ~ promises that often cannot be fulfilled with the limited funds that usually
are available. Since atmospheric or other field measurements occur at the beginning of a research
program, they often are completed at the expense of resources needed later for thorough analysis and
interpretation of data.
12) The tendency to buy what is cheap (or what is easy to do) rather than what is really needed (or may be
difficult to do). Atmospheric measurements are individually rather cheap. (For example, the elemental
composition of a single sample of fine particulate matter can be determined for about $120.) But the same
measurement made within the stack of a particular pollution source may cost many thousands of dollars
because of logistical difficulties and challenging environmental conditions (e.g., high temperatures).
13) Field measurement campaigns often are planned too soon after each other.
14) Granting agencies often assume that if they fund scientists to make field measurements that these
same scientists will find the time and energy for analysis and interpretation whether they are funded to do
so or not.
15) The too dominating role of tradition (what did we do last year?) as the "best guide" to wisdom in
budget and personnel decisions.
Four General and Thirteen Specific Things That Can Be Done About It
Social/political things:
1) Recognize that monitoring programs cannot be sustained unless there are "customers" who care about
the data, use the data often, and are willing to speak-out publicly about the constructive values they derive
from analysis and interpretation of the data.
Personal/psychological things:
2) Recognize that people want to enjoy the work they do. Making atmospheric measurements in the field
is both rewarding and fun. By contrast, collecting and tabulating real-world data regarding amounts of
emissions can be tedious or even boring. Not surprisingly, few people want to do the necessary but
boring part, and those who are pushed against their will usually do a sloppy job. Solution: Find ways to
make data analysis more fun and rewarding! For example, give high praise and other rewards to those
who willing to do what is necessary whether it is boring or not!
Scientific/intellectual things:
3) Formulate specific science and policy questions that are to be addressed and design measurement
systems to answer those specific questions.
4) Use available guidelines for formulation of statements of scientific findings to be used for policy
purposes ~ See Appendix III in the Oversight Board Report (NAPAP 1991).
-------
5) Place heavy emphasis on publication of results not only in peer-reviewed scientific journals, but also in
readily accessible professional, environmental, commercial, and public-interest publications and
electronically accessible information outlets.
Budgetary/logistical things:
6) Recognize how many significant scientific discoveries about the phenomena and effects of air pollution
have been made by analyzing "old" data.
7) Find and fund those scientists who really enjoy data analysis and interpretation.
8) Buy the research that is really needed, not what is cheap or convenient.
9) Recognize that data archiving, analysis, and interpretation are tedious, time consuming, intellectually
demanding, and expensive. An appropriate ratio for investment in measurements and investments in
analysis and interpretation generally is closer to 50:50 than the customary ratio of only 10:1 or even 20:1.
Also, it is valuable to use 2- to 5-year "roll-forward" mechanisms to ensure continuity of funding.
10) Recognize large measurement campaigns rarely can be sustained in successive years.
11) Plan research project and programs from the end backward to the start. First, specify the best data
analysis procedures necessary to answer important science and policy questions within the available funds
and time. Then, make or complete only those measurements that have a pre-specified place in the data-
analysis plan.
12) Try to balance uncertainties in the time, cost, and performance of research projects so as to keep all
three factors approximately on-target.
13) Initiate and maintain measurement programs to determine the effectiveness of environmental
management decisions.
GUIDELINES FOR THE FORMULATION OF SCIENTIFIC FINDINGS
TO BE USED FOR POLICY PURPOSES
The following guidelines in the form of checklist questions were developed by the NAPAP Oversight
Review Board to assist scientists in formulating presentations of research results to be used in policy
decision processes.
1) IS THE STATEMENT SOUND? Have the central issues been clearly identified? Does each
statement contain the distilled essence of present scientific and technical understanding of the
phenomenon or process to which it applies? Is the statement consistent with all relevant evidence-
evidence developed either through NAPAP [or SOS] research or through analysis of research
conducted outside of NAPAP [or SOS]? Is the statement contradicted by any important evidence
developed through research inside or outside of NAPAP [or SOS]? Have apparent contradictions or
interpretations of available evidence been considered in formulating the statement of principal
findings?
2) IS THE STATEMENT DIRECTIONAL AND, WHERE APPROPRIATE, QUANTITATIVE?
Does the statement correctly quantify both the direction and magnitude of trends and relationships in
the phenomenon or process to which the statement is relevant? When possible, is a range of
uncertainty given for each quantitative result? Have various sources of uncertainty been identified and
quantified, for example, does the statement include or acknowledge errors in actual measurements,
standard errors of estimate, possible biases in the availability of data, extrapolation of results beyond
the mathematical, geographical, or temporal relevancy of available information, etc. In short, are there
numbers in the statement? Are the numbers correct? Are the numbers relevant to the general meaning
of the statement?
3) IS THE DEGREE OF CERTAINTY OR UNCERTAINTY OF THE STATEMENT
INDICATED CLEARLY? Have appropriate statistical tests been applied to the data used in drawing
the conclusion set forth in the statement? If the statement is based on a mathematical or novel
conceptual model, has the model or concept been validated? Does the statement describe the model or
concept on which it is based and the degree of validity of that model or concept?
B-9
-------
4) IS THE STATEMENT CORRECT WITHOUT QUALIFICATION? Are there limitations of
time, space, or other special circumstances in which the statement is true? If the statement is true only
in some circumstances, are these limitations described adequately and briefly?
5) IS THE STATEMENT CLEAR AND UNAMBIGUOUS? Are the words and phrases used in the
statement understandable by the decision makers of our society? Is the statement free of specialized
jargon? Will too many people misunderstand its meaning?
6) IS THE STATEMENT AS CONCISE AS IT CAN BE MADE WITHOUT RISK OF
MISUNDERSTANDING? Are there any excess words, phrases, or ideas in the statement which are
not necessary to communicate the meaning of the statement? Are there so many caveats in the
statement that the statement itself is trivial, confusing, or ambiguous?
7) IS THE STATEMENT FREE OF SCIENTIFIC OR OTHER BIASES OR IMPLICATIONS OF
SOCIETAL VALUE JUDGMENTS? Is the statement free of influence by specific schools of
scientific thought? Is the statement also free of words, phrases, or concepts that have political,
economic, ideological, religious, moral, or other personal-, agency-, or organization-specific values,
overtones, or implications? Does the choice of how the statement is expressed rather than its specific
words suggest underlying biases or value judgments? Is the tone impartial and free of special
pleading? If societal value judgments have been discussed, have these judgments been identified as
such and described both clearly and objectively?
8) HAVE SOCIETAL IMPLICATIONS BEEN DESCRIBED OBJECTIVELY? Consideration of
alternative courses of action and their consequences inherently involves judgments of their feasibility
and the importance of effects. For this reason, it is important to ask if a reasonable range of alternative
policies or courses of action have been evaluated? Have societal implications of alternative courses of
action been stated in the following general form?:
"If this [particular option] were adopted then that [particular outcome] would be expected."
9) HAVE THE PROFESSIONAL BIASES OF AUTHORS AND REVIEWERS BEEN
DESCRIBED OPENLY? Acknowledgment of potential sources of bias is important so that readers
can judge for themselves the credibility of reports and assessments.
B-10
-------
Mr. Rich Poirot
EPA NAAMS Implementation Proposal, Review Comments, R. Poirot, 12/12/04
Generally, this is a very strong, carefully considered, well justified and timely proposal. Many
revisions are directly responsive to previous CASAC Subcommittee review comments.
Emphasis on use of continuous methods - for multiple, collocated PM & gaseous precursor
species, and de-emphasis on filter-based PM 2.5 and speciation is a resource-efficient means of
supporting multiple monitoring objectives, including compliance determination, improved
atmospheric model development & evaluation, and human health, ecological and welfare effects
research.
1. The CASAC has expressed its support for the Agency's proposal to redesign the
routine PM monitoring network to support PM precursor gas measurements (CO,
SCh, NO/NOy, NHs, HNOs) at NCore Level II multiple-pollutant sites, and for air
quality management decisions and to obtain relevant exposure data for research
programs. Given limited budgetary resources, does this represent both an
appropriate and adequate balance, as reflected by the relative resource allocations
provided in Section 11, "Draft Implementation Plan" of the Final Draft NAAMS
Document? In addition, are the relative adjustments in the training and guidance
approaches proposed in the draft implementation plan consistent with the overall
objectives of the Strategy?
The question illustrates what seems to me to be a rather proscriptive emphasis on a specific suite
of "precursor gasses" which is difficult to evaluate without additional detail on the specific siting
criteria and on the other "required" or optional measurements at these sites. Regarding siting
details, the general Level II formula (I/state, 2/state in moderately large states, 3/state in the
largest population states, and 12 rural) sounds to me like it will be a very predominantly "urban"
network, and that in addition there would likely be just 1 site per urban area in almost all cases.
These many " 1 per city" sites have limited value for supporting ecological effects research, for
enhancing understanding of atmospheric chemical processes, for model development and
evaluation, and for some of the proscribed gasses (like NO, CO & NH3) questionable value for
characterizing regional spatial patterns. Urban site locations for highest human exposures of
ozone, PM2.5 and some of the proscribed gasses differ by pollutants. So exactly what kind of
"representative" sites are we looking for? On a related point, the inclusion of detailed met
instruments is obviously desirable with all this continuous data, but what scale(s) of
representation are we looking for in the met measurements, and will or should this further effect
the siting criteria?
Regarding the other species measurements, I assume at minimum these include continuous ozone
and PM2.5. Will there also be filter-based speciation measurements, and if so will the new
measurements be added at existing STN sites, or will those "trends" sites be moved to the new
optimized multi-pollutant sites? How exactly will continuous PM2.5 mass be measured? Ideally
a single common method would be employed, rather than the current mix (all things being equal,
I'd suggest the FDMS TEOM, as it provides higher resolution information (separate volatile and
non-volatile), and may often capture volatile species partially lost from FRM filters).
B-ll
-------
Continuous (high sensitivity) SO2 is highly desirable, but the value of such data would be hugely
enhanced if collocated continuous SO4 were also measured. NOy is useful (especially at rural
sites), but could be substantially more useful if accompanied by continuous hydrocarbon
analyzers. Aethalometers (especially the dual wavelength - with its potential value as a wood
smoke monitor) could provide useful information with currently available, field-proven
technology, and/or continuous OC/EC (Sunset Labs for example) look promising. Will any of
these continues PM species measurements be included, or do we measure only the precursors?
The proposed monthly integrated NH3 and HNO3 measurements would substantially enhance
the status quo (almost no data), but would obviously provide minimal temporal resolution
(assuming the technologies work). A possible alternative or supplemental approach might be
through routine extraction of denuders at STN and IMPROVE sites. See additional detail on this
under question 3 below.
2. Does the Subcommittee have additional suggestions for addressing this need for
integration and communication to the broader community of "users," including
scientific researchers (ie., human health, atmospheric, ecological) and State, local
and Tribal (SLT) Agency representatives? More specifically, what is the most
effective manner for EPA both to reach-out to this broad user community and,
where appropriate, to incorporate their feedback and design input on such issues as
monitoring site locations and parameters?
Generally, the above "user communities" (human health, atmospheric, ecological, & SLTs) are
very disparate groups who don't very often interact. Also, the most relevant health and
ecological effects issues tend to vary on a regional basis, making "centralized" communication
awkward. One organizational concept that might be worth considering is to organize periodic
workshops on a regional basis, as there tend to be clustered integrated user communities who
focus on specific urban health effects or rural ecological effects monitoring/research sites.
Examples of the latter from my region include the Vermont Monitoring Cooperative at Mt.
Mansfield/ Underbill, VT, Hubbard Brook, NH, Harvard Forest, MA & Whiteface Mtn, NY
(unfortunately none of these is likely to become a Level II site, and I'm concerned my Underbill,
VT "SIP" IMPROVE site may also be on the NCore chopping block). About 15 years ago, the
International Air Quality Advisory Board of the IJC held a series of workshops on "Integrated
Transboundary Monitoring" at 5 locations along the US/Canada border, which were very well-
attended and productive. Possibly something like this might be considered on an EPA regional
basis or through the Regional Planning Organizations (RPOs).
3. One of the remaining technical issues relates to harmonizing rural- and urban-
based PM2.5 chemical speciation networks such that both categories of networks
utilize consistent sampling and analysis protocols. For example, EPA is considering
converting all of the Speciation Trends Network (STN) speciation sites to
Interagency Monitoring of Protected Visual Environments (IMPROVE) samplers
and IMPROVE laboratory and sample handling protocols. What are strengths and
weaknesses of this approach?
B-12
-------
Spatial & urban/rural consistency alone are good reasons to consider this, with the added
advantages that (I think) IMPROVE analytical, shipping & site operator costs are lower (though
the new equipment costs would be significant and STN "trends" - for some species - would be
disrupted).
What exactly is meant by IMPROVE methods? Samplers, filter media, "identical" copper anode
XRF systems, TOR carbon methods - which are currently being revised in IMPROVE, etc? If
IMPROVE PESA measurements for H are duplicated by STN, only a very few labs have that
capability.
IMPROVE is cited in the EPA Strategy as serving as a "core rural speciation trends network:
needed network adjustments are handled effectively through IMPROVE Steering Committee".
While I generally agree here, it should be cautioned that being a core rural speciation trends
network is serendipitous, but not an objective of IMPROVE, which is intentionally focused on
regional haze-related aspects of aerosols in specific class I areas, and in recent years specifically
focused on supporting the EPA Regional Haze Rule - which emphasizes long-term trends.
Consequently there is added inertia to change, or for adding supplemental measurements to
address non-haze-relevant issues.
Also, given the general tendency for class 1 areas to be located in mountainous terrain, the
IMPROVE monitoring objectives to measure regional rather than local aerosols, and to cover
each class I area (with funding constraints limiting sites to 1 per area), the IMPROVE network
tends to provide a "mountaintop" definition of rural speciation patterns. Caution is needed in
comparing urban sites to nearby rural counterparts (for example Washington DC to Shenandoah
NP) since we don't know how much of the DC "urban excess" (at 16 meters ASL) results from
Shenandoah (1100 m ASL) being often above the mixed layer - especially during winter. I
emphasize this to point out a need for some rural low elevation sites - which would also be of
benefit to IMPROVE, and might be best managed by adding lower elevation IMPROVE sites in
a few class 1 areas.
As the number of filter-based speciation sites is reduced, and as methods changes are
contemplated, I think it's possible that we might "think smarter" and squeeze additional
information out of the remaining sites. For example, the Canadian NAPS speciation program
(see brief summary from Tom Dann pasted at the end of my comments) measures more or less
everything we do at IMPROVE and STN sites, and by fairly similar methods to ours. But in
addition they denude for HNO3, SO2 and NH3 (STN does not do NH3 but should) and get (I
think) better quality NH4 data than STN does. Then they extract and analyze their denuders to
get HNO3, SO2 and NH3, averaged every 3rd sample day concurrent with the aerosol species.
This is less desirable than the ultimate goal of continuous, but way better than monthly means.
The Canadians also conduct additional analyses (on Teflon filters) for several organic ions
(oxalate, formate, acetate), which in turn have proven extremely valuable for source attribution.
Another difference in the Canadian protocol is that samplers are run SAM to SAM, rather than
midnight to midnight. If loss of volatiles is a concern, this seems like a much more intelligent
& efficient way of minimizing filter losses (during sampling) compared to the costly STN
requirements to ship exposed filters in iced coolers.
B-13
-------
Another example of outstanding "value added analysis" from existing (& expensively collected)
filter samples is in the recent molecular carbon analysis Ted Russell's group at GA Tech has
conducted on archived, composited quartz (carbon) filters from selected Eastern US STN and
IMPROVE sites - followed by CMB analysis using source profiles developed with similar
analytical methods. I'd like to see something like this conducted routinely (yearly with seasonal
compositing), for at least a subset of remaining STN & IMPROVE sites.
On a related topic, the stated emphasis on data analysis in the NAMS proposal is much
appreciated - although as usual, the funding level is minimal. A novel approach would be to
develop a reasonably detailed data analysis plan in advance of conducting new measurements -
and use it as a guide to what should be measured where, rather than just assuming "here are some
key species and sites from which we're bound to learn something".
An aside on this, is recent (5 minutes ago) note from our colleague Husar on his latest revisions
to the "Combined Aerosol Trajectory tool", which links aerosol data from the entire IMPROVE
and STN networks (through the end of 2003) to associated back trajectories (ATAD model run
by Kristi Gebhart at NFS) and allows all kinds of extremely powerful single and multi-site
analyses - and is also really fun! (once you get the hang of it). Check it out at:
http://webapps.datafed.net/dvoy_services/datafed.aspx?page=KittyC
This RPO-supported (with assistance from EPA) analysis tool is I think an excellent example of
the kind of data analysis that should be routinely funded as a key component of NAMS.
4. As EPA implements the National Ambient Air Monitoring Strategy to address
multiple monitoring objectives, it will be looking to spatially optimize the ambient
monitoring networks. This may mean that some redundant monitors in adjacent,
but separate, geopolitical areas (e.g., neighboring counties) are "divested" from a
given network. Although technically sound, these divestments could result in data
gaps which might, in turn, adversely impact regulatory decision-making. The
Agency is willing to adopt alternative approaches for assessing regulatory issues
such as non-attainment designations, so long as such approaches are scientifically
justifiable; hence, the rationale for initiating discussion of these issues with the
CASAC. Question: Is it scientifically acceptable to generate air quality surfaces
through modeled observations and/or integrated predictive/observational fields that
would be of appropriate uncertainty for use in the regulatory decision-making
process?
It has a stronger scientific basis than the current alternative of assuming each political
jurisdiction is represented by the single monitor within it. I think much could be learned by
efforts to develop and refine this king of "observation-based model", and at some point it may be
appropriate to use such methods to determine compliance status. A goal would be to push toward
applying such techniques in near-real-time.
B-14
-------
Attachment 1: Summary of Canadian PM-2.5 Speciation Sampling
(for more details contact Tom.Dann@ec.gc.ca)
Fine Particle Speciation Program
Sampling Equipment and Field Operation
Sampling sites will be equipped with R&P Partisol-Plus 2025-D sequential dichot particulate
samplers along with R&P Partisol Model 2300 Speciation samplers. These units share common
software and data storage systems. The speciation sampler uses Harvard designed Chemcomb®
cartridges which employ honeycomb glass denuders and filter packs with Teflon and Nylon
media.
Both samplers have sequential sampling features and up to 32 fine and coarse filters (16
sampling days) can be preloaded in the dichot Partisol while up to 12 Chemcomb cartridges (3
sampling days) can be preloaded in the speciation sampler. At this time, the protocol will be to
operate the samplers once every three days and to visit the sampling sites at least once every six
days. Samples will be collected over 24 hours. One fine and one coarse filter sample will be
collected on the dichot Partisol sampler and three Chemcomb cartridge samples will be collected
with the speciation sampler as described below. The Chemcomb cartridges are shipped to the
field completely assembled and sealed and require only mounting and leak-checking.
Field data sheets will be required but all instrument operating data will be downloaded to Palm
data systems (provided) and the RAM cartridges will be returned to Ottawa with exposed
samples.
Sample Types and Target Analytes
A description of the sample media and target species is provided below:
Unit
Dichot
Pariisol
Speeialion
Sampler
Module Description
f-'ino l-'raclion l-'iller
Coarse i-'raclion l-'illor
Module A (4 components)
Module 1)
Module C (2 components)
Media
47 mm Teflon
47 mm Teflon
Sodium Carbonate Denuder
(.'Uric Aciil Denuder
Tel'lon filter
Nylon l-'illor
Pro-fired Quart/ filter
Teflon filter
Pro-fired Quart/ filter
Function Analvtes
Mass. Metals"
Mass. Metals
S02& UNO.;
Ammonia
Sulphale and oilier major
Inorganic and Organic Ions*
Nilrale. Sulphale
Black carbon. Organic
carbon
Mass. Mela Is KM Check)
Organic carbon artifact
* see next Table
B-15
-------
Paniculate matter related metals (EDXRF) and ions (1C) measured
Sulphate
Nitrate
Aluminum
Silicon
Phosphorus
Sulphur
Chlorine
Potassium
Calcium
Scandium
Titanium
Vanadium
Chromium
Manganese
ron
Cobalt
Nickel
Copper
Zinc
Gallium
Germanium
Arsenic
Selenium
Bromine
Rubidium
Strontium
Yttrium
Zirconium
vliobium
Molybdenum
Palladium
Cadmium
ndium
Tin
Antimony
Tellurium
odine
Cesium
Barium
Lanthanum
_ead
Bromide
Chloride
slitrite
3hosphate
:luoride
Acetate
:ormate
Oxalate
Sodium
Magnesium
Cerium
3raseodymium
\leodynium
Tantalum
Tungsten
Mercury
Thallium
Bismuth
Sodium
Ammonium
Silver
B-16
-------
Mr. George Allen
To: Fred Butterfield, Designated Federal Officer
EPA SAB, Clean Air Scientific Advisory Committee (CASAC)
Ambient Air Monitoring and Methods Subcommittee
From: George Allen, AAMM subcommittee member, December 20, 2004
The following arc revised written comments for the December 15 meeting on "Implementation Aspects
of EPA's Final Draft National Ambient Air Monitoring Strategy (NAAMS)". A copy of these
comments is being sent to Dr. Phil Hopke, CASAC AAMM Subcommittee Chair. These comments
address the four "Charge" questions in the EPA OAQPS memo to the SAB dated November I1),
2004, and additional NAAMS topics brought up during the meeting.
Charge Question ?1. The five PM precursor gases listed here (CO. SO2, NO/NOy. NH,. HNO.,) are
useful to have at Level 2 NCore sites to aid in understanding the sources, transport, transformation and
fate of these PM-relcvant species. The first three pollutants listed here are likely to be enhancements to
existing measurements, not additional measurements. The need for significant training resources is still
important however, since practices for measurement, data acquisition, and data validation that arc
adequate for compliance-oriented monitoring often need substantial modification for "trace-level"
monitoring purposes.
At least for urban sites (which make up the vast majority of Level 2 sites), it may be worth considering
the use of NO and "true" NO, instead of NOy (or in addition to for NC)Z). Photolytic "true" NO,
methods are now becoming practical for routine monitoring, using near-UV LED converter
technologies: for an example see: Buhr. US Patent Application # 2004010X197 at
httn: annftl.usnto.uov netahtml PTO srehnum.html .
Monthly duration measurements of NlI, and 11NO, are of very limited value: as noted in the meeting,
12 daily samples year would be more useful. Where possible, 1 would like to encourage the use of
real-time methods (hourly or shorter measurement periods) rather than daily (24-hour) duration
integrated filter measurements for these gases. Daily measurements smear important temporal patterns.
and make it difficult to understand the underlying dynamics of these gas phase PM precursors (similar
to trying to understand ozone dynamics and chemistry using a daily 24-hour value). Real-time methods
for measurement of these gases at ambient concentrations are becoming more practical and may be
cost-effective in the longer run.
A network of Level 2 sites that measures these enhanced PM precursor gases begs the question of can
we also afford real-time PM speciation measurements (for carbon, nitrate, and sultate) at these sites?
The value and rationale for these real-time measurements arc the same as for NH, and UNO, above.
Keeping in mind the pyramid (not a wedding cake) NCore approach, these enhanced non-NAAQS
parameters could be made in or near areas where PM2.5 compliance is an issue, but not at all Level 2
urban sites. Another option could be to not deploy all of these additional measurements as a suite, but
to measure nitrate-related pollutants in areas where nitrate is a substantial contributor to PM2.5, and
(i. Alien - Re\ i.sed comments i-n NAAMS to (ASM AAMM Subcommittee-. December 20. 2004 Pa no I
B-17
-------
SCVSO4 where sulfate is a substantial contributor. In urban areas, carbon always is a major
contributor of course. Rural Level 2 sites should always have as complete a suite of measurements as
possible, to address a wide range of transport issues. The concept of a subset of enhanced Level 2
sites ("Level 1.5") with some or all of the real-time parameters discussed here becomes more important
if Level 1 sites continue to remain unfunded in the final NAAMS regulations.
I recommend that OAQPS require some performance-based acceptance process similar to the existing
reference or equivalent method designation used for NAAQS pollutants for these methods even though
the monitoring goal is not compliance with NAAQS and some of these parameters are not NAAQS
pollutants. An example of the need for this non-NAAQS based performance review is the NO
interference on trace SO2 analyzers: some vendors historically have traded a five-times worse NO
rejection ratio for a better SO2 sensitivity specification, rendering the method potentially useless in
urban areas.
Training is a critical component as new measurement methods and technologies are widely
implemented. The concept of an annual "National Monitoring Conference" (replacing and expanding
SAMWG) would be a iiood venue to use for training, from both vendors and experts in the monitoring
community (to give a user perspective). Regional versions of this can also be effective, for example the
joint MARAMA-NESCAUM continuous PM2.5 training session planned for spring 20115. I
discourage wide use of satellite-based training, since access to it is relatively limited at the local agency
level and it expensive in the context of current technologies: web-based training, downloadable or
streaming video training, and web-based video-conferencing are all more cost effective and widely
accessible remote training alternatives.
Charge Question #2. One approach to integration of these data into research programs is to work with
non-EPA funding agencies like DOE, HEI and NiEHS to insure their program funding opportunities
emphasize the leveraging of existing data. One recent example of this is the HE! program to make STN
and related data more accessible to those doing health effect research. If Level 1 sites are ever
implemented, it is essential that these sites be partnerships between research institutions and local air
agencies, and not stand-alone research programs. The technology transfer process and program
relevance to the local agencies are greatly enhanced by these partnerships. One recent example where
this has worked well is the NYC PM-Supersite.
Charge Question ?3. It is critical to harmonize the STN and IMPROVE networks as much as
practical, since it's difficult to perform analysis on a combined urban/rural data set when much of the
data from these two networks are often not directly comparable. I recommend modifying STN field,
laboratory (carbon, ions, and XRF-elements), and data handling (validation, uncertainty reporting,
below MDL treatment) protocols to bring them as close as possible to those of the IMPROVE:
network, consistent with the current EPA plans to convert STN to the IMPROVE TOR carbon
analysis method. Regardless of the debatable merits of either network, IMPROVE has a much longer
measurement history and is a regulatory trend network for the regional haze regulations; as such it can
< i. Allen - Rev ised eommenls on NAAMS to (ASA* AAMM Suheommiltee. December 20. 2004
B-18
-------
not be easily changed. It may not be necessary to convert all of STN to IMPROVE sampler
hardware, and it IN probably not practical to have the IMPROVE program simply "take over" the STN
operations. However some consideration should be given to reducing the variability of sampler types
across STN, and in the process minimize the major differences between IMPROVE and STN
samplers. Filter face velocity differences for carbon is an example of sampler parameters that could
cause differences across or within networks. Post-filter handling (time on sampler, stored shipped cold
or warm) is an example of a "method" parameter variable that is not related to either sampler type or
analytical method.
There was a brief discussion of the value of ammonium ion measurements in a speciation network
(IMPROVE or STN). While 1 agree those data are useful. 1 do not see a practical way of getting valid
ammonium ion data from either the Teflon or Nylon filters in these networks in environments where
ammonium nitrate is a dominant contributor to the ammonium ion concentration. Nitrate losses from
Teflon filters is substantial and well characterized. Even though they retain ammonium nitrate well.
based on the chemistry there is no reason to believe that Nylon filters quantitatively retain ammonium
ion from ammonium nitrate; since the Nylon filter media is basic it is likely that much or all of the
ammonium ion is lost as NHS. If ammonia is properly removed upstream of the filter, ammonia lost
from the Nylon filter could be measured down-stream, but this requires significant sampler changes.
additional filter extraction and analysis, and has the potential for substantial surface losses of the
volatilized ammonia within the sampler.
Charge Question °4. Generating ambient air quality surfaces by modeling processes or interpolation
between monitors can be useful in some situations, such as reducing exposure mis-classification for
health effects studies. In a NAAOS compliance context, this approach may have limited value given
the state of air pollution modeling science. The uncertainties in modeled or interpolated pollutant
concentrations that are acceptable for epidcmiological studies may be too large for "bright-line"
regulatory issues such as non-attainment designation. The potential value for this approach may be
limited to areas that arc expected to be both reasonably spatially uniform for the pollutant of interest
and where expected concentrations based on nearby monitoring sites are not near any regulator}' value
— e.g., either well above or below a NAAQS standard.
There was substantial discussion during the meeting about the value of the PAMS network data to date.
and the network's future design. The concept of replacing many of the PAMS sites with non-methane
hydrocarbon real-time measurements (as California has done) is worthy of further consideration:
NMHC could serve as a general trends indicator for PAMS parameters. The PAMS approach may
have substantial value in urban "enhanced" Level 2 NC'ore sites, run year round for air toxics and PM-
precursor use. in addition to the classic "summer only" ozone-centric operation. The new New 1laven
CT Criseulo Park site is an example where this has recently been implemented.
(r. Allen - Ro\ ised comments on NAAMS to ( ASA( AAMM Siiheommiltee. December 20. 2004
B-19
-------
Dr. Judith Chow
December 13, 2004
To: Fred Butterfield, Designated Federal Officer,
Clean Air Scientific Advisory Committee (CASAC)
From: Judith C. Chow, Research Professor, Desert Research Institute
CC: Phil Hopke, CASAC Ambient Air Monitoring and Methods (AAMM) Subcommittee
Chair
Subject: CACAS AAMM Subcommittee Charge on the Evaluation of the Final Draft of the
National Ambient Air Monitoring Strategy (NAAMS) document dated April 2004
The staff of the Office of Air Quality Planning and Standards did a remarkable job of assembling
the final draft National Ambient Air Monitoring Strategy (NAAMS) document and responses to
the 164 comments in the Addendum of the Strategy document. Conceptually, the NCore strategy
represents a major change from what has been done over the past 30 years. Its success or failure
will depend on building flexibility into, while establishing consistency within, the current
monitoring network. EPA is facing major challenges in coming years to assure that data from the
restructured NCore network will achieve the objectives for compliance and implementation of air
quality standards, air quality forecasting, atmospheric processes research, and the determination
of health, visibility, ecological, and radiative effects. The following review focuses on the four
assigned questions, with additional comments on NCore siting, Level 1 sites, and measurement
methods.
Question 1. Given limited budgetary resources, does this represent both an appropriate and
adequate balance, as reflected by the relative resource allocations provided in Section 11,
"Draft Implementation Plan, " of the Final Draft NAAMS Document? In addition, are the
relative adjustments in the training and guidance approaches proposed in the draft
implementation plan consistent with the overall objectives of the Strategy?
Table 11-3 listed the proposed redistribution of Federal resources for ambient monitoring. It is
difficult to evaluate resource allocation without adequate background on how these estimates
were formulated. It would be helpful to document the cost-estimation basis in categories of: 1)
initial hardware procurement; 2) initial installation and shakedown; 3) initial training; 4) spare
parts; 5) repairs; 6) normal operations labor; 7) rent/power/security; 8) quality control and
auditing; and 9) data processing and management. Based on my calculation, the allocation is
estimated on an average of-$60,000 per Level 2 NCore site, including capital investments for
equipment purchases and associated operating expenses, and $50,000 to $100,000 per Level 3
NCore site, depending on whether -1,000 or 500 sites are planned. (Richard Scheffe's
presentation and Figure 4-1 show >500 Level 3 sites. But Section 4.3.5 [p. 4-10] shows -1,000
sites.) Even if one factors in the infrastructure of the Level 2 site as being mostly established,
resource allocation to the Level 3 sites seems high. This defeats the objectives of the Level 3
sites, which focus on a subset of criteria pollutants (e.g., PM and Os).
While the training and guidance approach proposed in the draft implementation plan is consistent
with the overall objectives of the Strategy, the resource allocation seems low. Given a total of
B-20
-------
$198.85 million (note that there is an error in the proposed budget sum), the allocations of 1.6%
($3.2 million) to quality assurance (QA) and 0.1% ($200,000) to training are low, even though
they are additions to the current budget.
The performance evaluation program (PEP) in Table 7-3 estimates $3.6 million in QA— $0.4
million higher than the Table 11-3 estimate, of which -60% is allocated to PM2 5 PEP and
speciation. Listings of the current State and Tribal Air Grant (STAG) fund ($2.3 million in Table
7-3 and $1.9 million in Table 11-3) are also different. What about QA for existing PMi0, future
PMcoarse, newly added HNOs and NH3, and NOy measurements, and audits for meteorological
measurements? Even though some of the training and QA needs arising from the network
modification may be embedded into each pollutant category, the level of effort needed for QA
and training (probably at multiple locations) seems to be underestimated, especially for the first
2-5 years, during the transitional period.
The certification program is a good approach for ensuring consistent data quality. More detail is
needed on traceability from common primary standards, to transfer standards, and to in-station
QA standards. This is especially important for the non-standard measurements (e.g., PM, NH3,
NOy). There is a need for guidance to clearly state the corrective action for sites that fail to pass
the certification tests. More centralized QA, as recommended by the CASAC NAMS
subcommittee, with consistent, clearly stated data validation criteria and data formats, is a good
target.
It is encouraging to see "data analysis and interpretation" listed as a separate line item, even
though it accounts for only 1.1% ($2.2 million) of the total budget. If funding is available for
data analysis, a detailed data analysis plan is needed to specify data analysis objectives that
analyze pollution characteristics and trends, identify episodes, explore seasonal/annual trends
and spatial variations, develop control strategies, track progress of control measures, estimate
source attribution, as well as evaluate emissions inventory and air quality models (Chow et al.,
2002a). These analyses should be the tools that identify the needs to further refine the network to
meet multiple objectives (U.S. EPA, 1997a).
What Table 11-3 lacks (which may be embedded in network operations) are resources for
different levels of data validation. In the long run, it is more cost-effective to replace filter-based
measurements with in-situ continuous monitoring. The initial cost of the switch-over is high, and
continuous instruments are more labor-intensive in the field for routine calibration and
maintenance, and in the office for processing and evaluating short-term average (1- to 5-minute)
measurements. Substantial effort is needed for timely data validation to resolve data outliers and
to take necessary corrective actions to minimize the generation of additional invalid or suspect
data. Four levels or categories generally apply to validation of monitoring data (Chow et al.,
2002b):
B-21
-------
Level 0
Raw data right off the instrument (used for real-time alerts and
forecasting).
Level I
Routine checks made during the initial data processing and generation of
data, including proper data file identification, review of unusual events,
review of field data sheets and result reports, instrument performance
checks, and deterministic relationships.
Level II
Tests for internal consistency to identify values in the data which appear
atypical when compared to values of the entire data set.
Level III
Comparison of the current data set with historical data to verify consistency
over time. Tests for parallel consistency with data sets from the same
population (region, air mass, period of time, etc.) to identify systematic
bias. This level can be considered as part of the data interpretation or
analysis process.
Note that Level III, and possibly Level II, data validation are part of the data analysis and
interpretation process and may be grouped to the data analysis categories and allocated as
Research and Development rather than as monitoring funds. While it is desirable to have data
available soon after its collection—and real-time monitors permit this on a local basis—current
regulations requiring "submission to EPA 90 days after the calendar quarter in which the sample
was collected" are reasonable. This allows state and local agencies to conduct Levels 0 and I data
validation and to perform cross-comparisons to remove invalid data and flag suspect data.
The zero sum approach is challenging during the transition period since substantial costs may be
incurred for equipment procurement, operator training, and comparability testing (to retain
continuity with previous data). I disagree that a data validation protocol should be developed by
the modeling community (p. 26, response to Comment 127 of the Addendum). For advancement,
data validation protocols should benefit from lessons learned from those who conduct the field
measurements; modelers are not necessarily familiar with these measurements.
Question 2. Does the Subcommittee have additional suggestions for addressing this need for
integration and communication to the broader community of "users, " including scientific
researchers (i.e., human health, atmospheric, ecological) and State, local and Tribal (SLT)
Agency representatives? More specifically, what is the most effective manner for EPA to both
reach out to this broad user community and, where appropriate, to incorporate their feedback
and design input on such issues as monitoring site locations and parameters?
The Strategy formulates several good means (i.e., fact sheets, quarterly newsletters,
presentations, and brochures) for public outreach. However, from the public comment process (p.
9-4), it is clear that less than -10% of the local and state and 30% of the regional offices
responded during the comment period, with only two comments each from the public interest
groups and industry, and one from the tribes. It appears that it did not call enough attention to
those agencies that would be directly impacted by the proposed Strategy, and the 164 questions
raised may not be representative of the general population. More aggressive solicitation of
B-22
-------
comments is needed from the research community or data users in the field of atmospheric,
health, and ecological sciences. One good approach might be to put a notice on popular real-time
web sites (e.g., AIRNow) where people go to make decisions based on air quality data.
The majority of epidemiological studies rely on data from long-term compliance networks for
exposure assessment (Vedal, 1997). A need still exists to broaden the involvement of air quality
and health researchers to optimize the restructured national network while meeting the current
resource constraints.
If additional support can be obtained, the research community can assist in developing data
validation protocols, conducting data analysis, and evaluating the representativeness of the
NCore sites. As pointed out by Demerjian (2000), there is a need to develop appropriate
feedback between the measurement community and data analysts to assure data quality and data
applicability. This feedback should also cover additional data management issues, such as data
and metadata items and report formats to facilitate data analysis and interpretation.
Question 3. What are the strengths and weaknesses of converting all of the Speciation Trends
Network (STN) speciation sites to Interagency Monitoring of Protected Visual Environments
(IMPROVE) samplers and IMPROVE laboratory and sample handling protocols?
The lack of consistency between the IMPROVE and STN networks is a major shortfall, since the
two networks use different requirements for sample archiving, field blank collection, shipment,
blank subtraction, carbon analysis (Chow et al., 2001, 2004), and uncertainty estimation (NRC,
2004). Converting all STN sites to IMPROVE sites has the following advantages:
Data consistency:
A national approach is needed for field sampling, laboratory
analysis, and sample validation to ensure the consistency of data
quality. Switching STN over to the IMPROVE protocol would
minimize the discrepancies between the two current networks. This
will allow model comparison and evaluation between urban and
non-urban areas. Archived samples in the IMPROVE network have
been shared with the research community in the United States and
foreign countries or to be stored for special studies.
Cost savings:
IMPROVE samplers have been proven to be robust and easy to
operate since 1987-88. Recent modifications have further advanced
their utility. With the exception of capital investment for
IMPROVE samplers, the operating cost for STN following the
IMPROVE protocol can be substantially reduced. Approximately
80-90% of the current STN network uses Met One speciation
samplers (Met One Instruments, Inc., Grants Pass, OR), which are
labor intensive for filter loading and unloading. Shipping costs
could also be substantially reduced due to the stainless steel
housing of the Met One sampling cartridge. The pros and cons of
cold shipping requirements in the STN but not the IMPROVE
network also warrant further evaluation.
B-23
-------
Long-term value:
More than 100 publications have used IMPROVE data to evaluate
measurement systems, map spatial distributions, develop and apply
source apportionment models, and make important control strategy
decisions (e.g., Malm et al., 1989; Pitchford et al., 1999). The
IMPROVE network is just completing the baseline period required
by the regional haze rule (U.S. EPA, 1999; Watson, 2002) and
cannot experience major changes without substantial comparability
testing. The STN has no such requirements.
Question 4. Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integrated predictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision-making process?
Currently, network assessment starts at the national level. It uses interpolating methods between
measurements sites and uses error analysis to remove redundant sites. This type of spatial
analysis is adequate for flat terrain with a sufficient number of sites. It also works to evaluate
whether one site is redundant with other sites. It cannot extrapolate concentrations into areas
with insufficient monitoring. A good illustration is the IMPROVE network, where the addition
of Midwestern sites in 2001 revealed a large nitrate cloud that was not evident from the earlier
network. Integrated predictive/observational fields with adequate uncertainty estimates outweigh
any single modeling approach. Relying on modeled results alone may bias the decision-making
process.
As pointed out in a recent National Research Council (NRC) report (NRC, 2004), the enhanced
network should have the following three characteristics: 1) use continuous measurements of
appropriate indicators with real-time access; 2) represent less uniform micro- and middle-scale
exposures; and 3) encourage the development and use of continuous monitors for indicators
other than mass concentration. For areas that may have concentration gradients between the
Level 2 or 3 sites (i.e., not representative of micro- or middle-scale exposure), additional
monitors may be needed. In these cases, dense spatial monitoring (e.g., Chow et al., 1999,
2002c) over a short period of time (e.g., 2-4 weeks) is needed to add or remove sites, to assure
adequate spatial coverage, or to confirm observational or diagnostic modeling results. The "zone
of representation" experiments could be economically executed with portable instruments (e.g.,
Fujita et al., 2003). Their lower precision is still adequate to determine large spatial gradients.
Instrumentation used at Level 3 sites should be required (not "strongly encouraged", as stated on
p. 4-10) to be collocated with filter-based measurements for a set period of time to ensure
equivalence or comparability. Current requirements for the PM2.5 Federal Equivalent Method
(FEM) are: 1) collocated precision of 2 |ig/m3 or 5% (whichever is larger); 2) linear regression
slope of 1 ± 0.05; and 3) linear regression intercept of 0 ± 1 |ig/m3 and correlation coefficient (r)
of 0.97 (U.S. EPA, 1997b). For Regional Equivalent Monitors (REM), the proposed performance
criteria are bias (relative to a filter-based Federal Reference Method [FRM]) ± 10%, collocated
precision (continuous monitors) <10% of coefficient of variance (CV), with r of 0.93. These
criteria are too stringent to be met for currently available PM speciation monitors. I disagree with
relying on vendors to demonstrate equivalence (p. 20, response to Comment 92 of Addendum)
since comparability between instruments varies by location due to changes in emissions,
meteorology, and aerosol composition. There are only limited comparisons that can be supported
B-24
-------
by vendors. Depending on the complexity of the surrounding environment at Level 3 sites,
equivalence, comparability, or predictability (Watson and Chow, 2002) between the testing
instrument and the FRM should be established at each location for realistic comparison. Methods
with acceptable comparability should be allowed to facilitate the network transition. Level 3 sites
should also be used to understand the impact from source-oriented location (e.g., Zhu et al.,
2002a, 2002b) and concentration gradients between the Level 2 sites.
The Strategy document calls for multi-level network assessment every 5 years (Section 5). It
appears that a top-down approach is taken in that the national assessment is conducted prior to
regional or state/local assessment. In addition, the zero-sum approach is at the national level.
What will the incentive be for state/local agencies to voluntarily reduce their sites, consequently
cutting off their 103 grants and possibly having to reduce staffing? Is the Regional Assessment
Guideline Document that is expected to be completed by September 2004 (p. 5-6) available?
Were consistent statistical methods used at different levels of assessment? Step 3 in the
Statistical Analysis (under Section 5.4, Guidance for Future Regional Network Assessment) and
examples found at http://www.epa.gov/ttn/amtic/netamap.html are helpful, but specific criteria
are needed for national consistency.
Additional Comments
NCore Siting
U.S. EPA Guidance for Network Design and Optimum Site Exposure for PM2.5 and PMi0 (U.S.
EPA, 1997a) defines the following steps for community-oriented core sites and optional
community monitoring zones (CMZ): 1) locate emission sources and populations; 2) identify
meteorological patterns; 3) compare PM concentrations; 4) adjust CMZs to jurisdictional
boundaries; and 5) locate sites. These criteria should be considered for NCore siting rather than
relying on an "NCore design committee" (p. 4-12).
It is not clear how "Nearly 80 'representative' air quality regions that group populations based on
statistical geographical factors ..." (p. 4-11) were formulated for epidemiological studies, and
how "24 rural locations" were selected to support Community Modeling Air Quality System
(CMAQS) model evaluation. A total of 75 NCore Level 2 sites ranging from rural to urban areas
across all 50 states appears to be inadequate for human exposure assessment. In addition to
population density, source emissions, terrain features, and meteorological characteristics should
be factored in. A drastic reduction of speciation sites without adequate analysis of the chemical
composition data at the existing sites seems premature. Sites with long-term historical databases
and stable infrastructures are valuable for trend analyses and network design.
Level 1 NCore Sites
Much emphasis was given to the importance of Level 1 sites, but resources are insufficient. I do
not believe that Level 1 sites should be operated by EPA contractors or academia, or only for
short-term durations as suggested in the Strategy document (p. 4-7). One of their goals should be
to demonstrate how new instrumentation might be used at Level 2 sites. Investment to be made
in Level 1 sites, such as instrument procurement, trained operators, and the establishment of
B-25
-------
information technology for data transfers, requires a long-term commitment. For example, the
Fresno Supersite began operating around mid-1999 and has continued to be operated by the
California Air Resources Board (ARB) for more than 5 years. Several vendors, government
agencies (e.g., Battelle's Pacific Northwest National Laboratory), and academia (e.g., University
of California, Davis; University of California, Berkeley; and Brigham Young University) have
benefited from using the Fresno Supersite sampling platform for method testing and for research.
Concurrent epidemiological studies (i.e., FACES, Fresno Asthmatic Children's Environment
Study) have been conducted by taking advantage of 5-minute average continuous measurements
and 24-hour integrated chemical composition data acquired at the Supersite. As stated on p. 4-15,
Level 1 sites should be an integral long-term network component and operate with greater inter-
site consistency. Therefore, Level 1 sites should follow the same data validation and data
submittal criteria, and should be inclusive of Level 2 site's measurements.
Measurements at NCore Sites
Inconsistent NO or NO2 measurements have been listed throughout the Strategy document that
need to be clarified. I question the utility of month-long averaged NH3 and HNOs measurements.
What kind of denuders can accommodate such a wide range of concentration levels for NH3 and
HNOs without breakthrough? HNOs is a very unstable gas and might shift its equilibrium state
with particles under high temperatures. What will a month-long measurement or 12 samples per
year at 75 locations represent? Ideally, these measurements should be an integral part of the
speciation sampler. Or, at a minimum, they should be collected on a 24-hour basis corresponding
to one or several of the 24-hour speciation samples to understand the gas-to-particle relationship.
In fact, shorter averaging intervals (~1 to 3 hours) are desired for precursor gases to better
understand the gas-to-particle equilibrium under different temperatures and relative humidities.
These measurements may be considered as part of short-term special studies.
To facilitate the transition from filter-based speciation measurements to in-situ continuous mass
and chemical measurements (e.g., SO,f, NOs , carbon) requires a collocated comparison to
ensure the equivalence or comparability between the measurements (Fehsenfeld et al., 2003).
Recent comparisons (e.g., Drewnick et al., 2003; Fine et al., 2003; Weber et al., 2003; Harrison
et al., 2004) show that commercially available instruments still need to be modified to
demonstrate equivalence or comparability.
References
Chow, J.C.; Watson, J.G. Guideline on speciated particulate monitoring; prepared for U.S. EPA,
Research Triangle Park, NC, by Desert Research Institute: Reno, NV, 1998.
www.dri.edu.
Chow, J.C.; Watson, J.G.; Green, M.C.; Lowenthal, D.H.; DuBois, D.W.; Kohl, S.D.; Egami,
R.T.; Gillies, J.A.; Rogers, C.F.; Frazier, C.A.; Gates, W. Middle- and neighborhood-
scale variations of PMio source contributions in Las Vegas, Nevada; JAWMA 1999,
49(6), 641-654.
B-26
-------
Chow, J.C.; Engelbrecht, J.P.; Watson, J.G.; Wilson, W.E.; Frank, N.H.; Zhu, T. Designing
monitoring networks to represent outdoor human exposure; Chemosphere 2002a, 49(9),
961-978.
Chow, J.C.; Engelbrecht, J.P.; Freeman, N.C.G.; Hashim, J.H.; Jantunen, M.; Michaud, J.P.; de
Tejada, S.S.; Watson, J.G.; Wei, F.S.; Wilson, W.E.; Yasuno, M.; Zhu, T. Chapter one:
exposure measurements; Chemosphere 2002b, 49(9), 873-901.
Chow, J.C.; Watson, J.G.; Edgerton, S.A.; Vega, E.; Ortiz, E. Spatial differences in outdoor
PMio mass and aerosol composition in Mexico City; JAWMA 2002c, 52(4), 423-434.
Chow, J.C.; Watson, J.G; Chen, L.-W.A.; Arnott, W.P.; Moosmuller, H.; Fung, K.K.
Equivalence of elemental carbon by thermal/optical reflectance and transmittance with
different temperature protocols; Environ. Sci. Technol. 2004, 38(16), 4414-4422.
Demerjian, K.L. A review of national monitoring networks in North America; Atmos. Environ.
2000, 34(12-14), 1861-1884.
Drewnick, F.; Schwab, J.J.; Hogrefe, O.; Peters, S.; Husain, L.; Diamond, D.; Weber, R.;
Demerjian, K.L. Intercomparison and evaluation of four semi-continuous PM2.5 sulfate
instruments; Atmos. Environ. 2003, 37, 3335-3350.
Fehsenfeld, F.C.; Hastie, D.R.; Solomon, P.A.; Chow, J.C. Chapter 5: Gas and particle
measurements, in Particulate Matter Science for Policy Makers, A NARSTO Assessment,
Part 2; NARSTO: Pasco, WA, 2003, pp. 5-1-5-37.
Fine, P.M.; Jaques, P. A.; Hering, S.V.; Sioutas, C. Performance evaluation and use of a
continuous monitor for measuring size-fractionated PM2.5 nitrate; Aerosol Sci. Technol.
2003, 37(4), 342-354.
Fujita, E.M.; Watson, J.G.; Chow, J.C.; Zielinska, B.; Islam, M. Demonstration PM Study Plan
for Pune, Maharastra, India; prepared for U.S. Environmental Protection Agency Office
of International Affairs; Desert Research Institute: Reno, NV, 2003.
Harrison, D.; Park, S.S.; Ondov, 1; Buckley, T.; Kim, S.R.; Jayanty, R.K.M. Highly time
resolved fine particle nitrate measurements at the Baltimore Supersite; Atmos. Environ.
2004, 38(31), 5321-5332.
Malm, W.C.; Pitchford, M.L.; Iyer, H.K. Design and implementation of the Winter Haze
Intensive Tracer Experiment - WHITEX, in Transactions, Receptor Models in Air
Resources Management, Watson, J.G., Ed.; Air & Waste Management Association:
Pittsburgh, PA, 1989, pp. 432-458.
National Research Council of the National Academies. Research priorities for airborne
particulate matter: IV. Continuing research progress; The National Academy Press:
Washington, DC, 2004. www.nap.edu.
B-27
-------
Pitchford, M.L.; Green, M.C.; Kuhns, H.D.; Tombach, I.H.; Malm, W.C.; Scruggs, M.; Farber,
R.J.; Mirabella, V.A.; White, W.H.; McDade, C.; Watson, J.G.; Koracin, D.; Hoffer,
I.E.; Lowenthal, D.H.; Vimont, J.C.; Gebhart, D.H.; Molenar, J.V.; Henry, R.C.;
Eatough, D.A.; Karamchandani, P.K.; Yang, Z.; Seigneur, C.; Eldred, R.A.; Cahill, T.A.;
Saxena, P.; Allan, M.A.; Yamada, T.; Lu, D. Project MOHAVE, Final Report; U.S.
Environmental Protection Agency, Region IX: San Francisco, CA, 1999.
U.S. EPA. Guidance for network design and optimum site exposure for PM2.5 and PMi0; EPA-
454/R-99-022; US EPA, 1997a.
U.S. EPA. Revised requirements for designation of reference and equivalent methods for PM2.5
and ambient air surveillance for particulate matter - final rule; Federal Register 1997b,
62(138), 38763-38854.
U.S. EPA. 40 CFR Part 51 - Regional haze regulations: Final rule; Federal Register 1999,
64(126), 35714-35774.
Vedal, S. Critical review - Ambient particles and health: Lines that divide; JAWMA 1997,
47(5), 551-581.
Watson, J.G. Visibility: Science and regulation; JAWMA 2002, 52(6), 628-713.
Watson, J.G.; Chow, J.C. Comparison and evaluation of in-situ and filter carbon measurements
at the Fresno Supersite; J. Geophys. Res. 2002, 107(D21), ICC 3-1-ICC 3-15.
Weber, R.; Orsini, D.; Duan, Y.; Baumann, K.; Kiang, C.S.; Chameides, W.; Lee, Y.N.;
Brechtel, F.; Klotz, P.; Jongejan, P.; ten Brink, H.; Slanina, J.; Boring, C.B.; Genfa, Z.;
Dasgupta, P.; Hering, S.; Stolzenburg, M.; Butcher, D.D.; Edgerton, E.; Hartsell, B.;
Solomon, P. A.; Tanner, R. Intercomparison of near real time monitors of PM2.5 nitrate
and sulfate at the U.S. Environmental Protection Agency Atlanta Supersite; J. Geophys.
Res. 2003, 108(D7), SOS 9-1-SOS 9-13.
Zhu, Y.F.; Hinds, W.C.; Kim, S.; Shen, S.; Sioutas, C. Study of ultrafine particles near a major
highway with heavy-duty diesel traffic; Atmos. Environ. 2002a, 36(27), 4323-4335.
Zhu, Y.F.; Hinds, W.C.; Kim, S.; Sioutas, C. Concentration and size distribution of ultrafine
particles near a major highway; JAWMA 2002b, 52(9), 1032-1042.
B-28
-------
Mr. Bart Croes
U.S. EPA's National Ambient Air Monitoring Strategy Implementation
December 15, 2004 Consultation Meeting
CASAC AAMM Subcommittee Review Comments, Bart Croes
Overall, the Strategy document represents a welcome initiative by U.S. EPA to rethink the
nation's approach to ambient air quality monitoring in partnership with state, local and tribal
(SLT) agencies. The document provides a thorough description of the ad hoc process that led to
the current national air monitoring networks, explains why the networks need to be more
integrated and adaptable to changing needs, and provides a reasonable rationale for the many
proposed changes. I appreciate the degree to which U.S. EPA has responded to comments from
the CASAC NAAMS Subcommittee and others. My comments address the four charge
questions posed by Rich Scheffe in his November 19, 2004 memo to Fred Butterfield.
Question 1: Given limited budgetary resources, does the network redesign
represent both an appropriate and adequate balance, as reflected by the relative
resource allocations provided in Section 11, "Draft Implementation Plan," of the
Final Draft NAAMS Document? In addition, are the relative adjustments in the
training and guidance approaches proposed in the draft implementation plan
consistent with the overall objectives of the Strategy?
My comments are based on the premise that public health (and welfare) considerations should
inform the priority and funding allocation for what is measured, and that all measurements must
have clients that will use the data for their intended purpose.
The proposed future resource allocation (Table 11-3) has the right order in terms of the relative
priority of PM, ozone, and TACs. In California, we estimate 400 annual Statewide cancer cases
attributable to all TACs (primarily diesel PM, benzene, and 1,3-butadiene) versus 6,500 deaths
per year related to PM2.5 and 640 deaths per year for ozone. PM2.5 and PM10 are responsible
for the majority of the morbidity effects. The TAG mortality estimate is an overestimate since it
uses the 95% upper confidence limit while the PM2.5 mortality estimate could be up to three
times higher because there may not be a threshold. Thus, PM represents 80-90% of the known
health risk attributable to ambient air pollution and rightly deserves the majority of resources.
I recommend that U.S. EPA drop the PAMS VOC monitoring requirements entirely, as the vast
amounts of data collected each summer do not appear to have a client. Data collected in California
has had very little utility for trend analyses, emission inventory reconciliation, or air quality model
input and evaluation. A strategy of continuous NMVOC mass measurements (for trends to check on
the success of control programs and ozone data analysis), and VOC speciation during special studies
of ozone episodes likely to be used in SIP modeling, is sufficient. This would save at least $10 M
B-29
-------
that can be devoted to data analysis and interpretation (proposed for only $2.2 M), baseline funding
of the Level 1 sites, enhancement of the Level 2 sites, and environmental justice-oriented
monitoring.
Perhaps the resources ($10 M) devoted to PMc monitoring can be minimized. U.S. EPA and
SLT agencies have already invested huge resources into the current PM10 and PM2.5 monitoring
networks. Several states (i.e., California) have State ambient air quality standards for PM10 and
do not plan to follow U.S. EPA in adopting a coarse particle standard. Surely if a site meets the
PMc standard with PM10 monitoring data (uncorrected), then there is no need to deploy a PMc-
specific monitor at the site. While U.S. EPA has not yet promulgated a coarse particle NAAQS,
it has released a Staff Paper with a proposed range of possible standards for PM2.5 and PMc. As
a first-order estimate, data from the existing PM10 monitoring network should be compared to
the proposed lower and upper ranges of the coarse particle recommendations to determine if the
potential scope of a PMc monitoring network would be national in scale or restricted to a few
states. In these likely non-attainment areas, PM10 would primarily consist of the coarse fraction.
Sites that have collocated PM2.5 and PM10 monitors or SLT agencies that have operated dichot
samplers (e.g., California) provide more relevant data for this screening analysis.
The revised Strategy places importance on the Level 1 sites, but needs to devote resources to the
effort (with perhaps matching funds from SLT agencies and industry). Level 1 sites can serve as
a test bed for Level 2 instrumentation, and can appropriately by operated by highly trained SLT
agency personnel. This approach has worked extremely well at the Fresno PM Supersite, with
operations by the California Air Resources Board and funding from U.S. EPA. Major yearlong
air quality field studies (CRPAQS, CCOS), atmospheric researchers (BYU, DRI, UCB, UCD,
UCSD), government agencies (PNNL, U.S. EPA), and a major $7 M health study (i.e., Fresno
Asthmatic Children's Environment Study) have benefited from the Supersite.
The Level 2 sites should include continuous NMVOC (for trends to check on the success of
control programs and ozone data analysis) and CC>2 (for fuel-based emission inventories that are
a good check on MOBILE and EMFAC). Sites coordinated with health studies should include
particle counts or surface area to check health hypotheses and because they do not necessarily
correlate with PM mass like so many PM components. Some of the Level 2 sites should be
located for special purposes. These include roadway or tunnel sites (to measure the success of
the motor vehicle control program), sites to document Asian and Saharan dust events, global Os
trends, and conditions aloft (instrumented buildings).
Environmental justice concerns need some funding. Screening methods (i.e., low-cost, easy-to-use
monitoring technologies) should be developed and deployed to assess near-source exposures in low-
income communities and communities of color. Relatively low-cost passive monitoring
technologies exist for 63, NC>2, BTEX, and HCHO, and portable samplers are available for CO and
PM. These are not FRM-equivalent devices, but should be suitable for screening purposes.
Question 2: Does the Subcommittee have additional suggestions for addressing
the need for integration and communication of the NCore Level 2 network to the
broader community of "users," including scientific researchers (i.e., human
health, atmospheric, ecological) and State, local and Tribal Agency
B-30
-------
representatives? More specifically, what is the most effective manner for EPA to
both reach-out to this broad user community and, where appropriate, to
incorporate their feedback and design input on such issues as monitoring site
locations and parameters?
It is disappointing to read that only 29 comment letters were submitted on the September 2002
draft Strategy document, although that does not necessarily reflect poorly on U.S. EPA public
outreach process. However, as an active member of NARSTO and an attendee at the annual
conferences of the Health Effects Institute, I am not aware that NAAMS was presented to these
broad stakeholder communities. Perhaps a more active effort is needed to identify organizations,
conferences, journals, and other publications where U.S. EPA staff can present the Strategy.
If it has not already been done, an email list-serve should be developed and advertised on U.S.
EPA's monitoring-related websites. I know that Region 9 has a distribution list of SLT
representatives, and other regions likely do as well, that can serve as a starting point. The list-
serve can be used to release the final Strategy and periodic guidance documents, and to solicit
reviews.
A useful product to circulate widely is the maps and summaries of existing networks contained
in Chapter 3. A recent effort by epidemiologists at New York University to do national source
apportionment maps only found one California STN site in AIRS. There are many more sites as
part of STN, IMPROVE, and other networks, and easy-to-access network summaries with points
of contact will increase the use of network data.
Question 3: EPA is considering converting all of the Speciation Trends Network
(STN) sites to Interagency Monitoring of Protected Visual Environments
(IMPROVE) samplers and IMPROVE laboratory and sample handling protocols.
What are strengths and weaknesses of this approach?
There would be clear benefits from consistent sampling, analysis, data handling, and quality
assurance protocols for the STN and IMPROVE networks. My group is a major user of
speciation data for PM SIPs, Asian transport evaluations, determinations of the impacts of
shipping emissions, atmospheric deposition estimates for Lake Tahoe, etc., and the difficulty of
combining the two datasets constrains our analyses. I will leave it to my colleagues to advise
U.S. EPA whether the STN protocols, IMPROVE, or a hybrid should be employed by both
networks. Whatever the choice, an effort should also be made to develop source speciation
profiles consistent with the ambient data.
Question 4: Is it scientifically acceptable to generate air quality surfaces through
modeled observations and/or integrated predictive/observational fields that
would be of appropriate uncertainty for use in the regulatory decision-making
process?
Model results are not a substitute for measurements as they rely on highly uncertain emission
inventories. In general, model acceptability criteria allow much greater errors (on the order of
B-31
-------
30% for ozone) than is acceptable for ambient air monitoring (less than 15%). I doubt the public
would accept any substitute for observational data to determine non-attainment status because of
the risk for "gaming" the model. Ambient air quality standards link back to monitoring data, not
model results used in epidemiological studies. Could satellite data possibly be an alternative
approach?
B-32
-------
Dr. Kenneth L. Demerjian
iiuH. DoLViiilvr l.v 2OH-I
Revised l2/l(vH4
CASAC AAMM Subcommittee Consultation on Implementation Aspects of NAAMS
Deceniher 15. 2004 Meeting at Washington. IX'
Rex ie\\ and Comments: Kenneth I.. Demerjian
The NAAMS linal draft authors have incorporated the recommendations and addressed main of
the issues raised h\ NAAMS Subcommittee of the CASAC at the Julv 2003 review meeting, fhe
implementation aspects of the NAAMS. \\hieh xxas a subject of major concern at the .lulx 2003
rex iex\ lias been addressed in the preparation of "Section 1 1 - Draft Implementation Plan" in the
subject final draft and is the focus of discussion for this meeting. ( Kerall the dralt
implementation plan is xxell thought out and presents a reasonable framexxork and process for the
restructuring and reallocation of funding for lexel-2 and le\el-3 monitoring sites xxithin the
NCore eoneeptnal plan. Unlbrtunaleb. the implementation plan still does not address the lex el-1
funding issue, a subject of considerable concern at the Julv 2003 rex icvx meeting.
The discussion around lexel-1 monitoring actix ities has gone from an "unfunded" critical
element of the NCore strategx to a "possiblv fundable" (xia HPA's Science and Technology
resources) critical clement of the strategx. In a >omcxxhat disingenuous wax. the dralt strategy
suggests the importance of lex el-1 activities, but onlx if new S&T funds become axaiiable.
The fact is the NCore can operate uithout the level-1 component, fhe ultimate question is
\\helheror not the incorporation of lex el-1 sites prox ides suftlcient xalue added benefits to the
NCore strategx to justifx its consideration in the budgelarx redistribulion. fhe strategic plan
should prox ide an assessment oflhe trade offs and limitations such choices haxe on NCore
objectives. Blanket statements such as that made in section 12.1.2 "Resources for Lexel 1
measurements should not be extracted from the existing STAG resource pool, acknowledging the
need lor stable agenex and 1 ribal funding support." are not supportable \\ithout a more critical
assessment of the trade tiffs inxobed.
The AAMM Subcommittee xxas asked to locus their consultation around lour major questions
related to the NAAMS implementation:
1. Ciixen limited budgetary resources, does this represent both an appropriate and adequate
balance, as reflected b> the relatixe resource allocations proxided in Section I 1. "Draft
Implementation Plan" of the Final Draft NAAMS Document? In addition, are the relative
adjustments in the training arid guidance approaches proposed in the drat! implementation plan
consistent xxith the oxerall objectives of the Strategx?
fhe scenario provided in I able 1 1-3 indicating a proposed redistribution of federal resources is
a x er\ good start on defining budget allocations across monitoring programs, ll xxould he ex en
more persuasixe if the details of the FY2003 assessment performed bx Regional Offices xxere
available to determine hoxx disparate their recommendations are from national strategx. Are
these differences the consequences ofconllicting priorities of the stakeholders and hoxx do ih.e>
impact the Agenex "s critical monitoring, objectives? for reasons slated in {he preamble of this
critique. I believe the fable 1 1-3 redistribution should haxe included the NCore lex el-1 actix itx
wiih an appropriate assessment ot'iis impact (pros and cons) on the oxerall monitoring strategx.
B-33
-------
Dcnicnian. December 13.2004
Reused 12/1 MM
This discussion should he prov ided in section 6.4.1 of the report and should dra\\ upon results
and findings from the Supersitc programs. The lev el-1 discussion should also indicate the
direction this speciali/.cd expertise w ill likcK take in addressing needs and extending the \ ision
oi'the national strategy. In scetion 12.1.2 regarding NCore level 1 resources, it is not clear that
ihe use ol' S I AC i resources to support lex el-1 aetiv ities undermines the stability of state agcncv
support. I he case can he made that strong lev el-1 collaborative programs will have significant
valued added attributes to the overall monitoring program and 1'ar out weigh am monitoring
changes that might result from the budget redistribution.
There lias been significant criticism oi'the 1'AVIS monitoring network, specifically directed at
the lack of data anahsis and critical assessment of the uti itv of these data. I t'ullv support the
intent to divert funds a portion PAMS operational funds to support further data anuhsis
activities, hut I would suggest the hPA rev iew its past performance as to how it has expended
such data anuKses funds and the overall effectiveness of those activities. Here again, the lack of
easilv access data dissemination has limited participation and use of these data and stilled
innovations in anaKsCs and interpretation. An\ reductions in the PAMS network should be
rev iewed in terms of its contributions to not onlv its original objects (photochemical oxidant
precursor), but now in light of other contributions it can make w ith regard to air toxics and
PM Organic preeurors.
1 he relative adjustments in the training and guidance approaches proposed seem reasonable, but
I would defer to comments from slate/local entities who receive said services.
2. Does the Subcommittee have additional suggestions for addressing this need for integration
and communication to the broader community of "users." including scientific researchers (i.e..
human health, atmospheric, ecological) and State, local and Tribal (SLT) Agencv
representatives'.' More speeificallv. what is the most effective manner for 1-PA both to reach-out
to this broad user communitv and. where appropriate, to incorporate their feedback and design
input on such issues as monitoring site locations and parameters?
hngaging the scientific research eommuiiitv to participate in the analv sis of measurement data
generated from f.PA based monitoring remains a challenge. A competitive solicitation under the
hl'A (jranis program, highlighting innovative data analv sis approaches for demonstrating the
utililx of NC ore data should be considered. Another more directed approach might consider set
aside funds (e.g. STAG-105) for collaborative (Slate Ajienc} '[ niversilv) data analv sis projects
addressing specific policv relevant science questions utili/ing NCore data.
It has been mentioned main times oxer that the kc\ to engaging the various user communities is
through effective data dissemination. 1 he plans for enhanced data access and the evolution of the
AQS remain in im mind somewhat suspect xxith respect to their meeting the scientific
eommunitx "s interests. 1 will reserve critical comments upon rev iew ing the AQS Data Marl
product slated for deliver) in March 20(15.
3. What are the strengths and weaknesses of 1-T'A's consideration of converting all of the
Speciation 1 rends Network (SIN) ,->peciation sites to Interagencv Monitoring of"Protected Visual
B-34
-------
Denieniun. December I .V 2004
Reused 12/1 MM
Km ironments (IMPROVE) samplers and IMPROVK lahoratorv and sample handling protocols
U) assure consistency in sampling and analysis protocols and enhance harinoni/.atioti ol' rural-
arid urban-based PVh.s chemical spccialion networks?
Maintaining a good qualitv assurance program is the onl\ wav to guarantee dala qualitv and
hannoni/alion. Diversitv ol'samplers and methods helps to idemif\ measurement problems and
keeps the participating parties on their iocs. Gi\en that issues still remain amongst the tiller
based samplers and spcciation measurement methods. I \\ould view the observed differences in
these techniques as beneficial and insightful in resolv ing measurement biases and uncertainties.
At a minimum, before arts action \\ere So be considered regarding this conversion, a critical
rev ieu of the collocated S 1'N/IVIPROVT! results should be performed and present to the A A VIM
Subcommitlee as \\ell as a critical assessment of the various outstanding unresolved issues raised
regarding the differences between these sampling/anahsis measurement svstems.
4, Is it scientifically acceptable to generate air quality surfaces through imn!e/eiiobservations
and/or iniL'^nitt't.//>iv(./icii\v ohsL'rvuiionul fields thaf would he of appropriate uncertain!) for use
in the regulators decision-making process?
1 here is no absolute answer to this question. It depends on a \ ariet) of factors that related to the
qualitv and spatial representative of the observations and species specific emissions, the
predictive model and G1S s> steins and the performance evaluation statistics for the region under
stud). But most important!) it begs the questions as to what is meant b\ "appropriate
uncertain!)".
Other specific comments:
p. 4-9, in fable 4-1 T1 column third enm "continuous PV-12.5 mass" should read "continuous
mass/species"
p. 4-6. the v alue of inonthh average Ml l:i and I INC)-, is marginal and the claim it will stimulate
methods R&.D is questionable.
p.6-10. r' full paragraph last sentence. " 1 kmcver. vshen measuring K()v ..." the same
statement can be made for the current NO\ monitoring s\ stems.
p.l 1-3. in fable I 1-2 some comment in die notes section for 1V1PKOVK Section 105 Grants
would be informative (i.e. how are these funds distinguished from the IVIPROVK Section I0.i
Gant funds).
p.l 1-5. - I"' paragraph last sentence. An example of two as to how these discretionary funds are
spent In the Regional Offices would be informative.
B-35
-------
Dr. Delbert Eatough
Initial Comments by Delbert J. Eatough on
Implementation Aspects of EPA's National Ambient Air Monitoring Strategy
(NAAMS)
After careful reading of the Una! Draft NAAMS, aeeoinpam ing public comments. FPA's
response to these comments in the Addendum to the Final Draft and several of the relc.re.nccd
material in the NAAMS. \\ ith emphasis on the Re\ ision 2 of the Continuous Monitoring
Implementation Plans. I offer the ibikm ing general comments on the NAAMS. implications of
expected results \\ith respect to the successful launching of the Implementation Plan as current!)
outlined, and a feu specific comments on the charge questions.
General Comments
Continuous Monitors: A Lev focus of the NAAMS is a shift from the use of integrated monitors
to continuous monitors at the Level II sampling sites to be included in the NCORF Program of
the NAAMS. The NAAMS document correct!} points out in several sections lite \alue of such
data in providing input for use bv the scientific communit} in the understanding of peak
exposures, atmospheric processes and diurnal variations in the atmosphere. These data uil! all
he potential!} \aluable to the both the SLTs and the scientific communit} which \\ill also use the
data. Improving public access to and developing initial interpretation of this dala is a Lev part of
the Implementation Plan. A strong push for mo\ ing in this direction is the economic advantage
\\hich can be obtained using continuous monitors, coupled with the increased understanding of
atmospheric processes using the data, a potential win-win situation.
As pointed out in the NAAMS. there is also a clash between the monitoring needs of the
NAAMS as outlined in the current regulations \\liich define the fine paniculate FRM as the basis
of monitoring lor attainment and the problems in the FRM which are addressed, but not carefulK
considered in the NAAMS. What \\ ill and \\ ill not be obtained using the new suite of
continuous PM monitors is not considered in the NAAMS. however, the implication is there that
(he expected advances, which might accrue form the availability of continuous data will accrue.
The Continuous Monitoring Implementation Plan, on the other hand, outlined carellillv the
protocols which will be followed as the NAAMS Lex el II program is implemented. I have a
great concern that the philosophv given in the Continuous Monitoring Implementation plan and
the objectives for continuous monitors as outlined in the NAAMS cannot both be achieved.
The Continuous Monitoring Implementation Plan compared FRM and TFOM results
across the counirv. and notes areas of agreement and disagreement and suggests that aerosol
composition is responsible for the disagreements seen. The plan then proceeds to outline
protocols intended to assure that the continuous monitors to be implemented will be consistent
with data which would have been obtained w ith an FRM sampler. Substantial research has been
reported since the Version 2 Draft of the Continuous Monitoring Implementation Plan which
sheds additional light on implications of this approach.
F.PA is probabl} correct in identif} ing. the presence of semi-volatile material in fine
D.IF Comments. 11 December 2004 1
B-36
-------
paniculate matter as being related lo and responsible for the agreement or laek of agreement
between an l-'RM and a TKOM, We now. understand that both nitrate and organic inalerial can
contribute substantial!} to the SVM. Studies which have obtained 1:RM and TKOM data, as well
as correct I \ measuring both nitrate and organic SVM ha\e shown that agreement between the
KRM and TKOM monilors \\ill occur onK \\hcn either there is no SVM. or \\hen losses of SVM
is comparable lor the t\u> moniU)i-s. l;or example, in summer there is a tcndcnc} lor both
samplers to loss both nitrate and semi-volatile organic material and lor there to be agreement
belw ecu the two in 24-h average data. I low ever, in w inter, the I:'RM is often higher than she
I KOM because of better retention of SVM. Moreover, on a 24-h comparison basis, there is
usuall} good correlation in the two data sets over a given season or meteorological condition.
These correlations tend to exist between the l-'RM and TKOM on a 24-h average even when the
slope of such a comparison is different from unit}. These effects probablx account tor some of
the observed seasonal variations and general!} reasonable regression comparisons in the KPA
Continuous Monitoring Implementation Plan report. With such results, there is a temptation to
"correct" the THOM data so that agreement between the two s} stems is generalK seen. While
this is. potentiall\. an acceptable solution tor monitoring purposes, serious problems are
introduced w itli respect lo the uses of continuous data as proposed In KPA in the NAAMS.
Kven when there is reasonable correlation in 24-h data, comparison of l-h average
TKOM and more state-of-the art instrument (such as the KDMS TKOM) show quite a different
diurnal patterns. This is because the diurnal changes in the chemistn of the atmosphere which
lead to SVM not well measured hv the TKOM (and often In the KRM) are averaged out on a
da\-to-t!a\ 24-h comparison. The events which occur leading to significant atmospheric
chemistrv (and potential health risk and maximum exposure conditions) occur on a frequenc\
which can be seen in l-h data but not the 24-h data. 1'hese events are usual K missed b\ a
TKOM. Thus, using modified TKOM data w il! lead to the worst possible situation, believ ing we
have valid data on diurnal patterns because of agreement with 24-h KRM data, but complete!}
missing the diurnal features which will to aid in the understanding of atmospheric processes of
importance to exposure and possible risk. However, using an instrument which will measure
SVM (and hence catch these significant events) will produce data which do not meet the
agreement protocols (particular!} the + 10 °t> slope agreement) given in the Continuous
Monitoring Implementation Plan, preciselv because the, atmosphere is better monitored.
KPA ma} well be constrained lo not use these newer techniques because the} will not
agree with the KRM and thus fail monitoring legal requirements, especial!} in locations where
SVM is important and variable. I lowever. a consequence is that we will he producing data
which give us an inaccurate picture of the atmosphere and thus lead to incorrect decisions based
on continuous monitoring data. KPA needs to find a wa\ around this problem which can be
applied to the various Kev el II monitoring sites, allow ing for the adv ances in understanding
w hich form one of the major arguments tor nun ing low arc! continuous monitors. At a minimum.
some (if not main ) of the Level II sites should have both a KRM equivalent continuous monitor
and a state-of-the art continuous monitor which allows us to belter understand atmospheric
processes. The comparison between the two will give valuable insights on SVM in the
atmosphere and aid the health communit} in obtaining a better understanding of exposure.
D.IK Comments. 1 I December 2004 2
B-37
-------
The problem outlined in this section hears direct!) on Question 2 in the charge to the
CASAC AAMM Subcommittee. While these issue do not address the mechanisms of
communication actions. the} do address the application oftlie NCore Le\el 2 network in
supporting long-term health effects research and providing better support to ecosvslcin
assessments. We nun ven. well impede the progress \\e intend to assist.
Spcciation Sampling. One oftlie most valuable uses of the speeiation sampling results will be
the interpretation of processes occurring in the atmosphere \\hich lead to the observed
concentrations ofPM. The locus in the NAAMS Implementation is on the maintenance of 24-h
speeiation netuorks (uether STK or IMPROVL). 1 here are two signillcant downsides to this
approach: 1 ) cost sax ings possible \\ illi cotitinuous speeiation samplers are not reali/ed. 2)
There \\ ill be no wa\ to use the speeiation data to help us understand the atmospheric process
revealed h> the continuous monitor data.
LPA is correct in pointing out the current uncertainties in Hash volalili/alion specialion
results. 1 lowev er. new techniques liav e been introduced the past two v ears (real!} since the
rewrite of the NAAMS. and certain!} since the last version of the Continuous Monitoring
Implementation Plan) which appear to overcome the shortfalls of the Hash volatili/alkm
techniques. While these techniques are still in their inl'anc\. 1 believe the data is there to support
a reasonable introduction of the leclinolog} into the Level II sites, thus aliens ing the richness of
the combined continuous I'M and specialion results to be reali/ed. I his would be particular!}
true if the suggestion given above of locating both FRM consistent and state-of-the-art
continuous PM instruments at a reasonable number of Level II sites were implemented. This is
the onlv wa\ progress will be made on mov ing the science Ibnvard. In fact, inclusion of these
continuous monitors nun be a uav to 1111 in some of the limitations of the requirement of FRM
equivalenc} lor the continuous P.VI monitors. Such equivalent'} is not required oftlie speeiation
continuous monitors. I would hope that the level of competence of operators at the Level II sites
is such that a instrument such us an ion ehromatographv based measurement of unions and
cations ssould be feasible. This also opens the door to a better understanding of SV organic
material.
I would consider the use of funds in this direction to be a much better direction to go than
to replace all our STK sites with IMPROVL equipment which will give us constant1} across the
rural and urban sites (Charge Question 3) but will not realN mov e us ahead on the use of
continuous monitor PM data with respect to supporting long term health effects and atmospheric
process research (Charge Question 2.)
HNOj and NH3 Monitoring. Charge Question 1 includes Monitoring for these two species.
1 low ev er. w hile the} are essential species lo be monitoring lor both PM and o/one data
interpretation, and ecological effect understanding, the NAAMS Implementation Plan current!}
includes measuring this species onl\ on a month!} basis. It appears to be assumed that
modification of existing denuder techniques which have been extensive!} tested on a 24-h basis
will give v iahle long-term a\ erage data. I low ev er. there is no data giv en to support this
assumption. More important!}, the value of long-term averages of these two species is reullv no!
D.IL Comments. 1 I December 2004 3
B-38
-------
spelled out. It is tacit!) assumed the data will be valuable. I do not see how month!)-average
nitric acid and ammonia data can be combined with either 24-h PM. continuous PM or
continuous gaseous data to prov ide meaningful insights. The lime frames are too different for
the clala to be of am value other than provided an isolated long-term average. I he data mav
provide some insights in ecological effects hut will not be useful wilh respect to understanding
atmospheric processes.
While resources committed to this effort are admitted!) small, the benefits appear to be
equall) small. These are new continuous monitors lion chromatograpbicall) based, for example)
w hich would prov ide this data continuous!). A modest inv cstment in ev aluating these
techniques could add to the justification for inclusion of such instruments in the speciation
network at Level II siies (see comments above) and significami) aid both the objectives related
to the Charge Question \ And Charge Question 3.
Comments on Charge Questions 1 and 2.
As outlined above, serious problems are created when there is a mismaich belween the
measurements of a target species (e.g. PM) and the measurement of precursors (e.g. ammonia
and nitric acid). In addition, the measuremenl ofolher precursor gases is nol matched bv the
measurement of I'M species (e.g. sulfatc. nitrate) on a comparable time basis. I consider the
decision to avoid continuous speciation measuremenl and valid continuous PM mass
measurement a significanl imbalance in the value of the proposed XAAV1S Implementation Plan.
There are options for mov ing ahead w ith a modest inclusion of conlinuous measurement of all
pertinent species which would belter reveal the usefulness of She over-all data set and prov ide a
basis for cither expanding the program to all Level II sites or Devaluating wether the Spcciation
Monitoring data is meeting the intended objective with respect to making informed air qua!it)
management decisions and obtaining relevant exposure data for research programs. The voids
are potential!) too important to ignore as a major change in direction is implemented.
Comments on Charge Question 3.
The obvious strength of the proposed shift to IMPROVL protocols for all speciation sites
is that the data among all sites w ill be more comparable. The dow nsidc is that w c are ihcn apt to
forget that there are opcralional dcllnitions in the results and. hence believe, since there is
eoiiMstene). there is accurac). Perhaps the current situation of reminding ourselves, for
example, thai the carbon results do have problems (in both networks) is. in sonic \\a\s. better.
Neither the NIOS11 or IMPROVT. method for C anal) sis has been established as the superior.
However, it would be useful if LPA tried to incorporate tecitnologv into the specialkm network
which would allow for at Icasl an estimation ol'scmi-volalilc organic material. But then, even
hotlor)oh as outlined above, is making at least at effort toward continuous spcciation at some
Level II sites. This could include continuous C measurements for both nonvolatile and semi-
volatile carbon based on recent developments h) our group \\ith a modified Sunset monitor.
Thai modified monitor is now commercialh available from Sunset.
DJL Comments. 1! December 2004
B-39
-------
Comments on Charge Question 4.
While this seems a highly desirable direction to go. as it uill opthni/.e availability oi'
funds For important new objectives in the NAAV1S. I do not ha\e a complete understanding of
bou this objective would he accomplished. 1 also do not fully understand \\hat is considered a
suitable data base for the generation of the air quality surfaces. This nun well retlect a lack of
understanding ofthe expected process on my part, and not a deficiency in the KPA plans. 1 uill
look for further insights into this issue at the 15 December meeting. Some ofthe questions 1
have are: Will the generated fields be based on measurements ofthe regulated species
concentrations oniv. or uill reasonable projections of the changes in new source emissions (both
priman and of precursors) closer to the modeled site he included? Will the modeled fields be
generated \vith a suitable set of meteorological inputs to reasonable project expected transport.
chemistrs and deposition and variability in meteorological conditions at the modeled sites? How
will the uncertainly in the modeled fields he established based on the above, and probably other
points 1 do not see vet? If acceptable results can he demonstrated in trial locations with
variability in the above, this seems like a very acceptable direction for I-PA to go.
D.IK Comments. II December 2004
B-40
-------
Mr. Eric Edgerton
Response to CASAC AAMM Subcommittee Charge Questions on Implementation of
National Ambient Air Monitoring Strategy
Question 1: Given limited budgetary resources, does this represent both an appropriate
and adequate balance, as reflected by the relative resource allocations provided in Section
11, "Draft Implementation Plan," of the Final Draft NAAMS Document? In addition, are
the relative adjustments in the training and guidance approaches proposed in the draft
implementation plan consistent with the overall objectives of the Strategy?
It is difficult to answer the first question without: 1) knowing the fungibility of 103 and 105
grant dollars; 2) having the criteria pollutant network assessments from the Regions; and 3)
having an analysis of the PAMS network. The working assumption in Chapter 11 appears to be
that PM2.5 funds (2nd largest pool) are fair game for redistribution, criteria pollutant funds (by
far the largest pool) are not and PAMS funds (3rd largest pool) are somewhere in between.
Given this scenario, except for the inability to fund Level 1 sites, resource allocations seem
reasonable and adequate.
The PAMS network is a continuing enigma. On the one hand, it represents a tremendous
investment and a game attempt to collect high time resolution data for ozone assessments. By
my tally, the average operating cost for a PAMS site is nearly 375K. Many of the PAMS
measurements are analogous to what is needed for PM assessment. For example, the experience
of measuring 50-60 organic species by GC might come in handy when we want to measure 4-5
ions via 1C. On the other hand, there is little evidence the data have ever been used or even
scrutinized. My recommendation would be to conduct a forward-looking analysis of PAMS with
the following questions in mind: 1) Can the network be redesigned to serve needs for ozone, PM
and toxics assessments? and 2) Can a subset of PAMS sites serve as Level 1 NCORE sites?
Infrastructure at PAMS sites should provide considerable leveraging. Although PAMS is not
geographically positioned to provide all Level 1 sites, might it not underwrite 3 or 4, if the total
number of PAMS sites were reduced to 50 or 60? If the answers to these questions is "no", then
much deeper cuts in the current PAMS funding are warranted.
Regarding the second question, the QA component will need substantial funding for
development of standard reference materials in addition to lab and field audits. The training
component will need substantial funding to ensure the entire chain of data
collection/management understands the measurement objectives. For the Level 2 sites, a key
objective will be detection of secular trends, and this will require very careful attention to
analyzer response in the bottom 10% of the measurement range (5-20 ppb for SO2 and NOy;
100-300 ppb for CO).
B-41
-------
Question 2: Does the Subcommittee have additional suggestions for addressing this need
for integration and communication to the broader community of "users," including
scientific researchers (i.e., human health, atmospheric, ecological) and State, local and
Tribal (SLT) Agency representatives? More specifically, what is the most effective manner
for EPA both to reach-out to this broad user community and, where appropriate, to
incorporate their feedback and design input on such issues as monitoring site locations and
parameters?
As a long-term solution, analysis of data for Level 1, Level 2 and PAMS sites and publication of
results in peer-reviewed journals is hard to beat. It might also be worthwhile to convene
symposia or panel discussions at ISEEpi (International Society of Environmental
Epidemiologists, not to be confused with International Society of Explosive Engineers). There
could be a panel discussion at annual SLT monitoring workshops. The panel would include
recognized experts in the relevant fields and would discuss future monitoring needs and
opportunities. In the near-term, all data users should be alerted to the NCORE Strategy and
impending changes, and encouraged to provide feedback.
Question 3: What are strengths and weaknesses of converting all of the Speciation Trends
Network (STN) speciation sites to Interagency Monitoring of Protected Visual
Environments (IMPROVE) samplers and IMPROVE laboratory and sample handling
protocols?
Harmonization is seductive; however, I would recommend a thorough analysis of network
protocols before making a wholesale move to IMPROVE. Clearly, something needs to be done
about carbon and a switch to the IMPROVE protocol (TOR) makes sense in terms of current
operational definitions of OC/EC. Not so clear is whether STN measurement technologies and
sample handling protocols should be replaced. Does either network offer advantages in flow
control or Dp50 or sample preservation? For various reasons, IMPROVE does not measure
ammonium ion, an important component of PM2.5 mass, a nutrient species and an important
counter-ion to nitrate and sulfate. Instead, IMPROVE incorrectly assumes that nitrate is always
associated with ammonium and that sulfate is always fully neutralized by ammonium (molar
ratio = 2). We have to do better than this if we are ever going to understand dynamics of particle
formation, etc. STN, on the other hand, does measure ammonium and does have protocols in
place to ensure a certain level of data quality. If a change is in the cards, it will be important to
ensure it doesn't impair current data quality.
Question 4: Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integrated predictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision-making process?
From a scientific standpoint, you want to bring all information to bear. The key will be in
understanding uncertainties and properly factoring them into the decision-making process. If
this can be done, then there are significant advantages to the concept, including better definition
of attainment/non-attainment areas and potentially accelerated designation/re-designation. While
B-42
-------
the scientific advantages are obvious, using combined surfaces for regulatory decision-making
may be a very tough sell to stakeholders.
Eric S. Edgerton
12/19/04
Cary,NC 27513
B-43
-------
Mr. Henry (Dirk) Felton
Responses: Dirk Felton, NYSDEC (Submitted 12/14/04)
Charge to the CASAC AAMM Subcommittee
Dec 15th 2004 Meeting
General Comments follow the four questions to the committee:
1. The CASAC has expressed its support for the Agency's proposal to redesign the routine PM
monitoring network to support PM precursor gas measurements (CO, SCh, NO/NOy, NHs,
HNCh) at NCore Level II multiple-pollutant sites, and for air quality management decisions and
to obtain relevant exposure data for research programs.
Questions: Given limited budgetary resources, does this represent both an appropriate and
adequate balance, as reflected by the relative resource allocations provided in
Section 11, "Draft Implementation Plan," of the Final Draft NAAMS Document? In addition, are
the relative adjustments in the training and guidance approaches proposed in the draft
implementation plan consistent with the overall objectives of the Strategy?
No, the proposed extensive PM-2.5 non-attainment areas demonstrate that PM
monitoring and particularly PM-2.5 speciation cannot be reduced in many areas.
Additionally, it may be that the PM-2.5 annual and daily NAAQS will be lowered which
could require additional monitoring resources. This was evidenced after the
implementation of the 8-Hr Ozone NAAQS. More borderline attainment areas required
additional monitors and non-attainment areas required additional up and downwind
monitors. Establishing sites is difficult and expensive and it does not make sense to close
a site that may be needed a short time later for ozone, enhanced PM-2.5 or PM-coarse
monitoring. Many of the NCore principles are important and should be gradually
incorporated into the existing NAMS and SLAMS requirements.
The training and guidance approaches may be consistent with the strategy, however, it is
not generally helpful for the monitoring agency staff who need it most. Many of the staff
who will be operating the newer technologies are located in remote offices without the
ability to travel or to access many of these information resources.
2. The implementation plan proposes a series of communication actions to advance the
NCore Level 2 network, in order to more directly support long-term health effects research and
provide better support to ecosystem assessments through an increased level of coordination.
Questions: Does the Subcommittee have additional suggestions for addressing this need for
integration and communication to the broader community of "users," including scientific
researchers (i.e., human health, atmospheric, ecological) and State, local and Tribal (SLT)
Agency representatives? More specifically, what is the most effective manner for EPA both to
reach-out to this broad user community and, where appropriate, to incorporate their feedback and
design input on such issues as monitoring site locations and parameters?
B-44
-------
State agencies that have non-attainment concerns must be able to select site locations
based on how they believe they will be able to demonstrate the effectiveness of their SIP.
This monitoring, which is already aimed at protecting human health must take priority
over other research monitoring priorities. Health oriented researchers may be interested
in urban hotspot monitoring while a modeler may be interested in a boundary site or even
a monitoring location chosen at random. The monitoring objectives and operational
procedures may be too different for some of these groups to reconcile in one monitoring
location or in a comparable dataset.
State and Local Agencies face the very difficult task of operating quality-assured
monitoring networks consistently for long periods of time. Many research oriented
monitoring programs are designed to examine a particular issue and then publish results.
Some of these Science oriented monitoring programs such as CASTNET have
demonstrated problems in the long term operation of criteria instrumentation. One
CASTNET site in the lower Hudson Valley expanded to Ozone monitoring. Over time,
the operation of the instrument was not properly quality assured and the Regional EPA
office attempted to make the surrounding area non-attainment for ozone because of this
data.
It is advantageous for the various monitoring agency, health agency and research groups
to meet regularly to discuss instrument selection and limitations, data comparability,
multi-media pollutants public awareness and other issues.
3. One of the remaining technical issues relates to harmonizing rural- and urban-based
PM25 chemical speciation networks such that both categories of networks utilize consistent
sampling and analysis protocols. For example, EPA is considering converting all of the
Speciation Trends Network (STN) speciation sites to Interagency Monitoring of Protected Visual
Environments (IMPROVE) samplers and IMPROVE laboratory and sample handling protocols.
Question: What are strengths and weaknesses of this approach?
The urban IMPROVE study has not been completed and it is not apparent that the
IMPROVE protocols would be entirely successful in urban environments. For example,
inlets on IMPROVE sampler are cleaned annually while the STN inlets are cleaned either
Monthly or after each sample depending on the type of sampler. A better approach
would be to select the strengths of each network and design a new Nation-wide program.
Carbon sampling techniques can be drastically improved by minor changes in flow rates,
filter shipping containers and field procedures. Since the cost of analysis is so high for
this program, it makes sense to optimize the sampling equipment in an effort to raise the
quality of the data.
4. As EPA implements the National Ambient Air Monitoring Strategy to address multiple
monitoring objectives, it will be looking to spatially optimize the ambient monitoring networks.
This may mean that some redundant monitors in adjacent, but separate, geopolitical areas (e.g.,
neighboring counties) are "divested" from a given network. Although technically sound, these
divestments could result in data gaps which might, in turn, adversely impact regulatory decision
B-45
-------
making. The Agency is willing to adopt alternative approaches for assessing regulatory issues
such as non-attainment designations, so long as such approaches are scientifically justifiable;
hence, the rationale for initiating discussion of these issues with the CAS AC.
Question: Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integratedpredictive/observational fields that would be of appropriate
uncertainly for use in the regulatory decision-making process?
No, ambient data must be used in the regulatory decision making process. This is clear
from all of the recent cases where ambient data has been used to refute poor model
performance. Models are developed to work in a majority of usually simplified
applications. It is difficult enough to verify a model's overall predictive accuracy but
impossible to prove that it accurately covers all areas of a domain. Many areas of a State
or Territory may not fit into the acceptable range of a model and this could not be known
without actual data. Recent experiences have shown that the EPA is willing to use
models that are not accurate or verifiable.
For example, the EPA's accepted model for use in PM-10 SIP development calculates
road dust concentration based on the mass emitted from vehicle tailpipes. This model
was developed for older cars on roads with fewer vehicles than those of today. The
model substantially over-predicts the amount of vehicle generated PM-10 in dense urban
areas such as NYC. What is disappointing is the EPA's lack of concern or attention to
poor model performance even when it prevents State Agencies from meeting their
requirements.
Another example is the EPA's use of the Urban Excess model which has been
inappropriately applied in the urban Northeast. The EPA is currently using this model to
determine the extent of the PM-2.5 non-attainment areas. This model has been applied
without taking into account actual ambient data, transport, secondary particle formation,
meteorology or differences among monitoring programs.
General Comments on the National Air Monitoring Strategy:
The majority of people on the AAMM board work in the research field and most likely will be in
favor of the overall NCore strategy. Many of the NCore objectives are worthwhile such as trace
gas monitoring, a greater emphasis on data and program analysis and a better integration with
health and multimedia communities. There is more disagreement towards the overall objectives
of NCore from State and Local monitoring Agencies and perhaps even from the Regional EPA
offices. I see a greater loss from what is left out of NCore than what may be gained from its
adoption in its present form. I think the lack of enthusiastic support from some of the EPA
regional offices stems from their concern that when the spatial coverage of monitoring is
reduced, they will have a more difficult time responding to public issues and complaints.
My concerns with NCore include the melding of networks with different design principles such
as PM-2.5, Ozone and Toxics, the lack of emphasis on sources, characterization, controls and
B-46
-------
permits and the reduction in State Agency flexibility due to the tightening of SPM (Special
Purpose Monitor) regulation. The NCore funding strategy was determined by examining Nation-
wide statistics from AQS Criteria Data. The value of individual State networks or even
individual sites cannot be examined in this way. What is the value of a PM-10 instrument if it is
the only one downwind of a permitted facility? What will happen when the PM-10 standard is
rescinded before suitable PMc stack testing methods are developed. Data from many State and
Local Agency's monitoring sites are put into the AQS system even though these sites are not
Federally-supported SLAMs or NAMs monitors. In NY, for instance, almost one third of the
Ozone monitoring network operates at State funded Acid Rain sites. Closing these sites would
not create the funding opportunity that the NCore strategy indicates.
Many of the positive attributes of NCore can be implemented with less disruption to State and
Local agencies' existing monitoring and permit programs. Currently, State and Local monitoring
programs are evaluated by their respective EPA Regional offices annually. This network review
should be expanded to include representatives from the National OAQPS office, health officials
and if the case warrants; scientific specialties, such as atmospheric modelers, deposition
researchers, the Forest Service, the Park Service, NOAA, etc. The NCore monitoring objectives,
sampling technologies and operational experience should be available as a resource to the
stakeholders at these network design meetings.
Other goals of NCore may be achieved or at least improved upon through simple procedural
changes. Some State Agencies still upload /^ of their MDL (Minimum Detection Limit) to AQS
in place of their concentrations for parameters such as NO2 and toxics. This convention prohibits
modelers, risk assessors and data analysts from using low concentration data in their analysis.
Other low cost opportunities to improve low concentration data include lowering span
concentrations where appropriate, setting up dual range analog outputs, and through the use of
digital data logging.
The benefit to the State and Local monitoring agencies from the implementation of NCore is
supposed to be the flexibility that results from the reduction in criteria monitoring. The NCore
strategy suggests that 1000 may be an appropriate number of NCore Level III sites operating
Nationally. When this number is examined as representing the majority of the spatial component
of monitoring including PM-2.5, PM-10, PMc, Lead, Ozone, NO2 and Toxics, it is apparent that
there will not be any "surplus" Level III sites available for State and Local Agency specific
needs. The resources needed to establish monitoring for SIP development, control strategy
verification, toxics hot spot investigations or investigations of environmental justice are not
clearly defined in the Air Monitoring Strategy Document.
B-47
-------
Dr. Rudolf Husar
Response to the NAAMS Implementation Plan
Rudolf Husar, CAS AC AAMM Meeting, Updated December 17, 2004
Overall, the NAAMS implementation plan is as good as its parent conceptual plan. It
has been developed through the same responsive participatory process and it shows.
The freshness of the implementation plan sounds almost too good. What's the ketch?
Actual Implementation?
Question 0: Is the plan consistent with the NAAMS strategy?
To a large degree yes. A key implementation issue concerns the realization of the multi-tier,
monitoring strategy represented by the 3-level pyramid. Implicit in the strategy is that the levels
will be mutually supportive entities for the characterization of air quality. How will the tiers be
linked to form a coherent monitoring unit that provides useful, integrated data from the new
integrated network?
The 'pyramid' is actually three layers A
,i \
'Pillars' give the pyramid shape /
No pillars, no pyramid
No NAAMS
No
Level 3
Non- NCore Data
Question 4: Is it scientifically acceptable to extrapolate observations by physical or
statistical models?
Not only acceptable, but a necessity for integrating the multi-tier data. Thus, spatial data
extrapolation (say from L2 to L3 and below) should be an integral part of NAAMS strategy
implementation. It should be part of the monitoring process. 'Smart' extrapolation schemes can
now be used to constrain the estimates by the data as well as by physical laws. The derived
pollution 'surfaces' resulting from the monitoring process could then be used for regulatory as
well as for many other purposes.
Question 1: Is this implementation plan an appropriate reallocation of monitoring
resources?
The bulk of the reallocation makes sense. However, routine network assessment and level 2-3
data integration need specific implementation plan and line item in the budget. The $$ for these
B-48
-------
network operation activities could be from reduced PAMS and from the dubious $10M PMc
monitoring allocation.
Question 2: How to improve communication with user communities (health, atmospheric,
ecological)
By a more open NAAMS process (glasnost) and by closing communication feedback loops.
Question 3: Should the STN sites be converted to the IMPROVE protocol? IF (Since) the
STN approach does not show distinct advantages, it would make sense to adopt the IMPROVE
protocol for all routine speciated monitoring, so urban-rural differences can be reliably
quantified.
The implementation plan takes a very defensive posture toward linking with non-NCore
networks. There are several readily available real-time datasets that could significantly augment
the NCore aerosol characterization, e.g. ASOS and Satellites. As shown below, the spatial
coverage of real-time AIRNOW PM25 (-300 stations) and ASOS light scattering networks (1200
stations) operated by NWS, FAA and DoD would enrich the spatial texture AND the in situ
characterization
B-49
-------
AIRNOW Data-Not Official!
, (300 Stations, Hourly
2004-07-21114:00:
-125 -120 -115 -110 -105 -100 -95 -90 -85 -80 -75 -70
^rovider: US EPA Delivery: DataFed.Net F
METAR Surface Meteorology: Extinction Coefficient
521
ASOS Light
Scattering Sensor I
Stations,
-120 -115 -110 -105 -100 -95
-90 -85
-80
-75 -70
Delivery: DataFed.Net
B-50
-------
Satellites add spatial texture (MODIS, 250m), source identification (smoke, dust) and some
vertical information. The example shows a superposition of ASOS Bext and 1 km resolution
SeaWiFS reflectance and Aerosol Optical Thickness during the July 2002 Quebec Smoke event.
The augmenting the NCore 3-tier pyramid with such data should be encouraged and made
explicit in the implementation plan.
B-51
-------
B-52
-------
Dr. Kazuhiko Ito
Revised comments on the NAAMS final draft.
Kaz Ito, 12/20/04
General comments:
I came into this review process rather late (this Final Draft was the first draft I read regarding
NAAMS). Since most of the decisions regarding the network design have been already made,
some of my comments may be too late. I will nevertheless provide these comments hoping that
they may still be useful in the remaining period of implementation as well as in the future
revision of the Strategy.
The difficulty of balancing the budget with the most reasonable scientific decisions to re-
design the ambient monitoring network is overwhelming, but I was impressed with the level of
effort and progress made so far. The proposed NCore network is appealing. Its plan to collect
multi-pollutant data at each monitor accommodates the need of health effects research
community (such data were often lacking in the past network monitors). I understand that the
priorities of the state and local agencies are the most important factors in designing the new
network, but I also hope that the Strategy will take a closer look at the issues and the needs of the
health effects research community. In the past ten years, many observational epidemiological
studies utilized the data from these routine monitors, producing valuable inputs into the process
of setting the NAAQS. It is interesting to note that, while the main motivation to reduce the
number of monitors for CO, SCh, and NCh comes from the observation that most of these
monitors measure levels well below the NAAQS, many of the PM studies in the past ten years
reported associations with mortality and morbidity at levels well below the NAAQS. I am not
implying that I am against reducing redundant monitors (I am all for it). I am just suggesting
that we should do this carefully. I assume that determining the "redundancy" for CO, SO2, and
NO2 monitors would be much more complicated than for more regional pollutants such as O3.
Also, it seems that the decisions about the speciation monitors are being made without detailed
analyses of the new speciation network data (collected so far, 2001-2004). Some specific
comments related to these issues are given below.
Question 1: Given limited budgetary resources, does this represent both an appropriate and
adequate balance, as reflected by the relative resource allocations provided in Section 11, "Draft
Implementation Plan," of the Final Draft NAAMS Document? In addition, are the relative
adjustments in the training and guidance approaches proposed in the draft implementation plan
consistent with the overall objectives of the Strategy?
It is not easy for me to assess the adequacy of the balance without knowing the details of
the process to eliminate a certain fraction of monitors (for example, what reasoning was used to
reduce non-trend PM2.5 speciation sites from 160 to 80?). Given my limited knowledge on this
process, the general distribution of budget seems ok to me.
B-53
-------
Question 2: Does the Subcommittee have additional suggestions for addressing this need for
integration and communication to the broader community of "users," including scientific
researchers (i.e., human health, atmospheric, ecological) and State, local and Tribal (SLT)
Agency representatives? More specifically, what is the most effective manner for EPA both to
reach-out to this broad user community and, where appropriate, to incorporate their feedback and
design input on such issues as monitoring site locations and parameters?
The Strategy should obtain feedback from the epidemiology/public health researchers
who have been using the data from these routine monitors as well as those who have been
involved in setting up their own monitors for epidemiological studies. I do understand that the
existing and future air monitoring network is not necessarily for epidemiological studies but
mostly for regulatory and compliance purposes, and the budget and capacity of the SLT agencies
would determine the type and extent of the change in the network design. However, input from
the health effects research community should still be useful, especially because so many of the
observational epidemiological studies relied on the use of data from these routine air monitors
that were not developed for epidemiological study designs. Monitor site location needs for both
short- and long-term health effects study designs should be considered. I speculate that many of
the epidemiologists and public health scientists who have used the routine air quality network
data are unaware of the changes that are taking place through the Strategy. I would suggest a
workshop to identify and discuss the issues among the health effects/exposure community.
Federal Register is certainly not the medium that these researchers regularly seek information
from. At the NAAMS subcommittee meeting, Dr. Scheffe suggested that Health Effects Institute
may be a good place to have such a forum, but I think we need to reach a broader list of
researchers who should be informed on this issue. I know that some of us are even writing grant
applications (without knowing the changes in the network design) that rely on these air quality
data. Aside from getting feedback, at minimum, all the current and recent EPA grantees, as well
as those who have been involved in this type of research should be notified of the proposed
changes.
Question 3: What are strengths and weaknesses of converting all of the Speciation Trends
Network (STN) speciation sites to Interagency Monitoring of Protected Visual Environments
(IMPROVE) samplers and IMPROVE laboratory and sample handling protocols?
I think it would be easier to answer this question if we had a good summary of the STN
data that have been collected so far (2000-2004) so that we could compare various aspects of the
data with those for the IMPROVE database. The IMPROVE web site has a very good
descriptive analysis of their data. I haven't seen a similar analysis for the STN (I am sure many
of us, including myself, are trying to do this now). So, it is somewhat frustrating to make a
judgment without sufficient knowledge. With this limitation in mind, the strength of switching
to IMPROVE protocol would be that we would have one larger database that is measured in a
consistent way. At the 12/14/04 NAAMM meeting, I learned that they are collecting data using
co-located monitors with the STN and IMPROVE protocols. We should probably wait for these
data to be analyzed before we make decisions.
B-54
-------
Question 4: Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integratedpredictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision-making process?
This question is rather general. However, for a specific situation, I think there is always a
"scientifically acceptable" (the best available, anyway) way to use the modeled data. I assume
that EPA is referring to the surface depiction of some smoothed data such as Figure 5-1 on page
5-1 of the NAAMS final draft. In this case, the objective appears to be to identify areas where
removing existing monitors does not affect the predictive power of some statistics, in this case,
the 4th highest 8-hr daily maximum average Os value. The approach taken for the objective here
seems reasonable. The remaining question would be the model uncertainty associated with the
interpolation methods. For example, if we used another available method (e.g., kriging) to
estimate the surface, how much difference does it make? Also, how do you decide on the
acceptable (tolerance?) level of error?
At the 12/14/04 NAAMM meeting, I learned that these modeled surfaces were not being
used for determining monitoring "redundancy" to remove monitors, so I take back my initial
concern about the application of these models for SC>2 and NC>2 monitors. I maintain my concern
regarding the applicability of these models for non-regional pollutants. For regional pollutants
(e.g., O3 and sulfate), we can generally assume smoothness in spatial variability (except high
density traffic areas where NO may quench 63). Determining concentration surfaces for PM2.5
monitors may also be relatively less problematic in areas where spatially homogeneous
secondary sulfate dominates PM2.5. Estimating concentration surfaces for CO, SO2, and NO2 as
well as some of the PM2.s species would be more difficult because they are more strongly
influenced by primary emissions from local sources, and the monitor-to-monitor correlations are
much poorer both in terms of temporal correlation and mean levels. The problem with the
locally impacted pollutants that, if separation distance of two monitors is relatively far (> 20
miles), then the two monitors having similar average concentrations (or any annual statistics)
does not necessarily imply that a point between the two monitors would have a similar
concentration because they may simply reflect separate local sources that do not reach or
influence concentrations at the mid-point. Thus, the prediction error would be greater for these
pollutants than 63 or PM2.s in many of urban areas. I looked at monitor-to-monitor temporal
correlation for PMio and the gaseous pollutants in the nationwide data, and found that the overall
rankings in monitor-to-monitor correlation within the same Air Quality Control Region to be: Os,
NO2, and PMio, (r ~ 0.6 to 0.8) > CO (r < 0.6) > SO2 (r < 0.5), confirming the larger prediction
errors for locally impacted pollutants. We will have to deal with this issue for the PM2.s
speciation data. Some of the PM2.s species are more locally impacted than others. The
implication is that differential errors across the PM2.5 species from different source types (e.g.,
regional vs. local pollution sources) will affect source apportionment results and associated
prediction errors. This issue needs to be investigated using the current trend and non-trend PM2.s
speciation monitors' data. Ideally, such information should influence the decision as to which
non-trend monitors should (can) be removed (for the planned reduction of non-trend monitors).
B-55
-------
Dr. Donna Kenski
Response to CASAC AAMM Subcommittee Charge Questions on Implementation of
National Ambient Air Monitoring Strategy
Donna Kenski
Lake Michigan Air Directors Consortium
December 14, 2004
Question 1: Are the relative resource allocations appropriate and balanced?
The allocations presented in Table 11-3 seem mostly reasonable. However, the continuing
inability to confirm a funding mechanism for Level 1 sites is disturbing. These are an integral
part of the strategy and must be allotted adequate long-term funding. It is greatly encouraging to
see a line item for data analysis (two line items, in fact). The data analysis funding for the toxics
program has been a terrific investment that has and will continue to yield benefits for years to
come. That said, the strategy's funds for data analysis and interpretation seem a bit puny for the
size of the program - even the least generous estimates suggest that a minimum of 10% of
project funds should be spent on data analysis. Although some analysis will be undertaken by
SLTs or others, the costs are considerable to plan, perform, and integrate various analyses and
then make the results web-accessible. Inadequate funding will only leave us with unanswered
questions and broken or unsatisfactory public access to important data.
While the decrease in gravimetric PM measurements is justified, and moving to continuous PM
is advisable, the document doesn't discuss allocations to continuous speciation measurements,
except the modest bump-up in CASTNET funds for their 3 pilot sites. These continuous
speciation measurements are ultimately the most revealing about the nature of PM2.5, and more
investment on EPA's part is absolutely critical. These measurements are perhaps best left to the
Level 1 sites for now, but EPA needs a stronger commitment (financial and political) to
advancing the state of science of these measurements so that they can eventually be rolled out to
more sites. Since ORD isn't supporting this development, perhaps the strategy could take a
stronger stand.
Training and guidance adjustments were consistent with objectives; the emphasis on webcasts
and DVDs should be well received by states. It's nice of EPA to acknowledge the travel
restrictions that so many SLTs are subject to. I'm a little skeptical that these types of training
can really be adequate for some of the more complicated measurements - e.g., NOy. In general,
while the strategy's objectives were clear, I have some doubts that the existing technology (and
the states' willingness to operate it) are really ready for this broad dissemination. Here again, as
the strategy notes, EPA/ORD has failed to keep up with the needs of the monitoring community,
leaving a very large gap between our measurement needs and the technology available
(commercial or research) to meet those needs.
Question 2: How can EPA best reach the user community and incorporate their feedback?
Well...why not just ask them? EPA knows the user community. It's not that complicated or
expensive to put together a list of people likely to have something to say on these issues. Put
B-56
-------
together a short, focused set of questions, or even a draft proposal that gives EPA's rationale for
site selection, then get on the phone or email. You are much more likely to get meaningful
responses if you ask people directly. And don't forget to ask the stakeholders, too.
Question 3: What are the strengths and weaknesses of converting STN to IMPROVE?
From a data analysis standpoint, harmonizing these networks would be a huge improvement.
The urban/rural signals we're so anxious to tease apart are hopelessly obscured by these network
protocol differences. But so many issues are yet to be resolved! Blank correction - to do or not
to do? What's the right way? What role do inorganics and the various organic species play?
What's the effect of higher mass loadings on samplers designed for clean rural environments? Is
there a 'machine' effect at DRI or RTI - do different instruments yield different results for the
same protocol? This is not my area of expertise, but since both IMPROVE and STN protocols
are just that, and the EC/OC measurements are operationally defined, it seems silly to have the
two major networks be incompatible. We should choose one and be consistent. Even though
there are many outstanding questions (some of which should be answered by Supersite data and
other current studies), IMPROVE's very long and very reliable record make it the obvious
choice. EPA should make this switch part of the strategy, and get a plan for the transition in
place ASAP.
Question 4: Is it scientifically acceptable to generate air quality surfaces through modeled
observations or integrated predictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision-making process?
Of course it is scientifically acceptable to generate surfaces from modeled or observed data.
Defining 'appropriate uncertainty' is more difficult. Our interpolations should be based on a
fundamental knowledge of spatial variability of a given pollutant. In some cases we have a very
good understanding of pollutant emissions, dispersion, transport, and the resulting spatial
variability. But there are serious gaps in our knowledge of other pollutants. We don't have any
good idea of NH3 or HNO3 spatial variability across an urban area, so how can we have a
realistic estimate of modeled uncertainty for such pollutants? And if our modeled PM2.5
estimates depend on these precursor concentrations, then this uncertainty will propagate through
a model and perhaps lead to unacceptable uncertainty for decision making. This lack of data on
spatial variability (particularly on a microscale or in complex terrain) is a data gap that the
strategy doesn't really address. There really ought to be some provision for determining spatial
variability via short term saturation monitoring.
B-57
-------
Dr. Thomas Lumley
Comments on the National Ambient Air Monitoring Strategy.
Thomas Lumies
Associate Professor of BiosUitislles
l'ni\ersit\ of Washington.
General comments:
The KPA has proposed removing monitors thai Lire statistical!) redundant, moving
towards continuous-lime monitoring of major pollutants. impro\ ing data accessibility .
and seltini: aside specific 1 uiids lor data analysis. I hese are all excellent ideas if ihes can
lie achieved. Mv understanding is thai there is some disagreement about the accuracy of
continuous-time PM mass measurement. \\ hich ma\ reduce the usefulness of the
continuous-time data. There ma\ slill he benefits in automated telemeln of c\ en 24-
hoLirs a\ eratic data.
Spatial variability:
While the genera! use of monitors that are rcprcsentativ e of broad population exposure is
to he encouraged, there is a need to characteri/.e some small-scale spatial features.
The epidemioloijic evidence supporting the current PM standards is based lartzcK on
studies of acute effects of exposure. These l\ piculh compare changes o\ er time in I'M
concentrations to changes over lime In the rale of ad\ erse health e\ cuts.
For these sludies it is important onh thai the temporal pallern of concentrations at the
monitors tracks the temporal pattern of population a\ erage e.xposure. It is thus sufficient
that a monitor be broatlK representative of a population area.
More rcccnlK, there has been increasing interest in studies ol the eflects ol longer-term
exposure. These studies examine hou differences between people in exposure are related
to differences in risk. For these studies it is important that the variation in e.xposure
between individuals is uell estimated, I hese studies require much more accurate
information about spatial variability. While much of the detailed monitoring required for
lame epidemiologic studies can appropriately be carried out b\ sliid\ personnel, it v\ould
be useful to have some information about common spatial fealures. In particular, the
spatial pattern of pollutant concentrations near major roads in both rural and urban areas
is of interest. C'urrent monitoring f>ro\ ides good data on ('() in urban street camons, but
relaliveh little information on P\l ujadlenls near roads.
Data analysis and availability
hnprov ed ease of access to air qualilx data wo-uld be a \ aluable outcome of a ne\\
strategs. Providing complex data in a form thai can be readih used is difficult, however.
'The iHAPSS website produced at Johns Hopkins and funded b\ the Health Hffccts
Institute (lutp: wu w ,iiiLLpss.Jhsjili.edu) is an impressive model, but required substantial
B-58
-------
resources and is targeted to one specific use of the data, It mu\ be useful for sonic data
anal) sis resources to he made available specifically to support collaboration in ciala
anal) sis between HPA and non-KPA researchers.
The experience of the NHLBlTimcled Cardiovascular Health StutK (CHS) nun be
relevant here. As required h\ N1H. CHS has made large quantities of data publicalh
available and c\ en provided a part-time staff position to handle am enquiries about the
use of the data, Ver\ little use has been made of these public data. On the other hand.
main researchers \\ ilh no prior connection to CHS hav c found a ( 'HS sponsor and made
product! v e use of the data in that u a\.
Charge questions
1. I cannot comment on u heiher the resource allocations would be sufficient for the
given tasks. Assuming that thev are sufficient, the balance appears reasonable and
consistent with the objectives of the Strateg}
2. The Agenc\ and other Federal bodies alreatK require grand uncled researchers to
make certain data publiealK available. The Agenc\ should consider also
encouraging researching to share data anal) sis tools the} develop, w here this can
be done without undue effort or confitlentialitv, problems. In addition to lowering
the barriers for researchers tieu to the data this would improve the transparent'} of
research.
Some Agent1} recognition for successful and creativ c use of routine monitoring
data in public communication (eg good ueb displaxs) might increase awareness
of the value of HPA data.
3. The strength of a uniform IMI'KOVH protocol is that measurements will be more
comparable across space. One ob\ ious disucK antage is that thcv u ill be less
comparable over time, For epidemiologic research this is of secondan importance
as the main within-loealion comparisons that are of interest are short term, 1 note
that the greater uniformity need not imp!} greater accuracy — as far as I know it
is not clear which measurements of. eg. carbon fractions, are the most accurate.
4. It is in principle scientificall} acceptable to generate airqualtU surfaces b\
empirical fitting or mechanistic modeling that would be of satisfactory qualitv for
regulator) decision making. The current decision making procedure alread\ aims
to protect public health ov er a relative!} large spatial region from measurements
at a few points. In deciding which monitors to remove Irom a network it would
be useful to consider how close the air qualilv is to regulator} limits, both now
and in the near future. Fora given level of redundancy of information it would be
more desirable to remove a monitor that was far below a rcL.'ulator\ limit than
B-59
-------
one very close to tiie limit. In addition, as noted above, retaining a small number
of geographically close monitors would be useful to investigate small-scale spatial
gradients.
B-60
-------
Dr. Peter McMurry
Peter H. McMurry
University of Minnesota
December 18, 2004
Comments on EPA's plan for Implementing the National Ambient Air Monitoring Strategy
General Comments:
a. I compliment the EPA Staff who developed this strategy. They have been responsive to
previous recommendations by this committee, and have taken the initiative to create a coherent
vision that extends well beyond specific recommendations of the committee. NAAMS represents
a clear vision for a measurement strategy that is suitable for the 21st Century. NAAMS will lead
to more efficient use of the Agency's resources for carrying out measurements and providing
access to data. This strategy will play an important role in protecting the public's health and
welfare. It has been a pleasure to serve on this committee.
b. My primary concern with the NAAMS proposal is the lack of a plan for funding Level I sites.
My arguments in support of Level I sites are listed below:
• I have been involved in studies of atmospheric aerosols for more than 30 years. During this
period, the need for instrumentation that would measure, in real time, the composition of
aerosol particles was clearly understood. With funding from the EPA Supersites program
(and other programs that focused on airborne particles) in the 1990s, several different
instruments were developed to enable such measurements. This is terrific progress.
However, while several such systems are now commercially available, the technologies
have not yet grown to maturity. Level I sites would enable continued testing of such
instrumentation. I have no doubt that within the next decade, instruments that can routinely
and automatic measure the composition of atmospheric aerosols will be available. Level I
sites will play an enormous role in ensuring that this occurs.
• State and local agencies prefer to use instruments that operate continuously rather than filter
samplers, which are more expensive to operate and provide less useful data. Level I sites
will provide platforms for evaluating the performance of such instruments.
• Real-time measurements of composition offer the potential to provide much more useful
information on aerosol composition and at a much lower cost than can be achieved with
filter samplers. Real-time data is required to evaluate chemical transport models, to assess
human exposures, and to understand processes that affect size-resolved aerosol composition.
I am convinced that such data are needed to significantly advance our understanding of the
effects of atmospheric aerosols (especially their health effects). Level I sites would play a
substantial role in advancing measurement technology "to the next level."
B-61
-------
• Virtually all of the basic research and development on real-time instrumentation for
measuring the composition of atmospheric aerosols has been carried out in the U.S. Level I
sites will help to enable this industry sector to grow to world leadership.
• The measurements that are proposed for Level II and III sites do not include any
measurements of ultrafine particles (<100 nm) despite the fact that (1) concern has been
expressed about possible health effects of sub-100 nm particles, (2) planned changes in
emissions control (especially for diesel-powered vehicles) are likely to lead to significant
changes in concentrations of ultrafine particles, and (3) recent work has shown that such
particles can be measured well on a routine basis. Such measurements have been conducted
for extended periods (more than a decade, in several cases) in other countries, and these long-
term measurements are providing extremely valuable information on trends and
"climatology." Again, this is a case where the basic R&D was carried out in the U.S., but the
benefits of that work are largely being realized abroad.
• One of the greatest scientific challenges is understanding semi-volatile species (i.e., species
that are present in both gas and particle phases). Important among these are ammonium
nitrate, and many organic compounds. We cannot currently measure all semi-volatile
compounds on a routine basis (although techniques for measuring the gas and particle phases
or inorganics in real time are available.) Organics are an especially challenging problem.
Level I sites would enable continued work on this problem, which is likely to play a
significant role in potential health impacts of air pollutants, and which will require more
work before gas/particle distributions can be accurately represented in chemical transport
models.
c. Based on the discussion at our meeting I concluded that realizing the full benefits of NAAMS
will be hampered by EPA's organizational structure. As was pointed out by Professors Husar
and Hopke, it is important that NAAMS data (1) be available in real time, and (2) be easily
accessed by the public. The presentation by representatives in the Information Transfer and
Program Implementation Division made it clear that they do not currently plan to establish a data
access strategy that meets these needs. It appears that the Monitoring and Quality Assurance
Group is hampered because the Information Transfer and Program Implementation Division is
not responsive to their needs. This needs to be changed.
d. Recent studies provide growing evidence that those who live immediately downwind of
highways are especially likely to experience adverse health effects from air pollutants. The
NAAMS strategy does not explicitly deal with this issue. Some consideration should be given to
citing some of the NAAMS sampling stations in locations that would enable long-term studies of
such effects.
e. I served as one of three co-chairs on the NARSTO PM Assessment, and I wrote Chapter 11 of
that Assessment, which addressed needs for further research. I am delighted to note that many of
NARSTO's recommendations are included in the NAAMS program. NARSTO
recommendations that are not explicitly discussed include:
• The NAAMS proposal makes no mention of any consultations or discussions with
monitoring agencies in Mexico or Canada. One of the NARSTO recommendations is that
networks across North America should be "harmonized" to enable the integration of
B-62
-------
knowledge from ambient measurements, receptor models, chemical transport models as bases
for air quality management. I recognize that the U.S. cannot dictate to Canada and Mexico
how they should carry out air quality measurements. However, if they were informed about
U.S. plans, they would have an opportunity to work in parallel to develop measurement
programs that are reasonably compatible.
• NARSTO recommends that we aim to develop methods to assess benefits of emission
controls to air quality and the linkage to human health and welfare. NAAMS focuses
primarily on measurement, without explicit consideration given to the use that will be given
to such data.
• NARSTO recommends that ASOS and AWOS data be integrated into national air quality
monitoring programs. Such data are acquired routinely, but not by EPA. The public would
benefit substantially if such data were included in NAAMS.
My responses to questions given by Richard Scheffe in his November 19 memorandum are given
below:
Ql. Given limited budgetary resources, does this represent both an appropriate and adequate
balance, as reflected by the relative resource allocations provided in Section 11?
My primary concern is the lack of a specific plan to fund Level I sites, as explained above.
Other committee members have expressed other concerns that should be reviewed by EPA.
Q2. Does the Subcommittee have additional suggestions for addressing the need for integration
and communication to the broader community of "users."
Meet with Mexicans and Canadians to inform them of our plans, in hopes that they would
establish compatible measurement strategies and data distribution policies.
Q3. What are the strengths and weaknesses of converting STN to IMPROVE.
I will defer to others on the committee who are more knowledgeable than I about sampling and
analysis strategies for these networks.
Q4. Is it appropriate for EPA to use "models" for identifying non-attainment areas.
First, it is important to clarify what is meant by a "model." Is this a full-blown chemical
transport model (CTM) that includes everything that is known about meteorology, sources,
transport and transformations of air pollutants? If so, than it would be appropriate that such
models be used for regulatory purposes. CTMs would also provide useful information on diurnal
variations, etc., which is typically not obtained from sampling networks used for attainment
studies. Of course, different models will produce different results, so the choice of models that
are used for this purpose would be highly controversial.
B-63
-------
More empirical models that interpolate among data from sparse measurements would need to be
used with greater care. Such an approach would probably work reasonably well for species that
are distributed relatively uniformly. However, when species are emitted locally, then such
interpolations would likely lead to significant errors.
So, in summary, the use of "models" is needed, but implementing them for regulatory purposes
will not be straightforward.
B-64
-------
Dr. Armistead (Ted) Russell
Review of EPA's Final Draft National Ambient Air Monitoring Strategy and responses to
Charge Questions
Armistead (Ted) Russell
Georgia Institute of Technology
Atlanta, GA 30332-0512
Having read through the new Draft Implementation Plan section of the Final Draft National
Ambient Air Monitoring Strategy in detail, followed by the whole report, as well as attending the
AAMM meeting and presentations by EPA, my first comment is that I was pleased to note the
extent to which the strategy has evolved since our last review, and how they have addressed
many of the comments provided in the last round. In particular, they have increased their focus
on monitoring to address issues dealing with environmental endpoints, e.g., health and ecosystem
impacts. As noted below, there are lingering issues in this regard. Also, I note that they are
planning on adding more sensitive CO monitors: another plus. There is also increased emphasis
on analysis.
In this review, first, the four questions we have been charged are addressed. This is followed by
a more global discussion and summary.
Question 1. Given the limited budgetary resources, does this (the redesign) represent both an
appropriate and adequate balance? Are the adjustments in the training and guidance approaches
consistent?
As noted above, I am pleased to see that more sensitive CO measurements will be
pursued as part of the NCore Level II monitoring as this is important both for understanding of
PM sources, but also as a marker for the potential presence of elevated concentrations of other
automotive emission related species. In our recent source apportionment work, we found that
having SO2, NO/NOy and CO measurements available led to more stable and reliable results
when conducting PM source apportionment. The gas phase species act to constrain the solution
space. Further, these measurements can be made at a temporally finer scale than is currently
being used for the PM speciation.
In the preamble to this charge, the emphasis was on gas phase measurements. In regards
to PM measurements, the reduced emphasis (and resource allocation) on FRM PM mass
measurements is applauded as they help relatively little with understanding the sources and
solutions of the PM in a region. Continuous measurements help somewhat. However, it was not
apparent if the continuous measurements are primarily mass or speciated mass measurements.
Continuous mass measurements still fall short of providing the type of information needed to
really understand the sources impacting a region and for evaluating source apportionment study
results. Greater emphasis on semi-continuous speciated measurements by the emerging
techniques is suggested. Even if this does not cover the full range of species needed to conduct
CMB-type source apportionment, they will support other approaches, plus provide more
B-65
-------
information on the atmospheric dynamics. Further, along with the continuous gas-phase
measurements, one can utilize additional methods for assessing source impacts. Thus, if the
increased emphasis on continuous methods does not include continuous (or semi-continuous)
speciated measurements, I would rethink the resource allocation to include this important set of
measurements, at least in a limited, critical area, approach. This will also lead to identifying the
appropriate technologies to use elsewhere.
I would, in fact, go a bit further, and suggest that resources for non-continuous, speciated
measurements be reduced and those put in to continuous speciated measurements. One could
still, at a very reduced frequency, still conduct a full suite of speciated, integrated (filter-based)
measurements for use with CMB-type analyses, and also seasonal detailed OC analysis. I would
suggest that this source apportionment, along with the continuous gas and PM speciated
measurements will provide ample information to understand, even better, the sources impacting a
receptor, and provide much more information to those who might use the data later (e.g., the
health community).
The adjustment in training and guidance seems reasonable, though I do not have a good
feel for the level of resources it takes to effectively address this task.
Question 2: Does the Subcommittee have additional suggestions for addressing this need
(communication and coordination with health effects and ecosystem assessment communities),
including scientific and SLT representatives? What is the most effective manner for EPA to
reach-out, obtain and incorporate feedback?
Again, EPA is to be applauded for recognizing this need. However, the document was
rather vague in how this would be accomplished. I would be more proactive, identifying which
cities have been the basis for many of the recent health studies, as well as the researchers. Ask
them, specifically, what they would recommend. (A possible workgroup?) Likewise for
ecosystem researchers. Also, EPA should address the issue of site representativeness again. The
Supersite data, and that for other special studies and routine monitoring provides a wealth of data
to address the questions: How representative is a single site being used for health/ecosystem
assessments? What determines the representativeness of a site? How should one identify a site
that is a balance between having an historical record and is most representative of the exposures
(human/ecosystem) of concern? The issue of site representativeness for use in different purposes
has not been well addressed, particularly with attention to uncertainties. This is explored further
in the response to Charge 4.
Question 3: What are the strengths and weaknesses of this (harmonizing rural and urban-based
PM2.5 chemical speciation networks, e.g., using the IMPROVE EC/OC method) approach?
There is no easy answer here. In general, there are a variety of benefits to using consistent
methods, which usually outweigh some minor reasons for maintaining different approaches for
different applications. Data analysis is typically much more straightforward with less
uncertainties (note the word uncertainties, not uncertainty... one deals with the number of
uncertainties, the other with the total amount of uncertainty introduced) when the measurements
are conducted in a consistent fashion. On the other hand, there is some value in having two (or
B-66
-------
more) competing approaches in terms of more readily identifying weaknesses in the other
approaches, and if there is a known weakness in the approach that is to be adopted. In the case
of the IMPROVE vs. NIOSH method debate, both have weaknesses. My read is that the case is
very much open as to which one has greater weaknesses, in particular the consistency of results
for conducting source tests and the ensuing source apportionments. It would be a mistake to
adopt one method now if it makes if introduces a greater uncertainty in the analysis of the
measurements, e.g., if the diesel pm profile is more consistent using one method over the other.
Thus, it may be premature to switch right now. I have heard that the argument for choosing
IMPROVE over NIOSH is that IMPROVE has been around longer, and is not willing to change.
This is the wrong reason to switch the STN approach. EPA should show, and the community
agree, that IMPROVE has fundamental advantages over NIOSH. Without this, we should not
switch as that will close, possibly very adversely for future analyses, an important debate. If one
can develop better source profiles using NIOSH, then NIOSH should be used and vice versa.
Again, this is an area where the measurement community at large should discuss which method,
ultimately, is preferred and for what reasons. Absent from this debate should be that one method
or the other has been around for longer. That should come second.
In summary, I support harmonization of the methods used, but as an evolution to using
the best techniques, not as a convenience.
Question 4: Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integrated predictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision making process?
Great set of issues arising from this one question! First, however, I would not use
"modeled" observations in this process. Use simulated concentrations. That will upset fewer
people.
Next, should one use purely simulated concentrations? Surely not with today's level of
modeling accuracy/reliability. That was an easy answer. However, one could also answer is that
any worse than relying on a single observation in determining attainment/non-attainment and to
identify sources viewed as contributing to PM in a region. The measurements are representative
of a very limited region. This gets back to Question/Charge 2. While an observation showing
non-attainment very, very strongly suggests that the PM levels in a region are out of attainment
with the standards, it is less strong for determining attainment! The degree to which the
observation is below the standard is suggestive as to how confident one is that the region is fully
in attainment, but it has a limited spatial application to demonstrating that the region is in
attainment. A number of observations in a single region does provide further evidence to
determining attainment, but the observations all have limited spatial application and further
analysis is necessary to support their regional representativeness.
Thus, this leads to the use of integrated predictive/observational fields as being the
preferred approach. While there is much work to be done here, this approach will help tackle
multiple issues. First, it is probably the best way to extend an observation (or set of
observations) both spatially and temporally, if necessary. Second, it can be used in the process
of source apportionment (or vice-versa-: source apportionment can be used in extending the use
of observations). Third, it will help identify uncertainties in the representativeness of the
B-67
-------
observations at a monitoring location. Fourth, it will produce the type of information that can be
used by groups identified in Question 2. Thus, this is, in some ways, the silver bullet. However,
we still do not have the best approach laid out, and the enabling technologies developed (e.g.,
software/hardware environments to provide this information). However, this can probably be
tackled within three or for years. Such a process should utilize the observations available, both
in situ and remote (e.g., satellite) and PM modeling (with data assimilation). There should be a
feed-back loop where the information provided by the integrated system utilizes additional
approaches to assess the quality of the fields developed (e.g., data withholding, etc.). While this
may seem lofty, but EPA should set, as a goal, to have a field of source apportioned
daily/monthly/yearly PM for the U.S. by 2010 (e.g., target PM2.5 and coarse for 2008, with the
apportioned fields developed by 2010). The work should also include fields of the uncertainties
in the integrated daily PM levels and, at least, in the annual source apportionments. This should
be updated on an on-going basis. (Once it is done once, the second time is relatively easy.) If
the resources were currently available, I suspect this could actually be achieved by 2008 (using
2006 data).
I like what was presented at the meeting in this regard. EPA is going the right direction,
i.e., by developing surfaces using an integration of observations and modeling. They must take it
one step further, and that is to also calculate uncertainties. Folks who are against using modeled
concentrations for attainment/non-attainment decisions must realize that at present, we are not
even being so sophisticated. A number of counties are designated non-attainment without even
such information, and some are being designated non-attainment even with observations
suggesting attainment. Let's put some more rigor in the process, then struggle with the issue of
whether one should be 95% confident that a county is in/out of attainment to make
determinations.
General Discussion. As noted in previous discussion, I am generally pleased with the Draft
Implementation Plan and their responsiveness to prior recommendations. I am concerned about
the motivation to move to the IMPROVE approach without further discussion about the
disadvantages and advantages of the variety of methods. Given that I strongly recommend going
to more semi-continuous speciated measurements, this may be somewhat moot... instead of
discussing IMPROVE vs. NIOSH, lets discuss how to do the measurements semi-continuously
(if that is at all feasible), and the NIOSH vs. IMPROVE attributes as part of that discussion.
Given that the reason for conducting speciated measurements is to do source apportionment, the
question to address is which one is best suited to that task. This requires asking (and answering)
the question which method will provide the most consistent source profiles within a source
category and is least sensitive to interferences from atmospheric processing. (Difficult
question... ask the real experts.)
We were not directly charged with addressing the reduction in the funding for PAMS and the
allocation of resources directly for analysis of PAMS observations. PAMS data has been
plagued by both issues of quality and timeliness, and has been badly underutilized. To this end, I
would devote greater resources to the analysis of PAMS data, at the expense of other aspect of
PAMS, possibly reducing the coverage (spatially and temporally) of the PAMS network until we
milk the current PAMS data and understand what the network can really tell us. This will help
identify the future role of PAMS, and how resources should be allocated down the road. Until
B-68
-------
we really start using the PAMS data for what it was intended, let's spend some additional
resources to do more analysis of what we have with the concurrent reduction in resources
allocated for producing more unused data from monitoring.
Question 4 is a very important question in regards to policy and the need to develop the
appropriate techniques to achieve EPA's objectives in a scientifically defensible fashion. With
satellite observations, a much more powerful information technology infrastructure and new, but
not "reference method" measurement technologies coming on line, there is the possibility of
being able to estimate air pollutant fields, exposures, and source impacts, much better than can
be achieved from individual monitoring sites. This includes better identification of
"exceedences" defined more broadly as not specifically measured, but sufficiently likely given
the full range of information available. (Determining sufficiently likely will be fun.) Use of the
full range of sources can reduce the needed resources for pure monitoring, but this requires
developing the foundation that can undergo scientific and legal scrutiny. This activity will also
provide a foundation for better forecasting and control decisions. Don't be shy about tackling
the set of problems associated with blending observations (of whatever type), model results, and
other inputs.
Finally, the NCore Level 1 sites should get funds from sources other than Science and
Technology as part of the mission of these sites is regulatory, and their operation should not be
viewed as something that has little relevance to today's issues. Planning requires knowledge of
sources, which can come from NCore level 1 sites, as will attainment/non-attainment decisions.
Using the process EPA has developed to decide high-priority Level 2 sites, one could also locate
high-priority Level 1 sites, which then would be used, along with more detailed modeling, to
identify the regional sources.
B-69
-------
Dr. Jay Turner
CASAC AAIMIV1 Evaluation of Coarse Particle Monitoring Methods
Comments Submitted by Jay Turner on December 23, 2004
As Follow-Up to the December 15, 2004 Meeting
Per the November 19, 20O4 memorandum from R.D. Schcffe to F. Butterfield, the
Subcommittee has been charged with three questions: "< 1) given limited budgetary
resources, does this [the proposed network redesign] represent both an appropriate and
adequate balance, as reflected by the relative resource allocations provided in [...]":
(2) Does the Subcommittee have additional suggestions for addressing this need
[communication actions in the implementation plan] [...]": <-*) What the strengths and
weaknesses of his approach [converting all STN speciation sites to IMPROVE samplers
and IMPROVE laboratory and sample handling protocols; and (4) Is it scientifically
acceptable to generate air quality surfaces through modeled observations and, or
integratec! predictive observational field that would be of appropriate uncertainty for use
in the regulatory decision-making process? My written comments focus primarily on the
background materials provided prior to the meeting with some additional content added
in response to the meeting deliberations.
Question 111. (Hven limited hudgetarv resources, (foes /his ft he proposal network
redesign/ represent hoth an appropriate and attenuate halance. as reflected hv the
relative resource allocations provided in /.,./ "7 \ do not have the background to frame a
response in terms of the adequacy of relative resource allocations. However, the
following comments arc offered purely from the measurement {not resource) context.
Overall, the approach appears sound...
Regarding the upgrade to trace «ias measurements, in 2OOI or 2OO2 Warren White
conducted an analysis of criteria gas data at the Jefferson Street (Atlanta, GAt site and the
131'1 and Tudor ( East St. Louis, IL} site: it was striking how much added knowledge could
be gamed from the trace gas monitors compared to the conventional monitors. I
encourage you to contact Warren for these materials if EPA is interested in assembling a
concrete example of the added value in deploying trace measurements.
I echo the comments of several Subcommittee members that the lack of programmed
funds for Level I sites is disheartening. Both the Supersites program - and some current
activities which were catalyzed by the Supersites program demonstrate the intrinsic
value of having a few such sites. For example, while our measurement matrix at the St.
Louis Midwest Supersite is scaled back significantly from the core measurement period
(April 20O1 May 2O03K we have sustained out role in providing a platform for vendors
to beta-test instrumentation both in terms of collocated measurement data and on-site
field personnel. A key clement of this role is the extensive engagement of our site
personnel with the instrument developers, accomplished in large part because this is an
explicit part of our mission.
Current instruments for semi-continuous PM speciation jsulfate, nitrate, OC'EC) do not
have adequate challenge tests for determining field performance. There was discussion at
the meeting re
-------
of such feasibility, it will be imperative to have routine (at least monthly) filter samples
collected at such sites with rapid litriuu'ousul on the chemical analysis (mass, ions.
OCEC) as this data is needed us part of the routine instrument performance check.
The Final Draft of the National Ambient Monitoring Strategy states that detailed site
assessments will be conducted. I strongly support this effort us this metadata is critical
for the data users. Historically, network expansions to add additional measurements (in
contrast to new sites towards increasing measurement density) often capitalize on
existing infrastructure. However, the zone of representation for a site is necessarily
pollutant dependent and. depending on the measurement objective, what might be a
"good" PM speciation site might not be a "good" NATTS site. Detailed metadata is
needed to provide adeqtiate context for interpreting site-specific data. As nationwide data
becomes more-readily accessible, it is very important for the user to also have readily-
accessible access to important metadata (including narratives of the site characteristics.
aerial photographs of the sites, digital imagery, ...).
Question H2. Does the Sithconiiiiitft'c have (K/iiifioiutf sngxcsiioiis /or m/drcssiai; tlii.s
need {communication actions in the implementation plan/ /.../ "".' I have no concrete
suggestions to offer in this urea. Certainly coordination with f IE! would be prudent. I
understand they have funded someone to look at the STN network with a critical eye
towards its potential use in exposure and health effects studies (proposals were due in
Spring 20O4). This activity might shed light in a consolidated manner on what existing
sites / siting characteristics would be desirable. As forgetting out the message to
possible end users about the network redesign (and specifically, network pruning), in
addition to the stakeholders discussed at the meeting, perhaps another channel is the
respective metropolitan planning organizations (MPOs). They are often charged with
addressing the cross-section between air quality and transportation planning, and thus
often have air quality advisory boards with broad representation.
Question :!3, U'/u/l lite sirens/is and weaknesses aj his approach j convert ing all .S'7",V
spectaltoii silcs io IMPRO IT. samplers and IMPRO I ~E lahoratorv ami sample handling
protocols'.' From my perspective, the only glaring weakness is that we will be breaking a
time series of up to five years in some locations. Considering the advantages, however,
and especially in light of most of the intended data uses, this may indeed be justified.
Both networks have strengths and limitations, and currently the differences between the
protocols make it very difficult to integrate the data sets. I do not advocate a hybrid
approach which attempts to take the best elements from both STN and IMPROVE: even
if politically palatable, it would break the time series for two large networks. On the
other had, there certainly may be aspects of the STN program that work towards
providing a "better" representation of the true aerosol: it would be very important to
document such differences even if the "better" process is not adopted. An example
includes cold shipping and handling of all substrates - does it really make a difference, in
light of actual sampling conditions and post-sampling field latency'.' Such questions will
not necessarily be answered - at least not directly - through STN IMPROVE
mtcrcomparisons but it would be very fruitful to conduct small studies to address such
Turner
B-71
-------
issues (or simply summarize, in the context of background for the network transition the
data that has already been collected on such matters).
Finally, as discussed at the meeting, IMPROVE: integrates not just the sampling and
analytical work, but also the quality assurance. If the transition from STN to IMPROVE
is indeed implemented, this could be an opportune time to refine the QA strategy tor the
urban network and thereby ensure both its integrity as well as direct comparability to the
rural network.
Question !!4. /> il sciesiltfh-tilh' acceptable to generate air (jitulilv surface* through
modeled ohsermttons a ml? or integrated predictive--'ohsemitional field that would he of
appropriate uncertainty jor use in the regukitoi'v decision-makingprocess? My response
is a qualified "yes". Approaches such as model-generated air quality surfaces with
"nudging" by observed data could have substantial use, subject to the caveat that we have
adequate input data, fully understand the key physicochemical mechanisms that are built
into the model, and have an adequately dense monitoring network. A very good example
based strictly on air quality surfaces of observed concentration fields - is paniculate
matter nitrate in the Midwest and Central Plains. Recent expansion of the IMPROVE
network by adding several protocol sites to the C'ENRAP domain has dramatically
changed the shape of air quality surfaces for PM nitrate and surfaces generated prior to
the network expansion would have been misleading. While this is stating the obvious, it
hopefully reinforces the high bar that must be set.
Turner
B-72
-------
Dr. Warren H. White
Initial responses to the AAMM charge Warren H. White, 12/8/04
Question 1: Given limited budgetary resources, does this represent both an appropriate
and adequate balance, as reflected by the relative resource allocations ...?
I think it serves as a model for how to extract maximum utility from given resources. It is a
thoughtful initiative that has been carefully deliberated, and is well crafted to align the Agency's
air monitoring activities with the needs of the coming years.
Question 2: Does the Subcommittee have additional suggestions for ... communication to
the broader community of "users," including scientific researchers (i&, human health,
atmospheric, ecological)... to incorporate their feedback and design input on such issues as
monitoring site locations and parameters?
The Agency is already doing very well in this regard, but I might suggest adding measurement
quality to the list of issues on which inputs are explicitly solicited. I have in mind here such uses
as epidemiology, where critical information is carried by variations in concentration rather than
by concentration itself, or source apportionment, where it may be carried by small shifts in
concentration ratios, or by inter-species correlations. The real data needs of such uses can be
tricky to express in terms of the accuracy and precision of individual measurements. Of course,
every researcher you ask will just say "more is better!" And you will have to disappoint them.
What you learn by asking, though, is what kind of "more" they want; which aspects of
measurement quality are critical in their contemplated applications.
Question 3: EPA is considering converting all of the STN speciation sites to IMPROVE
samplers and IMPROVE laboratory and sample handling protocols. What are strengths
and weaknesses of this approach?
"If you want an exact answer, make just one measurement." That's the strength and weakness of
this approach. The core problem is that particulate matter has so many degrees of freedom. EPA
can simply define one particular measurement as truth, as it did in the PM2.5 FRM, but this
doesn't mean that important information won't later be found in the rejected alternatives, as the
existence of the speciation samplers acknowledges.
Any benefit of preserving dual networks would come from understanding their differences, and
that would require extensive overlap (collocated sampling), laboratory experimentation, and data
analysis. You have to sort out how often-subtle differences in sampling and analysis interact in
practice with the varieties of aerosols actually found in different environments. And this has to
be a sustained effort, if you are interested in trends over time; materials and practices inevitably
evolve through the years, and the differences can be expected to compound. I think there are
better uses for the Agency's limited resources. And absent the considerable resources to do it
right, running dual networks will yield nothing but problems; you will always be wondering
whether your differences arise in the atmosphere or the measurements.
B-73
-------
So I think the strengths of moving soon to the IMPROVE model substantially outweigh the
weaknesses. With that bottom line, here are some thoughts on specific strengths and
weaknesses.
Strengths of IMPROVE sampler: converting would clearly bring more consistency to
sampling, not just between the networks but also within STN. Moreover, the IMPROVE sampler
is field-proven and operator-friendly.
Weakness of IMPROVE sampler: it employs a critical orifice rather than active flow
control. While flows are measured accurately, they are allowed to depart from the nominal rate
that yields a 2.5 |j,m cut-point in aerodynamic particle diameter. Average cut-points lie outside
the 2-3 |j,m range in about 1 of 10 samples.
Strengths of IMPROVE protocols: there is no good reason for maintaining two non-
comparable systems for analyzing and reporting carbon measurements when both use the same
basic principle, thermal analysis. DRI (the carbon laboratory used by IMPROVE) has invested a
lot of research into the underlying method, and I trust their fundamental understanding of it.
Point of clarification: Would "a/7 of the STN speciation sites" include those oriented to SIPs
rather than trends? If it includes SIP sites, what sort of data quality objectives does the Agency
have for such trace XRF elements as Lanthanum, Yttrium, Cerium, Samarium, Niobium,
Europium, Terbium, Hafnium, Tantalum, Gallium, Indium, Tungsten, Iridium, Scandium,
Antimony, Gold, Cesium, Mercury, and Barium, all of which are currently reported by STN* but
not by IMPROVE? That is, how will it specify data quality for elements that are rarely/barely
detectable?
* Standard Operating Procedures for PM2.5 XRF Analysis, Revision 2, 1/20/2004.
Question 4: Is it scientifically acceptable to generate air quality surfaces through modeled
observations and/or integrated predictive/observational fields that would be of appropriate
uncertainty for use in the regulatory decision-making process?
Given that both satisfy the relevant data quality objectives, I don't see any scientifically
meaningful difference between modeled and measured data. I can't believe I said that!
B-74
-------
Dr. Yousheng Zeng
December 15, 2004 AAMM Subcommittee Advisory Meeting
Comments on the EPA NAAMS Implementation Plan
By Yousheng Zeng (AAMM Subcommittee Member)
Initial comments: December 14, 2004
Revised: December 20, 2004
Response to Charge Question 1:
Based on the NCore Level 2 parameter list in Table 4-1, if these PM precursor gas
measurements are not required for NCore Level 2 sites, it will leave only PM2.5, PM2.5
speciation, PMc, and ozone. This will almost downgrade Level 2 to Level 3.
Considering the usefulness of these measurements co-located with other Level 2
instruments and only 75 Level 2 sites nationally, I certainly consider it appropriate to
include these gaseous PM precursors at Level 2 sites. This seems barely adequate, if at
all.
By comparing Table 4-1 and Table 4-3, there seem to be some mismatch. Table 4-1 does
not include air toxic measurements as core parameters for Level 2 sites. However,
according to Table 4-3, air toxic trends are one of the objectives for Level 2 sites. How
can air toxic trends be assessed through Level 2 monitor network if air toxics are not
measured at Level 2 sites? How the measurements for air toxic trends be implemented?
Response to Charge Question 2:
My comment for this question is about presentation for communication purposes. I feel
that the representation of the big picture of the nation's ambient air quality monitoring
systems (before and after NCore) is still somewhat confusing. I think it will be helpful to
describe the landscape of the nation's monitoring systems before and after NCore, similar
to the approach used in the EPA presentation dated July 2003
(http://www.epa.gov/ttn/amtic/files/ambient/pm25/casac/jul03pre.pdf), page 27 (slide 27)
with further clarifications. Specifically, after NCore, what networks will still be in
existence? It may be helpful to readers if the nomenclature in the presentation
corresponds to the names of networks rather than "Core+PM Spec", "Core Spec", etc.
Does everything on the "Future Directions" side of the slide belongs to NCore and there
are other networks that will be coexist but not shown?
Response to Charge Question 3:
It sounds like a good idea.
B-75
-------
Response to Charge Question 4:
The context of this question is about divesting "redundant monitors in adjacent, but
separate, geopolitical areas". I think this approach is valid in principle; but there may be
some practical issues to be worked out. Modeled pollutant concentrations are used in
various regulatory processes from SIP demonstrations to permitting. Particularly in
permitting, modeled concentrations are used heavily as the basis for regulatory decisions
regardless of the geopolitical boundary. Some of these permitting decisions bear
important economical, political, and legal implications. In my opinion, air quality
surfaces generated based on a network of well placed monitors are not only scientifically
acceptable, they are better than using observed data from individual monitor(s) to
represent the air quality of the geopolitical area in which the monitor(s) are located.
There may be some practical issues to consider:
• How close should the monitors be to each other to be considered "adjacent" and
therefore a candidate for divestment? EPA may establish some general guidelines to
address this issue. This is also a function of uncertainty and reliability/confidence
level of the model to be used in place of the actual monitors.
• If a monitor is to be divested, which side of the boundary should maintain the other
monitor? How will that affect funding? Should the divesting side still contribute
some financially if they rely on the other side to provide the data for their regulatory
needs?
Additional General Comments:
Data Integration
There are significant ambient monitoring activities outside of the networks discussed in
the Strategy. They include special studies funded by State or local agencies and ambient
air monitoring programs funded by private industries as part of Beneficial Environmental
Projects (BEP). For example, the Louisiana DEQ recently ordered sixteen facilities in
the Baton Rouge ozone non-attainment area to monitor Total Non-Methane Organic
Compounds (TNMOC) and speciated ozone precursors. The initial order requires each of
the sixteen facilities to install four monitors, i.e., 64 monitors. Although the final number
of monitors may be less as a result of on-going negotiations, it will still represent a
significant increase in spatial resolution in the area and the monitoring will go on for at
least multiple years.
The data collected by these monitoring activities can be a very good complement to the
NCore and other monitoring networks covered by the Strategy. The data from many of
these monitoring programs are not captured in the EPA monitoring data system. The
EPA should consider developing some mechanism to bring the monitoring data into the
system. This will be a very cost effective way to improve the spatial and temporal
coverage. The quality control of these monitoring systems may or may not be as rigorous
as NCore network. This concern can be addressed by having code flags to indicate data
B-76
-------
quality confidence level (e.g., a code for NCore, codes for certain types of privately
funded monitors). In a data analysis, having a mixture of high and low quality data with
high density spatial coverage is often better than having accurate but sparsely populated
data.
Funding for Level 1 Sites
I agree with many AAMM Subcommittee members concerning the importance and
funding of Level 1 sites. Reducing PAMS sites in some east and west coastal areas (but
not in the Gulf Mexico coastal region where ozone problems require high spatial
resolution PAMS data) to fund Level 1 sites seems to be a good idea. There may be
another funding source (although I don't know how to make it work at this time). Based
on my observations and experience, some monitoring programs required as BEP are not
well thought-out. Sometimes the pollutants selected for monitoring are not critical.
Companies implement the monitoring program almost just for the sake of spending
money required by BEP elements of the settlement without too much concern as what
will be monitored. As a result, the monitoring data have very low value. It will be
interesting to investigate if there is an administrative mechanism to channel and
aggregate this type of BEP funds to support Level sites.
Atmospheric Processes That Were Previously Masked by High Pollutant Concentrations
Another comment I have is also related to spatial resolution. As the ozone situation
improves in some industrialized regions of the country, some previously less known
factors that have impact in a small area become more salient (i.e., as the overall ozone
level drops, some previously "buried" peaks will become visible. The ozone formation
due to highly reactive VOC (HRVOC) in Houston and Baton Rouge in recent years is an
example of such a condition. My concern is that the network spatial resolution may not
be high enough if we don't take these factors into the consideration.
B-77
-------
NOTICE
This report has been written as part of the activities of the Environmental
Protection Agency's (EPA) Clean Air Scientific Advisory Committee (CASAC), a
Federal advisory committee administratively located under the EPA Science Advisory
Board Staff that is chartered to provide extramural scientific information and advice to
the Administrator and other officials of the EPA. The CAS AC is structured to provide
balanced, expert assessment of scientific matters related to issue and problems facing the
Agency. This report has not been reviewed for approval by the Agency and, hence, the
contents of this report do not necessarily represent the views and policies of the EPA, nor
of other agencies in the Executive Branch of the Federal government, nor does mention
of trade names or commercial products constitute a recommendation for use. CAS AC
reports are posted on the SAB Web site at: http://www.epa.gov/sab.
B-78
------- |