# £* %

U3EZI

X PRO^&
-------
Enclosure

EPA Staff Response to the Clean Air Science Advisory Committee (CASAC)
Technical Subcommittee on Particle Monitoring

This letter provides a response to the CASAC letter of March 1, 2002 (EPA-SAB-
CASAC-LTR-02-001), which reviewed EPA's draft Continuous Monitoring Implementation
Plan. The review was based on a meeting with EPA in January 2002 at which EPA presented
details of the plan. The meeting was open to the public, and other presenters also had an
opportunity to provide input to CASAC.

The CASAC letter stated: "After discussion between the Subcommittee, the Office of
Air Quality Planning and Standards (OAQPS) staff, and others present at the meeting, we
(CASAC's Technical Subcommittee on Particle Monitoring) agreed that the document presented
a reasonable framework for the use of continuous monitors." In addition, there were several
comments and recommendations made in the letter from CASAC. The next iteration of the
continuous monitoring implementation plan (Revision 2) has been edited to address each of
these comments or recommendations. A summary of each comment or recommendation and a
response is provided below:

Responses to Comments and Recommendations:

a.	The Subcommittee is concerned that the process, as currently outlined, puts a
heavy burden on State or local air quality agencies to demonstrate the Regional
Equivalent Method (REM). This effort may be insurmountable for many such
organizations. Normally, the burden for equivalency demonstration is the
responsibility of equipment manufacturers, and the REM approach shifts that
effort to the state or local monitoring agency.

We will encourage collaboration between equipment manufacturers and
monitoring agencies; however, we recognize that much of the burden for this
approach will fall on the state or local monitoring agencies.

b.	There are other key issues regarding the regional approach that are difficult to
resolve, such as the definition of appropriate regional domains, and the potential
for atmospheric changes of such domains over time.

Since defining an appropriate regional domain may be impractical due to the
dynamics of changing aerosol concentration and composition, a simplified
approach to defining a regional domain will be made. Selection of regional
domains will be made based upon a simplified approach of starting with a state
network. If necessary, a sub-set of a state network may be approved where there
are areas of acceptable performance of the continuous monitor, or multiple states
may seek approval of a method in a coordinated effort.

-------
c.	The REM process only permits a simple model to convert the continuous monitor
data into values that would be defined as equivalent to the Federal Reference
Method (FRM) values. The model must be of the form: FRM = a * CM + b,
where FRM is the estimated FRM 24-hour mass concentration in |ig/m3, CM is
the measured 24-hour mass concentration (|ig/m3) estimated from the 24 one-hour
values, and a and b are empirical coefficients developed from measurements
within a region. The Subcommittee expressed the hope that it may be possible to
develop a similar semi-empirical model, based on physical/chemical principles,
that would provide adequate fits to the FRM data.

The continuous monitoring implementation plan will be edited to include semi-
empirical models to allow for multiple physical/chemical parameters, so long as
there is an appropriate system in place for assuring the quality of these additional
parameters. For instance, ambient temperature is readily available on most
continuous monitors as it is usedfor active flow control of the flow rate system.
Since ambient temperature is already a part of most continuous monitoring
methods, it is checked on a regular basis against a known independent standard.
Other considerations have also been made such as inclusion of Julian date or
utilizing two models with a breakpoint at a specific temperature.

d.	The Subcommittee expressed the opinion that the current requirements for the
REM designation are inadequate, in that they do not require a sufficient
correlation between the FRM and CM data. The subcommittee recommended
that, in addition to the requirements set out in the draft document, there be a
requirement that the squared correlation coefficient (r2) should be relatively high
with the value set by an appropriate Data Quality Objectives (DQO) process.

Initial identification of an appropriate correlation coefficient has been made. A
statistical analysis was performed which tests if true underlying correlations for
each category ofPM2 5 continuous monitors are at or below a certain level. For a
Correlated Acceptable Continuous (CAC) monitor category, a true underlying
correlation less than or equal to the square root of 0.7 was chosen. For a REM, a
true underlying correlation less than or equal to the square root of 0.8 was
chosen. These true underlying correlations are based upon an iterative process
of evaluating the error rates and sample sizes neededfor reasonable observed
correlations. Sample sizes were tied to sample frequencies of collocated FRM's,
which are typically l-in-6-day and l-in-3-day. Based upon this analysis, the
minimum observed correlation necessary for a CAC monitor was determined to
be 0.9 (squared correlation of 0.81) with at least 44 sample pairs spread over a
year. For a REM, a minimum observed correlation of 0.93 (squared correlation
of 0.87) with at least 96 sample pairs spread over a year was determined.

-------
The issue of monitoring for attainment of the 24-hour standard has not been
addressed. Although the Subcommittee understands that the annual average
standard will be the controlling standard, it is necessary to ascertain attainment or
nonattainment of the 24-hour standard.

EPA has developed a software tool that helps to address decisions for the 24-hour
standard. This software tool uses a series of power curves to help monitoring
organizations determine the potential for decisions errors at PM25 monitoring
sites. The power curves for a 98th percentile decision over three years
demonstrates that higher sampling frequencies result in reduced decisions errors.
This is especially important for potential use ofPM continuous monitors in that
these monitors will be delivering data each day. Increasing measurement
precision has little effect on decision errors for the 24-hour standard.

Because of some of the difficulties anticipated in the implementation of the REM
approach, the Subcommittee suggested an interim approach in order to begin
moving continuous monitors into the network while the details of the REM
process are being developed.

EPA has edited the continuous monitoring implementation plan to include a site
REM that would serve as the interim approach identified by CASAC when used
for National Ambient Air Quality Standards determinations. This approach
would allow for sample frequency relief and each year, if needed, a best-fit
function would be reset in a prospective mode. The continuous data could be
used for comparison to both the annual and the 24-hour standard. If the
continuous data were to be invalidated at a specific site for not meeting pre-
determined performance criteria, the FRM data would be available for
attainment decisions.

The Subcommittee recommends that EPA undertake a thorough DQO process to
determine the needs for monitors so that Federal Equivalent Method requirements
can be defined based on a clearly defined set of data quality needs.

EPA is planning to undertake this DQO exercise later this summer. Details of the
results are expected later this year.

-------