\
          UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                           WASHINGTON B.C. 20460
                                                           OFFICE OF THE ADMINISTRATOR
                                                             SCIENCE ADVISORY BOARD
                                  April 20, 2010

EPA-C AS AC-10-009

The Honorable Lisa P. Jackson
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, N.W.
Washington, D.C. 20460

       Subject: CAS AC Review ofP'articulate Matter Urban-Focused Visibility
               Assessment - Second External Review Draft (January 2010)

Dear Administrator Jackson,

       The Clean Air Scientific Advisory Committee (CASAC) Particulate Matter (PM)
Review Panel met on March 10-11, 2010 to review the Quantitative Health Risk
Assessment for Particulate Matter - Second External Review Draft (February 2010) and
Particulate Matter Urban-Focused Visibility Assessment - Second External Review Draft
(January 2010). In this letter, CASAC offers general comments on the Particulate
Matter Urban-Focused Visibility Assessment,  followed by our consensus responses to the
Agency's charge questions.  Comments from individual panelists are also enclosed.

       CASAC was favorably impressed with the quality and clarity of the second draft
Particulate Matter Urban-Focused Visibility Assessment (UFVA). As we indicated in our
letter of November 24, 2009 regarding the first draft of the UFVA, CASAC strongly
supports the introduction of a new PM light extinction indicator for a secondary standard
to protect against adverse effects on visibility. We concur with an hourly averaging time
as the most appropriate to represent the nearly instantaneous nature of human perception
of impaired visibility. We also agree with the 20 to 30 deciview (dv) range of light
extinction levels as a reasonable representation of the levels of visibility impairment
considered acceptable based on currently available studies.

       The UFVA presents logical and persuasive arguments for use of a PM light
extinction indicator, with a 1-hour (daylight) averaging time as a sound basis for a
secondary PM standard to protect urban visibility. The UFVA also presents and evaluates
reasonable ranges for the level (20 to 30 deciviews) and form (90th to 98th percentile)
that such a standard might take, and clearly illustrates the differences among the many
resulting optional combinations of levels and forms in terms of protectiveness and/or
stringency across the 15 diverse urban study areas selected for this analysis. The
approaches employed here, the ranges of options considered, and the estimates of visual

-------
air quality resulting from the various combinations of options considered should provide
a sound basis for more detailed consideration in the Policy Assessment.

       In addition, if the Agency is also considering options of using a PM2.5 light
extinction indicator or a PM2.5 mass indicator as the basis for secondary PM standard, it
would be important to see a more detailed comparison of the affected PM2 5
concentrations, species compositions and levels of visibility protection that would result
from the use of alternative PMi0 light extinction, PM2 5 light extinction or PM2 5 mass
indicators.

       In its second draft UFVA, EPA staff were responsive to previous CAS AC
recommendations, adding a detailed logit analysis to better understand similarities and
differences in results of the urban visibility preference studies,  recalculating estimated
extinction levels using a 90% relative humidity (RH) screen, and considering (90th, 95th
and 98th) percentiles based on all available daylight hours as well as on the single worst
daylight hour in each day.

       CAS AC recommends the addition of an integrative summary chapter to the
UFVA.  At present, it lacks a summary that brings main findings and their implications
together. This chapter might be included as an executive summary or as new chapter at
its conclusion.  CASAC has also identified needs for the next review cycle in terms of
further research on a number of topics related to urban visibility; these needs should be
assembled, highlighted and prioritized. In particular, there is a need for the Agency to
conduct additional urban visibility preference studies over a broad range of urban areas
and viewing conditions, to further evaluate and refine the range of visibility levels
considered to be acceptable in the  current assessment.

       We thank the Agency for the opportunity to provide advice on the UFVA, and
look forward to continued discussions with the Agency as we review the first draft Policy
Assessment for the Review of the Paniculate Matter National Ambient Air Quality
Standards (March 2010).
                                  Sincerely,

                                         /Signed/

                                  Dr. Jonathan M. Samet
                                  Chair
                                  Clean Air Scientific Advisory Committee
Enclosures A: CASAC Particulate Matter Review Panel Roster
           B: CASAC Responses to Charge Questions
           C: Individual Panelists' Responses to Charge Questions

-------
Enclosure A
                     Clean Air Scientific Advisory Committee
                         Particulate Matter Review Panel
CHAIR
Dr. Jonathan M. Samet, Professor and Chair, Department of Preventive Medicine,
University of Southern California, Los Angeles, CA

CASAC MEMBERS
Dr. Joseph Brain, Philip Drinker Professor of Environmental Physiology, Department of
Environmental Health, Harvard School of Public Health, Harvard University, Boston,
MA

Dr. H. Christopher Frey, Professor, Department of Civil, Construction and Environmental
Engineering, College of Engineering, North Carolina State University, Raleigh, NC

Dr. Donna Kenski, Data Analysis Director, Lake Michigan Air Directors Consortium,
Rosemont, IL

Dr. Armistead (Ted) Russell, Professor, Department of Civil and Environmental
Engineering, Georgia Institute of Technology, Atlanta, GA

Dr. Helen Suh, Associate Professor, Harvard School of Public Health, Harvard
University, Boston, MA

Dr. Kathy Weathers, Senior Scientist,  Gary Institute of Ecosystem Studies, Millbrook,
NY
CONSULTANTS
Dr. Lowell Ashbaugh, Associate Research Ecologist, Crocker Nuclear Lab, University of
California, Davis, Davis, CA

Prof. Ed Avol, Professor, Preventive Medicine, Keck School of Medicine, University of
Southern California, Los Angeles, CA

Dr. Wayne Cascio, Professor, Medicine, Cardiology, Brody School of Medicine at East
Carolina University, Greenville, NC

Dr. David Grantz, Director, Botany  and Plant Sciences and Air Pollution Research
Center, Riverside Campus and Kearney Agricultural Center, University of California,
Parlier, CA

-------
Dr. Joseph Helble, Dean and Professor, Thayer School of Engineering, Dartmouth
College, Hanover, NH

Dr. Rogene Henderson, Senior Scientist Emeritus, Lovelace Respiratory Research
Institute, Albuquerque, NM

Dr. Philip Hopke, Bayard D. Clarkson Distinguished Professor, Department of Chemical
Engineering, Clarkson University, Potsdam, NY

Dr. Morton Lippmann, Professor, Nelson Institute of Environmental Medicine, New
York University School of Medicine, Tuxedo, NY

Dr. Helen Suh Macintosh, Associate Professor, Environmental Health, School of Public
Health, Harvard University, Boston, MA

Dr. William Malm, Research Physicist, National Park Service Air Resources Division,
Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort
Collins, CO

Mr. Charles Thomas (Tom) Moore, Jr., Air Quality Program Manager, Western
Governors' Association, Cooperative Institute for Research in the Atmosphere, Colorado
State University, Fort Collins, CO

Dr. Robert F.  Phalen, Professor, Department of Community & Environmental Medicine;
Director, Air Pollution Health Effects Laboratory; Professor of Occupational &
Environmental Health, Center for Occupation & Environment Health, College of
Medicine, University of California Irvine, Irvine, CA

Dr. Kent Pinkerton, Professor, Regents of the University of California, Center for Health
and the Environment, University of California, Davis, CA

Mr. Richard L. Poirot, Environmental Analyst, Air Pollution Control Division,
Department of Environmental Conservation, Vermont Agency of Natural Resources,
Waterbury, VT

Dr. Frank Speizer, Edward Kass Professor of Medicine, Channing Laboratory, Harvard
Medical School, Boston, MA

Dr. Sverre Vedal, Professor, Department of Environmental and Occupational Health
Sciences, School of Public Health and Community Medicine, University of Washington,
Seattle, WA

SCIENCE ADVISORY BOARD STAFF
Dr. Holly Stallworth, Designated Federal Officer, EPA Science Advisory Board Staff
Office

-------
                                    NOTICE
This report has been written as part of the activities of the EPA's Clean Air Scientific
Advisory Committee (CASAC), a federal advisory committee independently chartered to
provide extramural scientific information and advice to the Administrator and other
officials of the EPA. CASAC provides balanced, expert assessment of scientific matters
related to issues 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 within the Executive
Branch of the federal government. In addition, any mention of trade names or
commercial products does not constitute a recommendation for use. CASAC reports are
posted on the EPA Web site at:  http://www.epa.gov/casac.

-------
Enclosure B

                     CASAC Responses to Charge Questions

1) In general, what are the Panel's views on the level of detail provided in the body
of the report and associated appendices? Does the Panel agree that all of this
information is useful to retain or is there material that the Panel would recommend
deleting? Does the Panel have any suggestions regarding the organization and
distribution of information throughout the document and in the appendices?

While a substantial level of detail has been provided in the UFVA, many of the
calculations included therein are complex, and a wide range of optional  combinations of
levels and forms is presented and compared.  The report is long, but its completeness and
detail are useful, particularly given that we are looking, for the first time, at a very
different indicator for this secondary standard. The discussion is clear, and the associated
appendices are valuable for examining questions raised by the main text. The CASAC
would not recommend deleting any of the  material. With the few exceptions noted below,
the organization is good as it stands, both in the main body of the report and in the
appendices.

Chapter 2 may contain more information than is required with respect to the individual
cities. Much of this material could be moved to the appendices. The memo of 2/3/2010
on the statistical analysis of the preference studies should also be placed in the
appendices, and consolidated with related  material from Chapter 2 to avoid redundancy.
Similarly, much of the material in sections 4.1.3 and 4.1.4 might be better suited to an
appendix. In any case, the transition into future monitoring site characteristics (middle of
page 4-2) is rather abrupt,  and might be more appropriate for inclusion in the Policy
Assessment, where it could be expanded based on recent feedback from the CASAC
Ambient Air Monitoring and Methods Subcommittee.

The SANDWICH method described in Chapter 3 does not include sea salt, which can be
an important contributor to PM at coastal sites. The sea salt component of PM may be a
source of uncertainty in the light extinction calculation at these sites. The SANDWICH
approach has other sources of uncertainty that should be mentioned. Relying on mass
closure to determine the hourly organic carbon assumes that all other species are
accurately determined. This is problematic for nitrate, as well as for sea salt. These
factors should be mentioned as areas of uncertainty that are associated with this method.

2) In the Panel's view, to what extent does the logit analysis presented in Chapter 2
of the second draft Visibility Assessment, and further expanded and described in the
supplemental memorandum, add value  to the urban visibility preference study
analysis and provide additional support for combining and comparing the results
from the four cities, as shown in Figure  2-16? What are the Panel's views on the
clarity and adequacy of the descriptions associated with such a combined
assessment and on the conclusions that can be drawn from the assessments? Please

-------
provide comments on any additional insights that might be drawn from these
analyses or on any additional caveats that should be considered.

The analysis presented in Section 2.6 and summarized in Figure 2-16 is quite reasonable.
The characteristics of the studies that produced the data are discussed. The modeling
approach is clearly presented and explained, with an appropriate level of detail, given that
this approach has not been previously described in EPA visibility documents. Figure 2-16
is very useful for comparing the four realizations of the model in comparison to the study
data.  Table 2-3 provides valuable information regarding  the statistical significance of
each model coefficient.  The results in Table 2-3 are appropriately explained.  The model
is appropriately critiqued in terms of goodness-of-fit, statistical significance of the
coefficients, and the interpretation of the coefficients. The limitations of the model with
respect to limitations of the underlying data are appropriately discussed.

The text does a good job of identifying limitations associated with the sample population
in the UBC study (all participants were university students and therefore had advanced
educational background), and discusses the broader socioeconomic distribution of the
population  participating in the Phoenix study.  Absent is a comparable discussion of the
demographic distribution in the second Washington DC study, other than noting the
relatively small size of the sample population, and limitation that all participants were
employees  of the research firm conducting the study.  It is therefore likely that all
participants were well educated, and may have understood the methods and goals of the
study better than the populations participating in the other studies.  Since this is a
preference  study, this point is worth acknowledging in the text.

The additional information presented in the supporting memo by Stratus Consulting is
very helpful and should be retained as an attachment or in an  appendix. The memo
evaluates several alternative specifications of a logit model, of which the one presented in
the assessment is Model 2. Confidence intervals on the predicted mean were estimated
using a numerical method, and verified with another numerical method. Models 1 and 2
are generally found to provide similar results.  Thus, the results of the assessment are not
substantially sensitive to the choice in functional form between these two models.
Overall, the use of the logit model is reasonable, since the output is a binary variable.
The model  is appropriately evaluated and reasonably represents the data to which it was
fit.  The limitations  of the underlying data are discussed qualitatively. The overall
findings are reasonably supported by the data, model, and appropriately take into account
limitations  of the data. The associated discussion presents a number of reasonable
hypotheses that could account for the significant differences in the acceptability response
curves among the different cities, and could explain why  some of the study results have
greater variability than others.

While it is appropriate to move forward on the basis of available data, CASAC believes
the data suggest some heterogeneity that needs to be explored with further research,
especially to evaluate potential differences among regions or  among cities that have
different inherent scenic characteristics. The need for additional urban visibility
preference  studies, using consistent and robust survey methods, over  a broader range of

-------
urban areas and viewing conditions, should be emphasized as a research priority. The
results from these studies could help refine a secondary PM light extinction standard in
future NAAQS review cycles.

3) What are the Panel's views on the extent to which the analysis of the frequency of
co-occurrences of hourly relative humidity values below and above 90 percent with
other meteorological events such as rain or fog (Chapter 3, section 3.3.5; Table 3-6)
provides scientific support for consideration of how to address relative humidity in
defining the form of a standard based on a PM light extinction indicator?

An informative, new analysis was presented in section 3.3.5 (as revised by the 3/4/10
memo from Philip Lorang and Mark Schmidt to the PM NAAQS Review Docket)
comparing hourly daytime relative humidity (RH) and the incidence of natural weather
visibility-impairing conditions (fog, rain, snow, etc.).  CASAC concurred that this
presentation provided a persuasive argument for the use of a 90% RH screen as an
effective and efficient way to discard most of the hours potentially affected by natural
weather conditions while removing only a small fraction of total daylight hours from
consideration.

CASAC also found that a 90% RH screen was not only a logical revision for use in the
UFVA assessment, but would also be an appropriate component of a new secondary
NAAQS based on PM light extinction. It would be an effective way to remove periods
when weather influences might be the dominant cause of visibility impairment from the
regulatory metric, and would also have other benefits relating to PM light extinction
measurements, such as accommodating spatial variability in RH and weather influences,
minimizing effects of RH measurement errors at high humidity, and allowing for
measurement modifications (such as a "smart heater" on a nephelometer) that could
substantially reduce instrument maintenance needs and improve data quality.

4) In response to CASAC recommendations, descriptions of current conditions and
results of just meeting NAAQS scenarios that considered all daylight hours were
added to those based on maximum daily 1-hour indicators. The 98th percentile form
was also included along with the 90th and 95th percentiles. Tile plots of hourly PM
light extinction (Figure 3-12)  and composition bar graphs  of the top 10% of days for
maximum daily 1-hour and aggregated individual  daylight hours (Figure 3-13) were
shown in part to help illuminate the similarities and differences between these
various indicators with respect to current conditions. Similarly, additional figures
and table entries were generated to  illustrate the characteristics of various PM light
extinction NAAQS scenario forms (Tables 4-2, 4-3, 4-5, and 4-7; Figures 4-1 through
4-3).

What are the Panel's views regarding EPA staff interpretations of these displays
included in the text? Are there supplemental or alternative interpretations the Panel
would suggest? Are there additional approaches the Panel would suggest regarding
ways to summarize, display, or assess the results of these analyses, including
similarities and differences between the various scenarios?

-------
The revisions to accommodate the 90% RH screen, the inclusion of a higher (98th)
percentile form, and consideration of an alternative way of calculating percentiles - based
on all daylight hours, rather than just the worst hour in a day - were valuable additions to
the UFVA. When these are combined with other options including 90th and 95th
percentiles and different levels (64, 112, 191 Mm"1) of PM light extinction, the many
resulting options are complex and difficult to communicate and compare. The various
new graphical displays are quite helpful in communicating this complex new information
clearly and illustrating the differences and similarities among the many options.

CASAC found that the tile plots were especially informative and greatly helped to show
the seasonal and diurnal patterns and similarities and differences across the different
study areas. The stacked bar charts of maximum hourly extinction on top 10% days vs.
compositions on top percentiles considering all daylight hours are useful for seeing
differences between cities, and the accompanying interpretation and discussion are
reasonable and informative. The plots also illustrate interesting differences in hourly
compositions within individual cities for the (relatively few) hours which represent the
single worst hours in a day. However, these comparisons  do not really allow for a direct
"apples vs. apples" comparison of the effect of calculating percentiles based on the single
worst daylight hour each day, vs. using all daylight hours to calculate the percentiles.

It seems likely that the two optional  approaches  may tend  to focus on different kinds of
visibility impairing PM species, sources, meteorological conditions, times of day, and/or
seasons. It would be useful to have a clearer picture of what their similarities and
differences really  are before recommending one approach  over the other.  Additional
suggestions on approaches for this are included in comment from individual panelists.
Several panelists also expressed concerns that some of the occasionally very high coarse
particle contributions in a few of the cities (St. Louis and Los Angeles) appeared
questionable, and might benefit from closer scrutiny  or warrant some cautionary language
in the UFVA.

-------
Enclosure C
           Compendium of Individual Preliminary Panelists Comments
     Urban-Focused Visibility Assessment for Particulate Matter (January 2010)
Comments from Dr. Lowell Ashbaugh	2
Comments from Dr. David Grantz	8
Comments from Dr. Philip Hopke	10
Comments from Mr. Rich Poirot	12

-------
Comments from Dr. Lowell Ashbaugh
January 2010 2nd External Review Draft Particulate Matter Urban-Focused Visibility
Assessment,
Charge question 1
       1)    In general, what are the Panel's views on the level of detail provided in the body
             of the report and associated appendices ? Does the Panel agree that all of this
             information  is useful to retain or is there material that the Panel would
             recommend deleting? Does the Panel have any suggestions regarding the
             organization and distribution of information throughout the document and in the
             appendices?
The report contains a lot  of detail, but it is presented well. The discussion is clear and the
associated appendices are valuable for examining questions raised by the main text. I
would not recommend deleting any of the material. The organization is good, both in the
main body of the report and in the appendices. I have some comments on the presentation
of material that I'll put later in this review.
The introductory material is excellent. It's short, but it's complete and it lays the
foundation for review of  the NAAQS. The overview of the science of visibility and PM
might be too brief for readers new to the field, but the references are sufficient to fill in
the  details.
 Charge question 2
       2)    In the Panel's view, to what extent does the logit analysis presented in Chapter 2
             of the second draft Visibility Assessment and further expanded and described in
             the supplemental memorandum add value to the urban visibility preference study
             analysis and provide additional support for combining and comparing the results
            from the four cities, as shown in Figure 2-16? What are the Panel's views on the
             clarity and adequacy of the descriptions associated with such  a combined
             assessment and on the conclusions that can be drawn from the assessments?
             Please provide comments on any additional insights that might be drawn from
             these analyses or on any additional caveats that should be considered.

The logit analysis is important to this assessment. It provides a common basis to analyze
the  results from the four studies in different cities and it sets forth a quantitative method
to establish boundaries on the CPL levels. The discussion of the hypotheses that might
explain the different VAQ acceptability responses is clear. There may be other factors
involved in the differences between  cities, but it would take more research to confirm or
refute them. For example, one hypothesis that was not discussed is the  possible
differences in population that inhabits the different cities that makes them more or less
sensitive to VAQ. Denver and BC are known for their outdoors activities; people who
live there may be more attuned to the visual environment and more sensitive to poor
VAQ. People living in Washington,  DC may be less attuned to the outdoor environment
and therefore less sensitive to poor VAQ.
Page 2-31 suggests that additional studies would be useful to gain further understanding.
Are there plans for such studies? Is there an outline or more comprehensive listing of the

-------
studies that would be useful? Such a list or discussion would be very useful if funding
becomes available at some time in the future.
Charge question 3
       3)     What are the Panel's views on the extent to which the analysis of the frequency of
             co- occurrences of hourly relative humidity values below and above 90 percent
             with other meteorological events such as rain or fog (Chapter 3, section 3.3.5;
             Table 3-6) provides scientific support for consideration of how to address relative
             humidity in defining the form of a standard based on a PM light extinction
             indicator?
       The analysis of the frequency of co-occurrences of rain and/or fog with relative
       humidity above or below 90 percent has a flaw in it. I don't believe it would
       change the conclusions, but on page 3-29, lines 11-12, is the statement "Based on
       this assessment, the 90% relative humidity cutoff criteria is effective in that on
       average less than 6% of the hours are removed from consideration..." This is not
       correct.  It's correct that (except for Tacoma) less than 6% of the hours with
       RN<90% co-occur with rain or fog. The number of daylight hours with relative
       humidity >90% and their percentage of all  daylight hours is shown below (data
       extracted from Table 3-6). On average,  about 10% of all hours in all study areas
       would be removed with the RH>90% criterion.  This percentage varies from city
       to city, with a low of 1% at Phoenix and a high  of 30% at Tacoma. Birmingham,
       Atlanta, Pittsburgh, Baltimore, and Philadelphia are also above 6%.
Study area
Tacoma
Fresno
Los Angeles
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
Overall total
Number of
Daylight
hours with
RH<=90%
18293
24245

26045
24989
25519


23826
23696

22254
22867
24302
24963
260999
Number of
Daylight
hours with
RH>90%
7987
1615

235
1291
761


2454
2584

4026
3413
1978
1314
27658
Percent of
Daylight
Hours with
RH>90%
30%
6%

1%
5%
3%


9%
10%

15%
13%
8%
5%
10%
       I can't reconcile the data in this table with Table G-l in Appendix G, but the
       statement on page 3-29 is also not consistent with Table G-l. Whatever the cause

-------
       of the inconsistency, it should be rectified so that the statements and tables are
       consistent. If I've misunderstood the presentation, it should be clarified.
       It's difficult to see the contrast in the distributions shown in Figure G-2. It's
       probably not important to see the difference (i.e. the distribution of extinction in
       the excluded hours), so I recommend the text be revised to bring the reader's
       attention to the important aspects of the plot. In particular, it appears that the
       horizontal bar on the display for each city shows the 90th percentile extinction
       design value, but it isn't mentioned in the text or in the table. It would be useful to
       highlight how the design value drops with the excluded hours.
Charge question 4
       4)    In response to CASAC recommendations, descriptions of current conditions and
             results of just meeting NAAQS scenarios that considered all daylight hours were
             added to those based on maximum daily 1-hour indicators. The 98th percentile
             form was also included along with the 90th and 95th percentiles. Tile plots of
             hourly PM light extinction (Figure 3-12) and composition bar graphs of the top
             10% of days for maximum daily 1-hour and aggregated individual daylight hours
             (Figure 3-13) were shown in part to help illuminate the similarities and differences
             between these various indicators with respect to current conditions. Similarly,
             additional figures and table entries were generated to illustrate the
             characteristics of various PM light extinction  NAAQS scenario forms (Tables 4-2, 4-
             3, 4- 5, and 4-7; Figures 4-1  through 4-3).

             What are the Panel's views regarding EPA staff interpretations of these displays
             included in the text? Are there supplemental  or alternative interpretations the
             Panel would suggest? Are there additional approaches the Panel would suggest
             regarding ways to summarize, display, or assess the results of these analyses,
             including similarities and differences between the various scenarios?
       The tile plots in Figure 3-12 are an excellent way to show the hourly light
       extinction. I would like to see all plots scaled  so the time periods directly
       correspond across the years. In particular, the 9-month plots for Houston and
       Phoenix should be scaled such that the 9 months shown correspond to the same
       period as in the 12-month plots,  i.e. not stretched out to fill the space. All the 12-
       month plots are currently scaled slightly differently so that the same time period is
       not directly opposite for  different years.
       The staff interpretation of the plots is good, but  the conditions at each  city could
       be more fully  explained.  Section 3.4.5 has highlights of the  extinction budgets for
       individual urban areas. I  would like to see something similar for Section 3.4.4.
       For example, the description at Tacoma could read: The hours excluded at
       Tacoma are primarily in the late fall and winter, and are generally in the
       morning. Occasional periods last all day. Few hours are excluded due to high RH
       in spring,  summer, and early fall. High light extinction periods also occur more
       often  in late fall and winter, and with few exceptions occur mostly in the early
       daylight hours in other seasons. Fresno could be described as: The hours excluded
       in Fresno due to high RH are exclusively in the  late fall and winter. This is the fog
       season in  California's Central Valley, and periods of high humidity can last for

-------
days at a time. The highest light extinction occurs during this period, also.
Episodes of high light extinction also occur in spring and summer, sometimes
lasting for many days. Many of these episodes last well into the afternoon. For
Los Angeles, the following description could apply: Most hours excluded due to
high RH at Los Angeles occur during the morning, and are primarily in the
spring, summer, and early fall. High light extinction periods occur at all times of
the year, but are more frequent in the spring and summer. These periods
generally last all day.  Each city has some characteristics that could be described
more fully; this would more completely establish the conditions of high RH and
light extinction in different urban areas of the country.
The extinction budgets shown in Figure 3-13 are very helpful in explaining the
causes of poor VAQ in the different cities. I particularly like the separate displays
of the top 10% of days and the top 10% of individual daylight hours. The staff
interpretation of these plots is very good.
I question the high contribution to light extinction at Los Angeles from coarse PM
during two of the top 10% of days. Having done considerable data validation of
IMPROVE data, I can attest to the possibility of invalid values entering the
database. I also know that hourly concentrations can be quite high under the right
conditions, but such extreme values warrant further investigation of their validity.
The explanation that the coarse PM is determined in part from measurements at
Victorville could well explain the high values, and would make these values
suspect for use at Los Angeles. There is no doubt that coarse PM measured at
Victorville would be unrepresentative of coarse PM at Rubidoux, and especially
in the wider South Coast Air Basin.
I have a few comments relating to particular sections of the assessment, and a few
editorial comments that I'll list separately. First, the comments:
Page
Line
Comment
3-16
        No suggestions -just a comment that this is an excellent description.
3-21
36-27
Sea salt is an important contributor at coastal sites. This may be a source
of uncertainty in the light extinction calculation at these sites.
3-25
31-36
The approach described here has other sources of uncertainty that should
be mentioned. Relying on mass closure to determine the hourly organic
carbon assumes that all other species are accurately determined. This is
problematic for nitrate, as pointed out just a few paragraphs earlier, and
it leaves out sea salt that is an important contributor to mass at coastal
sites. These factors should be mentioned as areas of uncertainty that are
introduced by this method.
3-26
14-15
Setting negative values to zero introduces a bias in the results. There is
inherent uncertainty in both PMio and PM2.5 measurements. During
periods of very low coarse mass, the PMio and PM2.5 measurements are
very close, and their uncertainties may cause the PM2.5 measurement to
be higher than the PMio measurement. It's important to retain negative

-------
                values as a measure of the uncertainty during low coarse mass periods.
3-32,
3-33
I have grave doubts about the outlier concentrations in these plots,
particularly for Los Angeles and St. Louis. I have seen many cases of
invalid high mass measurements of filter while performing data
validation of IMPROVE samples. I have no doubt that there are high
hourly PM concentrations for short periods, but the extreme values
shown, coupled with the caveats mentioned, cast doubt on their validity.
I have the following suggestions for editorial changes in the document.
The Table of Contents is hard to read with the mix of all caps and mixed case font
that is currently used. I suggest making Level 1 all caps, Level 2 small caps or
mixed case, and Level 3 mixed case. This would more effectively set off the
major and minor sections of the document in the Table of Contents. The
appendices are good as they are.
In the list of acronyms/abbreviations, the description for IMPROVE should have
an "s" at the end, i.e. "Interagency Monitoring of Protected Visual
Environments". The description of NARSTO is out of date. It was originally the
North American Research Strategy for Tropospheric Ozone, but is now just
NARSTO. This could be noted in the description. For SMOKE, "Kernel" is
misspelled.
Other editorial suggestions:
Page
1-1
2-11
2-20
2-28
2-30
3-19
3-32
3-33
Line
5
15
8
15
13
1
12-13

Suggestion
Change to ".. .which are to reflect accurately. .."
"Visibility" in the graph heading is misspelled
I suggest using different colors or symbols in Figure 2-11. These are the
same as are used in Figure 2-10, but they have different meanings. It
would be a good idea to review all graphics with this in mind, in
particular those that are adjacent to each other.
Do you mean to say "insignificant" as written, or "significant"? It looks
to me as if this term is significant.
Change "it" to "is"
For some reason this line that ends a sentence appears as a heading in
the Table of Contents. Please correct this.
Do you mean to say section 3.3. 1? I can't find the referenced step in
section 3.4.1.
Figure 3-7 might be better using a logarithmic scale on the y-axis. The
linear scale compresses the bulk of the data to the point that differences
between the cities are not distinguishable. A logarithmic scale would
show the outliers and still allow the lower values to be observable.

-------
Page
3-35
3-36
3-37
3-44
3-44
3-62
3-63
3-63
4-7
4-10
4-11
4-11
4-15,
4-16
App. F
G-l
1-1
Line
16-17
3

31
32
31
10
and
23
25
44
3
15
30


Para.
3
Para.
3
Suggestion
It looks like the references to Figures 3-8b and 3-8a are reversed.
Fresno is listed as a city without a preference for morning high PM light
extinction. The figures in Appendix E indicate otherwise. There are
other cities that show such a preference that aren't listed here.
There is no units label on Figure 3-8. 1 assume it should be Mm" , but it
should be on the graph.
Remove the first "a"
Change "than" to "that"
Change the comma to a semicolon
Remove "of after "as high as"
End the sentence with a period
Change to "... in one or more ..."

Figure 4-2 should be a square plot so the units have equivalent length on
the x- and y-axes.
Table 4-4 says Dallas meets the 15/35 NAAQS, not Salt Lake City.
Table 4-5 should be the next one in sequence.
The scale on Figures 4-3(a) and (b) is very high. Why? It compresses the
bulk of the data and makes it difficult to see the differences between the
cities.
The comments for Figure 4-3 also apply to the figures in Appendix F.
In the second line, change to "... prior to the their exclusion."
I believe the references to "top of figure" and "bottom of figure" should
be interchanged.

-------
Comments from Dr. David Grantz

3-16-10
Urban-Focused Visibility Assessment, Second External Review Draft
Charge Question 1) Level of detail provided in the body of the report and
appendices and suggestions regarding organization of the document.
This review draft of the UFVA appears to be of appropriate length and complexity, with
very few exceptions. Staff is to be commended for preparing an excellent document.
Chapter 2 may contain more information than is required with respect to the individual
cities. Much of this material could be moved to the Appendices. The memo of 2/3/2010
on the statistical analysis of the preference studies could also be placed in the
Appendices, and consolidated with related material from Chapter 2 to avoid redundancy.
Similarly, much of the material in sections 4.1.3 and 4.1.4 might be better suited to an
Appendix. In any case, the transition into future monitoring site characteristics (middle of
page 4 - 2) is rather abrupt.
Charge Question 2) Contribution of the logit analysis.
I remain concerned that despite the abundant statistical analysis of visual preference data
performed since the first external review draft, the amount of data is simply insufficient
to determine what level of air quality/visibility may be acceptable nationwide. The most
important conclusions from these preliminary studies are the closely related concepts that
(1) the range of haziness views presented to panelists is important, as suggested by Smith
and Howell, and (2) that there may be true regional differences in expected air quality,
due to the range of visibility to which residents have become accustomed. This is
considered explicitly at the bottom of page 2 - 30, although it may be premature to
attempt to identify the reasons for these differences. This has significant policy
implications. While the agency has every reason to move forward on this, these results
should be flagged throughout as tentative and subject to further research.
Charge Question 4) All daylight versus maximum daily 1-hour indicators.
The distinction between all daylight hours and maximum daily 1 hour indicators is not
made completely clear in the document. This could be defined more definitively
somewhere early in the text. In general, the abundant consideration of various indicators
and end points is useful and probably unavoidable.
Other editorial points.
The definition  of haziness in deciview units and light extinction in inverse mega meters
on page 1  6 is useful and well presented. However, visual range is used later in the
document (first appearing on page  1-10) but is not defined here.  Explanation of the
utility of each parameter would help the narrative. While visual range may be most
intuitive to many readers, it is dropped on page 1 - 15, in favor of explicit definition of
CPLs in terms  of light extinction. For general readability, explicit interconversion
between the different measures could be carried through the document.
Page 1-13, lines 4-8: The relationship between organic compounds, carbon mass, and
air mass aging needs to be better explained here.
Page 1 - 14, line 35: "indictor" or should be "indicator"
Page 3 - 44, line 32.  "such than average" should be changed to "such that average".

-------
Figures
Maps, Page 3-2. It is disconcerting to find Alaska, Hawaii, and Puerto Rico lumped
into the Gulf of Mexico. There may be room off the Pacific and Atlantic coasts,
respectively, if the legend were placed in the Gulf.
Figure 3-9, Page 3 - 40.  The X axis labels are unclear.

-------
Comments from Dr. Philip Hopke
1)  In general, what are the Panel's views on the level of detail provided in the body of the
    report and associated appendices? Does the Panel agree that all of this information is
    useful to retain or is there material that the Panel would recommend deleting? Does the
    Panel have any suggestions regarding the organization and distribution of information
    throughout the document and in the appendices?

Yes, it is long and somewhat dry, but its completeness and detail are useful particularly given
that we are looking at a very different indicator for this secondary standard. It is not obvious
to me that alternative organization or distribution of the information would be more
informative to the reader.
2)  In the Panel's view, to what extent does the logit analysis presented in Chapter 2 of the
    second draft Visibility Assessment and further expanded and described in the
    supplemental memorandum add value to the urban visibility preference study analysis
    and provide additional support for combining  and comparing the results from the four
    cities, as shown in Figure 2-16? What are the Panel's views on the clarity and adequacy
    of the descriptions associated with such a combined assessment and on the conclusions
    that can be drawn from the assessments? Please provide comments on any additional
    insights that might be drawn from these analyses or on any additional caveats that should
    be considered.

The logit analysis is useful in exploring these data.  One point that is not adequately made
regarding the Washington studies is that the DC photo is significantly different from the
others in the lack of significant scenic features at a distance. There is a brief mention of this
on the top of page 2-31 and only in comparison to Denver. However, all of the other studies
have mountains some tens of miles from the point of view. I suspect that the lack of a distant
object has significantly biased the acceptable  level values upward. In fact the distance to the
background features might well be a key factor in the acceptability ratings. Has there been an
effort to look at the relationship between the 50% acceptability level and the maximum
distance to visible objects in the pictures?  That might be informative.
3)  What are the Panel's views on the extent to which the analysis of the frequency of co-
    occurrences of hourly relative humidity values below and above 90 percent with other
    meteorological events such as rain or fog (Chapter 3, section 3.3.5; Table 3-6) provides
    scientific support for consideration of how to address relative humidity in defining the
    form of a standard based on a PM light extinction indicator?

This section seemed adequate.
4)  In response to CASAC recommendations, descriptions of current conditions and results
    of just meeting NAAQS scenarios that considered all daylight hours were added to those
    based on maximum daily 1-hour indicators. The 98th percentile form was also included
                                                                                   10

-------
    along with the 90th and 95thpercentiles. Tile plots of hourly PM light extinction (Figure 3-
    12) and composition bar graphs of the top 10% of days for maximum daily 1-hour and
    aggregated individual daylight hours (Figure 3-13) were shown in part to help illuminate
    the similarities and differences between these various indicators with respect to current
    conditions. Similarly, additional figures and table entries were generated to illustrate the
    characteristics of various PM light extinction NAAQS scenario forms (Tables 4-2, 4-3, 4-
    5, and 4-7; Figures 4-1 through 4-3).

    What are the Panel's views regarding EPA staff interpretations of these displays included
    in the text? Are there supplemental or alternative interpretations the Panel would suggest?
    Are there additional approaches the Panel would suggest regarding ways to summarize,
    display, or assess the results of these analyses, including similarities and differences
    between the various scenarios?

I would suggest there be more exploration of what if any potential problems would arise
if there would be for a PM2.5 visibility  standard only. There are substantial advantages
to establishing monitoring systems for fine particle scattering (nephelometers) that would
avoid the difficulties of the predominately forward scattering by coarse particles. In most
locations, it is likely that a PM10-2.5 primary standard could reduce the effect of this
fraction on extinction to the point where it is not a significant part of the visibility
problem.  Such an approach would be easier to implement on both a monitoring and
implementation basis.
                                                                                  11

-------
Comments from Mr. Rich Poirot

Review Comments on 2nd Draft January 2010 PM UFVA, R. Poirot, 2/4/10

I thought the first draft UFVA was well done, especially considering the limited time
schedule and complexity of the required calculations.  In the second draft, EPA staff has
been very responsive to CASAC recommendations, adding a detailed logit analysis to
better understand similarities and differences in the urban visibility preference studies,
recalculating estimated extinction levels using a 90% RH screen, and considering (90th,
95th and 98th) percentiles based on all available daylight hours as well as based on the
single worst daylight hour in each day.  These additions represent a substantial body of
new work, again conducted in a very limited time period, and as before the additional
calculations and results are explained, compared and illustrated in clear detail.

The UFVA presents logical and persuasive arguments for use of a PM light extinction
indicator with a 1-hour (daylight) averaging  time as a sound basis for a secondary PM
standard to protect visibility. The UFVA also presents reasonable ranges for the level
(20 to 30 deciviews) and form (90th to 98th percentile) that such a standard might take,
and clearly illustrates the differences among the many resulting optional combinations in
terms of the protectiveness and/or stringency across the 15 diverse urban areas selected
for this analysis.

I don't believe that any major revisions are required in this second draft, and that the
methods employed here - if not some of the  specific calculations - would make a sound
basis for recommending and comparing optional ranges of a secondary PM light
extinction standard in the upcoming Policy Assessment document.

1) In general, what are the Panel's views on the level  of detail provided in the body of
   the report and associated appendices? Does the Panel agree that all of this
   information is useful to retain or is there material that the Panel would recommend
   deleting? Does the Panel have any suggestions regarding the organization and
   distribution of information throughout the document and in the appendices?

While a substantial level of detail has been provided in the UFVA,  many of the calculations
included here are very complex, and a fairly wide range of optional combinations of levels
and forms are presented and compared. I think the level of detail provided is appropriate in
relation to the level of complexity and the range of options considered, allowing the
"interested reader" to understand just exactly how the many calculations have been made,
and often providing several different views for "seeing" the results  in several different
contexts. For the "disinterested reader" there are several summary tables toward the end
which condense most of the varied results to a single page or two. I don't believe the
document could be better organized or more succinctly written.

2) In the Panel's view, to what extent does  the logit analysis presented in Chapter 2 of
   the second draft Visibility Assessment and further expanded and described in the
   supplemental memorandum add value to the urban visibility preference study
   analysis and provide additional support for combining and  comparing the results
                                                                                12

-------
    from the four cities, as shown in Figure 2-16? What are the Panel's views on the
    clarity and adequacy of the descriptions associated with such a combined
    assessment and on the conclusions that can be drawn from the assessments? Please
    provide comments on any additional insights that might be drawn from these
    analyses or on any additional caveats that should be considered.

The logit analysis presented in chapter 2 helps in more clearly quantifying and understanding
the similarities and differences among results from the 4 urban visibility preference studies.
The associated discussion of the logit results and consideration of alternative hypotheses for
the differences and similarities across the different study areas (and different study
approaches) are also helpful. While the resulting 20 to 30 deciview range of visual air
quality levels meeting a 50% acceptability criteria, remains rather broad, I doubt that much
more can be learned or that this range could be substantially reduced by additional analyses
of these limited results, and think that selecting a level or levels from within this 20 to 30 dv
range is well justified for the initial establishment of a secondary PM standard.  Additional
studies using consistent methods and conducted over a range of different kinds of urban areas
and different views will help refine this range in  the future.

3)  What are the Panel's views on the extent to which the analysis of the frequency of
    co-occurrences of hourly relative humidity values below and above 90 percent with
    other meteorological events such as rain or fog (Chapter 3, section 3.3.5; Table 3-6)
    provides scientific support for consideration of how to address relative humidity in
    defining the form of a standard based on a PM light extinction indicator?

Eliminating hours when RH exceeds 90% is a good idea - for conducting this assessment as
well as for specifying conditions for and determining compliance with a secondary PM light
extinction standard. The analysis presented in the 2nd draft UFVA provides a compelling
demonstration that a 90% RH screen can efficiently eliminate periods when "natural"
visibility-impairing weather events are occurring, and when PM light extinction is not likely
to be the predominant cause of impaired visibility.  The spatial (& temporal) variability of
RH and weather (& PM species) are important related issues, in that there may be natural
weather event impairing visibility along the sight path but not observed at the location of the
RH (or PM or PM light extinction observation).  I don't think its warranted here, but it might
be possible to explore this possibility a bit by considering the RH and weather observations at
a range of stations within say a 30 mile  radius and looking for co-occurrences of RH above or
below 90% and weather events across all the sites.

In addition to screening out weather events, there are other benefits of employing this RH
screen.  It will help avoid any public perception,  however faulty, that the regulatory metric is
influenced by uncontrollable weather events.  RH can be difficult to measure, especially at
high levels, such that an indicated RH of 90% may actually be significantly higher (or lower)
than that. The hygroscopic growth curve gets very steep at high humidity such that a small
error in the RH or (sulfate or nitrate) PM species measurement or in its spatial or temporal
representativeness of the actual (or  in the case of this assessment, estimated) extinction
throughout the site path could result in a large error in estimated extinction, or in the
visibility relevance of extinction measured at a single location. An RH screen might also
allow for measurement options such as adding a  "smart heater" to a nephelometer that would
only kick in at high RH but would prevent larger water droplets from entering and soiling the
                                                                                   13

-------
measurement chamber,  reducing instrumental maintenance costs, and promoting better data
quality and capture efficiency.

Screening out the higher RH values will slightly "dry out" the regulatory metric, eliminate
some of the highest (but most uncertain) periods of estimated and measured PM extinction,
shift the distribution downward and reduce, somewhat, the differences between East and
West. It would also limit the extent to which the influences of hygroscopic sulfate and nitrate
species are enhanced relative to the other PM species. At 95% RH the extinction efficiencies
of ammonium sulfate and ammonium nitrate would be more than 22  m2/g, compared to
efficiencies of 10 m2/g for EC, 4 m2/g for organic matter, 1 m2/g for fine soil and 0.6 m2/g for
coarse particles. At 90% RH the maximum extinction efficiencies of ammonium sulfate and
ammonium nitrate would be limited to 12.5 m2/g. This would be an "artificial" limit, and
there will likely be hours of very poor visibility when natural weather is not the cause which
will not be considered with this RH screen, but I think it's use is well justified for the other
reasons indicated above, and that there are other benefits associated with having a slightly
dryer and more stable regulatory metric.

4) In response to CASAC recommendations, descriptions of current conditions and
   results of just meeting NAAQS scenarios that considered all daylight hours were
   added to those based on maximum daily 1-hour indicators. The 98th percentile form
   was also included along with the 90th and 95th percentiles. Tile plots of hourly PM
   light extinction (Figure 3-12) and composition bar graphs of the top 10% of days for
   maximum daily 1-hour and aggregated individual daylight hours (Figure 3-13) were
   shown in part to help illuminate the similarities and differences between these
   various indicators with respect to current conditions. Similarly, additional figures
   and table entries were generated to illustrate the characteristics of various PM light
   extinction NAAQS scenario forms (Tables 4-2, 4-3, 4-5, and 4-7; Figures 4-1 through
   4-3).

   What are the Panel's views regarding EPA staff interpretations of these displays
   included in the text? Are there supplemental or alternative interpretations the Panel
   would suggest? Are there additional approaches the Panel would suggest regarding
   ways to summarize, display, or assess the results  of these analyses, including
   similarities and differences between the various scenarios?

I think applying the 90% RH screen, including a higher (98th) percentile form and
considering an alternative way of calculating percentiles - based on all daylight hours ,
rather than just the worst hour in a day - were valuable additions to the UFVA. When
these are combined with other options including 90th and 95th percentiles, different levels
(64,  112, 191 Mm"1) of a secondary standard and assuming attainment of different levels
of a primary standard (15/35, 12/25 ug/m3), the many resulting options are complex and
difficult to communicate and compare. I think the UFVA does an excellent job of
presenting this information as clearly as possible.
I think the tile plots are very informative (not to mention beautiful!) and really help show
the seasonal and diurnal patterns and similarities and differences across the different
study areas. They also help show that the effect of the 90% RH screen tends to often be
elimination of the first  few early morning hours when fog is most likely and when its
                                                                                 14

-------
most difficult for human observers to discern the differences between natural and
manmade effects, and provide added confidence that the use of this screen is appropriate.
The plots of maximum hourly extinction on top 10% days vs. compositions on top
percentiles considering all daylight hours are useful for seeing differences between cities,
on different worst days (hours) and (to a lesser extent) for different high percentiles. I
think the accompanying interpretation and discussion are reasonable and informative. I
don't these graphs necessarily make for an effective direct comparison or evaluation of
whether percentiles calculated from the single worst hour of each day would be a better
regulatory metric. Personally, I think I prefer using all daylight hours to only using the
worst single hour each day,  since I think a day that's hazy all day long is more
objectionable than one where its hazy for just one humid hour just after sunrise. I also
don't think there are many 1-hr control strategies, especially for SOx and NOx (and
NHx) sources, which I think get additionally emphasized by the single worst hour
approach, with a consequent relative de-emphasis of carbonaceous pollutants.  I also
imagine that using all daylight hours would produce a more stable and somewhat dryer
metric, reducing, somewhat, the differences between East and West. I think this could be
important if the Agency finds it necessary to employ a traditional fixed "threshold-based"
approach for the secondary NAAQS, rather than the "progress-based" approach
recommended by the PM panel as scientifically preferable, and a conceptual approach to
consider during implementation - even though it may not currently be a practical option
for the Agency.
Possibly it would be informative to present side by side bar  graphs showing an average of
the compositions on the worst 10%, 5% and 2% hours calculated using the 2 different
approaches. Are the compositions really much different? Would it be informative to
know how many days would be involved using the 2 different approaches, or perhaps to
consider the days or kinds of pollution events that tend to get considered in one approach
but not the other?
Certain aspects of the discussion seem directed toward supporting the original worst hour
of day approach - for example a repeated observation that the 98th percentile using all
daylight hours is not much different from the 98th percentile using worst daylight hours
and expressing the results of both approaches in Table 4.7 in units of single worst hour of
day. As indicated, I think I conceptually prefer the all-hours approach but think it would
be useful to see more clearly what the differences are.  I don't think this is needed in the
Risk Assessment, but maybe the Policy Assessment could explore these options in more
detail.
                                                                               15

-------
Other Comments:
As indicated in comments on the 1st draft UFVA, I am very suspicious of the results for
St. Louis and especially the extremely high coarse mass concentrations there - which
often dominate the reconstructed extinction estimates during the haziest hours. I note that
the 2 estimated haziest hours at STL for which the estimated coarse mass is the major
contributor to extinction are 7/10/06 at 11:00 (estimated extinction of about 600 Mm"1)
and 5/17/06 at 15:00 (estimated extinction of about 800 Mm"1). However, the ASOS
visibility data from the STL airport (15 miles from downtown) indicates visibility was 10
miles or greater (Bext was 245 Mm-1 or less) during both of these time periods.  I
suggest discarding the STL estimates,  or using other PMio and PM2.5 data for making
600
600
400
3B
200
  KSTL Airport ASOS
       190 Mm1
{245 Mm-1 or 10    at
 1100 CST on 7/10/08
     UFA
6CO Mm-1  rrKHtiy coarse)
                        SURF MET:Bext
 l:
m:
1C
 2006-Jul  03 04 05 06 07
                on            be      by    of 1.3        so
                      of 190 Mm-1 =                   of 10
1000
 900
 800
 700
 600
 500
 400
 300
 200
                  "SURF MET:Bext
these estimates.  Note that these Bext data were extracted from the Datafed.net system, in
which a Koschmeider constant of 3000, rather than 3912 was used (incorrectly, I think) to
convert the visual range data to extinction.
                                                                                 16

-------