United States Office of Air Quality EPA-454/R-99'-QOS
Environmental Protection Planning and Standards April 1999
Agency Research Triangle Park, NC 27711
Ak
GUIDELINE ON DATA HANDLING
CONVENTIONS FOR THE PM NAAQS
-------
EPA-454/R-99-XXX
April 1999
GUIDELINE ON DATA HANDLING
CONVENTIONS FOR THE PM NAAQS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
Disclaimer
This guidance has been review and approved for publication by the U.S. Environmental
Protection Agency's Office of Air Quality Planning and Standards. Mention of trade names or
commercial products are not intended to constitute endorsement or recommendation for use.
-------
Contents
What does this guideline cover? 1
How is this guideline presented and organized? 1
Chapter 1. Comparing Your Data to the Standards 3
1. What are the standards for particulate matter? 3
2. How do I round my numbers? How many decimal places do I keep? 3
3. How do I compute the 3-year average annual mean for PM10? 5
4. How do I compute the 3-year average, spatially averaged, annual mean for PM2 5? ... 6
5. How do I compute the 3-year average 99th percentile for PM10? 8
6. How do I compute the 3-year average 98th percentile for PM25? 11
7. Is there another way to determine the 98th or 99th percentile? 12
8. How do I make sure my data is complete enough to meet the standards? 15
9. What if I want to show I meet the standards but I don't have complete data? 15
10. How do I fill in for missing data to show I meet the standards? 16
11. May I use data from a PM10 Monitor to show that I meet the PM2.5
standards? 22
12. May I ignore years with high concentrations if they have incomplete data? 22
13. How little data may I use to show I'm NOT meeting the standards? 23
14. May I fill in for missing data to show that I don't meet the annual standards? 23
15. What if I don't meet the minimum of 11 samples in a quarter? Can I still show
I don't meet the annual standards? 23
16. May I show that I meet the standards, or that I don't meet the standards, if I have only
1 or 2 years of data? 24
17. What do I do about a monitor that has stopped monitoring? 24
Chapter 2. Sampling Frequency 25
18. If I've collected more samples than were scheduled, may I use all the data to show I've
met or not met the 24-hour standards? 25
19. How do I calculate the 98th and 99th percentiles when my sampling frequencies are
seasonal? 27
20. How do I compute quarterly averages when parts of the quarter are sampled at
different frequencies? 30
21. Under what circumstances may I reduce the required sampling frequency at a site for a
year or season? 30
22. May I use a correlated acceptable continuous (CAC) monitor to reduce my sampling
frequency? 31
23. If I've missed a scheduled sample, may I make it up? 32
iii
-------
24. How do I use scheduled samples with make-up and other non-scheduled samples to
show that I meet or don't meet the standards? 33
Chapter 3. Monitoring Issues 35
25. Which monitors do I compare to which standards? 35
26. What are "community monitoring zones" (CMZs) and how do I decide which sites to
include in a spatial average? 37
27. If a monitor is reassigned to a different CMZ during the three-year period, which
CMZ assignment should I use to calculate the spatial average? .... 41
28. How many hourly values make up a valid 24-hour average for a continuous monitor?42
Chapter 4. Miscellaneous Issues 43
29. How do I handle data from uncontrollable or natural events? 43
30. How do I determine whether wildland and prescribed fires managed for resource
benefits significantly contribute to violations of the PM2 5 or PM10
NAAQS? 44
IV
-------
GUIDELINE ON DATA HANDLING CONVENTIONS FOR THE
PM NAAQS
What does this guideline cover?
This guidance provides information you, the state or local agency responsible for monitoring and
interpreting air quality, need to determine whether you are meeting the standards for particulate
matter in 40 CFR Part 50. It clarifies requirements for data handling and completeness in
Appendix N to 40 CFR Part 50. It tells you how to handle missing data, different sampling
frequencies, and calculating spatial averages for the PM2 5 annual standard.
How is this guideline presented and organized?
This guideline is organized in a question and answer format. Questions are sorted by topic into
chapters:
• Chapter 1 Comparing Your Data to the Standards
• Chapter 2 Sampling Frequency
• Chapter 3 Monitoring Issues
• Chapter 4 Miscellaneous Issues
Chapter 1: Comparing Your Data to the Standards
1. What are the standards for particulate matter?
2. How do I round my numbers? How many decimal places do I keep?
3. How do I compute the 3-year average annual mean for PM10?
4. How do I compute the 3-year average, spatially averaged, annual mean for PM2 5?
5. How do I compute the 3-year average 99th percentile for PM10?
6. How do I compute the 3-year average 98th percentile for PM2 5?
7. Is there another way to determine the 98th or 99th percentile?
8. How do I make sure my data is complete enough to meet the standards?
9. What if I want to show I meet the standards but I don't have complete data?
10. How do I fill in for missing data to show I meet the standards?
11. May I use data from a PM10 monitor to show that I meet the PM2 5
standards?
12. May I ignore years with high concentrations if they have incomplete data?
13. How little data may I use to show I'm NOT meeting the standards?
14. May I fill in for missing data to show that I don't meet the standards?
15. What if I don't meet the minimum of 11 samples in a quarter? Can I still
show I don't meet the annual standards?
1
-------
16. May I show that I meet the standards, or that I don't meet the standards, if
I have only 1 or 2 years of data?
17. What do I do about a monitor that has stopped monitoring?
Chapter 2: Sampling Frequency
18. If I've collected more samples than were scheduled, may I use all the data
to show I've met or not met the 24-hour standards?
19. How do I calculate the 98th and 99th percentiles when my sampling
frequencies are seasonal?
20. How do I compute quarterly averages when parts of the quarter are
sampled at different frequencies?
21. Under what circumstances may I reduce the required sampling frequency at
a site for a year or a season?
22. May I use a correlated acceptable continuous (CAC) monitor to reduce my
sampling frequency?
23. If I've missed a scheduled sample, may I make it up?
24. How do I use scheduled samples with make-up and other non-scheduled
samples to show that I meet or don't meet the standards?
Chapter 3: Monitoring Issues
25. Which monitors do I compare to which standards?
26. What are "community monitoring zones" (CMZs), and how do I decide
which sites to include in a spatial average?
27. If a monitor is reassigned to a different CMZ during the three-year period,
which CMZ assignment should I use to calculate the spatial average?
28. How many hourly values make up a valid 24-hour average for a continuous
monitor?
Chapter 4: Miscellaneous Issues
29. How do I handle data from uncontrollable or natural events?
30. How do I determine whether wildland and prescribed fires managed for
resource benefits significantly contribute to violations of the PM25 or PM10
NAAQS?
-------
Chapter 1
Comparing Your Data to the Standards
1. What are the standards for particulate matter?
Appendix N of 40 CFR Part 50 says air quality meets
• The annual PM10 standard whenever the 3-year average of the annual mean PM10
concentrations at each monitoring site is less than or equal to 50 ug/m3.
• The 24-hour PM10 standard whenever the 3-year average of the annual 99th percentile
values for PM10 at each monitoring site is less than or equal to 150 ug/m3.
• The annual PM25 standard whenever the 3-year average of the spatially averaged
annual mean PM2 5 concentrations (among designated monitors) is less than or equal to
15.0 ug/m3. [Designated monitors are sites designated for spatial averaging in a State
PM Monitoring Network Description. This applies only if you opt to do spatial
averaging in your area as discussed in 40 CFR Part 58. When you don't opt to do
spatial averaging, use the annual average of the single site.]
• The 24-hour PM2 5 standard whenever the 3-year average of the annual 98th percentile
values for PM2 5 at each monitoring site is less than or equal to 65 ug/m3.
Each standard is based on three consecutive, complete, calendar years of air quality data.
2. How do I round my numbers? How many decimal places do I keep?
• If you're doing an initial calculation with 24-hour average PM2 5 concentrations, such
as entering 24-hour averages into a computer program, use one decimal place.
Truncate any extra digits.
Example: 10.314 ug/m3 truncates to 10.3 ug/m3
10.37 ug/m3 truncates to 10.3 ug/m3
-------
If you're doing an initial calculation with 24-hour average PM10 concentrations, such
as entering 24-hour averages into a computer program, use the integer part. Truncate
any decimal parts.
Example: 45.29 ug/m3 truncates to 45 ug/m3
45.816 ug/m3 truncates to 45 ug/m3
If you're doing an intermediate calculation, such as a quarterly-average PM10 value
from the 24-hour averages, keep all available digits and decimal places on your
calculator. [Note that for the sake of brevity this Guideline does not include all
available digits in its examples.]
If you're comparing a result to a standard, which includes deciding whether to use
incomplete data with high concentrations, round as follows:
Annual PM25: Round to the nearest 0.1 ug/m3. Round decimals 0.05 or greater
up and those less than 0.05 down.
Example: 15.049 rounds to 15.0 |ug/m3 (not above the standard)
15.05 rounds to 15.1 |ug/m3 (above the standard)
24-Hour PM25: Round to the nearest 1 ug/m3. Round decimals 0.5 or greater
up and those less than 0.5 down.
Example: 65.49 rounds to 65 |ug/m3 (not above the standard)
65.5 rounds to 66 |ug/m3 (above the standard)
Annual PM10: Round to the nearest 1 ug/m3. Round decimals 0.5 or greater up
to and those less than 0.5 down.
Example: 50.486 rounds to 50 |ug/m3 (not above the standard)
50.51 rounds to 51 |ug/m3 (above the standard)
24-Hour PM10: Round to the nearest 10 ug/m3. Round integers of 5 or greater
up and those less than 5 down.
Example: 154.893 rounds to 150 |ug/m3 (not above the standard)
155.51 rounds to 160 |ug/m3 (above the standard)
-------
3. How do I compute the 3-year average annual mean for PM10?
Follow these steps:
a. Calculate the four quarterly means for each year. Add all of the 24-hour sample
concentrations within a quarter; then, divide by the number of samples in the quarter.
Example:
Suppose you took 87 24-hour average PM10 measurements (in ug/m3) during Quarter
1 of Year 1, of which the first eight are 39, 42, 58, 66, 45, 28, 36, and 27.
In this case, the
quarterly mean = (39 + 42 + 58 + 66 + 45 + 28 + 36 + 27 + ...) (ig/m3.
87
b. Calculate the annual mean from the four quarterly means. Add the four quarterly
means; then, divide by 4.
Example:
Assume the four quarterly means for Year 1 of the three-year period for which you
are to compare readings to the NAAQS are as follows: Quarter 1 - 43.23 ug/m3;
Quarter 2 - 44.72 ug/m3, Quarter 3 - 40.96 ug/m3, and Quarter 4 - 40.77 ug/m3.
Then, the annual mean = ( 43.23 + 44.72 + 40.96 + 40.77) ug/m3 = 42.42 ug/m3.
4
c. Calculate the 3-year average annual mean. Add the three annual average means; then,
divide by 3.
Example:
Assume you've also calculated the PM10 annual means for Years 2 and 3 as 57.17 and
53.23 ug/m3, respectively. If so, the 3-year average annual mean =
(42.42 + 57.17 + 53.23) ug/m3 = 50.94 ug/m3.
3
-------
d. Round the 3-year average annual mean. Because you're trying to calculate the 3-year
average annual mean for PM10, round to the nearest integer. Round up decimal parts
of 0.5 or greater; round down decimal parts less than 0.5.
Example:
Round 50.94 jug/m3 up to 51 jug/m3.
e. Compare your result to the standard. 51 ug/m3 is greater than 50 ug/m3, so this
example doesn't meet the PM10 annual standard.
4. How do I compute the 3-year average, spatially averaged, annual mean for PM2 5?
Spatial averages are computed across all designated sites in a Community Monitoring
Zone (CMZ) (see 40 CFR Part 58). If the CMZ has only one site, you can follow the procedures
described here or the slightly simpler ones in Question 3. If you follow the ones in Question 3,
round the PM25 three-year average annual mean to the nearest 0.1 ug/m3 before comparing it with
the standard.
Assume you've designated three monitors to use for computing the 3-year average,
spatially averaged, annual mean.
a. Calculate the four quarterly means for each site. Add all of the 24-hour sample
concentrations; then, divide by the number of samples.
b. Calculate the annual mean from the four quarterly means. For each site, add the four
quarterly means; then, divide by 4.
Example:
Assume the four quarterly means for one site for the first year are 11.6, 12.4, 15.1, and
12.1 ug/m3. If so, the annual mean is
(11.6+ 12.4+ 15.1 + 12.1) ug/m3 = 12.8 ug/m3.
4
Calculate the annual means for the other sites and other years using the same approach
to give the results shown in Table 4-1. Note that site 3 had insufficient data in Year 1
(fewer than 11 samples in one or more quarters, as indicated by "NA"), so the annual
mean is not available.
-------
Table 4-1. Using Annual Means (ug/m3) from Three Sites for Spatial Averaging.
Site Number
Year One Two Three
1 12.8 14.2 NA
2 13.0 13.5 12.9
3 15.2 14.8 17.1
c. Calculate the spatially averaged annual mean of the designated monitors in the area.
Add all values for a given year across all sites; then, divide by the number of sites with
data for that year.
Example:
Year 1. (12.8 + 14.2) ug/m3 = 13.5 jig/m3.
2
Year 2. (13.0 + 13.5 + 12.9) ug/m3 = 13.13 jig/m3.
3
Year 3. (15.2+ 14.8 + 17.1) ug/m3 = 15.7 jig/m3.
3
d. Calculate the 3-year average, spatially averaged, annual mean. Add the three spatially
averaged annual means and divide the total by 3.
Example:
13.50+ 13.13 + 15.70 ug/m3 = 14.11 jig/m3.
3
e. Round to the nearest 0.1 ug/m3. Round decimals 0.05 or greater up and those less
than 0.05 down.
-------
Example:
Round 14.11 down to 14.1 jug/m3.
f. Compare your result to the standard. 14.1 is less than 15.0 ug/m3, so this example
meets the PM25 annual standard of 15.0 ug/m3.
5. How do I compute the 3-year average 99th percentile for PM10?
This computation is explained with an example. If you've collected more than the
scheduled number of samples in a year, you will need to modify step c of this calculation as shown
in the example for Question 18.
Examine Table 5-1, which presents some sample data on 24-hour concentrations of PM10
collected once every three days over a three year sampling period. The table shows only the three
highest and three lowest values for each year. In this example, a total of 110 samples were
collected in Year 1; 98 samples in Year 2; and 100 samples in Year 3.
Table 5-1. Sample Values for Computing the 3-year Average 99th Percentile for PM10.
YEAR1
(110 total samples)
Value
(ug/m3)
YEAR 2
(98 total samples)
Value
(ug/m3)
YEAR 3
(100 total samples)
Value
(ug/m3)
88
130
120
85
128
87
•
Sort all data values
150
93
90
97
148
143
•
collected in each vear
40
52
144
140
48
147
•
from lowest to highest. The re
look like Table 5-2.
-------
Table 5-2. Arranging Data Values in Ascending Order.
YEAR1
Value
(ug/m3)
85
87
88
120
128
130
Assign a rank to each data value.
YEAR 2
Value
(ug/m3)
90
93
97
143
148
150
Assign rank 1
YEAR 3
Value
(ug/m3)
40
48
52
140
144
147
to the lowest 2
year, rank 2 to the second-lowest average, and so on. The results should look like
Table 5-3.
Table 5-3. Data Values with Assigned Ranks.
Rank
1
2
3
108
109
110
Year 1
Value
(ug/m3)
85
87
88
120
128
130
Rank
1
2
3
96
97
98
Year 2
Value
(ug/m3)
90
93
97
143
148
150
Rank
1
2
3
98
99
100
Year 3
Value
(ug/m3)
40
48
52
140
144
147
c. Calculate the rank of the 99th percentile value for each year.
• Multiply the number of samples taken in the year by 0.99.
-------
Example (from Table 5-3):
Year 1: 110x0.99 =108.9
Year 2: 98x0.99 = 97.02
Year 3: 100x0.99 = 99.0
• Take the integer part of the product and add 1. This step gives you the ranking
that corresponds to the 99th percentile:
Example:
Year 1. Result is 108.9; integer part is 108. 108 + 1 = 109.
Year 2. Result is 97.02; integer part is 97. 97 + 1 = 98.
Year 3. Result is 99.0; integer part is 99. 99 + 1 = 100.
The 99th percentile ranks are therefore 109, 98, and 100.
d. Find the value that corresponds to each rank.
Example:
Referring to Table 5-3,
Year 1. Rank 109 corresponds to 128 |ug/m3.
Year 2. Rank 98 corresponds to 150 |ug/m3.
Year 3. Rank 100 corresponds to 147 |ug/m3.
e. Calculate the 3-year average of all three values for the 99th percentile.
Example:
128 + 150+ 147 = 141.67 jig/m3.
3
f. Round to the nearest 10. In this case, round 141.67 down to 140 |ug/m3.
g. Compare your result to the standard. 140 is less than 150 ug/m3. This example meets
the 24-hour standard of 150 ug/m3 for PM10.
10
-------
6. How do I compute the 3-year average 98 percentile for PM2 5?
First, review answer 5, above, which is similar to this problem. Then, work through the
example below. However, if you've collected more than the scheduled number of samples in a
year, you will need to modify step c of this calculation as shown in the example for Question 18.
This example assumes you have data on 24-hour concentrations of PM25 collected on an
every day sampling schedule over a three year period: 281 samples in Year 1; 304 samples in Year
2; and 296 samples in Year 3.
a. Sort all data values collected in each year from lowest to highest.
b. Assign a rank to each data value. Assign rank 1 to the lowest 24-hour average in each
year, rank 2 to the second lowest, and so on.
The results for this example site after steps a and b are in Table 6-1, below.
Table 6-1. Rankings of Sample Values for One Site over Three Years.
Rank
1
275
276
277
281
Year 1
Value
(ug/m3)
40.7
58.9
59.0
62.2
64.4
Rank
1
296
297
298
304
Year 2
Value
(ug/m3)
40.6
54.3
57.1
63.0
65.8
Rank
1
290
291
292
296
Year 3
Value
(ug/m3)
45.1
66.0
68.4
69.8
70.9
c. Calculate the rank of the 98th percentile for each year.
• Multiply the number of samples taken in the year by 0.98.
Example (from Table 6-1):
Yearl. 281x0.98 = 275.38
Year 2. 304 x 0.98 = 297.92
Year 3. 296x0.98 = 290.08
11
-------
• Take the integer part of the product and add 1. This gives you the ranking that
corresponds to the 98th percentile:
Example:
Yearl. 275+1=276
Year 2. 297 + 1 = 298
Year3. 290+1=291
The 98th percentile ranks are therefore 276, 298, and 291.
d. Find the value that corresponds to each rank.
Example (from Table 6-1):
Year 1. Rank 276 corresponds to 59.0 ug/m3.
Year 2. Rank 298 corresponds to 63.0 ug/m3.
Year 3. Rank 291 corresponds to 68.4 jug/m3.
e. Calculate the 3-year average of all three values for the 98th percentile.
Example:
59.0 + 63.0 + 68.4 = 63.47 ug/m3.
3
f. Round to the nearest integer. Round decimals 0.5 or greater up and those less than 0.5
down.
Example:
Round 63.47 down to 63 |ug/m3.
g. Compare your result to the standard. 63 is less than 65 ug/m3, so this example meets
the 24-hour standard of 65 ug/m3 for PM25.
7. Is there another way to determine the 98th or 99th percentile?
Yes, you may also use Table 7-1, below, to calculate the 98th and 99th percentiles. Note
that for this method PM concentrations are ranked from highest to lowest (i.e., 1 = highest
concentration). If you've collected more than the scheduled number of samples in a year, see
Question 18 for an example of how to modify the calculations.
12
-------
Table 7-1. 98th and 99th Percentiles Defined by Appendix N.
Samples1
1-504
51-1005'6
101-1507
151-200
201-250
251-300
301-350
351-366
nth Max. for 98th
Percentile2
Value of n
1
2
3
4
5
6
7
8
nth Max. for 99th
Percentile3
Value of n
1
1
2
2
3
3
4
4
'Samples are all the monitored daily PM values in a year unless you've collected more than the scheduled number of samples. In that
case, see Question 18.
2nft Max for the 98th percentile is the nft highest value in a year that represents the 98th percentile.
(1 is the highest value measured in a year, 2 is the second highest value, etc.)
3nft Max for the 99th percentile is the nft highest value in a year that represents the 99th percentile.
41 in 6 day sampling achieving 75 to 85% data completeness would be in this category.
51 in 6 day sampling achieving 85 to 100% data completeness would be in this category.
61 in 3 day sampling achieving 75 to 82% data completeness would be in this category.
71 in 3 day sampling achieving 83 to 100% data completeness would be in this category.
Work through the example below, which uses the same PM25 data as the example for answer 6. In
years 1, 2, and 3, your data sets from the site contained 281, 304, and 296 samples, respectively.
a. Sort all of the data values collected in any given year from highest to lowest. The
results should look like Table 7-2, which is a rearrangement of Table 6-1 (the values
go from the highest to the lowest). This table has the n"1 highest value listed for each n
so it is easy to use it with Table 7-1.
b. Find n from Table 7-1 and find the r^1 highest value from Table 7-2. For PM2 5.
calculate the 98th percentile. For PM10, calculate the 99th percentile.
13
-------
Table 7-2. Maxima of Sample PM25 Values from One Site for a Three-Year Period.
Year 1
IIth highest value
n (ug/m3)
1 64.4
2 63.7
3 63.1
4 62.8
5 62.2
6 59.0
7 57.9
8 57.3
9 54.7
• •
Year 2
IIth highest value
(ug/m3)
65.8
65.2
64.9
64.5
64.1
63.8
63.0
57.1
54.3
•
Year 3
IIth highest value
(ug/m3)
70.9
70.8
70.3
70.0
69.8
68.4
66.0
65.6
65.6
•
Example:
In Year 1 there were 281 PM25 samples. For PM25, you calculate the 98th percentile.
From Table 7-1, n equals 6. From Table 7-2, the n"1 highest value equals 59.0 |ug/m3.
Similarly, for Years 2 and 3, n equals 7 and 6, and the n"1 highest values equal 63.0 and
68.4 |ug/m3, respectively.
c. Calculate the 3-year average of all three values for the 98th or 99th percentile.
Example:
59.0 + 63.0 + 68.4 = 63.47 jig/m3.
3
d. Round to the nearest integer for PM2 5 and to the nearest multiple of 10 for PM10 For
PM2 5, round decimals 0.5 or greater up and those less than 0.5 down. For PM10, round
integers 5 or greater up and those less than 5 down.
14
-------
Example:
Round 63.47 down to 63 |ug/m3.
e. Compare your result to the standard. 63 is less than 65 ug/m3 so this example meets
the 24-hour standard of 65 ug/m3 for PM25.
8. How do I make sure my data is complete enough to meet the standards?
Table 8-1 summarizes how complete data must be to show you meet the standards.
Table 8-1. How Complete Data Must Be to Show that an Area Meets the NAAQS for PM.
Standard
Data Completeness to Show You Meet the Standards
Daily PM2 5 Single site: at least 75% of the scheduled sampling days per quarter
Daily PM10 Single site: at least 75% of the scheduled sampling days per quarter
Annual Single site: if each quarter has at least 75 % of the scheduled sampling days, the
PM2 5 annual mean for that year and site is valid.
Community monitoring zone: In each of the three years, at least one site must
have a valid annual mean. The valid sites may be the same every year, or may
vary from year to year.
Annual
PM,n
Single site: at least 75% of the scheduled sampling days per quarter
9. What if I want to show I meet the standards but I don't have complete data?
Appendix N says you may have compelling reasons to use less complete data, but the
Regional Administrator must approve it. The Regional Administrator may want to consider filling
in for missing scheduled sampling days using the procedures in Question 10 if you
• Have at least 50% of the scheduled number of samples for each quarter for all three
years.
15
-------
• Show that the emissions and meteorology for the substitute quarters compare to the
emissions and meteorology for the quarters in question.
• Meet the standards based on the incomplete data.
10. How do I fill in for missing data to show I meet the standards?
First, you should meet the criteria from Question 9. Then, you may use either of two
approaches to fill in for missing scheduled sampling days:
(1) Replace missing data with collocated data for the same year and quarter.
(2) Replace missing data with the maximum data value across all three years for the same
quarter.
Approach No. 1: How you may use collocated data to substitute for missing data
You may use collocated data in either of two ways:
(1) If you have collocated PM10 or TSP data from a monitor for the same year and
quarter, you may use it to replace missing PM10 or PM2 5 values on scheduled
sampling days.
(2) If you have collocated PM25 data from a monitor for the same year and
quarter, you may use it to replace missing PM2 5 values on scheduled sampling
days.
The following notes apply in either case:
• You must substitute for all missing scheduled sampling days where collocated data
is available, not just for selected days in that quarter.
• If you didn't collect the collocated data on the same day as the scheduled sampling
day, you can use collocated data from the nearest day (within two days before or
after) to replace missing PM concentrations. The emissions and meteorology for
the substitute day must compare to the emissions and meteorology for the missing
day.
• The collocated PM10 or PM2 5 monitor must use a Federal Reference Method or
Equivalent Method, or must be a collocated QA monitor.
16
-------
Example: How to Substitute Collocated PM10 Data for Missing Data and Use It to
Calculate the Annual PM25 Standard
For this example, assume
• Your PM2 5 monitor is on a once every three days schedule, and it has a collocated
PM10 monitor that has collected three years of data and measured it using a Federal
Reference Method.
• The site is the only site in a community monitoring zone.
• In Years 1 and 3 of sampling, the annual average concentrations of PM25 are 12.84
and 12.43 ug/m3, respectively.
• In Quarter 3 of Year 2 (see Table 10-1), you got only 23 samples from the 31
scheduled sampling days.
• The total PM25 concentration of the 23 samples is 292.1 ug/m3.
• You have collocated PM10 data for only three of the eight scheduled sampling days
that are missing PM25 data. The three PM10 values are 30, 21, and 54 ug/m3.
• You meet the criteria from Question 9. [To show that you meet the annual PM2 5
standard based on the incomplete data, repeat steps b to f below without using the
substituted data. In step b, the average for Quarter 3 of Year 2 is 12.70 ug/m3
without using the substituted data. Then, in step e, the resulting rounded 3-year
average annual mean is 12.8 ug/m3, which meets the annual PM25 standard of 15.0
ug/m3.]
For Quarter 3 of Year 2, you had 31 scheduled sampling days but only 23 samples. 75%
of 31 equals 24 (rounding up to the next integer). So you don't meet the data completeness
requirement of 24 samples. You need at least one more sample.
17
-------
Table 10-1. PM25 Quarterly Statistics for Year 2 (in ug/m3).
Quarter
1
2
3
4
Number of
Samples
28
29
23
28
* Calculated from available PM2 5 concentrations.
a. Combine the three PM10 samples with the
Total*
332.9
426.0
292.1
369.9
23 PMo s samples
Quarterly Average*
(*,)
11.89
14.69
12.70
13.21
collected in Quarter 3 of
Year 2. This total of 26 samples constitutes a valid quarter because 26 is at least 75%
of 31. Thus, you can use the combined data from this quarter to show you've met the
standard.
b. Calculate the average for Quarter 3 of Year 2.
292.1+(40.1x8)
x = —
23+8
292.1+ 320.8
31
= 19.77 g/nr
c. Calculate the average for Year 2.
11.89+14.69+19.77+13.21
= 14.89 g/m
3
d. Calculate the 3 -year average.
12.84+13.77+12.43
y= =13.01 g/m
3
18
-------
e. Apply the rounding convention. Round the 3-year average annual mean for PM2 5 to
the nearest 0.1, so 13.01 |ug/m3 rounds down to 13.0 |ug/m3.
f. Compare your result to the standard. Because 13.0 ug/m3 is below the annual standard
for PM25 of 15.0 ug/m3, you've met the standard.
Example: How to Substitute Collocated PM10 Data for Missing Data and Use It to
Calculate the 24-Hour PM2 5 Standard
Consider the same example (see Table 10-1) and also assume that
• In Years 1 and 3 of sampling the annual 98th percentile values are 58.7 and 54.3 ug/m3,
respectively.
• In Year 2, the highest three of the 108 measured PM25 concentrations are 75.7, 70.4,
and 50.9 ug/m3.
• You meet the criteria from Question 9. [To show that you meet the 24-hour PM2 5
standard based on the incomplete data, repeat steps b to f immediately below without
using the substituted data. In step c, the number of samples equals 108, n equals 3
from Table 7-1, and the 3rd highest value equals 50.9 ug/m3 without using the
substituted data. Then, in step e, the resulting rounded 3-year average 98th percentile is
55 ug/m3, which meets the 24-hour PM25 standard of 65 ug/m3.]
Follow these steps.
a. Combine the three PM10 samples with the 23 PM2 5 samples collected in Quarter 3 of
Year 2. This total of 26 samples constitutes a valid quarter because 26 is at least 75%
of 31. Thus, you can use the combined data from this quarter to show you've met the
standard. In Year 2 there are 108 measured PM25 concentrations. The combined data
has a total of 111 samples in Year 2.
b. Sort the combined data from highest to lowest. For Year 2, the original three highest
values are 75.7, 70.4, and 50.9 ug/m3. The three substituted PM10 samples are 30, 21,
and 54 ug/m3. Then, the highest four of the 111 samples in the combined data are
75.7, 70.4, 54, and 50.9 ug/m3.
c. Find n from Table 7-1 and find the n111 highest value. There are 111 samples. From
Table 7-1, n equals 3. The 3rd highest value equals 54 jug/m3.
19
-------
d. Calculate the 3-year average of all three values for the 98th percentile.
58.7 + 54 + 54.3 = 55.67 jig/m3.
3
e. Round to the nearest integer for PM2 5i For PM2 5, round decimals 0.5 or greater up
and those less than 0.5 down.
Round 55.67 up to 56 |ug/m3.
f. Compare your result to the standard. 56 is less than 65 ug/m3 so this example meets
the 24-hour standard of 65 ug/m3 for PM25.
Approach No. 2: How you may use maximum observed values to substitute for missing
data
Replace each missing scheduled sampling day in an incomplete quarter with the maximum
observed value from the same site and the same quarter (from any of the three years).
Example: How to Substitute Maximum Observed Values for Missing Data and Use
Them to Calculate the Annual PM25 Standard
Consider the same set of PM25 observations used in the example above (see Table 10-1).
Assume
• You have no collocated data available, so you must substitute for each of the eight
scheduled PM25 sampling days that are missing in Quarter 3 of Year 2.
• Emissions and meteorology of the quarter in question are typical for Quarter 3
throughout the three years.
With these assumptions, you can substitute data from Quarter 3 of all three years to show
you meet the annual PM2 5 standard.
a. Combine eight substituted maximum values with the 23 PM2 5 samples collected in
Quarter 3 of Year 2. Substitute the maximum PM2 5 concentration observed in Quarter
3 in any of the most recent three years for all eight missing samples. The maximum
may occur on one of the 23 non-missing days in Quarter 3 of Year 2, or in Quarter 3
of Year 1 or Year 3.
20
-------
You review Quarter 3 data for all three years, and find the maximum value is 40.1
ug/m3. It occurred in Quarter 3 of Year 2. (You would get the same results if the
maximum for that quarter occurred in Year 1 or Year 3). Substitute 40.1 ug/m3 for
each of the eight missing sampling days scheduled in Year 2, Quarter 3.
b. Calculate the average for Quarter 3 of Year 2.
292.1+(40.1x8)
23+8
292.1+ 320.8
31
c. Calculate the average for Year 2.
= 19.77 g/nr
11.89+14.69+19.77+13.21 , ,
=14.89 g/m3
d. Calculate the 3-year average. Because the averages for Year 1 and Year 3 (from
the previous example) are
^1= 12.84 jig/m3 and ^3 = 12.43 jig/m3,
the three-year average is
12.84+14.89+12.43 , 3
y= =13.39 g/m3
e. Apply the rounding convention and compare your result to the standard. This three-
year average of 13.39 |ug/m3 rounds to 13.4 |ug/m3, which is below the PM2 5 annual
standard. Thus, this example meets the PM2 5 annual standard.
Them to Calculate the 24-Hour PM25 Standard
Example: How to Substitute Maximum Observed Values for Missing Data and Use
Them to Calcul
Continue the same example.
21
-------
a. Combine the eight substituted maximum values with the 23 PM2 5 samples collected in
Quarter 3 of Year 2. In Year 2 there are 108 measured PM25 concentrations. The
combined data has a total of 116 samples in Year 2.
b. Sort the combined data from highest to lowest. For Year 2, the original three highest
values are 75.7, 70.4, and 50.9 ug/m3. The eight substituted maximum values are all
equal to 40.1 ug/m3. Then, the highest three of the 116 samples in the combined data
are also 75.7, 70.4, and 50.9 ug/m3.
c. Find n from Table 7-1 and find the n111 highest value. There are 116 samples. From
Table 7-1, n equals 3. The 3rd highest value equals 50.9 ug/m3.
d. Calculate the 3-year average of all three values for the 98th percentile.
58.7 + 50.9 + 54.3 = 54.63 ug/m3.
3
e. Round to the nearest integer for PM2 5. For PM2 5, round decimals 0.5 or greater up
and those less than 0.5 down.
Round 54.63 up to 55 jug/m3.
f. Compare your result to the standard. 55 is less than 65 ug/m3 so this example meets
the 24-hour standard of 65 ug/m3 for PM25.
11. May I use data from a PM10 Monitor to show that I meet the PM25 standards?
Yes, if the monitor meets PM2 5 siting requirements and the acceptable PM2 5 sampling
schedule, you may treat the PM10 data as if it were PM25 data and compare it to the standards as
in Questions 4 and 6. However, you may not use PM10 data to show that you do not meet the
PM25 standards.
12. May I ignore years with high concentrations if they have incomplete data?
No. For the 24-hour standards, Appendix N says you must include data from years with
quarters which have less than 75% completeness if the resulting annual 98th or 99th percentile
(rounded under the conventions in Question 2) exceeds the standard. For the annual standards,
Appendix N says you must include data from years with less than 75% completeness but at least
11 samples per quarter, if the resulting annual mean (rounded under the conventions in Question
2) exceeds the standard.
22
-------
13. How little data may I use to show I'm NOT meeting the standards?
Generally, you must use data from years meeting the rules for completeness in Table 8-1
above. Plus, Appendix N says you must keep data from years with high values that don't meet
those rules. The minimum data you need to show you're not meeting the standards follows:
• For the 24-hour standards: You must keep any year where the annual percentile
exceeds the standard, even if there was only one measurement in that year and that
single measurement exceeded the standard.
• For the annual PM10 standard: You must keep any year where
the annual mean exceeds the standard, and
there were at least 11 samples in each quarter.
• For the annual PM2 5 standard: For one monitor, the rule is the same as for the annual
PM10 standard. When there are multiple monitors designated for spatial averaging in a
CMZ, you must keep any year where
the spatially averaged annual mean exceeds the standard, and
at least one of the monitors has in each quarter at least 11 samples.
Of course, in all of these cases, data from three years is averaged.
14. May I fill in for missing data to show that I don't meet the annual standards?
No, except in some of the situations described in the answer to Question 15, where the
annual mean exceeds the standard but you don't meet the minimum of 11 samples in a quarter.
15. What if I don't meet the minimum of 11 samples in a quarter? Can I still show I
don't meet the annual standards?
Appendix N says that situations may arise in which there are compelling reasons to retain
years containing quarters which do not meet the minimum number of 11 samples, and the use of
less than complete data is subject to the approval of the Regional Administrator.
If, for example, a site is missing part or all of a quarter, the Regional Administrator may
want to consider examining data from the same quarters in the other two years and/or data from
nearby monitors during the same quarter for all 3 years. Consistent meteorology and emissions
data between the quarter with missing data and the one from which data are substituted would
also be a factor to examine.
23
-------
If, as another example, a site is missing part or all of a quarter, the Regional Administrator
may want to consider substituting 0 (zero) for the missing data to demonstrate that the site has an
annual average that does not meet the standard. The Regional Administrator could also consider
substituting the historically lowest 24-hr concentration observed at the site. Since this is a
compelling argument, the Regional Administrator might consider using this approach for several
quarters with no data.
In any situation, the Regional Administrator could combine these approaches to strengthen
the argument for a determination that the site does not meet the standards.
16. May I show that I meet the standards, or that I don't meet the standards, if I have
only 1 or 2 years of data?
• For the 24-hour PM standards and the annual PM10 standard, you must wait for the
third year of data if your site is still operating.
• For the annual PM2 5 standard, a site may contribute only 1 or 2 years to the spatial
average across a CMZ, as long as in each of the 3 years, at least one site contributes.
For more on this method, see Question 4 of this guidance or Example 1 in Appendix
N.
17. What do I do about a monitor that has stopped monitoring?
The appropriate Regional Office must be notified prior to shutting down a NAMS or
SLAMS site that has exceeded the level of the particulate matter NAAQS. The air quality status
at a NAMS or SLAMS site that has stopped monitoring after two years must be handled on a
case-by-case basis. Factors to consider are the reasons that the site stopped monitoring, the
magnitude of the particulate matter concentrations measured, and the likelihood that additional
monitoring data may be available in future years.
24
-------
Chapter 2
Sampling Frequency
18. If I've collected more samples than were scheduled, may I use all the data to show
I've met or not met the 24-hour standards?
Yes, but the 98th and 99th percentiles are based on the applicable number of samples,
rather than the actual number of samples.
For the 24-hour standards, you won't receive credit for more samples than the maximum
number of scheduled sampling days in a quarter. For each quarter, the applicable number of
samples is the lower of the actual number of samples and the scheduled number of samples. The
percentile will be calculated as in Question 7 using the applicable number of samples for the year
in Table 7-1. This policy ensures the annual 98th or 99th percentiles aren't biased low by extensive
sampling over a short period when values are low. If "extra" samples are collected when values
are high, those samples could contribute to a violation. If this occurs for an uncontrollable or
natural event, see Question 29 for additional data handling considerations.
If you prefer to use the calculations from Questions 5 and 6, then you must modify step c.
To calculate the rank of the 99th percentile for each year: Multiply the applicable number of
samples by 0.99. Take the integer part of the product, add 1, and then add the number of extra
samples beyond the number of scheduled samples. To calculate the rank of the 98th percentile for
each year: Multiply the applicable number of samples by 0.98. Take the integer part of the
product, add 1, and then add the number of extra samples beyond the number of scheduled
samples.
The answers to this question and to Question 20 show that the calculations for the 24-
hour and annual standards use all the samples that were collected. Therefore, all measured sample
concentrations must be reported to AIRS, including any extra samples beyond the number of
scheduled samples.
25
-------
Example:
Assume that in Year 1 the scheduled sampling frequency for PM2 5 was once every three
days. Assume that the numbers of samples collected and scheduled were as shown in
Table 18-1. Assume that the four highest values in the year (among all 228 samples) were
66.7, 62.2, 62.1, and 62.1 ug/m3.
Table 18-1. Actual and Applicable Number of Samples for Year 1.
Quarter Actual Number of Scheduled Number Applicable Number
Samples of Samples of Samples =
Minimum (Actual,
Scheduled)
1
2
3
4
Total
90
89
25
24
228
30
30
30
31
121
30
30
25
24
109
Using Table 7-1
a. Calculate the applicable number of samples. For each quarter, the applicable number of
samples is the lower of the scheduled number of samples and the actual number of
samples. For each year, the applicable number of samples is the total of the applicable
number of samples in each quarter. From Table 18-1, in quarters 1 and 2 the scheduled
number of samples was 30, but you collected 90 samples in quarter 1 and 89 samples
in quarter 2. For each of those quarters you will only be credited with 30 samples. For
quarters 3 and 4 you collected less than the scheduled number of samples, so the
applicable number of samples is the actual number of samples. The total number of
applicable samples for Year 1 is 109.
b. Find n from Table 7-1 and the n111 highest value from the sorted values. For PM10 you
calculate the 99th percentile. For PM2 5 you calculate the 98th percentile. The total
number of applicable samples is 109. The value of n from Table 7-1 is 3. The 98th
percentile for Year 1 is the 3rd highest value from all 228 samples, i.e. 62.1 jug/m3.
26
-------
Using the Calculated Rank
a. Calculate the applicable number of samples. As before, from Table 18-1 the total
number of applicable samples for Year 1 is 109.
b. Multiply the number of applicable samples by 0.98 or 0.99. For PM10 you multiply by
0.99. For PM2 5 you multiply by 0.98.
109 x 0.98 = 106.82.
c. Take the integer part of the product, add 1. and add the number of extra samples
beyond the number of scheduled samples. From Table 18-1, the number of extra
samples was 228 actual samples minus 109 applicable samples, i.e., 119. (You get the
same number of extra samples if you add the 60 extra samples in quarter 1 to the 59
extra samples in Quarter 2.)
106+ 1 + 119 = 226.
The value with rank 226 is the 226th lowest value, which equals 62.1 jug/m3.
19. How do I calculate the 98th and 99th percentiles when my sampling frequencies are
seasonal?
You calculate the percentiles by finding the smallest measured concentration, x, that
makes W(x) > .98 (or .99) where:
_ _ _ _
j j High \ j j Low
dHigh +dLOW
W(x\-
rr \A ) —
dHigh = number of calendar days in the "High" season,
dLow = number of calendar days in the "Low" season, (dHigh + dLow = days in a year),
number of samples in season a that are < x
Fa(x) = 7 7 r~- '
number oj samples in season a
such that a can be either High or Low,
x is the measured concentration, and
igh + dLow) and dLow/(dm h + dLow) are constant and are called seasonal "weights."
27
-------
The following example shows how you find the percentile using this formula. Suppose you
expect high PM2 5 values during January, February, November, and December, but you expect low
PM concentrations the rest of the year. You now know that dHigh =120 and dLow = 245, so you
can calculate the seasonal weights as 0.329 and 0.671.
You schedule samples to be taken every day during the High months and once every 3
days the rest of the year. In the year in question, suppose you actually got 105 samples out of a
possible 120 in the High season and you got 70 samples out of a possible 82 for the Low season.
Rank or sort the samples from each season. The highest concentrations have the highest
ranks. Suppose the results are as shown in Table 19-1.
Table 19-1. Ranking of High Values in High and Low Seasons.
Concentrations in ug/m3.
High!
Rank of
Samples
105
104
103
102
101
100
99
98
Season
X
Concentration
74.3
71.2
70.0
67.5
64.8
59.2
55.1
50.0
Low
Rank of
Samples
70
69
68
67
66
65
64
63
Season
X
Concentration
60.9
58.7
43.6
35.7
28.5
25.0
20.3
20.1
Ranking will allow you to calculate FHigh(x) and FLow(x). Looking at the first line in the table, you
105
calculate FHigh(74. 3) as = 1, because 105 of 105 samples are • 74.3. Similarly, calculate
70 104
F£ow(60.9)as 70 = 1. Likewise, calculate FHigh(7 1.2) as 105 = 0.990 and F£ow(58.7) as
69
70 =
0.986. Your results are in Table 19-2, below.
28
-------
Table 19-2. Distribution Functions of High Values in High and Low Seasons.
Concentrations in ug/m3.
High Season
Rank of x
Samples Concentration FHigh(x)
Low Season
Rank of x
Samples Concentration FLow(x)
105
104
103
102
101
100
99
98
74.3
71.2
70.0
67.5
64.8
59.2
55.1
50.0
1.000
0.990
0.981
0.971
0.962
0.952
0.943
0.933
70
69
68
67
66
65
64
63
60.9
58.7
43.6
35.7
28.5
25.0
20.3
20.1
1.000
0.986
0.971
0.957
0.943
0.929
0.914
0.900
Next, you rank or sort all the values in a year (both seasons) to help you calculate W(x). If
a concentration value wasn't measured in the Low (or High) season, you find the value ofFLow
(or FHigh) from the next lowest measured concentration. For example, 74.3 wasn't measured in
the Low season; the next lowest measured concentration in that season was 60.9. FLow(74.3) =
FLow(60.9) = 1.000. As another example, 7^(58.7) = 7^(55.1) = 0.943. Table 19-3 shows the
results with the calculated values of W(x).
Table 19-3. Calculating Weighted Percentiles for Two Seasons.
Concentrations in ug/m3.
Concentration
74.3
71.2
70.0
67.5
64.8
60.9
59.2
58.7
55.1
50.0
Seasonal
Weight
0.329
0.329
0.329
0.329
0.329
0.329
0.329
0.329
0.329
0.329
Fffiril(x)
1.000
0.990
0.981
0.971
0.962
0.952
0.952
0.943
0.943
0.933
Seasonal
Weight
0.671
0.671
0.671
0.671
0.671
0.671
0.671
0.671
0.671
0.671
FLow(x)
1.000
1.000
1.000
1.000
1.000
1.000
0.986
0.986
0.971
0.971
W(x)
1.000
0.997
0.994
0.990
0.988
0.984
0.975
0.972
0.962
0.959
29
-------
Using the formula for W(x), calculate W(743) as 0.329x1.000 + 0.671x1.000 = 1.000.
You calculate W(71.2), W(70.G), and so on, in the same way. Now you can find the smallest x for
which W(x) is > 0.98 by moving up the right hand column until you find the first number greater
than but not equal to 0.98 (0.984). This value corresponds to the concentration of 60.9 ug/m3,
which is the 98th percentile for this year.
This method has two advantages. First, the seasons don't have to be contiguous months
and can even be planned episodic monitoring with concurrence from the appropriate regional
administrator. Second, if you set one of the seasons to have 0 days, the method will calculate a
percentile identical to one calculated using the procedures in Appendix N.
20. How do I compute quarterly averages when parts of the quarter are sampled at
different frequencies?
You should use the same calculation as in Question 3 (for PM10) or Question 4 (for
PM25). The quarterly average is the arithmetic mean of the measured concentrations in that
quarter, so you add all the sampled values and divide by the number of samples. [Note that
"extra" samples beyond the scheduled sampling days are included when you calculate the annual
mean.] The arithmetic mean should represent the true average concentrations during that quarter
reasonably well, even if the scheduled sampling frequencies change.
21. Under what circumstances may I reduce the required sampling frequency at a site
for a year or season?
Review our Interim Guidelines for Granting Schedule Exemptions for PM10 Monitoring,
(attachment to Waivers for PM10 Sampling Frequency memorandum, from William F. Hunt, Jr. to
EPA Regional Office Air Program Directors, December 2, 1997) which describes methods and
decision rules for shifting certain sites from every-day and one-in-three-day sampling for PM10
schedules to one-in-six-day sampling. You must show there is little or no chance of exceeding the
daily standard. A brief summary follows. [Note that EPA is developing similar guidance for PM25
sampling frequencies. For initial guidance on PM25 sampling frequencies see Guidelines for
Granting Exceptions for Daily PM25 Monitoring memorandum, from William F. Hunt, Jr. to EPA
Regional Office Air Program Directors, April 9, 1998. For information on PM25 sampling
frequencies for calendar year 1999 only, see PM25 Site Types and Sampling Frequency During
CY-99 memorandum, from William F. Hunt, Jr. to EPA Regional Office Air Program Directors,
May 18, 1998.]
30
-------
Requesting a Year-Round Exemption for PM1C
The Regional Administrator may grant a year-round exemption that allows you to sample
only once in six days. To request this year-round exemption, you must, at least, show one of the
following:
• The mean 99th percentile for the most recent three years is statistically significantly
lower than the 24-hour standard. Apply a statistical t-test at the 10 percent significance
level.
• The annual standard is the controlling standard. Show that the three-year average
annual mean divided by the annual standard level is greater than the three-year average
annual 99th percentile divided by the 24-hour standard level.
• The 99th percentile estimated using an exponential distribution is lower than the 24-
hour standard. Fit an exponential distribution to the top 25 percent of the
concentrations measured over the most recent three years. Show that the fitted 99th
percentile is below the daily standard level.
Requesting a Seasonal Exemption for PM,
JO
The Regional Administrator may grant a seasonal exemption to allow your site to sample
once in three days or once in six days instead of every day, but you must show
• The site has met Appendix N's minimum requirements for data completeness.
• Every value during that season for the past two years was below the level of the 24-
hour standard.
22. May I use a correlated acceptable continuous (CAC) monitor to reduce my sampling
frequency?
You may use the CAC monitor to reduce the required PM2 5 sampling frequency from
every day to once in three days at a core monitor under the following conditions, as given in 40
CFR58.13(f).
(1) You must show the CAC monitoring data is correlated with PM2 5 data from a
collocated monitor using a Federal Reference Method or Equivalent Method.
(2) In a Priority 1 PM monitoring area, you must collect at least two complete years of
PM2 5 monitoring data using a Federal Reference Method or Equivalent Method before
31
-------
being considered for permission to reduce the sampling frequency. A Priority 1 PM
monitoring area, as defined in 40 CFR Part 58(f)(2):
• Is a Metropolitan Statistical Area with a population of 1 million or more
• Has an average annual 98th percentile PM2 5 greater than or equal to 80% of
the 24-hour standard level for PM25.
23. If I've missed a scheduled sample, may I make it up?
EPA publishes schedules for each calendar year that give the dates on which you should
collect particulate matter samples at State and Local Air Monitoring Stations (SLAMS) or other
monitoring sites. EPA publishes these schedules for one-in-three-day sampling and for one-in-six-
day sampling. Your scheduled sample might be missed or invalidated for various reasons,
including sampler malfunctions and power outages. EPA encourages but does not require you to
make up missing scheduled samples by collecting samples on other days, referred to as
replacement sampling days. You may want to use these make-up samples to help meet the
requirements for data completeness.
The EPA Regional Office is responsible for ensuring that uses of make-up samples avoid
bias by being consistent with this and other EPA guidance and with the data validation procedures
in the State's Quality Assurance Project Plan (QAPP). If you decide to use make-up samples,
You must
• Make up the sample no later than one week after the scheduled sampling day
You should
• Discuss the use of make-up sampling in the QAPP
• Report the reason why the scheduled sample was missing or invalidated to the
Regional Office
• Include the reason why the scheduled sample was missing or invalidated in the State's
internal database
You must not
• Intentionally invalidate or fail to collect scheduled samples
32
-------
• Select whether a missing scheduled sample will be made up based on the level of the
expected concentrations
• Select replacement sampling days based on the level of the expected concentrations
You should not
• Make-up 6 or more scheduled samples in any calendar quarter
• Consistently have to use make-up samples to meet requirements for data completeness
The preferred approach is to make up all missed or invalid scheduled samples whenever
practicable within the one week time limit. If the QAPP does not include such a plan, the EPA
Regional Office may not want to approve make-up samples on days that are not representative of
the corresponding scheduled sampling day. For example, this might include make-ups after a
severe stagnation or after a brief period of extremely high emissions.
The replacement sampling day should be chosen as follows:
Preferred Approach: Sample before the next scheduled sampling day
• If your monitor samples once every six days, and if other sites in the same network
sample once every three days, choose the next scheduled one-in-three-day sampling
day. This gives additional spatial resolution and is likely to be most convenient.
• Otherwise, choose the earliest possible day before the next scheduled sampling day.
This increases the probability that the replacement day has similar meteorological
conditions.
Alternative Approach: Sample exactly one week later
• This approach reduces potential biases due to the variation of emissions patterns with
the day of the week.
24. How do I use scheduled samples with make-up and other non-scheduled samples to
show that I meet or don't meet the standards?
Treat make-up samples and other non-scheduled samples exactly the same as the
scheduled samples for the calculations showing that you meet or don't meet the particulate matter
standards:
33
-------
The comparisons with the standards use all valid 24-hour samples except for data
affected by natural or uncontrollable events and excluded under Appendix N
provisions (see Question 29). The actual number of samples (in Question 18) is the
total number of
samples on scheduled sampling days
make-up samples on non-scheduled sampling days, as described in the
answer to Question 23
samples substituted for missing data as described in the answers to
Questions 10 and 15
any other non-scheduled samples
For quarters with one-in-three-day or one-in-six-day sampling, the scheduled number
of samples is the number of sampling days in the published schedule for the sampling
frequency
For quarters with every day sampling, the scheduled number of samples is the number
of days in the calendar quarter
Data completeness percentages are given by the ratios of the actual number of samples
to the scheduled number of samples
34
-------
Chapter 3
Monitoring Issues
25. Which monitors do I compare to which standards?
• Is the PM monitoring data collected using a Federal Reference Method or Federal
Equivalent Method?
• Does the monitored data meet the Part 58 requirements for data quality assurance,
sampling frequency, monitor siting, and, if necessary, spatial averaging?
If so, you may compare that data with the applicable PM10 or PM2 5 standards, as
described below. If not, you generally may not compare that data to the PM standards in order to
decide whether or not you've met the standards. Part 58 and its appendices discuss some
exceptions.
Handling PM1C Data:
Compare your PM10 data with the annual and 24-hour standards for PM10 if you collect PM10 data
• At State and Local Air Monitoring Stations (SLAMS) or at National Air Monitoring
Stations (NAMS)—a subset of the SLAMS network.
• At a Special Purpose Monitor (SPM) that the State intends to use to show that the
PM10 standards are met (i.e., they are included in the PM Monitoring Network
Description).
35
-------
Handling PM^ < data:
•2.5-±
Follow Table 25-1 below to determine how to compare your PM25 data to the standards.
Table 25-1. How to compare your PM25 data to the standards.
If your monitor
is a ...
SLAMS or
NAMS or SPM*
And if the
location is ...
population-
oriented
And if the
spatial scale
(see Part 58,
App D, 2.8) is
neighborhood,
urban, or
regional
And if the
location
represents
Then compare with
these PM2 5 standards
annual and 24-hour
SLAMS or population-
NAMS or SPM* oriented
micro or many
middle scale locations in
the area
annual and 24-hour
SLAMS or
NAMS or SPM*
SLAMS or
NAMS or SPM*
population-
oriented and a
local hot spot
a background
site
micro or a unique
middle scale area
neighborhood,
urban, or
regional
24-hour only
annual and 24-hour, but
address failures to meet
the NAAQS at these
sites as part of the PM
implementation
program
*Special purpose monitors that meet Part 58 requirements will be exempt from NAAQS
comparisons with the PM25 NAAQS for the first two calendar years of their operation to
encourage PM25 monitoring initially (see 40 CFR 58.14(b)).
36
-------
Table 25-1. How to compare your PM25 data to the standards.
If your monitor
is a...
And if the
location is ...
And if the
spatial scale And if the
(see Part 58, location
App D, 2.8) is represents
Then compare with
these PM2 5 standards
SLAMS or
NAMS or SPM*
a transport site
urban or
regional
annual and 24-hour, but
address failures to meet
the NAAQS at these
sites as part of the PM
implementation
program
SLAMS or
NAMS or SPM*
not population
oriented, and
not a
background or
transport site
neither standard
SPM with at
most two years
of data
population-
oriented
neither standard, but if
your data violates the
NAAQS, report the
results in the State's
annual monitoring
report and consider
whether these sites
should become SLAMS
*Special purpose monitors that meet Part 58 requirements will be exempt from NAAQS
comparisons with the PM25 NAAQS for the first two calendar years of their operation to
encourage PM25 monitoring initially (see 40 CFR 58.14(b)).
26. What are "community monitoring zones" (CMZs) and how do I decide which sites
to include in a spatial average?
A community monitoring zone (CMZ) is an area that contains one or more sites with
relatively similar concentrations of PM25 that are impacted by similar emission sources. The states
define CMZs in their PM Monitoring Network Description, based on 40 CFR Part 58. For the
PM2 5 annual standard, the states may spatially average the annual means over the sites in the
37
-------
CMZ (each site-year with data is given equal weight in the computation of the annual spatial
average). They compare the average of the three annual spatial averages to the level of the
standard. CMZs are defined to meet the following three conditions, only the first of which is
required (see Part 58, Appendix D, §2.8.1.6.1):
• The annual average concentrations at every site in the CMZ must be no less than 80%
or more than 120% of the annual spatial average [required]
• The 24-hour average concentrations should have strong correlations (a correlation
coefficient of 0.6 or greater) [recommended]
• Emissions from the same source or types of sources of PM2 5 affect the entire CMZ
[recommended]
State and local agencies that regulate air quality must develop descriptions for the PM2 5
network. In those descriptions, they must group sites into CMZs if they choose to do spatial
averaging. If the CMZ's data does not meet the three conditions above, then states may need to
redefine the CMZs before comparing the data to the NAAQS. To decide which sites to include in
the spatial average, follow these steps:
a. Calculate the annual mean for each year and site. Average the four quarterly means.
Don't use data for a year and site that does not meet the criteria for data completeness
of the annual mean.
b. Calculate the spatially averaged annual mean for each year. Average the annual means
across the designated sites in the CMZ.
c. Check each year to see whether all sites' annual means are greater than or equal to 80
percent and are less than or equal to 120 percent of the spatially averaged annual mean
for that year.
• YES. The annual averages meet the first condition for a CMZ. Check the other
two conditions.
• NO. The annual averages do not meet the first condition for a CMZ. Redefine the
CMZ.
d. Calculate the spatial correlations between the 24-hour average concentrations.
Compute the Pearson correlation coefficients between the 24-hour average
concentrations at each pair of sites for each year. Suppose the n pairs of
concentrations at sites X and Y measured on the same day in a year are (X1? Yj),(X2,
38
-------
Y2), (X3, Y3), . . . (X^ Yn). Then the Pearson correlation coefficient, rxy, for sites X
and Y in a year is given by the formula:
n n
^ ^i±l n
;=i n
( ( » V^l
• M
i^\ ' n
( ( " }2}
y 72 v ,-=1 )
&*' n
J \
e. Check to see whether all the spatial correlations between the 24-hour average
concentrations are at least 0.6.
• YES. The 24-hour average concentrations meet the second condition for a CMZ.
Check the third condition.
• NO. The 24-hour average concentrations do not meet the second condition for a
CMZ. Consider redefining the CMZ.
f. Check to see if emissions from the same source or source-types affect the CMZ. For
help with this, see the Guidance for Network Design and Optimum Site Exposure for
PM25 andPM10, December 15, 1997, Section 5.5.
• YES. The emissions meet the third condition for a CMZ.
• NO. The emissions do not meet the third condition for a CMZ. Consider
redefining the CMZ.
Example: How to check site annual averages across a CMZ for consistency
For this example, examine Table 26-1 below. Assume the CMZ contains 5 sites, numbered
1 to 5. Assume each site and quarter is at least 75% complete.
a. Calculate the annual means. The annual means are as shown in Table 26-1.
39
-------
b. Calculate the spatially averaged annual mean. The spatially averaged annual means
(10.80, 14.08, and 13.99 |ug/m3) for each year are shown in the row "Spatial mean."
For example, in Year 1, the spatially averaged annual mean is
10.38 + 13.71 + 10.62 + 10.50 + 8.79 = 10.80 jig/m3.
5
Check whether each site's annual means are between 80 and 120 percent of the
spatially averaged annual mean. The spatially averaged annual means multiplied by 0.8
and by 1.2 are shown in the rows "spatial mean x 80%" and "spatial mean x 120%."
For example, in Year 1, 80% of the spatially averaged annual mean is 10.80 x 0.8 =
8.64 jug/m3. In year 2, all five sites have annual means between 80 and 120 percent of
the spatially averaged annual mean (i.e., between 11.26 and 16.90). However, in Year
1, site 2 is more than 120% of the spatially averaged annual mean (13.71 > 12.96). In
Year 3, site 1 is less than 80% of the spatially averaged annual mean (9.71 < 11.19). In
Year 3, site 2 is more than 120% of the spatially averaged annual mean (20.36 >
16.79). Thus, the first condition for a CMZ is not met for this example. Redefine the
CMZ before evaluating whether or not the NAAQS is met.
Table 26-1. Example of Consistency Checks for the CMZ's Annual Average. (PM25
concentrations in ug/m3).
Site
1
2
3
4
5
Spatial mean
Spatial mean x 80%
Spatial mean x 120%
Year 1
10.38
13.71
10.62
10.50
8.79
10.80
8.64
12.96
Year!
13.21
16.32
12.50
12.51
15.84
14.08
11.26
16.90
Year3
9.71
20.36
16.11
11.66
12.10
13.99
11.19
16.79
40
-------
Example: How to check spatial correlations between 24-hour average concentrations
across a CMZ for consistency
d. Calculate the spatial correlations between the 24-hour average concentrations.
Suppose that a CMZ has 3 sites, numbered 1, 2, and 3. Table 26-2 shows correlation
coefficients for one year calculated by the equation given in step 4 above:
Table 26-2. Example of Consistency Checks for the CMZ's Spatial Correlations.
Correlations Site 1
Site 1 X
Site!
Site3
Site!
0.68
X
Site3
0.71
0.63
X
f.
Check to see whether all the spatial correlations between the 24-hour average
concentrations are at least 0.6. Since the correlations are all greater than 0.6, sites 1, 2,
and 3 in this year meet the second condition for a CMZ. To complete the test of the
condition for strong spatial correlations, check whether all other correlations between
pairs of sites in the rest of the years are all greater than or equal to 0.6.
Check to see if emissions from the same source or source-types affect the CMZ.
Assume that using the guidance provided in the document sited in step 6 above shows
that PM2 5 emissions are from the same source type throughout the CMZ. The third
condition for a CMZ is met.
27. If a monitor is reassigned to a different CMZ during the three-year period, which
CMZ assignment should I use to calculate the spatial average?
Apply the most recent definition of the CMZ to all three years.
Example:
Suppose that in Year 1, sites 1, 2, and 3 formed one CMZ, and site 4 formed another. In
Year 3, the state or local agency reassigns site 2 to the second CMZ. Calculate whether you meet
the standard based on having sites 1 and 3 in one CMZ and sites 2 and 4 in the other CMZ. Use
all valid annual means for the three-year period. Apply the three conditions for a CMZ described
in the answer to Question 26 to each of these CMZs, so you can decide whether you need to
subdivide one or both CMZs.
41
-------
28. How many hourly values make up a valid 24-hour average for a continuous
monitor?"
• If measurements are available for at least 75 percent (18 or more) of the hours during
the 24-hour period, the 24-hour average is valid. You'll compute it by summing the
hourly concentrations and dividing by the number of hourly measurements.
• If measurements are available for less than 75 percent (17 or less) of the hours during
the 24-hour period, you must treat the 24-hour average as invalid, unless the
concentrations are too high to be ignored, as determined by the following calculation:
a. Compute the lower bound. Substitute zero for each missing hour and
compute the average over all 24 hours.
b. Round the lower bound. Round the lower bound to the nearest 1 ug/m3 for
PM2 5 concentrations (round up decimals 0.5 or greater). Round the lower
bound to the nearest 10 ug/m3 for PM10 concentrations (round up integers
of 5 or greater).
c. Compare the rounded lower bound to the standard. If the rounded lower
bound exceeds the level of the 24-hour standard, use the unrounded lower
bound as the valid 24-hour average. Otherwise, treat the 24-hour average
as invalid.
Example: How to calculate 24-hour averages from hourly concentrations
Suppose the hourly PM10 measurements are as follows, where "M" denotes a missing
value: M, M, M, M, M, M, 112, M, 109, 95, 110, 113, 140, 150, 160, 160, 160, 165, 160, 170,
140, 130, 135, 140 ug/m3. You have 17 hourly concentrations and 7 missing values. Because
there are fewer than 18 valid hours, you must calculate the lower bound. The lower bound equals
(0 + 0 + 0 + 0 + 0 + 0+112+ 0+109 + ... + 135 + 140) 724 = 97.9 ug/m3. Rounding to the
nearest 10 gives a rounded value of 100 ug/m3. This value is lower than the level of the 24-hour
standard, 150 ug/m3, so the 24-hour average for this day is invalid.
"Note: See 40 CFR Part 50 Appendix J (for PM10) and Appendix L (for PM2 5) for the
rules on determining how many hours make a valid 24-hour average for an integrated 24-hour
measurement.
42
-------
Chapter 4
Miscellaneous Issues
29. How do I handle data from uncontrollable or natural events?
First, you should review the following three documents for guidance on how to address
PM data that are affected by uncontrollable or natural events.
• Guideline on the Identification and Use of Air Quality Data Affected by Exceptional
Events, EPA 450/4-86-007, July, 1986.
• Areas Affected by PM-10 Natural Events, Memorandum from Mary D. Nichols,
Assistant Administrator for Air and Radiation, to EPA Regional Office Air Program
Directors, May 30, 1996.
• Interim Air Quality Policy on Wildland and Prescribed Fires, Memorandum from
Richard D. Wilson, Acting Assistant Administrator for Air and Radiation, to EPA
Regional Administrators, May 15, 1998.
The Guideline on the Identification and Use of Air Quality Data Affected by Exceptional
Events addresses uncontrollable events such as structural fires, high pollen count, chemical spills
and industrial accidents, and activities that temporarily affect a nearby monitor. The Natural
Events Policy for the Paniculate Matter National Ambient Air Quality Standards addresses the
treatment of data that are affected by volcanic and seismic activities, unwanted wildland fires
(wildfires), and high wind events. The Interim Air Quality Policy on Wildland and Prescribed
Fires addresses the treatment of air quality data that are affected by wildland and prescribed fires
that are managed to achieve resource benefits. Actions to be taken depend upon whether the
wildland and prescribed fires managed for resource benefits significantly contribute to violations
of the PM NAAQS; the answer to Question 30 shows you how to decide this.
Appendix N generally sets forth what is needed to determine whether the PM standards
are met, based on three consecutive, complete years of air quality data. Section l.O(b) of
Appendix N allows EPA to give special consideration to data affected by uncontrollable or natural
events. While all valid ambient air quality data should be submitted to the EPA Aerometric
Information Retrieval System (AIRS), appendix N provides that in some cases it may be
appropriate for you to exclude such data from the calculations because they could result in
inappropriate values to compare with the levels of the PM standards. In other cases, you should
use all the data to calculate the comparison with the standards and you should then follow the
regulatory response determined to be appropriate by EPA, which could include approaches
43
-------
outlined in the Natural Events and Interim Air Quality policies. The appropriate Regional
Administrator is responsible for approving the exclusion, adjustment, or retention of data affected
by uncontrollable or natural events.
30. How do I determine whether wildland and prescribed fires managed for resource
benefits significantly contribute to violations of the PM2 5 or PM10 NAAQS?
The Interim Air Quality Policy on Wildland and Prescribed Fires addresses the flagging
(in section VII.B) and treatment of air quality data that are affected by wildland and prescribed
fires that are managed to achieve resource benefits. Actions to be taken depend upon whether the
wildland and prescribed fires managed for resource benefits significantly contribute to violations
of the PM NAAQS. You may consider fires managed for resource benefits to have significantly
contributed to a violation of the PM25 or PM10 NAAQS if the corresponding standard was not
met and if
• For the 24-hour standards: 25 percent or more of all the PM concentrations that are
above the level of the standards have been flagged as being due to fire impacts.
• For the annual standards: the sum of all measured PM concentrations flagged as being
due to fire impacts, divided by the total number of sample days (fire days plus non-fire
days) is greater than or equal to 25 percent of the annual standard.
Follow the calculations in these examples, which also show how you should apply the
rounding conventions:
Example: How to Decide if Fires Managed for Resource Benefits Have Significantly
Contributed to a Violation of the 24-hour PM,C NAAQS.
Steps a to e of this example follow the steps given in Question 7 to decide if the site meets
or does not meet the 24-hour standard for PM10 If the site does not meet the standard, follow
steps f and g to decide whether fires managed for resource benefits have significantly contributed
to a violation of the PM10 NAAQS.
a. Sort all of the data values collected in any given year from highest to lowest. Consider
the set of PM10 observations given in Table 30-1. The values have been sorted from
highest to lowest. Assume 111 values were measured in Year 1, 99 values in Year 2,
and 101 values in Year 3. Assume one of the values (in Year 2) has been flagged as
being due to fire impacts, as shown by the superscript f.
44
-------
Table 30-1. Maxima of Sample PM10 Values from One Site for a Three-Year Period.
Year 1 Year 2 Year 3
111 values 99 values 101 values
n IIth highest value n"1 highest value n"1 highest value
(ug/m3) (ug/m3) (ug/m3)
1
2
3
4
5
6
7
8
9
•
170
130
128
120
95
90
90
88
86
•
203 f
155
152
143
133
120
100
97
90
•
175
147
144
140
133
120
100
99
99
•
f Concentration flagged as being due to fire impacts.
b. Find n from Table 7-1 and find the n111 highest value from Table 30-1. The 99th
percentile value for Year 1 is the 2nd highest value (130 ug/m3), for Year 2 is the
highest value (203 ug/m3), and for Year 3 is the 2nd highest value (147 ug/m3).
c. Calculate the 3-year average of all three values for the 99th percentile.
130 + 203 + 147 = 160.0 jig/m3.
3
d. Round to the nearest multiple of 10 Round integers of 5 or greater up and those less
than 5 down. Round 160.0 to 160 |ug/m3.
e. Compare your result to the standard. 160 is greater than 150 ug/m3 so this example
does not meet the 24-hour standard of 150 ug/m3 for PM10.
f. Count the number of rounded concentrations that exceed the standard (exceedances).
For the 24-hour standard for PM10, concentrations are rounded to the nearest 10
before being compared to the standard of 150 ug/m3. Therefore the exceedances are
any concentrations that are greater than or equal to 155 ug/m3. Table 30-1 shows 4
exceedances (in bold).
45
-------
g. Calculate the percentage of flagged concentrations among the exceedances. lout of 4
exceedances are flagged as being due to fire impacts, which is 25 percent. Since 25
percent or more of the exceedances standard have been flagged as due to fire impacts,
fires have caused or significantly contributed to the 24-hour PM10 standard being
violated in this area.
Example: How to Decide if Fires Managed for Resource Benefits Have Significantly
Contributed to a Violation of the 24-hour PM2 5 NAAQS.
Steps a to e of this example follow the steps given in Question 7 to decide if the site meets
or does not meet the 24-hour standard for PM25 If the site does not meet the standard, follow
steps f and g to decide whether fires managed for resource benefits have significantly contributed
to a violation of the PM25 NAAQS.
a. Sort all of the data values collected in any given year from highest to lowest. Consider
the set of PM25 observations given in Table 30-2. The values have been sorted from
highest to lowest. Assume 281 values were measured in Year 1, 304 values in Year 2,
and 296 values in Year 3. Assume 13 values have been flagged as being due to fire
impacts, as shown by the superscripts f.
Table 30-2. Maxima of Sample PM2 5 Values from One Site for a Three-Year Period.
Year 1 Year 2 Year 3
281 values 304 values 296 values
n n411 highest value n"1 highest value n"1 highest value
(ug/m3) (ug/m3) (ug/m3)
1
2
3
4
5
6
7
8
9
•
64.9 f
64.6 f
64.5 f
64.4
64.2
64.0
57.9
54.0
54.0
•
72.0 f
70.0 f
68.4 f
67.8 f
67.2 f
66.5
65.5
65.4
54.3
•
73.0 f
71.0 f
70.5 f
70.0 f
69.8 f
68.4
68.4
66.0
62.7
•
f Concentration flagged as being due to fire impacts.
46
-------
b. Find n from Table 7-1 and find the IIth highest value from Table 30-2. The 98th
percentile value for Year 1 is the 6th highest value (64.0 ug/m3), for Year 2 is the 7th
highest value (65.5 ug/m3), and for Year 3 is the 6th highest value (68.4 ug/m3).
c. Calculate the 3-year average of all three values for the 98th percentile.
64.0 + 65.5 + 68.4 = 65.97 jig/m3.
3
d. Round to the nearest integer Round decimals of 0.5 or greater up and those less than
0.5 down. Round 65.97 to 66 |ug/m3.
e. Compare your result to the standard. 66 is greater than 65 ug/m3 so this example does
not meet the 24-hour standard of 65 ug/m3 for PM25.
f. Count the number of rounded concentrations that exceed the standard (exceedances).
For the 24-hour standard for PM2 5, concentrations are rounded to the nearest integer
before being compared to the standard of 65 ug/m3. Therefore the exceedances are
any concentrations that are greater than or equal to 65.5 ug/m3. Table 30-2 shows 15
exceedances (in bold).
g. Calculate the percentage of flagged concentrations among the exceedances. 10 out of
15 exceedances are flagged as being due to fire impacts, which is 66.7 percent. Since
25 percent or more of the exceedances standard have been flagged as due to fire
impacts, fires have caused or significantly contributed to the 24-hour PM2 5 standard
being violated in this area.
Example: How to Decide if Fires Managed for Resource Benefits Have Significantly
Contributed to a Violation of the Annual PM1C NAAQS.
This example uses the same data as the example given in Table 30-1.
a. Use the calculations given in Chapter 1 to decide if the site meets or does not meet the
annual standard for PM10. Assume for this example that the site does not meet the
annual standard for PM10.
b. Calculate the sum of all measured PM10 concentrations flagged as being due to fire
impacts. From Table 30-1, only one concentration is flagged, and this sum equals 203
ug/m3.
47
-------
c. Divide by the number of sample values. Include fire days and non-fire days. For this
example, the ratio equals
d. Compare the result to 12. 25 percent of the annual standard for PM10 equals 12 ug/m3
(after rounding to the nearest integer). Since 0.65 is less than (and not equal to) 12
ug/m3, fires have not caused or significantly contributed to the annual PM10 standard
being violated in this area.
Example: How to Decide if Fires Managed for Resource Benefits Have Significantly
Contributed to a Violation of the Annual PM2 5 NAAQS.
This example uses the same data as the example given in Table 30-2. Assume that the area
has not been designated for spatial averaging and has a single site.
a. Use the calculations given in Chapter 1 to decide if the area meets or does not meet
the annual standard for PM2 5i Assume for this example that the area does not meet the
annual standard for PM2 5.
b. Calculate the sum of all measured PM2 5 concentrations flagged as being due to fire
impacts. If the area has been designated for spatial averaging, sum across all flagged
days and across all sites in the CMZ. From Table 30-2, 13 concentrations were
flagged, and their sum equals
64.5 + 64.6 + 64.9 + 67.2 + 67.8 + 68.4 + 70.0 + 72.0 + 69.8 + 70.0 + 70.5 + 71.0 +
73.0 = 893.7 jig/m3.
c. Divide by the number of sample values. Include fire days and non-fire days. If the area
has been designated for spatial averaging, sum across all sites in the CMZ. For this
example, the ratio equals
893J 1.01 g/m'
281+304+296
d. Compare the result to 4. 25 percent of the annual standard for PM2 5 equals 4 ug/m3
(after rounding to the nearest integer). Since 1.01 is less than (and not equal to) 4
48
-------
ug/m3, fires have not caused or significantly contributed to the annual PM2 5 standard
being violated in this area.
49
------- |