20220110_Wy_DAQ_PM2.5_Powder_River_Report

Miller, David
From: Cara Keslar <[e-mail]>
Sent: Monday, January 10, 2022 1:23 PM
To: Miller, David
Subject: [External] Re: PM 2.5 from mines
Attachments: 20200810_PRB_Buckskin_&_Black Thunder_Close_Out_Summary.pdf
CAUTION: External email. Do not click links or open attachments unless you verify. Send all suspicious email as an attachment to
Report Spam.
Hi David,
I am not exactly sure which studies you are referring to. We collected data at the Black Thunder Mine, Belle
Ayr Mine, Buckskin Mine, and Antelope Mine from the late 1990's until 2018/2020. The AQD uses the
Antelope Mine station as the background PM2.5 concentration for the PRB ( AQS ID 56-009-0009; you could
ask one of your ambient monitoring people to pull the AQS quicklook report for 2018-2020; that would give
you the annual and 24 hour maximum values)
You can also query data from AQS at the EPA's Air Data website hns://www.epa.gov/outdoor-air-quality-data
I have also attached our station closeout memo for the Buckskin and Black Thunder stations, that has
information on trends and important info on what years were severely impacted by wildfire.
You may also view permits and emission inventories from coal mines at AQD's OpenAir website:
OOpenAir.wvo.gov
Choose "facility type: Surface coal Mine"
Once you choose a facility ( Belle Ayr, Antelope, Black Thunder might be good examples) you will see a menu
on the left where you can choose permits or emission inventories.
In the permits, anytime there is an increase in coal tonnage there is likely a permit analysis that lists a
background concentration for all applicable pollutants.
If you need more assistance, please let me know.
Cara Keslar, P.E.
Monitoring Section Supervisor
Wyoming DEQ - Air Quality Division
[phone] (office)
[phone] (cell)
cara.keslarp3yyo.gov

On Wed, Jan 5, 2022 at 7:50 AM Miller, David <david.miller2ncdenr.gov> wrote:
Good Morning Cara,
I trust your meeting went well on Monday.
I am just following up concerning PM 2.5 emissions from mining operations. If you recall I told I had found
the Black Thunder Mine study from 2012 through 2014, but the study lacked the ambient PM 2.5 background
levels for the area.
You said that you could supply me that information but there had been additional studies done, which were
better and you could supply me.
By chance have you had a chance to get those study?
David Miller, PE
State Mining Engineer
NC Department of Environmental Quality
Division of Energy, Mineral, and Land Resources
Mining Section
1C1PAU6Yy
Raleigh, NC 27699
Phone: [phone]
E-mail: David. Millerna,ncdenr.gov

E
Q��
L'��µ�.t. n: ul.n FrtntMnMOul Nlillly
Email correspondence to and from this address is subject to the North Carolina Public Records Law and may
be disclosed to third parties unless the content is exempt by statute or other regulation.
E-Mail to and from me, in connection with the transaction
of public business, is subject to the Wyoming Public Records
Act and may be disclosed to third parties.

Wyoming Department of Environmental Quality
Air Quality Division
Memorandum
TO: Nancy E. Vehr
THROUGH: Darla Potter, Air Quality Resource Management Program Manager
Cara Keslar, Monitoring Section Supervisor t lic
FROM: Kristina Hooper -Barden, Monitoring Project Manager,I?z- 4'➢�fR¢ ,t4.
CC: Tanner Shatto, District Engineer
DATE: August 10, 2020
SUBJECT: Wyoming Department of Environmental Quality, Air Quality Division
Powder River Basin's Black Thunder and Buckskin Monitoring Station's
Closeout Data Summary
Backeround
The Black Thunder (AQS 1D#56-005-0891) and Buckskin (AQS ID#56-005-1899) PM2.5
Monitoring Stations were part of the Powder River Basin (PRB) PM2.5 Monitoring Network in
the PRB of northeast Wyoming. PM2.5 data have been collected in the PRB since 1993. The
Black Thunder Station was the first site established, by Arch Coal Company. The PRB PM2.5
Monitoring Network was expanded in 1996 by the Wyoming Mining Association (WMA) as a
research project to characterize fine particulate ambient concentrations in the PRB before the
establishment of the first PM2.5 National Ambient Air Quality Standards (NAAQS) in 1997. In
1999, operation funding was a collaboration between the Office of Surface Mining's Abandoned
Mine Lands Research Program, RAG, Inc., Kennecott Energy Company, Powder River Coal
Company, Rupprecht and Patashnick Company, Inc., Thunder Basin Coal Company, Triton Coal
Company, the State of Wyoming Air Quality Division (AQD), and Inter -Mountain Laboratories,
Inc. (IML). In 2002, the monitoring objective of the network changed from research to
determination of compliance with the PM2.5 NAAQS and the operation of the network was
transferred to the AQD.
The stations were considered Special Purpose Monitors (SPMs) and were operated in a State and
Local Air Monitoring Stations (SLAMS) equivalent fashion as described in 40 CFR part 50 and
58. More background details can be found in Appendix A: 2018 PRB PM2.5 Network Data
Analysis.

The 2018 PRB PM2.5 Network Data Analysis encompassed data from 2013-2017. The Black
Thunder and Buckskin close out summary will focus on data collected from 2002-2019 because
this is the timeframe where the monitoring objective of the stations was determination of
compliance with the PM2.5 NAAQS.
Data Collection
A summary of the data completeness for the Black Thunder and Buckskin PM2.5 monitors for
2002 through 2019 will be discussed in this section. For PM2.5 data associated with an Ambient
Air Quality Standard, 40 CFR part 58 states that the data collection must meet 75% data
completeness to be comparable to the NAAQS. Detailed breakdown for years that did not meet
data completeness of 75% or greater are provided below for each station.
Black Thunder
2006: Data completeness was 72%.
• 1Q: Data completeness was 43% due to machine malfunction that resulted in 17 days of
sample data invalidated. No additional details were provided in the 2006 quarterly
reports.
• 2Q: Data completeness was 84% this quarter.
• 3Q: Data completeness was 68% due to a machine malfunction that resulted in 8 days of
sample data invalidated. No additional details were provided in the 2006 quarterly
reports.
• 4Q: Data completeness was 94% this quarter.
2010: Data completeness was 55%.
• 1Q: Data completeness was 37% due to filter exchange errors and machine repairs that
resulted in 19 days of sample data invalidated.
• 2Q: Data completeness was 0% due to continual power issues at the site.
• 3Q: Data completeness was 75% this quarter.
• 4Q: Data completeness was 67.4% due to a machine malfunction after the audit and the
machine was not reset until 28 days later.
2011: Data completeness was 53%.
• 1Q: Data completeness was 80%.
• 2Q: Data completeness was 64.8% due to power failure and measurement instabilities
that resulted in 32 days of invalid data.
• 3Q: Data completeness was 51.1% due to overloaded sensor filter, failing pump, power
failure, ambient temperature/RH probe malfunction, and measurement instabilities that
resulted in 44 days of invalid data.
• 4Q: Data completeness was 0% due to failed leak checks that resulted in data invalidated
back to the third quarter's leak check.
2012: Data completeness was 46%.
• 1Q: Data completeness was 0% due to out of bounds flow rates that resulted in data
invalidated back to last successful check that occurred in 2011.
• 2Q: Data completeness was 0% due to failed leak check that resulted in data invalidated
back to last successful check that occurred in 2011.
• 3Q: Data completeness was 89.1 %.

• 4Q: Data completeness was 93.5%.
2013: Data completeness was 54%.
• IQ: Data completeness was 96.7%.
• 2Q: Data completeness was 53.8% due to a failed zero air calibration that resulted in the
sampler being returned to the factory; therefore, the analyzer was offline for 38 days and
was not online until 3rd quarter. An additional 3 days of data were invalid due to machine
malfunction.
• 3Q: Data completeness was 39.1% due to failed precision checks due to build up on the
vane, which then punched a hole in the tape and caused a major leak and flows that
would not pass specifications.
• 4Q: Data completeness was 23.9% due to a power failure that resulted in data invalidated
from October 2 through November 19. Additionally, the sampler failed the precision
check in December and the manufacturer determined the vane needed to be replaced.
Buckskin
2002: Data completeness was 68%. Details are not available regarding the exact situation. See
Appendix B.
2011: Data completeness was 34%.
• IQ: Data completeness was 0% due to the sampler being down for repairs since late
December of 2010. It began regularly collecting data again in April 2011.
• 2Q: Data completeness was 42.9% due to a failed leak check in April which resulted in
data invalided back to last successful check in December 2010.
• 3Q: Data completeness was 91.3%.
• 4Q: Data completeness was 7.6% due to failed leak checks throughout the quarter.
2012: Data completeness was 45%.
• 1Q: Data completeness was 20.9% due to a failed leak check in January, which resulted
in invalid data back to last successful leak check in 2011. The sampler resumed normal
operation in February but had machine malfunctions for an additional two days. The
sampler had out of bound flow rates in March, resulting in invalid data back to last
successful verification.
• 2Q: Data completeness was 0% due to an out of bound flow rate in the third quarter that
resulted in invalid data back to last successful check in the first quarter.
• 3Q: Data completeness was 57.6% due to an out of bound flow rate that resulted in
invalid data back to the last successful verification in the first quarter.
• 4Q: Data completeness was 98.9%.
2019: Data completeness was 70%.
• 1Q: Data completeness was 52.7% due to machine malfunction where the BAM tape was
not advancing properly. The ball slide was replaced in February.
• 2Q: Data completeness was 94.1 %. Monitor was decommissioned in May 2019.
Audits
This section summarizes PM2.5 audits performed at the Black Thunder and Buckskin monitoring
stations from 2002 through 2019. Resources used for the audit history include EPA's AQS

system (AMP256), available hard -copy and digital quarterly reports in the Wyoming Air Quality
Division (AQD) records, AQD Audit Reports, and AQD Audit Forms.
Black Thunder
2002: No audits occurred as per EPA's AQS system and the quarterly reports.
2003: Four (4) audits were performed during 2003 according to EPA's AQS reports. The
quarterly reports do not note that audits occurred, but twelve (12) Sampler Verifications were
performed.
2004: EPA's AQS system indicates twelve (12) audits occurred during 2004. Text in hard -copy
reports indicated that no audits occurred, but twelve (12) Sampler Verifications were performed.
2005: No audits occurred as per EPA's AQS system. Only the 3' quarter report remains in the
file, no audit is mentioned in this report. Details from the other quarterly reports are not
available, see Appendix B.
2006: No audits occurred as per EPA's AQS system. The quarterly reports do not note that
audits occurred, but do indicate that monthly Sampler Verifications were performed in each
month except for February, April, May, June, and August.
2007: EPA's AQS system indicates that one (1) audit occurred in 3rd quarter 2007. The 3rd
quarter report in the file is incomplete; Sample Verifications and audit information for July,
August, and September are not documented. The first, second, and fourth quarter have no record
of an audit occurring at this site during 2007.
2008: EPA's AQS system indicates that two (2) audits occurred during 2008, one in the 2"d
quarter and one in the 3rd quarter. The 2"d quarter report indicates an audit occurred in June 2008
with no issues noted during the audit. The 31 quarter report in the file is incomplete, so Sampler
Verifications and audit information for July, August, and September are undocumented. The
other quarterly reports indicate that no audits occurred and that the other monthly Sample
Verifications were performed. No issues were noted.
2009: EPA's AQS system indicates that two (2) audits occurred during 2009, one in the 1"
quarter and one in the 3rd quarter. The first quarter report has documentation for an audit that
occurred in Mach 2009. Notes state that the sampler failed the external leak check due to a
cracked cassette and that the leak check was repeated later in March. The third quarter report
includes documentation for an audit performed in August 2009. No issues noted during this
audit.
2010: EPA's AQS system indicates that four (4) audits occurred during 2010. The quarterly
reports indicate that the Black Thunder manual samplers were shut down on April 28, 2010 and
that a continuous TEOM was installed on the same day. The first quarter report indicated that a
shut down audit occurred on the manual samplers in March 2010. No issues were noted during
this audit. The second quarter report indicated that a start up audit was performed on the TEOM
in April. The third quarter report includes documentation for an audit performed in September.
4

No issues were noted. The fourth quarter report has documentation for an audit that occurred in
October. No issues were noted.
2011: EPA's AQS system indicates that two (2) audits occurred during 2011. The quarterly
reports indicate that Black Thunder was audited in January and in March. No issues were noted.
2012: EPA's AQS system indicates that no audits occurred during 2012; however, the quarterly
reports indicate that audits occurred in the 1st and Yd quarters. No issues were noted on these
audits.
2013: EPA's AQS system indicates that three (3) audits occurred in 2013, two in 2nd quarter and
one in 41' quarter. Black Thunder's TEOM was discontinued on May 24, 2013 and a BAM
began collecting data on July 16, 2013. The quarterly reports indicate that shutdown and startup
audits occurred the 2nd quarter on the TEOM and BAM respectively.
2014: EPA's AQS system indicates that two (2) audits occurred during 2014. ARS began as
station contractor in 2014 and this is the first year that the AQD began keeping internally created
Audit Forms. Previous audits were summarized into reports and submitted by an external
contractor. The second quarter audit was performed on May 10 and the Audit Form indicated
that no issues were noted. The fourth quarter audit was performed on November 19 and the
Audit Form indicated that no issues were noted.
2015: EPA's AQS system indicates that two (2) audits occurred in 2015. This is also the first
year that this site was included in the AQD Audit Report. The second quarter AQD Audit
Report for the May 28 audit indicates that no issues were noted. The fourth quarter AQD Audit
Report for the November 16 audit indicates that no issues were noted.
2016: EPA's AQS system indicates that two (2) audits occurred in 2016, one in the 2nd quarter
and one in the O quarter. The AQD Audit Reports were discontinued this year and Audit Forms
began to be uploaded into the IMPACT system. The May 19 Audit Form indicates no issues
were noted. The November 17 Audit Form indicates no issues were noted.
2017: EPA's AQS system indicates that two (2) audits occurred in 2017, one in the 1st quarter
and one in the 3rd quarter. The first quarter audit was performed on March 7 and the Audit Form
indicates that no issues were noted. The third quarter audit was performed on August 18 and the
Audit Form indicates that no issues were noted.
2018: EPA's AQS system indicates that two (2) audits occurred in 2018, one in the 2nd quarter
and one in the 4 h quarter. The second quarter audit was performed on May 16 and the Audit
Form indicates that no issues were noted. The fourth quarter audit was performed on October 10
and the Audit Form indicates that no issues were noted.
2019: EPA's AQS system indicates that one (1) audit occurred in 2019. This sampler was
decommissioned in May of 2019, thus only one audit was required. The audit was performed on
May 30 and the Audit Form indicates that no issues were noted.
5

Buckskin
2002: No audits occurred as per EPA's AQS system and the quarterly reports.
2003: Four (4) audits were performed during 2003 according to EPA's AQS reports. The
quarterly reports do not note that audits occurred, but twelve (12) Sampler Verifications were
performed.
2004: EPA's AQS system indicates twelve (12) audits occurred during 2004. The quarterly
reports indicate that no audits occurred, but twelve (12) Sampler Verifications were performed.
2005: No audits occurred as per EPA's AQS system. Only the 3rd quarter report remains in the
file, no audits are mentioned in this report. Details from the other quarterly reports are not
available, see Appendix B.
2006: No audits occurred as per EPA's AQS system. The quarterly reports note that one (1)
audit occurred in December and that monthly Sampler Verifications were performed in each
month except for April, May, and June. No issues were noted.
2007: EPA's AQS system indicates that one (1) audit occurred in 41 quarter 2007. The 3rd
quarter report in file is incomplete, so Sample Verifications nor audit information for July,
August, and September are documented. The other quarterly reports note that one (1) audit
occurred in December and that the other monthly Sample Verifications were performed. No
issues were noted.
2008: EPA's AQS system indicates that three (3) audits occurred in 3rd quarter 2008. The 3rd
quarter report in the file is incomplete, so Sampler Verifications and audit information for July,
August, and September are undocumented. The other quarterly reports indicate that no audits
occurred and that the other monthly Sample Verifications were performed.
2009: EPA's AQS system indicates that two (2) audits occurred in 2009 in the Is' and 3rd
quarters. The quarterly reports verify that audits were performed in the ls' and 3rd quarters. No
issues were noted.
2010: EPA's AQS system indicates that one (1) audit occurred in 2010 in the 3 rd quarter. The
quarterly reports indicate that Buckskin was temporarily shut down on January 17, 2010 due to
break downs and commenced data collection on April 28, 2010 when a TEOM was installed.
The quarterly report further indicate that audits occurred in the 2nd, 3rd, and 4" quarters. No
issues were noted.
2011: EPA's AQS system indicated that no audits occurred in 2011. However, the quarterly
reports indicate that audits occurred in the Is' and 3' quarter. No issues were noted.
2012: EPA's AQS system indicates that no audit occurred in 2012. The quarterly reports
indicate that audits occurred in the 1 It and 31d quarters. No issues were noted.
6

2013: EPA's AQS system indicates that three (3) audits occurred in 2013, two in 2nd quarter and
one in 4" quarter. Buckskin's TEOM was discontinued on May 22, 2013 and a BAM was
installed on June 10, 2013. The quarterly reports indicated that shutdown and startup audits
occurred the 2"d quarter on the TEOMs and BAMs, respectively.
2014: EPA's AQS system indicates that two (2) audits occurred in 2014, one in 2"d quarter and
one in 4" quarter. ARS began as station contractor in 2014 and this is the first year that the AQD
began keeping separate Audit Forms. The second quarter audit was performed on May 12 and
the Audit Form indicates that no issues were noted. The fourth quarter audit was performed on
November 17 and the Audit Form indicates that issues were noted.
2015: EPA's AQS system indicates that two (2) audits occurred in 2015 one in the 2"d quarter
and one in the 4' quarter. This is also the fast year that this site was included in the AQD Audit
Report. The second quarter AQD Audit Report for the May 28 audit indicates that no issues
were noted. The fourth quarter AQD Audit Report for the November 17 audit indicates that no
issues were noted.
2016: EPA's AQS system indicates that two (2) audits occurred in 2016, one in the 2"d quarter
and one in the 4d' quarter. The AQD Audit Reports were discontinued this year and Audit Forms
began to be uploaded into the IMPACT system. The May 19 Audit Form indicates no issues
were noted. The November 15 Audit Form indicates no issues were noted.
2017: EPA's AQS system indicates that two (2) audits occurred in 2017, one in the I" quarter
and one in the 3`d quarter. The first quarter audit was performed on March 8 and the Audit Form
indicates that no issues were noted. The third quarter audit was performed on August 17 and the
Audit Form indicates that no issues were noted.
2018: EPA's AQS system indicates that two (2) audits occurred in 2018, one in the 2"d quarter
and one in the 4`h quarter. The second quarter audit was performed on May 16 and the Audit
Form indicates that no issues were noted. The fourth quarter audit was performed on October 10
and the Audit Form indicates that no issues were noted.
2019: EPA's AQS system indicates that one (1) audit occurred in 2019. This sampler was
decommissioned in May of 2019; thus, only one audit was required. The audit was performed on
May 30 and the Audit Form indicated that no issues were noted.
Data Summary
From 1997 to 2012, the annual primary and secondary NAAQS for PM2.5 were met at Black
Thunder and Buckskin as the annual arithmetic means, averaged over three years were less than
15 µg/m3. Beginning on January 15, 2013, the annual primary NAAQS was updated to 12 µg/m3
and the secondary annual NAAQS of 15 µg/m3 was retained. These annual primary and
secondary standards continue to be calculated as the annual arithmetic mean, averaged over three
years and are the current annual NAAQS.

From 1997 to 2006, the 24-hour primary and secondary NAAQS for PM2.5 were met at Black
Thunder and Buckskin as the 981 percentile of the 24-hour averages, averaged over three years
were less than or equal to 65 µg/m3. Beginning on October 17, 2006, the 24-hour primary and
secondary NAAQS were updated to 35 µg/m3. These 24-hour primary and secondary standards
continue to be calculated as the 98' percentile of the 24-hour averages, averaged over three
years.
Black Thunder
Between 1999 and 2019, no annual average concentrations were above the annual standard of 15
µg/m3 or 12.0 µg/m3. The highest annual average was 7.0 µg/m3, which was the average for the
year 1999.
16.0
14.0
M
Nuo
Black Thunder Annual PM2.5 Averages, 1999-2019
of 12
Q1 O e4 N KI C N �O h O] Q� O '1 N KI O V1 N n M O1
m o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
e4 N N N N N N N N N N N N N N N N N N N N
Annual Average (ug/m3) —Annual NAAQS
0

No 24-hour averages in the monitoring period (1997-2006) where the 24-hour NAAQS was set
at 65 µg/m3 exceeded the NAAQS. During the monitoring period (2006-current) where the
NAAQS was set at 35 µg/m3, eight (8) days showed concentrations above the 24-hour standard.
The AQD suspects that seven (7) of these days were Exceptional Events (EE) due to wildfire
smoke affecting the area. The highest 24-hour average was 64.7 µg/m3 on September 1, 2017.
Black Thunder 24-Hour PM, s Averages, 1999-2019
1.
s0
10
R
s0
e .m
a
w
to
0
I! ) t l l t 2 t t 2 ] 2 1
9 0 0 U 0 0 0 O I] 0 U U 0 0 1 0 n i,
9U 0 0 0 0 0 0 0 11 0 1 1 1 1 1 1 1 1 I
90 1 ] 1 < $ 1 ] 99 . 1 1 3 .1 b ) X
}aJbw Avv raNe �NNPCb
Exceedance and Exceptional Event Summary for Black Thunder
Date
Value
(p m8)
Status
8/28/2015
36
EE due to wildfire smoke. No demonstration submitted. RT flag placed on
the data in AQS.
9/1/2017
65
EE due to wildfire smoke. Demonstration submitted to EPA
9/2/2017
50
EE due to wildfire smoke. Demonstration submitted to EPA
9/3/2017
81
EE due to wildfire smoke. Demonstration submitted to EPA
9/13/2017
38
EE due to wildfire smoke. Demonstration submitted to EPA
9/14/2017
48
EE due to wildfire smoke. Demonstration submitted to EPA
8/19/2018
39
EE due to wildfire smoke. Demonstration was not submitted to EPA. IT
flag placed on the data in AQS.

Buckskin
Between 1999 and 2019, no annual average concentrations were above the annual standard of 15
µg/m3 or 12.0 µg/m3. The highest annual average was 6.2 µg/m3, which was the average for the
year 2000.
Buckskin Annual PMz s Averages, 1999-2019
16
14
12 NAAQS of 12
co 10
12
s
v
C
6
c
C
Q
4
2
0
N N N N N N N N N N N N N
Annual Average lµgpn3l —Annual NAAQS
0111

24-hour averages in the monitoring period (1997-2006) where the 24-hour NAAQS was set at 65
µg/m3 exceeded the NAAQS. During the monitoring period (2006-current) where the NAAQS
was set at 35 µg/m3, seven (7) days showed concentrations above the 24-hour standard. The
AQD suspects that six (6) of these days were Exceptional Events (EE) due to wildfire smoke
affecting the area. The highest 24-hour average was 71.1 µg/m3 on September 3, 2017.
Buckskin 24-Hour PM... Averages, 1999-2019
as
)0
� 30
W
in
n
..8 E 3 8 S S
—14 Hpn A..v41g, �NAAj
Exceedance and Exceptional Event Summary for Buckskin
Date
Value
(µ rn)
Status
6/29/2011
42
Conflicting information exists on whether this exceedance was valid due to
failed flow checks. AQD did not investigate this exceedance further.
8/28/2015
41
EE due to wildfire smoke. rt (wildfire smoke) flag placed on the data in AQS.
Demonstration not submitted to EPA.
9/1/2017
47
EE due to wildfire smoke. Demonstration submitted to EPA.
9/2/2017
61
EE due to wildfire smoke. Demonstration submitted to EPA.
9/3/2017
71
EE due to wildfire smoke. Demonstration submitted to EPA.
9/13/2017
46
EE due to wildfire smoke. Demonstration submitted to EPA.
11

8/19/2018 43 1 EE due to wildfire smoke, IT (informational wildfire smoke) flag placed on
the data in AQS. Demonstration not submitted to EPA.
Conclusion
PM2.5 monitoring at Black Thunder and Buckskin began as industrial owned monitoring projects
in the early 1990s. Over the years, the network was transferred to the AQD to continue
monitoring for PM2.5 in the PRB. Data analysis that occurred in 2018 determined that the AQD
could decommission the Black Thunder and Buckskin PM2.5 monitors, while retaining the
Antelope and Belle Ayr PM2.5 monitors to monitor air quality in the PRB area. The close out
summary shows that during the life of these monitors, there were data collection issues due to
miscellaneous equipment malfunctions, which resulted in frequent years where data collection
was not greater than 75%. This summary also shows that data collected at these sites did not
record frequent exceedances and many of the exceedances that were recorded, the AQD believes
are due to Exceptional Events.
The Black Thunder and Buckskin monitors have sufficiently fulfilled their objective of
characterizing and monitoring air quality in the PRB.
12

1-A
W
o A a y
at
N �Q"
co
o 00 o Q
0 o w OC
A
w w p �
oaa ►y
w 00 N Zs
0 0 �
00
A N p
O00 f�
`1 p
J 0000 00
N
00
N
,%0 00 N p
0 o O II�
�O
O
w vi to p �v
O �r
� w N
.A. 46 N
0 o O
N
� � O
0 o W
N %0 N
0 o A
N � p
No
O
�A N N
J
0 o 000
�t �O N
O " O
0 0 �

'bt
sis,91guv v4loQ 8IOZ :V xlpuadd-V

Mark Gordon, Governor
Department of Environmental Quality
To protect conserve and enhance the quaAty of Wyoming's
envIronmentforthebe ItofcurrentandfuturegeneraGons
10
TO: Nancy Vehr, Administrator ty/pAV • +1*-11
THROUGH: Darla Potter, AQRM Program Manager*
+11 -M
FROM: Cara Keslar, Monitoring Section Supervisor 19
CC: Dan Sharon, Project Manager �J, 41161 /9
Kristina Hooper -Barden, Project Manageryos 4/,,/,q
DATE: April 15, 2019
RE: Powder River Basin PM2.5 Network Data Analysis
1. Background
Todd Parfitt, Director
The EPA first established National Ambient Air Quality Standards (NAAQS) for PM2.5 in 1997:
an annual standard of 15 µg/m3, based on the 3-year average of annual arithmetic mean; and a
24-hour standard of 65 µg/m3, based on the 3-year average of the 98" percentile. In 2006, the
EPA retained the PM2.5 annual standard at 15 µg/m3 and revised the level of the 24-hour PM2.5
standard to 35 µg/m3. In 2012, EPA revised the level of the primary annual PM2.5 standard to 12
µg/m3, retained the secondary annual PM2.5 standard of 15 µg/m3, and retained the level of the
24-hour at 35 µg/m3.
PM2.5 data have been collected in the Powder River Basin (PRB) since 1993. The first site was
established at Black Thunder Mine by Arch Coal Company. The PM2.5 network was expanded
by the Wyoming Mining Association (WMA) in 1996 as a research project to characterize fine
particulate ambient concentrations in the PRB before the establishment of the first PM2.5
NAAQS in 1997. Beginning in 1999, funding the operation of the network was a collaboration
between Office of Surface Mining's Abandoned Mine Lands Research Program, RAG, Inc.,
Kennecott Energy Company, Powder River Coal Company, Rupprecht and Patashnick
Company, Inc., Thunder Basin Coal Company, Triton Coal Company, the State of Wyoming Air
Quality Division (AQD), and IML Air Science. In 2002, the objective of the network changed
from research to determination of compliance with the PM2.5 NAAQS and the operation of the
network was transferred to the AQD.
The network has expanded and contracted with instrumentation changing frequently over the
ensuing years. The promulgation of the 1997 PM2.s NAAQS prompted the upgrading and
expansion of the network to characterize the spatial and temporal distribution of the pollutant in
the PRB and to evaluate various sampling methods against the Federal Reference Methods
200 West 17th Street, Cheyenne, WY 82002 • http://deq.wyoniing.gov • Fax [phone]
ADMIWOUTREACH ABANDONED MINES AIR QUALITY INDUSTRIAL SITING LANDQUALITY SOLIDSHALWASTE WATER QUALITY
[phone] [phone] [phone] [phone] [phone] [phone] (307)7T7-T781

(FRM) used. From 1999 to 2010, the network consisted of 5 R&P Partisol-Plus 2025 sequential
samplers at 4 sites, 1 continuous monitor, and 2 Partisol samplers. Data recovery during this
period was marginal to poor (data recovery information is available upon request to the AQD).
In April 2010, all manual particulate samplers in the network were replaced with continuous
Thermo Scientific 1405-DF TEOM samplers. Data recovery during this period was also
typically poor and the data quality are questionable due to known issues with the 1405-DF
samplers. In the second quarter of 2013 the TEOMs were replaced with BAM 1020 continuous
monitors and data quality improved. However, the remote nature of these sites and inconsistent
power availability have resulted in quarters and years where data collection does not meet EPA
requirements for NAAQS comparability.
The stations are operated in a State and Local Air Monitoring Stations (SLAMS) -equivalent
fashion and are considered Special Purpose Monitors (SPMs). They are therefore funded
through the state general fund and are not required or funded by the EPA. The AQD contracts
out the operation of these sites through a competitive RFP process.
As a result of data collection difficulties at the sites, the operation of these monitors has been
very time consuming. Considering the two and a half decades of data collection, the shifting of
AQD priorities, persistent operational difficulties, emerging monitoring needs elsewhere in the
state, and the evolution of funding for these sites, the AQD is initiating the following analysis to
evaluate the effectiveness and efficiency of the PRB PM2.s network in its current configuration.
The current station locations and station equipment are shown in the figure and table below.
21Pa�g,e

Site
Name
A S ID
Lat/Lou2
Analyzer
PRBA-AN
Antelope
56-009-0009
43.4256,-105.3886
Met One BAM-1020
PRB5
Black Thunder
56-005-0891
43.6504,-105.2140
Met One BAM-1020
PRB3
Belle Ayr A
56-005-0892, POC-3
44.0968,-105.3431
Met One BAM-1020
PRB3
Belle Ayr B
56-005-0892, POC-4
44.0968,-105.3431
Met One BAM-1020
PRB1
Buckskin
56-005-1899
44.5027,-105.5398
Met One BAM-1020
Because of the poor data completeness and uncertainty associated with the data collected prior to
mid-2013, the following analyses focus only on the data collected after the station
instrumentation changed to BAM 1020s.
31Pa.-e

2. Data Collection and Ouality Assurance
In 2013 the Black Thunder and Belle Ayr sites had data collection below 75%. Black Thunder
data were invalidated during 2013 due to a flaw in the vane which caused the flow to be out of
bounds. The Belle Ayr TEOM failed most quarterly maintenance checks in the fourth quarter of
2012 and was removed from service until the BAMs were installed in June 2013. Antelope Site
3 data were not reported starting on June 21, 2013 due to construction activities in the vicinity.
Antelope Site 3 was not reporting throughout 2014 and was relocated to Antelope Site 7 on
February 18, 2015.
r" riviz,s ivionnor
yam uouection i able
Site
2013
2014
2015
2016
2017
Antelope
90%
NC
98%
96%
75%
Belle Ayr
55%
88%
94%
90%
99%
Black Thunder
54%
94%
84%
76%
92%
Buckskin
94%
92%
92%
97%
94%
NC=Not Collected
Audits are performed on the PRB PM2.5 monitors twice a year. From 2013 to 2017 all audits
were performed by the AQD QA Coordinator and all audits passed.
41 Pituc

3. Data Trends
3.1 Primary Standard -Weighted Annual Mean
The annual average data from 2013-2017 for all four sites were plotted to identify any trends in
the data during the 5-year period.
Antelope Annual PM,.; Averages, 2013-2017
14
12
10
v = 0Q5;14x - INN
S R==0.4192
6 •
2
0
2013 2014 201 2016 2017
Year
• Site 3 Annual Average (ughn3)
— Auntal \AAQS
p-value of trend line = 0.3516
51Pa-e
• Site - Ai uual Avemee (tie 113)
••••• Liuear (Antelope Mutual Average (ug�nu))

, 119
9I I I £Z'0 = aujl puall jo onlan-d
((f«(i-on) aaeaa.%V IenuuF) aean17 .........
sa'rtl\ lenuti (fut/Sn) aoeaa.kt lencnctr
lea.0
LIOZ 910Z iIOZ t10Z f10Z
s4Zt'0 = "d
ZL c+19iO=A
......................0......................a
L I OZ_` I OZ `saAsJand i `lAId leninld aapunul xDelg
6I98LZ'0= aujj puau;o anjun-d
(Quvin) aaeaa.%v lennuv) aeam-I .........
s(%FF\ Ienuu (futrn) as"ejaeF junmiV •
jea_C
LIOZ 910Z SIOz t10Z f10Z
0
Z —
..............................................*.....................R....................... V
9
s
01 c
Z I r�
tl `'
L I OZ_` I OL `sQSP-IQnd i'zAld junuu V .iXV 2jpS

Buckskin Annual PM,... Averages, 2013-2017
la
?, 12
10
S
Z` 6 , •
1.............................................................................................
v=-0.13x-4.53�
• R= = 0.016
0
2013 2014 2015 2016 201
Year
• Annual Avetnge (tie 1113) Annual \AAQS
......••• Linear (Annual Avetnae (uJ1113))
p-value of trend line = 0.839298
Data for the Buckskin and Belle Ayr sites showed slightly decreasing trends over the 5-year
period, while data for the Antelope and Black Thunder sites showed slightly increasing trends
over the same period. However, even though Antelope and Black Thunder show increasing
trends, the RZ values of the trend lines was below 0.5, indicating poor ability of the trend lines to
explain the variability in the data set. The p-values of all trend lines were greater than 0.2,
indicating low statistical significance of the trends compared with the data set. It should be
noted that 2015 and 2017 had several days influenced by wildfire smoke. Section 5 discusses an
analysis of the trends and data when exceedances due to wildfire smoke are excluded.
71Pag

3.2 Primary and Secondary Standard-24-Hour Average
The 24-hour average data from 2013-2017 for all four sites were plotted to identify any trends in
the data during the 5-year period.
Antelope 24-Hour PM,. Averages, 2013-2017
90
SO
-,0
60
u
0
50
40
2 30
20 c=0.000Ss =Z.S937
fl 6
-10
2 1 > - 1
0 0 0 u 0
1 1 1 1
3 4 i
Year
• Site324-HOW A% eraeetuvm31
—24-Hwr \A.AQS
p-value of trend line = 0.0007909
8 P .11,_ �
• Site - 24-Hour Averaue (ugm3)
-^•••••• Lincar(.Ante[V 24-Hay AVaraze(ns/n1!))

1
ti 16
(0 Ala^II0a33H) LO-EIZZ'8 = gull puaal jo onlen-d
(limn) aacn.+t=14-filsatu-I......... sbt'V.*':=H-1:— (su in) Aviaw MH-rC
9
1 1 I 1
I
0 0 0 0
0
O b
y��•
• t
••
Oi _
OS
9910'0-:ii
Of �i
F6i'f - sry00'0 = .i
5
Oi c
09 e_
0-
os
06
LIOc £IOZ 'saRPJ,?:�V'iK'd anoH-rZ japuny-l- `oujg
99£ 10'0 = ouil puail jo amen-d
((Fttt zm PiUMAp moH-t[i.can1]......... sZIFY.-oWH•ri— i(tit,m i a,Sw%F' itk H•ti
AvI,
0
i
0
i
0
WWI)=:a
LION iIOL 'sa'01UO-AV'•IKd m0H-fZ 1AV?lREl
I
1
0
or
o „
of
or
of
or
of
r
09
0- Z
09 =
06

I
Buckskin 24-Hour PM, _< Averages. 20li-2017
00
so
60
<0 y=5E-0h - 1.142
ti
40 R'= 2E-05
20
10
-10
bear
• 24-How Average(ua m3 i —24-Hour NAAQS •......•• Lirwar(24-Hour Average(uem3))
p-value of trend line = 0.8431
The data for all four sites yielded trend lines with nearly flat slopes and W values close to or at
zero. The trend lines for all sites except for Buckskin produced p-values less than 0.05. The low
RZ values indicate that these trend lines do not explain the observed variability in the data set.
The low p-values for Antelope, Belle Ayr, and Black Thunder indicate that, despite seasonally
elevated values at all sites due to wildfires in the late summers of 2015 and 2017, the trends are
statistically significant compared with the data set.
4. Design Values
4.1 Primary Standard -Weighted Annual Mean
The Primary Annual NAAQS for PM2.5 is met when the annual mean, averaged over 3 years, is
less than or equal to 12 µg/m3. Valid Design Values must have a minimum data collection of
75% per calendar quarter. The summers of 2012, 2015, and 2017 had several days that the AQD
has requested exclusion of the data as due to Exceptional Events. Section 5 discusses these days
in further detail. The following tables of Design Values include columns that include and
exclude AQD-flagged Exceptional Events.
101P 1c

Antelope Site 3
Design Value
Weighted Mean
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)*
Period
Value (µg/m3)
Value (µjm3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
2.8
2011-2013
4.6
4.5
N
2014
N/A
2012-2014
5.4
5.3
N
2015
N/A
2013-2015
2.8
2.8
N
2016
N/A
2014-2016
N/A
N/A
N/A
2017
N/A
2015-2017
N/A
N/A
N/A
*Antelope site 3 ended data collection on June 21, 2013
Antelope Site 7
Design Value
Weighted Mean
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)*
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
N/A
2011-2013
N/A
N/A
N/A
2014
N/A
2012-2014
N/A
N/A
N/A
2015
4.2
2013-2015
4.2
4.0
N
2016
2.7
2014-2016
3.4
3.3
N
2017
5.8
2015-2017
4.2
4.2
N
*Antelope Site 7 began data collection on February 18, 2015.
Belle Ayr (Primary)
Design Value
Weighted Mean
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m)
Value (µg/m)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
6.4
2011-2013
6.5
6.3
N
2014
5.3
2012-2014
6.5
6.3
N
201
1 4.9
2013-2015
5.5
5.5
N
2016
4.1
2014-2016
4.7
4.7
Y
2017
5.4
2015-2017
4.8
4.7
Y

Black Thunder
Design Value
Weighted Mean
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m)
Value (µgW)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
4.2
2011-2013
4.1
4.1
N
2014
3.9
2012-2014
4.3
4.3
N
2015
4.9
2013-2015
4.3
4.3
N
2016
3.5
2014-2016
4.1
4.1
Y
2017
5.5
2015-2017
4.6
4.6
Y
Buckskin
Design Value
Weighted Mean
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
4.8
2011-2013
5.2
5.2
N
2014
5.5
2012-2014
5.4
5.4
N
2015
2.2
2013-2015
4.2
4.1
Y
2016
2.6
2014-2016
3.4
3.4
Y
2017
5.6
2015-2017
r 3.5
3.4
Y
As can be seen in the tables above, no PRB PM2,5 station has recorded a violation of the Primary
Annual NAAQS in the past 5 years. Due to the 3-year averaging that goes into calculating
design values and data completeness being so poor prior to the installation of the BAM 1020s in
the second quarter of 2013, only the most recent few years have valid 3-year design values. The
Antelope sites have not had a valid design value in the analysis period due to the relocation of
the station in 2015, continued operational issues, and the inability to meet completeness criteria.
121Pitu,c

3.2 Primary and Secondary Standard-24-Hour Average
The Primary and Secondary 24-Hour NAAQS for PM2.5 are met when the 98I' percentile of daily
24-hour averages, averaged over 3 years, is less than or equal to 35 µg/m3. Valid Design Values
must have a minimum data collection of 75% per calendar quarter. The following tables of
Design Values include exceedances recorded at each site.
Antelope Site 3
Design Value
98Ih Percentile
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)*
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
8.0
2011-2013
14
13
N
2014
N/A
2012-2014
17
16
N
2015
N/A
2013-2015
8
8
N
2016
N/A
2014-2016
N/A
N/A
N/A
2017
N/A
2015-2017
N/A
N/A
N/A
*Antelope Site 3 ended data collection on June 21, 2013.
Antelope Site 7
Design Value
98th Percentile
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)*
Period
Value (µWm3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
N/A
2011-2013
N/A
N/A
N/A
2014
N/A
2012-2014
N/A
N/A
N/A
2015
18.5
2013-2015
19
16
N
2016
9.6
2014-2016
14
13
N
2017
20.1
2015-2017
16
15
N
*Antelope Site 7 began data collection on February 18, 2015.
Belle Ayr (Primary)
Design Value
981h Percentile
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
13.5
2011-2013
30
19
N
2014
10.5
2012-2014
26
16
N
2015
18.5
2013-2015
14
14
N
2016
13.7
2014-2016
14
14
Y
2017
23.4
2015-2017
19
18
Y
1311)Liu e

Black Thunder
Design Value
981h Percentile
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
13.6
2011-2013
14
14
N
2014
9.9
2012-2014
13
13
N
2015
21.6
2013-2015
15
14
N
2016
11.0
2014-2016
14
13
Y
2017
25.5
2015-2017
19
18
Y
Buckskin
Design Value
98" Percentile
3-Year DV
3-Year Design
3-Year Design
Valid
Year
(µg/m3)
Period
Value (µg/m3)
Value (µg/m3)
DV?
including
excluding
Exceptional
Exceptional
Events
Events
2013
13.7
2011-2013
16
16
N
2014
12.2
2012-2014
15
15
N
2015
21.0
2013-2015
16
14
Y
2016
9.4
2014-2016
14
13
Y
2017
26.0
2015-2017
19
18
Y
As can be seen in the tables above, no PRB PM2.5 station has recorded a violation of the Primary
and Secondary 24-Hour NAAQS in the past 5 years. Again, due to the 3-year averaging that
goes into calculating design values and data completeness being so poor prior to the installation
of the BAM 1020s in the second quarter of 2013, only the most recent few years have valid 3-
year design values. The Antelope sites have not had a valid design value in the analysis period
due to the relocation of the station in 2015, continued operational issues, and the inability to meet
completeness criteria.
14111itue

5. Exceedances
An exceedance is one occurrence of a measured concentration that exceeds the level such
standard for the averaging period (See 40CFR Part 50.1(1). Exceedances and their apparent
cause(s) for each site from 2013-2017 were assessed.
Antelope
Date
Exceedance
Explanation
Value m3
5/10/2015
38
No apparent cause of this exceedance could be determined, the data was
marked with a "3" qualifier code for "field issue"
7/6/2015
41.7
Starting in early July 2015, smoke from many large fires in Alaska and
Canada began affecting AQD monitors across the state of Wyoming and
much of the western United States. Smoke effects lasted throughout the
summer of 2015, intensifying in late August. The AQD has flagged
these data with RT Wildfire-U.S.) exclusion flags in AQS.
9/14/2017
43.7
Starting in mid -July 2017, smoke from many large fires in the northwest
United States began affecting AQD monitors across the state of
Wyoming and much of the western United States. Smoke effects lasted
throughout the summer of 2017, intensifying in early September. The
AQD plans to flag these data with RT (Wildfire- U.S.) exclusion -flags in
AQS.
Belle Ayr
Date
Exceedance
Value m3
Explanation
8/28/2015
38.2
Starting in early July 2015, smoke from many large fires in Alaska and
Canada began affecting AQD monitors across the state of Wyoming and
much of the western United States. Smoke effects lasted throughout the
summer of 2015, intensifying in late August. The AQD has flagged
these data with RT (Wildfire-U.S.) exclusion flags in AQS.
9/1/2017
58.1
Starting in mid -July 2017, smoke from many large fires in the northwest
9/2/2017
51.1
United States began affecting AQD monitors across the state of
9/3/2017
79.9
Wyoming and much of the western United States. Smoke effects lasted
9/14/2017
37
throughout the summer of 2017, intensifying in early September. The
AQD plans to flag these data with RT (Wildfire- U.S.) exclusion flags in
AQS.
151Pa_ue

Black Thunder
Date
Exceedance
Value m3
Explanation
'
8/28/2015
35.9
Starting in early July 2015, smoke from many large fires in Alaska and
Canada began affecting AQD monitors across the state of Wyoming and
much of the western United States. Smoke effects lasted throughout the
summer of 2015, intensifying in late August. The AQD has flagged
these data with RT Wildfire-U.S. exclusion flags in A S.
9/ 1 /2017
64.7
Starting in mid -July 2017, smoke from many large fires in the northwest
9/2/2017
49.5
United States began affecting AQD monitors across the state of
9/3/2017
81
Wyoming and much of the western United States. Smoke effects lasted
9/13/2017
38
throughout the summer of 2017, intensifying in early September. The
9/ 14/2017
47.8
AQD plans to flag these data with RT (Wildfire- U.S.) exclusion flags in
AQS.
Buckskin
Date
Exceedance
Value m3
Explanation
8/28/2015
40.7
Starting in early July 2015, smoke from many large fires in Alaska and
Canada began affecting AQD monitors across the state of Wyoming and
much of the western United States. Smoke effects lasted throughout the
summer of 2015, intensifying in late August. The AQD has flagged
these data with RT Wildfire-U.S.) exclusion flags in A S.
9/1/2017
47
Starting in mid -July 2017, smoke from many large fires in the northwest
9/2/2017
60.6
United States began affecting AQD monitors across the state of
9/3/2017
71.1
Wyoming and much of the western United States. Smoke effects lasted
9/14/2017
46.1
throughout the summer of 2017, intensifying in early September. The
AQD plans to flag these data with RT (Wildfire- U.S.) exclusion flags in
AQS.
Based on the tables above, all exceedances during the analysis period were due to smoke impacts
from North American wildfires, with the exception of the May 10, 2015 Antelope exceedance,
where the AQD was not able to determine a cause of the exceedance. The graphs in Section 2
were rerun with the fire -flagged exceedances removed. This had no major effect on the station
trends outlined in Section 2 or their statistical significance, with the exception of the p-value for
the trend line of 24-hour PM2.5 at Buckskin. While the p-value was greatly reduced following
the removal of exceedance values from the dataset, it was still greater than 0.05 indicating low
statistical significance of the trend compared with the data set. The rerun graphs are shown
below.
161 P -.

'-r'�IILT
8901.0 = onjen-d
SaFb:\ lennntr (ctu,4n) eaeraat lenmrN •
IN 910i >10< floz Mz
ftf9'0=.'d
9f'9 + -"tr-o- = .S
...................... .....................0.......................
(saZ)uLpaa.3xa ou)
LIOZ-cli0Z `sane.>anV''`Wd lEnuuy .lid ailag
((fuuon) A.W.1v lenmrt• Molalnt)jeaul-7 .........
(cm an) aarimt lemma' _ ails •
1eaA
[.IOZ 9101 iloi
1_t�z- �9srf o= •+
£8£b'0 = amen-d
Sb F'\ lenmid —
(fugan) aoma.�t lennnt f al!S •
rlol FIN
.............................0
(samepaaax) paaaLu-aiu ou)
L I OL-fl OL 'SZOE AV INd iL'nuuV adopiud

^t-d18I
111669'0 = anlsn-d
((fmrn) zaaia.�tl lennaH) ivaa�7 .........
sadd\ lunuu (Fmrn) za"eta.%y lentntb •
LIOz 91oz 51OF f10Z FIoz
.....................
•
6950'O = z'l
(saZ)uupaa3xa ou)
LIOG_`i0Z `saae.Tany''`Wd lunuuV uilsxang
8LO18£'0 = onlsn-d
((Fuyin) aoaul. tV lunucn).teaut-I .........
satiti\luntmv (Fnt/on)as"Etaay lummb •
tuz k
LIOZ 91OF iloz riot Floz
l6=z'0=zd
r6'F - ,9l'O = A
•............................................. *.....................o
(saz)uepaa3xa ou)
L IOZ-£IOZ 'saDeiany ' `Ind lunuuy iopuiilgl 13ulg
e

'n rd161
(0 ,ClanllaaJJg) 80-361Z'9 = anlen-d
iffy Snlz¢e.izat'moH-Y.l iezn!l ••••••••• Satt:C �H•ti— ffSu 8nl zovx+t �re+H-G .
mzl
I
0
i
610'0=:a
i 196.f - %1100,0•
(sa:MFp isa ou) L I ON] OZ ''L'd moH-tL JAV allafl
((qm on) z4va�p nwH-t: tdilzlnt'I rezn(� .........
(fill ails
L885000.0 = onlsn-d
SOY V N WH-tZ —
@'tw 1q zovzaP moH•t^, t zt(S
9
1 I I 1 1
0 0 0 0 0
01-
......................................................... ow
0
iiw�+ • •• • . 60I0'0=:N 01
. Oi
Z
• Ot f7
Oi
�r
09 r.
0_ 'C
OS
06
(saaaepaaaxa paaaHO-aly ou) LIOc £IOL 'saouuaeV" Kd .InoH-tL adolamy

u
90
30
1, 'o
60
_u
r 40
< 40
g 30
`— 20
— 10
rt 0
9P;
Black Thunder 24-Hour PM,_, Averages- 201;-2017 (no exceedances)
c = 0.0003x - 3.5313,,
W=0.0105
0 0
1 1
0
1
a 0
- +
1 1
6 ,
Year
• 24-Hour ACerage(ug,InO) —2a-Hour \AAQS •••••••••Linear (24•Hour Acerase (ugnn3))
p-value = 9.614E-05 (Effectively 0)
Buckskin 24-Hour PM,., Averages, 2013-2017 (no exceedances)
90
so
3r -0
60
:n
3 50 v=•0.0003s +4.3392
40 R2=0.0019
2 30
20
10
Y
0
-to
2 , 2 ,
0 0 0 0 0
1 1 1 1 t
3 1 5 6
Year
• 24-Hour Average(ugnU)—2J•Hour\.AAQS ......••• Linear(24-Hour Average(ugan3))
p-value = 0.0745
201Pa-e

6. Correlations
The table below shows the Pearson Correlation Coefficients (PCC) for each of the station pairs
in the PRB PM2.5 network. The scale of the PCC is -1 (negative linear correlation) to +1
(positive linear correlation) with a value of 0 indicating no correlation between two data sets.
The daily data collected by each of these stations are generally highly correlated with other
stations, with PCC values ranging from 0.62 to 0.96. The Black Thunder station data especially
appear to correlate highly with all other stations with PCC values around 0.9, indicating potential
redundancy of this station.
2017 PMzs Data Correlation Chart for Powder River Basin Monitoring Network
Antelope
Belle Ayr
Black Thunder
Buckskin
Antelope
I
Belle Ayr
0.71
1
Black Thunder
0.87
0.96
1
Buckskin
0.62
0.95
0.91
1
The figure below displays data correlation for 2017 as a time series plot to highlight the
correlation of exceedance data collected by the stations during a series of smoke -related events in
September.2017. Individual station data during this time generally tracked well with other
stations. The highest value recorded at each station is labeled below.
Powder River Basin PM, 5 24-Hour Averages, 2017
85 9/3/2017, 81
9/3/2017, 79.9
75
9/3/2017, 71.1
A 65
u 55
> 45 9114/2017, 43.7
"i 35
f `
25 ,
15
J.r
ry0
\� ,��) �(`• h�i cry 1\^) �\ 1\\ 0 \\\ ^\ `\
Date
Antelope • Belle:x%r
21111au, e
Black Ihundcr , BuaAskin —24-1lour NA.AQS

Another effective way to compare data is a notched box and whisker plot (hereafter referred to as
a boxplot). The boxplots below were generated using Program R, an open -source statistical
software package. These boxplots can be used to assess statistical significance between two or
more sets of data. They can be used to assess the trend in concentrations at a monitoring station
by evaluating data collected in different years. These boxplots can also be used to determine
whether two or more chosen stations monitor significantly different concentrations. The
boxplots are best utilized when there are multiple years of data for a given pollutant at a given
site to compare and study. The figure below illustrates what a boxplot is and provides
instructions for interpretation.
How to Interpret Notched Box -Whisker Plots
A notched box -whisker plot illustrates the distribution of concentrations. The
notch is centered on the median concentration, widening to the size of the box to
illustrate the 95% confidence interval in the median concentration value. The
edges of the box illustrate the 25th and 75th percentile concentrations. The
whiskers indicate values that are 1.5 times the interquardle range (IQR). Star
outliers fall between 1.5 and 3 times the IQR. Circle outliers are greater than 3
times the IQR.
outlier more than 3 times IQR from the mean (extreme out w)
outna more than 1.5 time, the IQR
7P pemer,We
The notch and extents of the notch Indicates the 95% confidence
median s-
interval; when comparing notched box -whisker plots, if the notch of
one box does not overlap with the notch of another box, the
250percenth
median values are statistically significantly different at the 95%
boxindkatesft IQR
confidence interval. If the notches overlap, the median valves are
whiskerends-1.5UnmtMIQR
not statistically significantly different.
2211'age

2013 Boxplots for PRB PM2.5 Monitors
go
M O
E
m
c
0 0
m v
b
C
N
U
C
O O
Antelope Belle Ayr Black Thunder Buckskin
Site Name
2014 Boxplots for PRB PM2.5 Monitors
Pq
M O
E c°
7
C
O p
�—Ca CT
b
C
N
U
C
O O
6�
0
Antelope Belle Ayr Black Thunder Buckskin
Site Name
2311's_

2015 Boxplots for PRB PM2.5 Monitors
11
0
0 0 0
8
e
® a
A
Antelope Belle Ayr Black Thunder Buckskin
Site Name
2016 Boxplots for PRB PM2.5 Monitors
M CD
E
8
O
0
Antelope Belle Ayr Black Thunder Buckskin
Site Name
241�'_ _

c
0 0
m v
is
c
m
0 0
AS
2017 Boxplots for PRB PM2.5 Monitors
0 0
0
0
0 °
0 8
0
0 0
0 0 0 0
o
� � T
Antelope Belle Ayr Black Thunder Buckskin
Site Name
The yearly boxplots above illustrate that the interquartile ranges (IQR, bracketing the 25-75
percentiles of the data collected) for each station and are all below 10 µg/m3 and generally
overlap, indicating overall agreement between stations. Station data outliers are typically within
the same magnitude as other stations, with the exception of Antelope in 2016.
251P-,i--:

Comparison of PRB PM25 Network
i
201224-hour PAl.:\A4QS
_,: _L1
V.ar
Legend
Antelope
Belle Ayr
Black Thunder
n
Buckslun

The figure on page 22 displays box plots for each station for each year in the assessment period
to show long-term trends between station data. Again, all IQRs are below 10 µg/m3 and
generally overlap with a few exceptions. Outliers with greater magnitude can be seen in 2015
and 2017 due to active fire seasons in those years, though even these extreme values show
general agreement between stations.
7. Conclusion
Results of the PRB PM2.s analysis show the PRB PM2.5 network as a resource of historical
source -oriented data that characterizes PM2.5 conditions at or near PRB coal mines. The analysis
discussed moderate issues with data collection in detail, primarily stemming from site
accessibility, power concerns, and instrumentation issues. However, even with these concerns,
valid design values are less than 50% of the annual NAAQS and less than 60% of the 24-hour
NAAQS. Given the AQD's experience with long-term monitoring of PM2.5, the AQD does not
believe these monitors are in jeopardy of violating the current NAAQS. Given the low Design
Values recorded by the stations the AQD also believes that these station have a low likelihood of
violating potentially lower future NAAQS. For the most recent 3-year period, Belle Ayr has the
highest annual design value, 4.8 µg/m3 (including AQD-flagged Exceptional Events) and 4.7
µg/m3 (excluding AQD-flagged Exceptional Events). For the most recent 3-year period. Belle
Ayr, Buckskin and Black Thunder all had equal 24-hour design values o m3 (including
Exceptional Events) an 18 µg/m exc u ng D-flagged Exceptional Events).
Many of the highest values were caused by smoke impacts during high wildfire years. The AQD
is confident in this causation and has either flagged or submitted requests to exclude the data for
several exceedances under the Exceptional Events Rule to EPA Region 8 for concurrence. The
EPA has not acted on any of these requests, saying that the exceedances are not anticipated to be
used in any pending regulatory decisions by EPA. If the EPA were to act on those
demonstrations, the design values would be at an even lower level. The data also show that
when taking out wildfire -related exceedances, the trend in PM2.5 has less than 0.5 µg/m3 increase
or decrease per year.
The analyses also showed the data collected by monitors in the network to be highly correlated to
each other, with all but one correlation above 70%. This result signifies redundancy within the
network. The Belle Ayr and Black Thunder monitors have the highest correlation with all other
monitors in the network and are highly correlated with each other. The Antelope monitor has
lower correlations with the other monitors however, which is to be expected given Antelope's
siting as a background monitor and the reduced amount of data available given siting issues.
Even with Antelope's lower correlation, the AQD still considers it moderately correlated with
other monitors in the network.
The Monitoring Section has discussed with New Source Review (NSR) how the data from the
PRB PM2.5 network are used in the permitting process. NSR will use the data from these
monitors as a representation of facilities' compliance with the PM23 NAAQS or add them as a
2x

background to modeled concentrations. In either case, using values from a monitor that is
showing higher concentrations is a more conservative practice. NSR is supportive of retaining
the Belle Ayr monitors for this purpose. NSR also indicated that due to differing meteorological
conditions in the southern PRB, retaining the Antelope PM2.5 monitor will be beneficial for use
as background data in permit analyses.
The Monitoring Section also considered findings from the most recent Network Assessment and
emerging issues to determine if continuous PM2.5 data could make the AQD's population —based
network more efficient and effective. For example, during the 2017 and 2018 wildfire season, the
AQD noted significant gaps in real-time PM2.5 data around Wyoming since many monitors in
populated areas are currently filter -based. These continuous monitors might serve Wyoming
more effectively if they were relocated to populated areas of the state that currently do not have
continuous PM2.S monitoring. Relocating these samplers would be a step to achieving a more
efficient monitoring network for the people of Wyoming.
Given these analyses, it is the Monitoring Section's conclusion that the PRB PM2.5 network
shows redundancy. The AQD's objective is to have the AQD network optimized for efficient
and effective operation. Therefore, the. Monitoring Section is supporting decommissioning of the
Buckskin, Antelope, and Black Thunder monitoring locations and continuing operations at Belle
Ayr. Belle Ayr is chosen as a location that has the collocated PM2.5 monitor, as well as also
having PMio and NOX collected at the site. Additionally, the AQD is confident that Belle Ayr
will produce concentrations that are representative of the entire PRB based on correlation
statistics, while being conservative as one of the stations with higher past design values in the
network.
During the process of evaluating the PRB network, the AQD reached out to the PRB coal mines
and WMA at a meeting on January 14, 2019 to discuss the AQD's conclusions. The AQD
received a comment letter from WMA on January 21, 2019 requesting that the AQD retain the
Belle Ayr and Buckskin station, with Buckskin acting as a background station. The AQD has
elected not to keep the Buckskin station because it has a 95% correlation with the Belle Ayr
station (2017 data) and it is not sited for background monitoring.
281P:1 o C

ST
aal;aq pioaag EjoQ :g xipuaddV

19LL-Lu Q00 ZSLL-LLL (LOC) 99LL-LLL (LOC) 69CL-LLL (Loe) ML-LLL (LOC)
ALIIVn0 H31VM 31SVM'ZVH Y onoS ALIIVnO ONVI SNUIS NRllSnONI ALIIVno NIV
"LVSC9(L0C) xej . n06.6u)woAm-bepp:d74y • ZOOZB AM `BUUMay0
SMILL (LOC) LCBL-LLL (LOC)
S3NIW 03NOONVSV HoV3HLnoINIWOV
laanS 43L13saM OOZ
utgltm pauteluoo Iles st s)sp aql `algsltsns )ou an; splooai Sdoo- as a l no aseq�sP Sad aql
3o 6usdwoo a of polelai uogeuuo;ul pantgois Isnlos ay) `ianamog auuarf q uslaua enm panit *
ipeaile3o satdoo atam sp[ooaJ vepuags aq) lsg) pawaas )t `spxooai 8uuoinzow auaiquie panrgois3o
matnaL IsRtut us uo passg �aCoid Sutuus0 )otgstp >faZnooQ agl io3 suot)sisdaid 8vunp pa io>;sap
SpvauanpsuL aaam `y00Z Vnoap 9661 91UM3 own aql 8uu2noo `spi000i xdoo-p.Leq 8uuo)tuOW
)ual4w'd/S'Z Y�Id vTseg iantg sapmod aq) `610Z )sn8ny ut aogjo uspuagS alp o) )Isin a BuunQ
610Z `Ll iagoloo : ILLV(l
b00Z-566I spiooag 8uuoltuoW luatgwd/S•Z wd ul ll cant -I japmod :.LjaraaS
( )stlutoadS )uawa8suslnl MgG Pus splooaN `uuuw
Iqn}I atle)eN :w011.3
b 1. u - A l�J 'osi vadns 8uuo)tuolnl uotssnug pus nv luatgwy `-tslsax sjeo :Os
wauNvaowaw
101WJ:p rilyed OFOL .
}`.��.e .ou:anoJ 'uoPLo'J Amity
suogeiaua6 aLnify pue fuauno Lo Lyauaq ayj iof lueujuevinua
s 6unuor(M Lo IPIMb apf soueyue pue avasuoo 7oalcud oZ _T
�i�enp leluewuoiinu3 }o }uewl.leda(l
I