Estimating Secondary Pollutant Impacts from Single Sources
Kirk Baker
4/27/2012
All material shown is preliminary
1
Note
• All
All information presented is considered information presented is considered
preliminary by the U.S. EPA and provided to describe and illustrate potential approaches
describe and illustrate potential approaches and complex ideas
• This work is evolving and we are continuing to This work is evolving and we are continuing to
learn and improve these techniques so some or all of the information presented in this
or all of the information presented in this presentation may change
4/27/2012
All material shown is preliminary
2
Fine Scale/Single Source Modeling
Fine Scale/Single Source Modeling
•
•
Table below shows model scales, application types, and example models currently used for these needs (not all models shown)
Evaluate models consistently against available observation data and in the context the models will be applied
Application type and scale
Application
type and scale
Single source fenceline
Single source urban scale
Hybid single source/all sources "hot spot" analysis
All sources urban scale
All sources urban scale
All sources urban to regional scale
Single source long range transport assessments
All sources regional to continental scale
4/27/2012
SO2, NO2, Primary PM2.5
Primary PM2.5
AERMOD
AERMOD
Secondary PM2.5
CAMx, CMAQ
CAMx
CMAQ
CAMx, CMAQ
CALPUFF
CAMx, CMAQ
CAMx, CMAQ
CAMx
CMAQ
CAMx, CMAQ
CALPUFF
CAMx, CMAQ
All material shown is preliminary
Ozone
CAMx, CMAQ
CAMx
CMAQ
CAMx, CMAQ
CAMx, CMAQ
3
Focus Areas
Focus Areas
• How
How well do we represent plume placement and well do we represent plume placement and
chemistry near the source and far downwind?
• Single source modeling of long range transport
– Ozone
Ozone, PM2.5, deposition at Class I areas for PSD and PM2 5 deposition at Class I areas for PSD and
NSR programs
• Single source modeling on urban scale
Single source modeling on urban scale
– Ozone and PM2.5 (maybe visibility in the future) impact assessments for PSD and NSR programs
p
p g
4/27/2012
All material shown is preliminary
4
Single Source – Long Range Transport
Single Source Long Range Transport
• Long range transport of ozone and PM2.5 (air quality or AQ) and deposition (air quality related values or AQRVs) to Class I areas for PSD/NSR assessments
• Looking at existing and alternative modeling systems for long Looking at existing and alternative modeling systems for long
range transport assessments of PM2.5, AQRVs, and ozone
– Currently working with other Federal Agencies (USFS), contractors (
(ENVIRON/UNC) and internally to evaluate and compare modeling /
)
y
p
g
systems
• Mesoscale Model InferFace tool developed to convert WRF or MM5 output directly to CALPUFF, SCICHEM/SCIPUFF, and
MM5 output directly to CALPUFF, SCICHEM/SCIPUFF, and AERMOD
– beta release of the MMIF tool in February 2012 (EPA SCRAM site)
– AQMG team actively supporting the MMIF tool along with ENVIRON
AQMG team actively supporting the MMIF tool along with ENVIRON
4/27/2012
All material shown is preliminary
5
Consequence Analysis
Consequence Analysis
• Draft report “Comparison of Single‐Source Air Quality Assessment Techniques for Ozone, PM2.5, other Criteria Pollutants and AQRVs”
– will be available on the EPA SCRAM site when complete
• Compares
Compares modeled estimates of air quality (PM2.5) and air
modeled estimates of air quality (PM2 5) and air‐quality
quality related values (deposition) of various LRT modeling systems at Class I area receptors in the western U.S.
• Includes the application of CAMx, the regulatory version of I l d th
li ti
f CAM th
l t
i
f
CALPUFF using CALMET and MMIF, and the most recent version of CALPUFF that includes additional PM chemistry options
• Discussed in more detail in Bret Anderson’s Wednesday presentation
4/27/2012
All material shown is preliminary
6
Inert Tracer Experiment Evaluation
Inert Tracer Experiment Evaluation
• Draft final report “Documentation of CALPUFF and Other Long p
g
Range Transport Models using Tracer Test Experiment Data”
– http://www.epa.gov/ttn/scram/reports/EPA‐454_R‐12‐003.pdf
• R
Report details the evaluation of multiple LRT dispersion t d t il th
l ti
f
lti l LRT di
i
modeling systems using multiple tracer field studies: CALPUFF, SCIPUFF, HYSPLIT, FLEXPART, CAMx, and CALGRID
• Examines sensitivity of CALPUFF to variations in CALMET application and MMIF tool
• Sensitivities of HYSPLIT, CALPUFF, and CAMx
S ii ii
f HYSPLIT CALPUFF d CAM to alternative l
i
configuration options • Work done by US EPA/USFS; ENVIRON performed QA of Work done by US EPA/USFS; ENVIRON performed QA of
results and generated the report
4/27/2012
All material shown is preliminary
7
Regional Inert Tracer Field Experiments
Regional Inert Tracer Field Experiments
• Inert tracer release experiments provide a useful independent evaluation of long range transport and dispersion algorithms
– A known amount of tracer gas is released
– Release characteristics are typically well known
– Measured at downwind receptor sites
• Several used by US EPA for 1998 evaluation study of y
y
CALPUFF v5.8
– Savannah River Laboratory 1975 (SRL75)
– Great Plaines 1980 (GP80)
(
)
• 1983 Cross‐Appalachia Tracer Experiment (CAPTEX) used extensively to test MM5 and WRF FDDA
• 1994 European Tracer Experiment (ETEX) 1994 European Tracer Experiment (ETEX)
4/27/2012
All material shown is preliminary
8
Comparisons to 1998 CALPUFF Evaluation
Comparisons to 1998 CALPUFF Evaluation
Great Plains 1980 (GP80) Tracer Field Experiment
• CALPUFF “SLUG” near‐field option needed to reproduce “good” model performance on 600 km arc from 1998 EPA study
• Using different valid CALMET configurations, the maximum CALPUFF estimated tracer concentrations vary by a factor of 3
• Less variation in CALPUFF tracer estimates using MMIF, but there are few options with MMIF
Savannah River Lab (SRL75) Tracer Field Experiment
• Fitted Gaussian plume evaluation approach can be flawed
4/27/2012
All material shown is preliminary
9
CAPTEX 1983
CAPTEX 1983
•2 of the 5 tracer releases from CAPTEX‐83 are used to evaluate LRT models
•LRT models evaluated include CALPUFF, SCICHEM, HYSPLIT, FLEXPART, and CAMx
•All models applied using the same underlying 36 km MM5 output
4/27/2012
All material shown is preliminary
10
CAPTEX 1983
CAPTEX 1983
• CALPUFF sensitivities using CALMET show CALMET approach that most closest matches observations provides comparatively (among CALPUFF/CALMET )
g
sens.) worst agreement with tracer observations
• Using 36 km prognostic data only composite
only, composite performance metrics shown for both releases at right for multiple LRT models
multiple LRT models
CTEX3
CTEX5
*Figures courtesy of ENVIRON
4/27/2012
All material shown is preliminary
11
ETEX 1994
ETEX 1994
• 1994 European Tracer Experiment
4/27/2012
2.0
1.6
1.2
(1‐KS/100)
0.8
FMS/100
((1‐FB/2)
/ )
04
0.4
R^2
All material shown is preliminary
CAMx
FLEXPART
HY
YSPLIT
0.0
SC
CIPUFF
• LRT models evaluated include CALPUFF, SCICHEM, HYSPLIT, FLEXPART, and CAMx
• All models applied using the same underlying 36 km MM5 output
MM5 output
2.4
CA
ALPUFF
– Two releases from NW France in Oct. and Nov. 1994
– Tracer measured at 168 monitor sites
Rank (RANK) (Perfect = 4)
*Figure courtesy of ENVIRON
12
ETEX 1994
ETEX 1994
Observations (+36 hours)
SCIPUFF
CAMx
Observations (+60 hours)
b
(
h
)
SCIPUFF
CAMx
*Figures courtesy of ENVIRON
4/27/2012
All material shown is preliminary
13
Great Plains 1980
Great Plains 1980
• Maximum field measurements of tracers f
shown at right for all monitors in the study
• Used to evaluate CAMx
U dt
l t CAM
sensitivities to various Kz
schemes and horizontal advection algorithms (BOTT
advection algorithms (BOTT and PPM) in LRT report
• Currently building on that work by evaluating variety of
work by evaluating variety of horizontal and vertical advection approaches in CAMx (96 unique variations) 4/27/2012
All material shown is preliminary
14
Great Plaines 1980
Great Plaines 1980
•CAMx using YSU Kz values (left) does well at capturing plume placement
•Kz
Kz values using OB70 approach (right) advect
values using OB70 approach (right) advect mass too far to the SE at 600 km arc
mass too far to the SE at 600 km arc
•Performance of tracer suggest underlying WRF wind fields are well characterized
4/27/2012
All material shown is preliminary
15
Single Source – Urban scale
Single Source Urban scale
• Need near field plume chemistry to estimate single source p
y
g
impacts on ozone and secondarily formed PM2.5 for PSD & NSR programs
– Currently AERMOD is used to assess impacts of directly emitted PM2.5 Currently AERMOD is used to assess impacts of directly emitted PM2 5
• EPA granted Sierra Club petition with commitment to update Appendix W to address O3 and secondary PM2.5 impacts
• Need for technical basis to establish interpollutant trading ratios for PM2.5 to inform NSR offsets and SIP measures
• Develop modeling guidance for assessing single source D l
d li
id
f
i
i l
impacts on secondary pollutants such as ozone and PM2.5 relevant for PSD/NSR
4/27/2012
All material shown is preliminary
16
Single Source – Urban scale
Single Source Urban scale
• NACAA recommendation of a multi‐tier approach to assessing pp
g
single source impacts of PM2.5 (did not address ozone) that ranges from simple (screening level) to complex
1) Use
Use AERMOD for primary PM2.5 with offset ratios to approximate AERMOD for primary PM2 5 with offset ratios to approximate
secondary PM2.5; location specific offset ratios difficult to estimate
2) Use a Lagrangian model with plume chemistry (e.g. CALPUFF, SCICHEM)
3) Use a photochemical modeling system (e.g. CAMx, CMAQ)
• Source sensitivity, source apportionment, sub‐grid plume treatment?
• U.S. EPA needs to understand how these different approaches d
d
dh
h
d ff
h
are comparable and how best to apply models for this p p
purpose to develop guidance and review future permits pg
p
4/27/2012
All material shown is preliminary
17
Single Source Screening Level Tool
Single Source Screening Level Tool
Are screening level tools for single source estimates of g
g
secondary PM2.5 and ozone feasible? •
•
A screening tool would ideally provide a quick, reasonable, credible, and appropriately conservative assessment of single source secondary impacts
appropriately conservative assessment of single source secondary impacts before more complex applications are required
ENVIRON presented a reduced form single source screening model that estimates ozone impacts from single source emissions of VOC and/or NOX
estimates ozone impacts from single source emissions of VOC and/or NOX based on CAMx‐HDDM
– http://www.epa.gov/ttn/scram/10thmodconf/presentations/2‐21‐
Morris_Ozone_Screen_New_Srcs_EPA_10th_AQMC_Mar_2012.pdf
•
•
AQMG plans to explore this approach for ozone and PM2.5 to support single source NSR/PSD screening assessments
This modeling may provide some information for developing appropriate g yp
p g pp p
interpollutant trading ratios for PM2.5
4/27/2012
All material shown is preliminary
18
Plume Chemistry & Placement
Plume Chemistry & Placement
What are the best approaches when screening level assessments are not enough?
• Draft report “Evaluation of Chemical Dispersion Models using Atmospheric Plume Measurements from Field Experiments”
p
p
– Will be available on the EPA SCRAM site when complete
• Compares modeled estimates of air quality (ozone) and key p
precursors of various LRT modeling systems using two plume g y
g
p
measurement field experiments: 1999 TVA and 2006 Oklaunion
• Includes the application of CAMx, SCICHEM, the regulatory version of CALPUFF using CALMET and MMIF, and the most recent version g
,
of CALPUFF that includes additional PM chemistry options
• Additional detail on EPA’s SCICHEM application discussed in more detail in Jim Kelly’s presentation on Wednesday
y p
y
4/27/2012
All material shown is preliminary
19
Plume Chemistry & Placement
Plume Chemistry & Placement
• Application of photochemical modeling systems pp
p
g y
for single source O3 and PM2.5 assessments using a plume measurement field study (1999 TVA helicopter observations)
TVA helicopter observations)
• Source sensitivity: CMAQ and CAMx brute force and DDM emissions sensitivity
• Source contribution: CMAQ and CAMx source apportionment
• CAMx
CAM flexi‐nesting & sub‐grid plume treatment
fl i
i & b id l
• When available, CMAQ sub‐grid plume treatment (APT approach developed by EPRI)
(APT approach developed by EPRI)
4/27/2012
All material shown is preliminary
20
Modeling Setup
Modeling Setup
• 1999 hour specific CEM emissions for TVA Cumberland
i i
f TVA C b l d
–
http://camddataandmaps.epa.gov/gdm/index
.cfm?fuseaction=emissions.wizard
• 1999 hour specific biogenics
estimated with BEIS model
ti t d ith BEIS
d l
• 2001 NEI based anthropogenic emissions
• Meteorological inputs Meteorological inputs
generated using the WRF model version 3.3
• Photochemical models used: CMAQ v.4.7.1 and CAMx v5.40
• Domains (34 layers): 36 km CONUS  12 and 4 km
4/27/2012
All material shown is preliminary
21
2 week (July 1999) episode maximum impact on NOX (NO+NO2) from source NOX
*red dot indicates source location 4/27/2012
All material shown is preliminary
22
2 week (July 1999) episode maximum impact on ozone from source NOX
Ozone tracking has not been
not been implemented yet in CMAQ
*red dot indicates source location 4/27/2012
All material shown is preliminary
23
2 week (July 1999) episode maximum impact on elemental carbon from source EC
*red dot indicates source location Current version of CAMx only l
includes DDM g
for gases
4/27/2012
All material shown is preliminary
24
2 week (July 1999) episode maximum impact on PM2.5 sulfate from source SOX
*red dot indicates source location Current version of CAMx only l
includes DDM g
for gases
4/27/2012
All material shown is preliminary
25
Source – Receptor Proximity Issues
Source Receptor Proximity Issues
•
Source‐receptor spatial relationships are sometimes not resolved at 1 km
– In
In Figure below (from Illinois EPA), the source is outlined in green and the receptor is Figure below (from Illinois EPA), the source is outlined in green and the receptor is
labeled ‘A’ (the meteorological station is labeled ‘B’)
4/27/2012
All material shown is preliminary
26
Sub‐grid
Sub
grid Plume Treatment
Plume Treatment
• What types of situations require the use of sub‐grid plume treatment? – Source‐receptor proximity may be such that they are in the g
y
g g
p
same grid cell or only a single grid cell apart
– How to configure plume‐in‐grid for PSD/NSR types of applications?
– What is the best way to track source contribution and/or y
/
sensitivity when combining sub‐grid and grid‐resolved techniques?
• When
When would the use of sub
would the use of sub‐grid
grid plume treatment be plume treatment be
inappropriate? • What grid resolution (horizontal and vertical) could preclude the need for sub grid plume treatment?
preclude the need for sub‐grid plume treatment?
4/27/2012
All material shown is preliminary
27
Source – Receptor Proximity Issues
Source Receptor Proximity Issues
• Plume measurement studies provide one opportunity to evaluate the utility of sub‐grid
the utility of sub
grid plume treatments for PSD/NSR
• Modeling system and sub‐
d l
d b
grid plume evaluation work is critical to develop p
and improve modeling guidance for these situations
4/27/2012
200 M
1 km w/ PiG
/
200 M
1 km w/ PiG
All material shown is preliminary
*Graphics from Illinois EPA
28
Last Slide
Last Slide
• OAQPS interested in input and participation from State and Local Agencies as we move forward to better understand single source modeling for secondary p
pollutants (O3 and PM2.5)
(
)
• Also looking for contributions from US EPA Regional offices and other government agencies
• Special session at this year
Special session at this year’ss CMAS meeting in Chapel CMAS meeting in Chapel
Hill (October 2012) on modeling secondary impacts from single sources (www.cmascenter.org)
• Acknowledge the contributions of Bret Anderson, Jim A k
l d
h
ib i
fB A d
Ji
Kelly, Tyler Fox, ENVIRON, CSC, OAQPS/AQAD emissions and modeling groups
4/27/2012
All material shown is preliminary
29