vvEPA
United Mates
Environmental Protection
Air Quality Modeling Technical Support Document:
Proposed Utility NESHAP
-------
EPA-456/R-11-002
January 2011
Air Quality Modeling Technical Support Document:
Proposed Utility NESHAP
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC 27711
-------
I Introduction
This document describes the air quality modeling performed by EPA in support of the proposed
utility NESHAP. A national scale air quality modeling analysis was performed to estimate the
impact of the sector emissions changes on future year: annual and 24-hour PM2.5
concentrations, 8-hr maximum ozone, total mercury deposition, as well as visibility impairment.
Air quality benefits are estimated with the Community Multi-scale Air Quality (CMAQ) model.
CMAQ simulates the numerous physical and chemical processes involved in the formation,
transport, and destruction of ozone, particulate matter and air toxics. In addition to the CMAQ
model, the modeling platform includes the emissions, meteorology, and initial and boundary
condition data which are inputs to this model.
Emissions and air quality modeling decisions are made early in the analytical process. For this
reason, it is important to note that the inventories used in the air quality modeling and the
benefits modeling may be slightly different than the final utility sector inventories presented in
the RIA. However, the air quality inventories and the final rule inventories are generally
consistent, so the air quality modeling adequately reflects the effects of the rule.
II. Photochemical Model Version, Inputs and Configuration
Photochemical grid models use state of the science numerical algorithms to estimate pollutant
formation, transport, and deposition over a variety of spatial scales that range from urban to
continental. Emissions of precursor species are injected into the model where they react to form
secondary species such as ozone and then transport around the modeling domain before
ultimately being removed by deposition or chemical reaction.
The 2005-based CMAQ modeling platform was used as the basis for the air quality modeling for
this rule. This platform represents a structured system of connected modeling-related tools and
data that provide a consistent and transparent basis for assessing the air quality response to
projected changes in emissions. The base year of data used to construct this platform includes
emissions and meteorology for 2005. The platform is intended to support a variety of regulatory
and research model applications and analyses. This modeling platform and analysis is described
below.
A. Model version
The Community Multi-scale Air Quality (CMAQ) model v4.7.1 (www.cmaq-model.org) is a
state of the science three-dimensional Eularian "one-atmosphere" photochemical transport model
used to estimate air quality (Appel et al., 2008; Appel et al., 2007; Byun and Schere, 2006).
CMAQ simulates the formation and fate of photochemical oxidants, ozone, primary and
secondary PM concentrations, and air toxics over regional and urban spatial scales for given
input sets of meteorological conditions and emissions. CMAQ is applied with the AERO5
1
-------
aerosol module, which includes the ISORROPIA inorganic chemistry (Nenes et al., 1998) and a
secondary organic aerosol module (Carlton et al., 2010). The CMAQ model is applied with
sulfur and organic oxidation aqueous phase chemistry (Carlton et al., 2008) and the carbon-bond
2005 (CB05) gas-phase chemistry module (Gery et al., 1989).
B. Model domain and grid resolution
The modeling analyses were performed for a domain covering the continental United States, as
shown in Figure II-l. This domain has a parent horizontal grid of 36 km with two finer-scale 12
km grids over portions of the eastern and western U.S. The model extends vertically from the
surface to 100 millibars (approximately 15 km) using a sigma-pressure coordinate system. Air
quality conditions at the outer boundary of the 36 km domain were taken from a global model
and vary in time and space. The 36 km grid was only used to establish the incoming air quality
concentrations along the boundaries of the 12 km grids. Only the finer grid data were used in
determining the impacts of the emission standard program changes. Table II-l provides
geographic information about the photochemical model domains.
Table II-l. Geographic elements of domains used in photochemical modeling.
Map Projection
Grid Resolution
Coordinate Center
True Latitudes
Dimensions
Vertical extent
Photochemical Modeling Configuration
National Grid
Western U.S. Fine Grid
Eastern U.S. Fine Grid
Lambert Conformal Projection
36km
12km
12km
97degW, 40degN
33degNand45degN
148x112x14
213x192x14
279 x 240 x 14
14 Layers: Surface to 100 millibar level (see Table II-3)
-------
Figure II-1. Map of the photochemical modeling domains. The black outer box denotes the 36
km national modeling domain; the red inner box is the 12 km western U.S. grid; and the blue
inner box is the 12 km eastern U.S. grid.
C. Modeling Time-period
The 36 km and both 12 km modeling domains were modeled for the entire year of 2005. Data
from the entire year were utilized when looking at the estimation of PM2.5, total mercury
deposition, and visibility impacts from the regulation. Data from April through October is used
to estimate ozone impacts.
D. Model Inputs: Emissions, Meteorology and Boundary Conditions
The 2005-based modeling platform was used for the air quality modeling of future emissions
scenarios. In addition to the photochemical model, the modeling platform also consists of the
base- and future-year emissions estimates, meteorological fields, as well as initial and boundary
condition data which are all inputs to the air quality model.
1. Emissions Input Data
The emissions data used in the base year and future reference and future emissions adjustment
case are based on the 2005 v4.1 platform. The emissions cases use different emissions data for
some pollutants than the official v4 platform to use data intended only for the rule development
and not for general use. Unlike the 2005 v4 platform, the configuration for this modeling
application included mercury emissions from the National Air Toxics Assessment Inventory and
some industrial boiler sector mercury emissions more consistent with the engineering analysis
for the Industrial/Commercial/Institutional Boilers and Process Heaters NESHAP. Emissions for
the future years for the EGU sector utilized information collected from the utility MACT
information collection request. The information collection request informed existing HCL, NOX,
HG, and PM controls. In addition this data was used to supply HCL removal rates from selected
-------
control technology. Emissions are processed to photochemical model inputs with the SMOKE
emissions modeling system (Houyoux et al., 2000).
The 2016 reference case is intended to represent the emissions associated with growth and
controls in that year projected from the 2005 simulation year. The United States EGU point
source emissions estimates for the future year reference and control case are based on an
Integrated Planning Model (IPM) run for criteria pollutants, hydrochloric acid, and mercury in
2016. Both control and growth factors were applied to a subset of the 2005 non-EGU point and
non-point to create the 2016 reference case. The 2005 v4 platform 2014 projection factors were
the starting point for most of the 2016 SMOKE-based projections. The mercury projections for
non-EGU point sources accounted for emission reductions expected in the future due to
NESHAP for various non-EGU source categories that were finalized or expected to be finalized
prior to the Utility proposal including the Boiler MACT, Gold Mine NESHAP and Electric Arc
Furnace NESHAP. The 2016 reference scenario for the utility sector includes growth and control
for the industry from the 2005 sector emissions estimates. The estimated total anthropogenic
emissions and emissions for the utility sector used in this modeling assessment are shown in
Table 112.
Table II.2 Estimated total inventory and EGU sector emissions for each modeling scenario.
Scenario
2005cr_hg
2016cr2
2016cr2 hg control 1
Scenario
2005cr_hg
2016cr2
2016cr2 hg control 1
Sector
EGU(PTIPM)
All
EGU(PTIPM)
All
EGU(PTIPM)
All
Sector
EGU(PTIPM)
All
EGU(PTIPM)
All
EGU(PTIPM)
All
VOC
40,950
17,613,543
40,845
14,390,421
38,217
14,387,792
HG2
21
33
7
16
2
11
NOx
3,726,459
22,216,093
1,769,764
15,019,836
1,618,199
14,868,270
HGO
30
64
21
42
5
26
Emissions
CO
601,564
83,017,436
691,310
59,148,384
656,245
59,113,319
Emissions
HG PM25
1.6
8.5
0.7
5.9
0.4
5.6
tons/year)
SO2
10,380,786
15,050,209
3,577,698
7,245,595
1,220,379
4,888,276
tons/year)
HCL
351,592
429,223
74,089
140,638
8,802
75,351
PM10
615,095
13,031,716
523,504
12,772,091
358,165
12,606,752
CL2
99
6,409
6,050
6,050
PM2.5
508,903
4,400,680
384,320
4,022,846
291,044
3,929,570
NH3
21,684
3,762,641
36,655
3,897,033
36,982
3,897,360
As part of the analysis for this rulemaking, the modeling system was used to calculate daily and
annual PM2.5 concentrations, 8-hr maximum ozone, annual total mercury deposition levels and
visibility impairment. Model predictions are used in a relative sense to estimate scenario-
specific, future-year design values of PM2.5 and ozone. Specifically, we compare a 2016
reference scenario, a scenario without the utility sector controls, to a 2016 control scenario which
-------
includes the adjustments to the utility sector. This is done by calculating the simulated air
quality ratios between any particular future year simulation and the 2005 base.
These predicted ratios are then applied to ambient base year design values. The design value
projection methodology used here followed EPA guidance for such analyses (USEPA, 2007).
Additionally, the raw model outputs are also used in a relative sense as inputs to the health and
welfare impact functions of the benefits analysis. Only model predictions for mercury deposition
were analyzed using absolute model changes, although percent changes between the control case
and future baselines are also estimated.
2. Meteorological Input Data
The gridded meteorological input data for the entire year of 2005 were derived from simulations
of the Pennsylvania State University / National Center for Atmospheric Research Mesoscale
Model. This model, commonly referred to as MM5, is a limited-area, nonhydrostatic, terrain-
following system that solves for the full set of physical and thermodynamic equations which
govern atmospheric motions. Meteorological model input fields were prepared separately for
each of the three domains shown in Figure II-l using MM5 version 3.7.4. The MM5 simulations
were run on the same map projection as shown in Figure II-l.
All three meteorological model runs were configured similarly. The selections for key MM5
physics options are shown below:
• Pleim-Xiu PEL and land surface schemes
Kain-Fritsh 2 cumulus parameterization
• Reisner 2 mixed phase moisture scheme
• RRTM longwave radiation scheme
Dudhia shortwave radiation scheme
Three dimensional analysis nudging for temperature and moisture was applied above the
boundary layer only. Analysis nudging for the wind field was applied above and below the
boundary layer. The 36 km domain nudging weighting factors were 3.0 x 104 for wind fields and
temperatures and 1.0 x 105 for moisture fields. The 12 km domain nudging weighting factors
were 1.0 x 104 for wind fields and temperatures and 1.0 x 105 for moisture fields.
-------
Table II-3. Vertical layer structure (heights are layer top).
CMAQ Layers
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
MM5 Layers
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Sigma P
1.000
0.995
0.990
0.985
0.980
0.970
0.960
0.950
0.940
0.930
0.920
0.910
0.900
0.880
0.860
0.840
0.820
0.800
0.770
0.740
0.700
0.650
0.600
0.550
0.500
0.450
0.400
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0.000
Approximate Height
(m)
0
38
77
115
154
232
310
389
469
550
631
712
794
961
1,130
1,303
1,478
1,657
1,930
2,212
2,600
3,108
3,644
4,212
4,816
5,461
6,153
6,903
7,720
8,621
9,625
10,764
12,085
13,670
15,674
Approximate Pressure
(mb)
1000
995
991
987
982
973
964
955
946
937
928
919
910
892
874
856
838
820
793
766
730
685
640
595
550
505
460
415
370
325
280
235
190
145
100
All three sets of model runs were conducted in 5.5 day segments with 12 hours of overlap for
spin-up purposes. All three domains contained 34 vertical layers with an approximately 38m
deep surface layer and a 100 millibar top. The MM5 and CMAQ vertical structures are shown in
Table II-3 and do not vary by horizontal grid resolution. The meteorological outputs from all
three MM5 sets were processed to create model-ready inputs for CMAQ using the MCIP
processor.
Before initiating the air quality simulations, it is important to identify the biases and errors
associated with the meteorological modeling inputs. The 2005 MM5 model performance
evaluations used an approach which included a combination of qualitative and quantitative
analyses to assess the adequacy of the MM5 simulated fields. The qualitative aspects involved
comparisons of the model-estimated synoptic patterns against observed patterns from historical
weather chart archives. Additionally, the evaluations compared spatial patterns of estimated to
observed monthly average rainfall and checked maximum planetary boundary layer (PEL)
heights for reasonableness.
Qualitatively, the model fields closely matched the observed synoptic patterns, which is not
unexpected given the use of nudging. The operational evaluation included statistical
-------
comparisons of model/observed pairs (e.g., mean normalized bias, mean normalized error, index
of agreement, root mean square errors, etc.) for multiple meteorological parameters. For this
portion of the evaluation, five meteorological parameters were investigated: temperature,
humidity, shortwave downward radiation, wind speed, and wind direction. The three individual
MM5 evaluations are described elsewhere (Baker, 2009a, b, c). It was ultimately determined that
the bias and error values associated with all three sets of 2005 meteorological data were
generally within the range of past meteorological modeling results that have been used for air
quality applications.
3. Initial and Boundary Conditions
The lateral boundary and initial species concentrations are provided by a three-dimensional
global atmospheric chemistry model, the GEOS-CHEM model (standard version 7-04-11). The
global GEOS-CHEM model simulates atmospheric chemical and physical processes driven by
assimilated meteorological observations from the NASA's Goddard Earth Observing System
(GEOS). This model was run for 2005 with a grid resolution of 2.0 degree x 2.5 degree
(latitude-longitude) and 30 vertical layers up to 100 mb. The predictions were used to provide
one-way dynamic boundary conditions at three-hour intervals and an initial concentration field
for the 36 km CMAQ simulations. The 36 km photochemical model simulation is used to supply
initial and hourly boundary concentrations to the 12 km domains. The future base conditions
from the 36 km coarse grid modeling were used as the initial/boundary state for all subsequent
future year 12 km finer grid modeling scenarios.
E. Base Case Model Performance Evaluation
1. PM2.5
An operational model performance evaluation for the speciated components of PM2.5 (e.g.,
sulfate, nitrate, elemental carbon, organic carbon, etc.) was conducted using 2005 state/local
monitoring data in order to estimate the ability of the modeling system to replicate base year
concentrations. The evaluation of PM2.5 component species includes comparisons of predicted
and observed concentrations of sulfate (SO4), nitrate (NO3), ammonium (NH4), elemental
carbon (EC), and organic carbon (OC). PM2.5 ambient measurements for 2005 were obtained
from the Chemical Speciation Network (CSN) and the Interagency Monitoring of PROtected
Visual Environments (IMPROVE). The CSN sites are generally located within urban areas and
the IMPROVE sites are typically in rural/remote areas. The measurements at CSN and
IMPROVE sites represent 24-hour average concentrations. In calculating the model performance
metrics, the modeled hourly species predictions were aggregated to the averaging times of the
measurements.
Model performance statistics were calculated for observed/predicted pairs of daily
concentrations. The aggregated metrics and number (N) of prediction-observation pairs are
shown by chemical specie and quarter in Table II-4. The "acceptability" of model performance
was judged by comparing our 2005 performance results to the range of performance found in
recent regional PM2.5 model applications for other, non-EPA studies. Overall, the mean bias
(bias) and mean error (error) statistics shown in Table II-4 are within the range or close to that
-------
found by other groups in recent applications. The model performance results give us confidence
that our application of CMAQ using this modeling platform provides a scientifically credible
approach for assessing PM2.5 concentrations for the purposes of this assessment.
TABLE II-4. Model performance metrics for speciated PM2.5 averaged by quarter.
Metric Quarter
Number
Mean Observed
(Hg/m3)
Mean Predicted
(ug/m3)
Bias
(ug/m3)
Error
(Hg/m3)
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
Ammonium Organic Elemental
Sulfatelon Nitrate Ion Ion Carbon Carbon
6,218
6,537
6,108
5,894
2.69
3.80
5.43
2.45
2.37
3.45
4.39
5,870
6,146
5,828
5,757
2.39
0.83
0.45
1.28
2.36
1.21
0.48
2.31 1.84
-0.27
-0.28
-0.88
-0.08
1.05
1.25
1.73
0.74
0.12
0.49
0.07
0.63
1.37
0.89
4,045
4,075
3,869
3,763
1.72
1.49
1.84
1.26
1.50
1.47
1.39
1.32
0.05
0.33
-0.12
0.36
0.77
0.81
0.46 0.80
1.17
0.66
6,368
6,552
6,061
5,697
0.51
0.48
0.52
0.61
0.82
0.64
0.66
0.81
0.34
0.15
0.16
0.24
0.47
6,305
6,516
6,152
5,792
0.40
0.66
0.78
0.52
5.20
4.06
4.69
5.46
4.87
3.50
4.03
5.01
4.90
0.35 3.65
0.35
0.44
4.24
5.05
2. Ozone
An operational model performance evaluation for hourly and eight-hour daily maximum ozone
was conducted in order to estimate the ability of the modeling system to replicate the base year
concentrations. Ozone measurements were taken from the 2005 State/local monitoring site data
in the Air Quality System (AQS) Aerometric Information Retrieval System (AIRS). The ozone
metrics covered in this evaluation include one-hour and eight-hour average daily maximum
ozone bias and error. The evaluation principally consists of statistical assessments of model
versus observed pairs that were paired in time and space on an hourly and/or daily basis,
depending on the sampling frequency of each measurement site (measured data). This ozone
model performance was limited to the prediction-observation pairs where observed ozone
exceeded or equaled 60 ppb. This cutoff was applied to evaluate the model on days of elevated
ozone which are more policy relevant. Aggregated performance metrics by quarter are shown in
Table II-5.
-------
TABLE II-5. Model performance metrics for dail
Quarter
2
3
Quarter
2
3
N
24,357
27,430
N
24,370
27,447
1-hr Daily
Observed
0.0
0.0
8-hr Daily
Observed
0.0
0.0
Maximum Ozone (ppb)
Predicted Bias
0.0
0.0
0.0
0.0
Maximum Ozone (ppb)
Predicted Bias
0.0
0.0
0.0
0.0
y maximum ozone by quarter.
Error
0.0
0.0
Error
0.0
0.0
*metrics estimated where observed ozone > 60 ppb
3. Mercury Wet Deposition
Model estimated weekly mercury wet deposition is compared to observation data to assess model
skill simulating this component of mercury deposition. Mercury wet deposition measurements
are weekly totals taken at sites that are part of the Mercury Deposition Network
(http://nadp. sws.uiuc.edu/MDN/) which operates under the National Atmospheric Deposition
Program. In addition to mercury wet deposition, the network sites also collect rainfall data which
is also evaluated against estimates used by the photochemical model from prognostic
meteorological model output.
TABLE II-6. Model performance metrics for mercury wet deposition and rainfall by quarter.
Total Mercury Wet Deposition (u.g/m2)
Observed Predicted Bias Error
Quarter
N
1
2
3
4
Quarter
1
2
3
4
798
853
840
748
N
1,073
1,082
1,118
1,037
0.15
0.26
0.31
0.13
F
Observed
17.27
19.41
20.64
20.18
0.23
0.28
0.16
-0.80
ainfall (mrr
Predicted
18.10
23.50
26.51
17.22
0.14
0.06
-0.12
-1.27
]_
Bias
0.83
4.09
5.87
-2.97
0.18
0.24
0.24
1.47
Error
8.34
14.45
19.77
10.38
-------
III. References
Appel, K.W., Bhave, P.V., Gilliland, A.B., Sarwar, G., Roselle, S.J., 2008. Evaluation of the community multiscale
air quality (CMAQ) model version 4.5: Sensitivities impacting model performance; Part II - paniculate matter.
Atmospheric Environment 42, 6057-6066.
Appel, K.W., Gilliland, A.B., Sarwar, G., Gilliam, R.C., 2007. Evaluation of the Community Multiscale Air Quality
(CMAQ) model version 4.5: Sensitivities impacting model performance Part I - Ozone. Atmospheric Environment
41, 9603-9615.
Baker, K., Dolwick, P., 2009a. Meteorological Modeling Performance Evaluation for the Annual 2005 Continental
U.S. 36-km Domain Simulation. US Environmental Protection Agency OAQPS.
Baker, K., Dolwick, P., 2009b. Meteorological Modeling Performance Evaluation for the Annual 2005 Eastern U.S.
12-km Domain Simulation. US Environmental Protection Agency OAQPS, RTF.
Baker, K., Dolwick, P., 2009c. Meteorological Modeling Performance Evaluation for the Annual 2005 Western U.S.
12-km Domain Simulation, in: EPA, U. (Ed.). US Environmental Protection Agency OAQPS.
Byun, D., Schere, K.L., 2006. Review of the governing equations, computational algorithms, and other components
of the models-3 Community Multiscale Air Quality (CMAQ) modeling system. Applied Mechanics Reviews 59, 51-
77.
Carton, A.G., Bhave, P.V., Napelenok, S.L., Edney, E.D., Sarwar, G., Finder, R.W., Pouliot, G.A., Houyoux, M.,
2010. Model Representation of Secondary Organic Aerosol in CMAQv4.7. Environmental Science & Technology
44, 8553-8560.
Carton, A.G., Turpin, B.J., Altieri, K.E., Seitzinger, S.P., Mathur, R., Roselle, S.J., Weber, R.J., 2008. CMAQ
Model Performance Enhanced When In-Cloud Secondary Organic Aerosol is Included: Comparisons of Organic
Carbon Predictions with Measurements. Environmental Science & Technology 42, 8798-8802.
Gery, M.W., Whitten, G.Z., Killus, J.P., Dodge, M.C., 1989. A Photochemical Kinetics Mechanism for Urban and
Regional Scale Computer Modeling. Journal of Geophysical Research-Atmospheres 94, 12925-12956.
Nenes, A., Pandis, S.N., Pilinis, C., 1998. ISORROPIA: A new thermodynamic equilibrium model for multiphase
multicomponent inorganic aerosols. Aquatic Geochemistry 4, 123-152.
USEPA, 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality
Goals for Ozone, PM2.5, and Regional Haze, RTF.
10
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-456/R-11 -002
Environmental Protection [Name of Division] January 2011
Agency Research Triangle Park, NC
------- |