ITiH
W""*" m***
J^t
Air Quality Modeling Technical Support
Document: Changes to the Renewable Fuel
Standard Program
-------
EPA 454/R-10-001
February 2010
Air Quality Modeling Technical Support Document:
Changes to the Renewable Fuel Standard Program
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC 27711
February 2010
-------
I. Introduction
This document describes the air quality modeling performed by EPA in support of the
final revisions to the National Renewable Fuel Standard rule (commonly known as RFS2). A
national scale air quality modeling analysis was performed to estimate the effect of the rule on
future year: annual and 24-hour PM2.5 concentrations, daily maximum 8-hour ozone
concentrations, annual nitrogen and sulfur deposition levels, and select annual and seasonal air
toxic concentrations (formaldehyde, acetaldehyde, ethanol, benzene, 1,3-butadiene, acrolein).
To model the air quality benefits of this rule we used the Community Multiscale Air Quality
(CMAQ)1 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.
It is critical to note that a key limitation of the air quality modeling analysis is that it
employed interim emission inventories, which were somewhat enhanced compared to what was
described in the proposal, but due to the timing of the analysis did not include some of the later
enhancements and corrections of the final emission inventories presented in the FRM (see
Section 3.3 of the RIA). Most significantly, our modeling of the air quality impacts of the
renewable fuel volumes required by RFS2 relied upon interim inventories that assumed that
ethanol will make up 34 of the 36 billion gallon renewable fuel mandate, that approximately 20
billion gallons of this ethanol will be in the form of E85, and that the use of E85 results in fewer
emissions of direct PM25 from vehicles. The emission impacts and air quality results would be
different if, instead of E85, more non-ethanol biofuels are used or mid-level ethanol blends are
approved. There are additional, important limitations and uncertainties associated with the
interim inventories that must be kept in mind when considering the results. These limitations
and uncertainties are described in more detail in Section 3.4.1.3 of the RIA.
II. CMAQ Model Version, Inputs and Configuration
The 2005-based CMAQ modeling platform was used as the basis for the air quality
modeling of the two future baselines and the RFS2 future control scenario for this final 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 was developed by the U.S. EPA's Office of Air Quality
Planning and Standards in collaboration with the Office of Research and Development and is
intended to support a variety of regulatory and research model applications and analyses. This
modeling platform and analysis is fully described below.
1 Byun, D.W., and K. L. Schere, 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, Volume 59, Number 2 (March 2006), pp. 51-77.
-------
A. Model version
CMAQ is a non-proprietary computer model that simulates the formation and fate of
photochemical oxidants, primary and secondary PM concentrations, acid deposition, and air
toxics, over regional and urban spatial scales for given input sets of meteorological conditions
and emissions. The CMAQ model version 4.7 was most recently peer-reviewed in February of
2009 for the U.S. EPA.2 The CMAQ model is a well-known and well-respected tool and has
been used in numerous national and international applications.3'4'5'6 This 2005 multi-pollutant
modeling platform used the latest publicly-released CMAQ version 4.77 with a minor internal
change made by the U.S. EPA CMAQ model developers intended to speed model runtimes when
only a small subset of toxics species are of interest.8 This version reflects updates in a number of
areas to improve the underlying science, including:
1) an enhanced secondary organic aerosol (SOA) mechanism to include chemistry of
isoprene, sesquiterpene, and aged in-cloud biogenic SOA in addition to terpene,
2) an improved vertical convective mixing algorithm;
3) an improved heterogeneous reaction involving nitrate formation, and
4) an updated gas-phase chemistry mechanism, Carbon Bond 05 (CB05), with extensions
to model explicit concentrations of air toxic species as well as chlorine and mercury.
This mechanism, CB05-toxics, also computes concentrations of species that are involved in
aqueous chemistry and that are precursors to aerosols. Chapter 3 of the RIA discusses in detail
2 Allen, D., Burns, D., Chock, D., Kumar, N., Lamb, B., Moran, M. (February 2009 Draft Version). Report on the
Peer Review of the Atmospheric Modeling and Analysis Division, NERL/ORD/EPA. U.S. EPA, Research Triangle
Park, NC
3 Hogrefe, C., Biswas, J., Lynn, B., Civerolo, K., Ku, J.Y., Rosenthal, J., et al. (2004). Simulating regional-scale
ozone climatology over the eastern United States: model evaluation results. Atmospheric Environment, 38(17),
2627-2638.
6 United States Environmental Protection Agency. (2008). Technical support document for the final
locomotive/marine rule: Air quality modeling analyses. Research Triangle Park, N.C.: U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, Air Quality Assessment Division.
5 Lin, M., Oki, T., Holloway, T., Streets, D.G., Bengtsson, M., Kanae, S. (2008). Long-range transport of acidifying
substances in East Asia-Part I: Model evaluation and sensitivity studies. Atmospheric Environment, 42(24), 5939-
5955.
6 Lin, M., Oki, T., Holloway, T., Streets, D.G., Bengtsson, M., Kanae, S. (2008). Long-range transport of acidifying
substances in East Asia-Part I: Model evaluation and sensitivity studies. Atmospheric Environment, 42(24), 5939-
5955.
7 CMAQ version 4.7 was released on December, 2008. It is available from the Community Modeling and Analysis
System (CMAS) as well as previous peer-review reports at: http://www.cmascenter.org.
8 CMAQ version 4.7 was released on December, 2008. It is available from the Community Modeling and Analysis
System (CMAS) as well as previous peer-review reports at: http://www.cmascenter.org.
-------
the chemical mechanism, SOA formation, and details about the improvements made to the SOA
mechanism within this recent release of CMAQ.
B. Model domain and grid resolution
The CMAQ 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 did not change over the simulations. In turn, 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 RFS2 program changes.
Table II-l provides some basic geographic information regarding the CMAQ domains.
Table II-l. Geographic elements of domains used in RFS2 modeling.
Map Projection
Grid Resolution
Coordinate Center
True Latitudes
Dimensions
Vertical extent
CMAQ 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-l. Map of the CMAQ modeling domain. The black outer box denotes the 36 km
national modeling domain; the red inner box is the 12 km western U.S. fine grid; and the
blue inner box is the 12 km eastern U.S. fine grid.
C. Valid Modeling Days
The 36 km and both 12 km CMAQ modeling domains were modeled for the entire year
of 2005.9 For the 8-hour ozone results, we are only using modeling results from the period
between May 1 and September 30, 2005. This 153-day period generally conforms to the ozone
season across most parts of the U.S. and contains the majority of days with observed high ozone
concentrations in 2005. Data from the entire year were utilized when looking at the toxics
impacts from the regulation.
Normally, all 365 model days would also have been used in the estimation of PM2.5 and
visibility impacts; however during the RFS2 modeling, an error was discovered in the aqueous
9 We also modeled 10 days at the end of December 2004 as a modeled "ramp up" period. These days are used to
minimize the effects of initial conditions and are not considered as part of the output analyses.
-------
phase chemistry routines of CMAQ v4.7. This error10 caused simulated hourly sulfate
concentrations to increase sporadically and in an unrealistic manner over a very limited number
of grid-cell hours over the RFS2 simulations. While this artifact has subsequently been removed
from CMAQ v4.7, the RFS2 modeling schedule did not allow for the simulations to be redone.
Instead, we simply invalidated any day that contained evidence of the aqueous phase problem
and used the remaining data to determine the "true" model signal from the RFS2 scenarios. The
following invalidation criteria were used. Any day in which there were five or more grid cell-
hours that had greater than 50 ug/m3 difference in sulfate concentrations between the future base
and control cases was invalidated. Additionally any day with a single grid cell-hour difference
exceeding 250 ug/m3 was invalidated. Based on these invalidation criteria, nine days were
removed from the EUS12 analysis and two days were removed from the WUS12 analysis11.
D. Model Inputs: Emissions, Meteorology and Boundary Conditions
The 2005-based CMAQ modeling platform was used for the air quality modeling of
future baseline emissions and control scenarios. As noted in the introduction, in addition to the
CMAQ model, the modeling platform also consists of the base- and future-year emissions
estimates (both anthropogenic and biogenic), meteorological fields, as well as initial and
boundary condition data which are all inputs to the air quality model.
1. Base Year and Future Baseline Emissions: The emissions modeling TSD, found in the
docket for this rule (EPA-HQ-OAR-2005-0161), contains a detailed discussion of the emissions
inputs used in our air quality modeling as well as Section 3.1 in the final RFS2 RIA. We have
provided a brief summary of the base year and future baseline emissions used for the air quality
modeling. The emissions data used in the base year and each of the future base cases are based
on the 2005 v4 platform. The RFS2 cases use some different emissions data than the official v4
platform for two reasons: (1) the RFS2 modeling was done prior to the completion of the
platform and (2) the RFS2 modeling used data intended only for the rule development and not
for general application. The US EGU point source emissions estimates for all 2022 future year
base cases are based on an Integrated Planning Model (TPM) run for criteria pollutants,
hydrochloric acid, and mercury in 2020. The year 2020 was used since it was the year closest to
the 2022 modeling year supported by the IPM model. Both control and growth factors were
applied to a subset of the 2005 non-EGU point and nonpoint to create each of the 2022 future
base cases. The 2002 v3.1 platform 2020 projection factors were the starting point for most of
the RFS2 year 2022 SMOKE-based projections. Ethanol plant replacements and additions were
included in the 2005 base and 2022 future baselines as well as biodiesel additions and portable
fuel containers.
It should be noted that the emission inventories used in the air quality and benefits
modeling were enhanced compared to what was described in the proposal, but did not include
10 This model artifact is discussed in more detail in an August 5th, 2009 document prepared by Shawn Roselle and
Ann Marie Carlton. This document has been placed in the rule docket (EPA-HQ-OAR-2005-0161-DRAFT-2902).
11 The days to be removed for the EUS12 are: 1/25, 2/25, 3/04, 3/05, 3/13, 3/14. 12/08, 12/09, 12/12. The days to
be removed for the WUS12 are: 1/04 and 1/27.
-------
some of the later enhancements and corrections of the final emission inventories presented in this
FRM.
2. RFS2 Modeling Scenarios: As part of our analysis for this rulemaking, the CMAQ
modeling system was used to calculate daily and annual PM2.5 concentrations, 8-hour ozone
concentrations, annual and seasonal air toxics concentrations, and total nitrogen and sulfur
deposition levels for each of the following emissions scenarios:
2005 base year
2022 baseline projection (RFS1 Mandate) of 7.5 billion gallons of renewable fuels
2022 baseline projection (Annual Energy Outlook (AEO) 2007) volume of
roughly 14 billion gallons of renewable fuels
2022 control case projection (implementation of RFS2; also referred to as EISA
(Energy Independence and Security Act of 2007)
Model predictions are used in a relative sense to estimate scenario-specific, future-year
design values of PM2 5 and ozone. This is done by calculating the simulated air quality ratios
between any particular future year simulation and the 2005 base. These predicted change ratios
are then applied to ambient base year design values. The design value projection methodology
used here followed EPA guidance12 for such analyses. 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. Model predictions for air toxics as well as nitrogen and sulfur deposition were
analyzed for an absolute change and percent change between the control case and two future
baselines.
3. Sensitivity analyses looking at impacts of chosen speciation profiles: During the
course of the RFS2 modeling, two issues arose concerning the approaches used to speciate
certain classes of mobile source emissions into the chemical mechanism used by CMAQ. In
order to determine what effect, if any, these particular RFS2 speciation assumptions may have
had on the modeling results, a limited set of sensitivity modeling runs were performed and are
summarized below.
The first analysis considered the impacts of an error in the emissions processing of
nonroad gasoline emissions. Inadvertently, the speciation profiles for highway sources, which
reflect a mix of pre-/post-Tier 2 vehicles and a mix of EO, E10, and E85 gasoline, had also been
applied to nonroad gasoline engines which do not have similar advanced Tier-2 emissions
controls, nor do they use E85 gasoline. The concern was that this error would result in potential
overestimates of ethanol and potential underestimates of acetaldehyde in the control case. The
corrected RFS2 emissions contained 9.1% less ethanol and 1.1% more acetaldehyde than what
was modeled in the original scenario. The RFS2 control case was remodeled with the
appropriate speciation profiles for four months in 2005 (January, April, July, and October). The
12 U.S. EPA, Guidance on the Use of Models and Other Analyses in Attainment Demonstrations for the 8-hour
Ozone NAAQS; EPA-454/R-05-002; Research Triangle Park, NC; October 2005.
-------
sensitivity modeling showed that the original simulations (i.e., the ones summarized herein) do
overestimate the expected ethanol changes between the baselines (i.e., AEO and RFS1) and the
control case (RFS2). However, the CMAQ modeling indicated that the impacts of the speciation
fixes were very small for ozone, PM25, and key toxics species. Outside of ethanol, the impacts
of the fix were generally at least one order of magnitude smaller than the differences between the
RFS1 base and the control scenario. As a result, it was determined that the original modeling
was sufficient for isolating the impacts of RFS2.
The second analysis evolved out of initial comparisons of the RFS1 mandate reference
case with the RFS2 control case, where the modeling showed decreases in acetaldehyde
concentrations in the summer and winter in urban areas. Decreases are less pronounced in winter
when there is less secondary formation of acetaldehyde. The main reason for the decrease in
urban areas is determined to be due to reductions in emissions of certain acetaldehyde
precursors. In particular, reductions in alkenes (olefins) were noted, driven by differences in the
EO gasoline headspace speciation profiles used for the control case and the reference cases, as
discussed in Section 3.4.1.3 of the RFS2 RIA. Headspace profiles are used to speciate
hydrocarbon emissions from gasoline storage, gasoline distribution, and gas cans. After the
initial modeling was completed, EPA noticed that the headspace profiles used in the reference
case scenarios exhibited a reduction in alkene levels going from EO to E10 that was inconsistent
with what one would expect as a result of increased ethanol use. In these cases, the EO gasoline
headspace profile has 13% of the VOC as alkenes and the E10 profile has an alkene content of
4%. To address this issue, EPA conducted a sensitivity analysis by adjusting the EO headspace
profile in the RFS1 mandate reference case for the Eastern U.S. modeling domain13 (based on the
assumption that the emissions have an alkene content of 4%, consistent with the percent alkene
content of the E10 headspace profile14). A sensitivity analysis was conducted for the month of
July and EPA compared results with the control case for the following two cases:
1) RFS1 case with no change in alkene levels between headspace profiles for EO and
E10 (i.e., adjusted EO profile)
2) RFS1 case with higher alkene levels for EO headspace profile
Because of these uncharacteristic differences, EPA reran the control case using the
adjusted EO gasoline headspace profile. Due to time constraints, we were not able to make this
improvement for the reference cases. Thus, alkene levels associated with the EO use are lower in
the control case than the reference cases, leading to a reduction in secondarily formed
acetaldehyde.
The results of the sensitivity analysis showed that acetaldehyde levels were significantly
higher for the comparison between Case 1 and the control case than for the comparison between
Case 2 and the control case. The sensitivity analysis thus confirmed that the decrease in these
13 Details of the sensitivity run are discussed in the emissions modeling TSD, found in the docket for this rule (EPA-
HQ-OAR-2005-0161).
14 U.S. Environmental Protection Agency. 2010. Hydrocarbon Composition of Gasoline Vapor Emissions from
Enclosed Fuel Tanks. Draft Report EPA-420-D-10-001, January 2010.
-------
acetaldehyde precursors between the reference cases and the control case EO headspace profile is
driving the decrease in ambient concentrations of acetaldehyde in urban areas. Thus, while the
air quality modeling results presented in the RFS2 RIA and in Section III.C. 1 below suggest
impacts of increased renewable fuel use on ambient acetaldehyde are not substantial and there
may be decreases in urban areas, there is considerable uncertainty associated with these results.
In fact, if the reference cases were rerun with revised EO headspace profiles, some of the
observed decreases could become increases.
4. 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.15 Meteorological model input
fields were prepared separately for each of the three domains shown in Figure II-1 using MM5
version 3.7.4. The MM5 simulations were run on the same map projection as CMAQ.
All three meteorological model runs 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.
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.
Table II-3. Vertical layer structure for MM5 and CMAQ (heights are layer top).
CMAQ Layers
0
1
2
o
J
MM5 Layers
0
1
2
3
Sigma P
1.000
0.995
0.990
0.985
Approximate
Height (m)
0
38
77
115
Approximate
Pressure (mb)
1000
995
991
987
15 Grell, G., J. Dudhia, andD. Stauffer, 1994: A Description of the Fifth-Generation Perm State/NCAR Mesoscale
Model (MM5), NCAR/TN-398+STR., 138 pp, National Center for Atmospheric Research, Boulder CO.
10
-------
A
c
6
7
8
1 f\
12
1 "3
1 /I
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
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
154
232
310
389
469
550
631
712
794
961
,130
,303
,478
,657
,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
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
The meteorological outputs from all three MM5 sets were processed to create model-
ready inputs for CMAQ using the Meteorology-Chemistry Interface Processor (MCIP), version
3.4, to derive the specific inputs to CMAQ.16
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 monthly
average rainfall and monthly maximum planetary boundary layer (PEL) heights. 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
16
Byun, D.W., and Ching, J.K.S., Eds, 1999. Science algorithms of EPA Models-3 Community Multiscale Air
Quality (CMAQ modeling system, EPA/600/R-99/030, Office of Research and Development).
11
-------
evaluation, five meteorological parameters were investigated: temperature, humidity, shortwave
downward radiation, wind speed, and wind direction. The three individual MM5 evaluations are
described elsewhere.17'18'19 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.
5. Initial and Boundary Conditions: The lateral boundary and initial species
concentrations are provided by a three-dimensional global atmospheric chemistry model, the
GEOS-CHEM20 model (standard version 7-04-1121). 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 future base
conditions from the 36 km coarse grid modeling were used as the initial/boundary state for all
subsequent 12 km finer grid modeling.
E. CMAQ Base Case Model Performance Evaluation
1. PM2.s: An operational model performance evaluation for PM2 5 and its related
speciated components (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 CMAQ
modeling system to replicate base year concentrations. In summary, model performance
statistics were calculated for observed/predicted pairs of daily/monthly/seasonal/annual
concentrations. Statistics were generated for the following geographic groupings: domain wide,
Eastern vs. Western (divided along the 100th meridian), and each Regional Planning
Organization (RPO) region22. The "acceptability" of model performance was judged by
17 Baker K. and P. Dolwick. Meteorological Modeling Performance Evaluation for the Annual 2005 Eastern U.S.
12-km Domain Simulation, USEPA/OAQPS, February 2, 2009.
18
Baker K. and P. Dolwick. Meteorological Modeling Performance Evaluation for the Annual 2005 Western U.S.
12-km Domain Simulation, USEPA/OAQPS, February 2, 2009.
19
Baker K. and P. Dolwick. Meteorological Modeling Performance Evaluation for the Annual 2005 Continental
U.S. 36-km Domain Simulation, USEPA/OAQPS, February 2, 2009.
20 Yantosca, B., 2004. GEOS-CHEMv7-01-02 User's Guide, Atmospheric Chemistry Modeling
Group, Harvard University, Cambridge, MA, October 15, 2004.
21 Henze, O.K., J.H. Seinfeld, N.L. Ng, J.H. Kroll, T-M. Fu, DJ. Jacob, C.L. Heald, 2008. Global modeling of
secondary organic aerosol formation from aromatic hydrocarbons: high-vs.low-yield pathways. Atmos. Chem. Phys..
8, 2405-2420.
22 Regional Planning Organization regions include: Mid-Atlantic/Northeast Visibility Union (MANEVU), Midwest
Regional Planning Organization - Lake Michigan Air Directors Consortium (MWRPO-LADCO), Visibility
Improvement State and Tribal Association of the Southeast (VISTAS), Central States Regional Air Partnership
(CENRAP), and the Western Regional Air Partnership (WRAP).
12
-------
comparing our CMAQ 2005 performance results to the range of performance found in recent
regional PM2.s model applications for other, non-EPA studies23. Overall, the fractional bias,
fractional error, normalized mean bias, and normalized mean 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 the RFS2 assessment. A detailed summary of the 2005 CMAQ model performance
evaluation is available in Appendix B24.
Table II-4. 2005 CMAQ annual PM2.s species model performance statistics.
CMAQ 2005 Annual
PM25
Total Mass
Sulfate
STN
IMPROVE
STN
IMPROVE
CASTNet
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
No. of
Obs.
11797
3440
2318
2977
2960
2523
2826
9321
10411
571
2339
1694
2376
9258
13897
3920
2495
3441
3499
2944
3157
9034
10002
531
2253
1685
2350
8896
3170
NMB (%)
-0.8
-10.0
8.5
10.8
-13.7
-6.3
-10.9
-6.4
-19.9
1.3
12.1
-17.6
-13.3
-22.9
-12.3
-17.0
-5.2
-7.7
-14.5
-22.5
-15.5
-15.2
-14.4
-12.7
-7.7
-17.8
-23.5
-10.5
-19.3
NME (%)
37.6
45.0
35.2
41.6
34.0
39.8
46.1
41.6
44.6
36.8
47.7
37.5
41.7
44.8
33.2
42.3
34.0
32.1
30.9
37.2
45.8
34.5
41.0
32.5
34.3
32.7
36.6
42.3
24.8
FB (%)
-2.4
-9.5
9.2
9.9
-14.5
-9.8
-10.6
-7.2
-21.9
1.3
8.1
-16.9
-11.9
-23.5
-9.2
-7.8
0.7
-4.0
-12.2
-19.5
-6.7
-6.2
2.7
-5.0
-0.2
-12.2
-16.1
5.0
-18.5
FE (%)
38.7
44.4
33.5
38.8
37.1
44.3
45.0
43.6
48.4
37.2
44.3
43.1
46.2
48.6
35.9
42.8
34.8
33.9
33.4
41.6
44.0
38.9
44.5
34.6
37.3
36.2
40.7
45.1
27.8
23 These other modeling studies represent a wide range of modeling analyses which cover various models, model
configurations, domains, years and/or episodes, chemical mechanisms, and aerosol modules.
24 U.S. Environmental Protection Agency, Air Quality Modeling Technical Support Document: Changes to
Renewable Fuel Standard Program, Appendix B: CMAQ Model Performance Evaluation for Ozone, Paniculate
Matter and Toxics. January, 2010 (EPA-454/R-10-001A).
13
-------
Nitrate
Total
Nitrate
(NO3 +
HNO3)
Ammonium
Elemental
Carbon
STN
IMPROVE
CASTNet
STN
CASTNet
STN
IMPROVE
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
1142
615
786
1099
300
1091
12741
3655
2495
3442
3499
1812
31339
9027
9987
531
2248
1685
2350
8881
3170
1142
615
786
1099
300
1091
13897
3893
2495
3498
3882
3059
3130
3170
1142
615
786
1099
300
1091
14038
3814
2502
3479
3877
3221
3015
8668
9495
602
-24.6
-16.7
-14.2
-21.5
-32.2
-23.7
25.5
-41.7
29.8
37.0
13.1
5.2
-47.3
34.2
-29.7
20.7
66.4
42.5
22.8
-45.7
27.3
-3.3
26.1
40.6
25.7
10.4
-4.7
6.7
-14.9
14.3
14.7
0.8
-4.3
-20.5
-1.3
-14.5
8.1
6.2
-13.6
-4.2
-20.8
25.9
31.1
18.7
37.7
10.7
48.1
38.7
-25.9
-10.2
-16.8
33.6
23.4
22.6
24.7
34.3
33.8
70.4
65.3
64.7
73.9
78.2
58.8
65.4
86.9
72.5
69.8
106.9
106.6
72.7
75.6
40.3
34.5
37.4
46.0
41.4
33.9
35.7
42.3
55.3
42.0
44.3
39.1
43.6
59.0
34.6
38.8
34.6
37.2
32.6
35.4
39.1
66.0
77.7
51.7
70.5
59.1
86.3
82.8
49.1
57.2
41.8
-16.6
-14.8
-12.3
-23.6
-33.6
-15.8
-8.1
-70.8
17.1
3.2
-27.5
-6.2
-79.1
-31.6
-93.8
-7.3
-1.0
-37.5
-20.5
-102.3
21.9
6.4
26.9
34.4
17.2
7.3
6.4
12.6
7.1
22.3
25.0
6.5
-0.2
5.8
0.4
-12.1
12.8
11.3
-13.7
-1.1
-13.9
18.2
19.5
20.1
26.7
8.1
26.5
21.0
-28.3
-17.8
-28.3
35.2
24.8
24.8
27.7
38.4
35.2
78.1
97.5
63.9
73.6
86.0
71.6
99.9
101.6
124.0
82.1
95.4
105.2
95.0
128.4
38.1
39.9
34.1
42.4
39.2
33.3
40.5
45.4
55.0
42.2
46.3
42.0
50.4
57.2
35.8
40.1
33.3
35.9
36.5
39.6
40.2
54.3
62.5
47.1
54.7
49.4
64.3
65.1
56.0
60.4
50.4
14
-------
Organic
Carbon
STN
IMPROVE
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
2117
1584
2123
8518
12619
3582
2380
3323
3802
2259
3060
8662
9495
601
2116
1587
2123
8518
-4.8
-46.1
-31.3
-9.6
-35.1
-32.1
-37.3
-17.4
-45.7
-38.8
-31.7
-29.9
-24.0
-33.1
-7.7
-37.8
-42.5
-22.3
48.8
51.4
49.5
58.2
52.6
56.7
51.9
52.7
52.9
53.4
57.6
50.6
57.1
43.6
52.4
46.7
54.2
57.3
-15.5
-51.5
-30.1
-18.2
-32.5
-28.2
-31.0
-13.7
-48.7
-37.3
-27.8
-34.3
-29.4
-39.3
-14.4
-48.0
-46.8
-28.1
54.1
62.8
56.5
61.5
63.9
61.3
62.5
60.3
66.7
66.9
61.4
59.1
62.8
53.2
53.5
60.2
65.2
62.7
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 CMAQ modeling system
to replicate the base year concentrations for the 12-km Eastern and Western United States
domain shown in Figure II-1. Ozone measurements from 1194 sites (817 in the East and 377 in
the West) were included in the evaluation and 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 daily maximum ozone
concentrations and eight-hour daily maximum ozone concentrations. 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 ozone
season (May through September) that was modeled for the RFS2 final rule. Appendix B
contains a more detailed summary of ozone model performance over the 12km Eastern and
Western U.S. grid. A summary of the evaluation is presented here.
As with the national, annual PM2.5 CMAQ modeling, the "acceptability" of model
performance was judged by comparing our CMAQ 2005 performance results to the range of
performance found in recent regional ozone model applications (e.g., EPA's Proposal to
Designate an Emissions Control Area for Nitrogen Oxides 25 and the Clean Air Interstate Rule26).
Overall, the normalized mean bias and error (NMB and NME), as well as the fractional bias and
error (FB and FE) statistics shown in Tables II-5 and II-6 indicate that CMAQ-predicted 2005
hourly and eight-hour daily maximum ozone residuals (i.e., observation vs. model predictions)
are within the range of other recent regional modeling applications. The CMAQ model
25 U.S. Environmental Protection Agency, Proposal to Designate an Emissions Control Area for Nitrogen Oxides,
Sulfur Oxides, and Paniculate Matter: Technical Support Document. EPA-420-R-007, 329pp., 2009.
(http://www.epa.gov/otaq/regs/nonroad/marine/ci/420r09007.pdf)
26 U.S. Environmental Protection Agency; Technical Support Document for the Final Clean Air Interstate Rule: Air
Quality Modeling; Office of Air Quality Planning and Standards; Research Triangle Park, NC; March 2005.
15
-------
performance results give us confidence that our applications of CMAQ using this modeling
platform provide a scientifically credible approach for assessing ozone concentration changes
resulting from the final RFS2 emissions reductions.
Table II-5. 2005 CMAQ one-hour daily maximum ozone model performance statistics
calculated for a threshold of 40 ppb.
CMAQ 2005 One-Hour Maximum Ozone:
Threshold of 40 ppb
May
June
July
August
September
Seasonal Aggregate
(May - September)
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
No. ofObs.
21394
9631
4418
4102
6424
4328
8294
19517
9056
4639
4148
4644
4062
7737
19692
9443
4923
4445
4733
3521
8168
19643
9562
4549
4139
5303
3589
8357
18085
8725
4002
3667
5259
3286
7530
98331
46417
22531
20501
26363
18786
40086
NMB (%)
-1.6
-3.4
0.8
5.4
-3.6
-6.4
-3.5
-3.5
-3.7
-4.6
-1.0
-2.7
-6.2
-4.0
1.2
0.4
0.4
4.2
4.2
-3.8
0.2
0.1
-0.8
0.2
0.2
3.6
-4.1
-1.0
-2.2
-3.6
-3.6
-1.8
-0.1
-6.1
-4.1
-1.2
-2.1
-1.4
1.4
0.1
-5.4
-2.3
NME (%)
11.5
12.8
10.0
11.8
11.3
13.4
12.9
12.8
13.0
12.3
14.1
12.5
13.2
13.1
14.2
16.0
12.7
15.2
15.1
14.8
16.2
13.9
15.5
12.2
13.2
14.9
16.2
15.7
12.0
14.1
10.7
11.3
12.1
14.5
14.3
12.9
14.3
11.7
13.3
13.1
14.4
14.5
FB (%)
-0.8
-2.8
1.0
5.9
-3.0
-5.5
-3.0
-2.8
-3.2
-4.0
-0.1
-2.2
-5.4
-3.6
1.8
1.0
0.9
4.8
4.6
-3.1
0.7
0.8
-0.6
1.0
1.2
3.9
-2.9
-1.0
-1.3
-3.2
-3.0
-0.7
0.8
-5.1
-3.8
-0.5
-1.7
-0.8
2.3
0.7
-4.4
-2.1
FE (%)
11.6
12.7
10.2
11.7
11.5
13.4
12.8
12.9
13.0
12.4
14.2
12.6
13.3
13.1
14.1
15.8
12.6
14.9
14.8
14.9
16.0
13.8
15.5
12.3
13.1
14.5
16.1
15.7
12.0
14.3
10.8
11.3
12.1
14.5
14.4
12.8
14.2
11.7
13.1
13.0
14.4
14.4
16
-------
Table II-6. 2005 CMAQ eight-hour daily maximum ozone model performance statistics
calculated for a threshold of 40 ppb.
CMAQ 2005 Eight-Hour Maximum
Ozone: Threshold of 40 ppb
May
June
July
August
September
Seasonal Aggregate
(May - September)
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
No. ofObs.
19310
8445
3858
3528
6019
3927
7234
17404
8102
4324
3590
3924
3663
6889
17045
8556
4429
3856
3806
3057
7407
16953
8523
4027
3530
4447
3096
7469
15190
7465
3265
2856
4647
2798
6446
85902
41091
19903
17360
22843
16541
35445
NMB (%)
-1.0
-1.6
0.2
5.2
-2.1
-5.8
-1.8
-2.1
-1.9
-3.8
0.3
-0.3
-5.5
-2.2
3.3
3.7
1.8
6.6
7.4
-2.3
3.5
1.9
1.6
0.9
1.4
7.4
-3.4
1.4
-1.8
-2.4
-4.2
-2.3
1.5
-6.5
-2.9
0.1
0.0
-0.9
2.4
2.3
-4.8
-0.2
NME (%)
10.9
12.0
10.0
11.4
10.5
12.8
12.1
11.9
11.9
11.6
13.1
11.4
12.1
12.1
13.4
15.0
11.8
14.6
15.0
13.2
15.1
12.9
13.9
11.3
12.3
14.7
14.4
14.1
11.2
13.4
10.2
10.6
11.2
13.6
13.7
12.1
13.3
11.1
12.6
12.3
13.2
13.5
FB (%)
-0.4
-1.2
0.7
5.4
-1.6
-5.2
-1.5
-1.5
-1.6
-3.4
1.0
0.1
-5.0
-2.0
3.6
3.9
2.3
6.8
7.3
-2.1
3.6
2.2
1.5
1.4
2.0
7.2
-3.1
1.2
-1.3
-2.6
-4.0
-1.8
2.1
-6.1
-3.1
0.5
0.1
-0.5
2.9
2.6
-4.4
-0.3
FE (%)
11.0
12.0
10.2
11.2
10.6
13.0
12.0
12.0
11.9
11.8
13.2
11.5
12.3
12.1
13.3
14.7
11.8
14.3
14.5
13.5
14.9
12.9
13.9
11.4
12.2
14.1
14.8
14.0
11.3
13.9
10.4
10.7
11.2
14.0
14.1
12.1
13.3
11.2
12.4
12.2
13.4
13.5
17
-------
3. Hazardous air pollutants
An operational model performance evaluation for daily, monthly, seasonal, and annual
specific air toxics (formaldehyde, acetaldehyde, benzene, acrolein, and 1,3-butadiene) was
conducted in order to estimate the ability of the CMAQ modeling system to replicate the base
year concentrations for the 12-km Eastern and Western United States domains. Toxic
measurements from 471 sites in the East and 135 sites in the West were included in the
evaluation and were taken from the 2005 State/local monitoring site data in the National Air
Toxics Trends Stations (NATTS). Similar to PM2.5 and ozone, the evaluation principally
consists of statistical assessments of model versus observed pairs that were paired in time and
space on daily basis. Appendix B contains a more detailed summary of air toxics model
performance over the 12km Eastern and Western U.S. grid. A summary of the evaluation is
presented here.
Model predictions of annual formaldehyde, acetaldehyde and benzene showed relatively
small bias and error percentages when compared to observations. The model yielded larger bias
and error results for 1,3 butadiene and acrolein based on limited monitoring sites. As with the
national, annual PM2.5 and ozone CMAQ modeling, the "acceptability" of model performance
was judged by comparing our CMAQ 2005 performance results to the limited performance
found in recent regional multi-pollutant model applications.27'28'29 Overall, the normalized mean
bias and error (NMB and NME), as well as the fractional bias and error (FB and FE) statistics
shown in Table II-7 indicate that CMAQ-predicted 2005 toxics (i.e., observation vs. model
predictions) are within the range of recent regional modeling applications.
27 Phillips, S., K. Wang, C. Jang, N. Possiel, M. Strum, T. Fox, 2007: Evaluation of 2002 Multi-pollutant Platform:
Air Toxics, Ozone, and Paniculate Matter, 7th Annual CMAS Conference, Chapel Hill, NC, October 6-8, 2008.
28 Strum, M., Wesson, K., Phillips, S., Cook, R., Michaels, H., Brzezinski, D., Pollack, A., Jimenez, M., Shepard, S.
Impact of using lin-level emissions on multi-pollutant air quality model predictions at regional and local scales. 17th
Annual International Emission Inventory Conference, Portland, Oregon, June 2-5, 2008.
29 Wesson, K., N. Farm, and B. Timin, 2010: Draft Manuscript: Air Quality and Benefits Model Responsiveness to
Varying Horizontal Resolution in the Detroit Urban Area, Atmospheric Pollution Research, Special Issue: Air
Quality Modeling and Analysis.
18
-------
Table II-7. 2005 CMAQ annual toxics model performance statistics
CMAQ 2005 Annual
Formaldehyde
Acet aldehyde
Benzene
1,3-Butadiene
Acrolein
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
No. of Obs.
6365
1928
771
1982
1246
1815
1746
6094
1892
703
1969
1231
1640
1709
11615
3369
1425
2589
2426
4737
2333
8102
1976
516
1902
1226
4142
1082
1660
783
n/a
850
278
n/a
592
NMB (%)
-55.5
-28.4
-77.1
-30.5
-66.2
-43.5
-25.5
-4.2
-19.2
-12.6
-9.5
0.4
1.8
-20.4
-32.6
-38.4
-8.3
21.6
-41.1
-47.0
-30.5
-74.7
-51.9
-78.7
-41.6
-85.4
-66.5
-40.8
-94.4
-95.7
n/a
-90.4
-97.0
n/a
-95.9
NME (%)
65.3
52.1
85.4
51.3
72.2
51.0
52.3
62.0
53.7
58.0
62.8
63.5
57.0
54.1
66.8
60.8
72.7
53.3
68.6
68.3
61.2
85.6
82.1
86.2
55.5
86.4
85.9
77.5
95.0
95.7
n/a
91.5
97.0
n/a
95.9
FB (%)
-39.2
-30.1
-25.8
-28.5
-51.3
-41.4
-26.0
-8.2
-19.5
-12.1
-9.0
-6.2
-4.3
-20.1
-13.5
-30.4
25.2
18.1
-17.2
-32.7
-19.2
-49.4
-34.5
-48.3
-54.8
-106.2
-20.0
-41.9
-131.3
-168.1
n/a
-120.5
-156.4
n/a
-177.6
FE (%)
65.6
60.7
74.0
61.6
70.4
61.5
59.8
60.3
59.6
60.0
63.7
62.2
51.1
60.6
62.8
63.9
62.4
46.8
59.8
69.4
63.4
91.6
91.7
81.9
71.3
111.5
89.2
85.3
142.2
170.4
n/a
134.2
157.0
n/a
177.6
III. CMAQ Model Results
A. Impacts of RFS2 Changes on Future PM2.s Levels
It is important to remember that there are uncertainties and limitations related to the air
quality modeling (see Section 3.4.1.3 in RFS2 RIA), including the projected amount of E85 in
19
-------
use. The modeled projected usage of E85 is higher than what was included in the final rule
inventory, which could overestimate the decreases in PM2.5. These differences in the air quality
modeling inventories and the final rule inventories are discussed in detail in Section 3.3 of the
RFS2 RIA.
After the air quality modeling was complete an error was found in the PM inventory for
locomotives, therefore only design value changes over all 577 modeled counties are reported. A
large majority of the modeled counties will have relatively minor annual average PM2 5 design
value changes of between -0.05 |ig/m3 and +0.05 |ig/m3. On a population-weighted basis, the
average modeled future-year annual PM2.5 design values are projected to decrease by 0.002
|ig/m3 when compared with the RFS1 mandate or AEO reference case.30 Likewise, daily PM2 5
design values show the majority of the modeled counties will experience changes of between -
0.25 |ig/m3 and +0.25 |ig/m3. On a population-weighted basis, the average modeled future-year
daily PM2.5 design value is projected to decrease by 0.06 |ig/m3 when compared with the RFS1
mandate scenario or 0.05 |ig/m3 when compared with the AEO scenario.
The changes in ambient PM2 5 described above are likely due to both increased emissions
at biofuel production plants and from biofuel transport, and reductions in SOA formation and
reduced emissions from gasoline refineries. In addition, decreases in ambient PM are predicted
because our modeling inventory assumed large volumes of E85 use and also that E85 usage
reduces PM tailpipe emissions. As mentioned previously and in more detail in Section 3.4 of the
RIA, these direct PM emission reductions would not occur with final rule inventory assumptions.
B. Impacts of RFS2 Changes on Future 8-Hour Ozone Levels
This section summarizes the results of our modeling of ozone air quality impacts in the
future due to the required renewable fuel volumes. Our modeling indicates that the renewable
fuel standards will result in increases in ozone design value concentrations in many areas of the
country as well as decreases in ozone design value concentrations in a small number of areas.
Figures III-l and III-2 display the projected county-level, 8-hour ozone design value changes
expected when the RFS2 control scenario is compared to the RFS1 mandate reference case and
the AEO 2007 reference case respectively.31 The air quality modeling of the expected impacts of
the final rule shows that in 2022, most counties with modeled data, especially those in the
southeast U.S., will see increases in their ozone design values. The bulk of these design value
increases are less than 0.5 ppb. On a population-weighted basis, the average modeled future-
year 8-hour ozone design values are projected to increase by 0.15 ppb in 2022 when compared
with the RFS1 mandate reference case and increase by 0.27 ppb when comparing with the AEO
reference case. On a population-weighted basis those counties that are projected to be above the
30 Note that the change in annual average PM2 5 for design values differs from the change in national population-
weighted annual average PM25 discussed in Sections I and VIII of the preamble and Chapter 5 of the RIA. The
discussion of national population-weighted annual average PM2 5 with respect to health impacts in Sections I and
VIII of the preamble and Chapter 5 of this RIA is based on modeling data from all grid cells rather than just those
counties with monitors. It finds that there is a small increase in annual average PM2 5.
31 The air quality modeling used a different speciation profile for E10 gasoline headspace emissions in the RFS2
control case than was used for the RFS1 and AEO reference cases. This inconsistency is described in Section 3.4.1.3
intheRFS2RIA.
20
-------
2008 ozone standard in 2022 will see decreases of 0.18 when compared with the AEO reference
case and 0.17 ppb when compared with the RFS1 mandate reference case.
When comparing the changes in projected ozone it is important to note the differences in
the inventories used for the air quality modeling and the inventories presented in the RFS2 final
rule. The most important difference and uncertainty has to do with the fact that the modeled
inventory assumes increases in NOx for vehicles using E10 fuel. The air quality modeling
indicates that the NOx increases required from the renewable fuel volumes contribute to the
ozone increases in NOx-limited areas as well as the ozone decreases in VOC-limited areas.
Figure III-l. Model-projected change in annual 8-hour Ozone design values between the
RFS2 Control Scenario and the RFS1 Mandate Scenario in 2022. Units are ppb.
-0 75 to <= -0.50
-0.50 to <= -0.30
-030to<=-010
-0 10to<0.10
>= 0.10 to < 0.30
>= 0.30 to < 0.50
>= 0.50 to< 0.75
>=0.75to< 1.0
>=1.00
Difference in 8-hour ozone DV- EISA_rr minus RF51
21
-------
Figure III-2. Model-projected change in annual 8-hour Ozone design values between the
RFS2 Control Scenario and AEO Scenario in 2022. Units are ppb.
O/fference in 8-hour ozone DV - EISA_rr minus AEO
C. Impacts of RFS2 Changes on Toxic Air Pollutant Levels
This section summarizes the results of our modeling of ambient air toxics impacts in the
future from the renewable fuel volumes required by RFS2. Specifically, we compare the RFS1
mandate and AEO reference scenarios to the RFS2 control scenario for 2022 (see Section 3.3 of
the RIA for more information on the scenarios).32 Our modeling indicates that, while there are
some localized impacts, the renewable fuel volumes required by RFS2 have relatively little
impact on national average ambient concentrations of the modeled air toxics. An exception is
increased ambient concentrations of ethanol. Since the overall impacts are relatively small, we
concluded that assessing exposure to ambient concentrations and conducting a quantitative risk
assessment of air toxic impacts was not warranted. Although, we developed population metrics,
including the population living in areas with increases or decreases in concentrations of various
magnitudes. We also estimated aggregated populations above and below reference
concentrations for noncancer effects.
We used a different speciation profile for E10 gasoline headspace emissions in the RFS2 control case than was
used forthe RFS1 and AEO reference cases. This inconsistency is described in Section 3.4.1.3 of the RIA.
22
-------
1. Acetaldehyde
Overall, the air quality modeling does not show substantial nationwide impacts on
ambient concentrations of acetaldehyde due to the renewable fuel volumes required by this rule.
Figure III-3 shows the annual percent changes in ambient concentrations of acetaldehyde are less
than 1% for most of the country. Several urban areas show decreases in ambient acetaldehyde
concentrations ranging from 1 to 10%, and some rural areas associated with new ethanol plants
show increases in ambient acetaldehyde concentrations ranging from 1 to 10% with RFS2. In
Figure III-4, the annual absolute changes in ambient concentrations of acetaldehyde are generally
less than 0.1 |ig/m3. As noted above, the results show that the largest increases in ambient
acetaldehyde concentrations with RFS2 volumes occur in areas associated with new ethanol
plants. This result is due to an increase in emissions of primary acetaldehyde and precursor
emissions from ethanol plants not included in the RFS1 mandate reference scenario.
Figure III-3. Acetaldehyde Annual Percent Change in Concentration Between the RFS2
Mandate Reference Case and the RFS2 Control Case in 2022
SFSf,
23
-------
Figure III-4. Acetaldehyde Annual Absolute Changes in Ambient Concentrations Between
the RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ug/m3)
Figures III-5 and III-6 show the comparison of the RFS1 mandate reference case with the
RFS2 control case for summer and winter shows decreases in ambient acetaldehyde
concentrations in urban areas. Decreases are less pronounced in winter when there is less
secondary formation of acetaldehyde (Figures III-6). As stated above, the main reason for the
decrease in urban areas is reductions in certain acetaldehyde precursors, primarily alkenes
(olefins) that are related to the differences in the EO gasoline headspace speciation profiles used
for the control case and the reference cases. While the air quality modeling results presented
here and in the RFS2 RIA suggest impacts of increased renewable fuel use on ambient
acetaldehyde are not substantial and there may be decreases in urban areas, there is considerable
uncertainty associated with these results. Thus, if the reference cases were rerun with revised EO
headspace profiles, some of the observed decreases could become increases. Additional research
is underway to address these uncertainties, e.g., measurement of representative fuels to create
better headspace speciation profiles (Section 3.4.1.3 in the RFS2 RIA) and improvements in
other speciation profiles based on additional results from the EPAct emissions test program.33
33. EPAct Phase I II, and III Testing: Comprehensive Gasoline Light-Duty Exhaust Fuel Effects Test Program to
Cover Multiple Fuel Properties. EPA Contract: EPC-07-028EPA. Southwest Research Institute, San Antonio, TX.
Phase III of the EPAct emission test program is scheduled for completion in 2010.
24
-------
Figure III-5. Summer Changes in Acetaldehyde Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes
and (b) Absolute Changes (ug/m3)
a b
Figure III-6. Winter Changes in Acetaldehyde Ambient Concentrations Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes and (b)
Absolute Changes (jig/m3)
a
2. Formaldehyde
Our air quality modeling results do not show substantial impacts on ambient
concentrations of formaldehyde from the renewable fuel volumes required by this rule. As
25
-------
shown in Figure III-7, most of the U.S. experiences a 1% or less change in ambient
formaldehyde concentrations. Decreases in ambient formaldehyde concentrations range between
1 and 5% in a few urban areas. Increases range between 1 and 2.5% in some rural areas
associated with new ethanol plants; this result is due to increases in emissions of primary
formaldehyde and formaldehyde precursors from the new ethanol plants. Figure III-8 shows that
absolute changes in ambient concentrations of formaldehyde are generally less than 0.1 |ig/m3.
Figure III-7. Formaldehyde Annual Percent Change in Concentration Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022
26
-------
Figure III-8. Formaldehyde Annual Percent Changes in Ambient Concentrations Between
the RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ug/m3)
-
r RfSf . for F
3. Ethanol
Our modeling projects that the renewable fuel volumes required by this rule will lead to
significant nationwide increases in ambient ethanol concentrations. Figure III-9 shows increases
ranging between 10 to 50% that are seen across most of the country. The largest increases (more
than 100%) occur in urban areas with high amounts of onroad emissions and in rural areas
associated with new ethanol plants. Absolute increases in ambient ethanol concentrations are
above 1.0 ppb in some urban areas (Figure 111-10). The location of these localized increases is
limited by uncertainties in the placement of the new ethanol plants, as discussed in Section
3.4.1.3 of the RFS2 RIA. It should be noted here that these increases are overestimated because
the speciated profile combination used for modeling nonroad emissions was misapplied. While
sensitivity analyses suggest that the impact of this error was negligible for other pollutants, it
resulted in overestimates of ethanol impacts by more than 10% across much of the modeling
domain. Details on the ethanol impacts are discussed in the emissions modeling TSD, found in
the docket for this rule (EPA-HQ-OAR-2005-0161).
27
-------
Figure III-9. Ethanol Annual Percent Changes Change in Concentration Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022
Figure 111-10. Ethanol Annual Absolute Changes in Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ppb)
28
-------
4. Benzene
Our modeling projects that the renewable fuel volumes required by this rule will lead to
small nationwide decreases in ambient benzene concentrations. Figure III-l 1, show decreases in
ambient benzene concentrations that range between 1 and 10% across most of the country and
can be higher in a few urban areas. Figure 111-12 indicates absolute changes in ambient
concentrations of benzene show reductions up to 0.2 |ig/m3.
Figure III-l 1. Benzene Annual Percent Change in Concentration Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022
-• BSMjrr tor
29
-------
Figure 111-12. Benzene Annual Absolute Changes in Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ug/m3)
5. 1,3-Butadiene
The results of our air quality modeling show small increases and decreases in ambient
concentrations of 1,3-butadiene in parts of the U.S. as a result of the renewable fuel volumes
required this rule. Overall, as seen in Figure HI-13, decreases occur in some southern areas of
the country and increases occur in some northern areas and areas with high altitudes. Percent
changes in 1,3-butadiene concentrations are over 50% in several areas; but the changes in
absolute concentrations of ambient 1,3-butadiene are generally less than 0.005 |ig/m3 (Figure III-
14). Annual increases in ambient concentrations of 1,3-butadiene are driven by wintertime
rather than summertime changes (Figures III-15 and III-16). These increases appear in rural
areas with cold winters and low ambient levels but high contributions of emissions from
snowmobiles, and a major reason for this modeled increase may be deficiencies in available
emissions test data used to estimate snowmobile 1,3-butadiene emission inventories. These data
were based on tests using only three engines, which showed significantly higher 1,3-butadiene
emissions with 10% ethanol. However, they may not have been representative of real-world
response of snowmobile engines to ethanol.
30
-------
Figure 111-13. 1,3-Butadiene Annual Percent Change in Concentration Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022
Figure 111-14. 1,3-Butadiene Annual Absolute Changes in Ambient Concentrations
Between the RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ug/m3)
;ikl"^-
-------
Figure 111-15. Summer Changes in 1,3-Butadiene Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes
and (b) Absolute Changes (ug/m3)
a
Figure 111-16. Winter Changes in 1,3-Butadiene Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes
and (b) Absolute Changes (ug/m3)
a
b
32
-------
6. Acrolein
Our air quality modeling shows small regional increases and decreases in ambient
concentrations of acrolein as a result of the renewable fuel volumes required by this rule. As
shown in Figure III-17, decreases in acrolein concentrations occur in some eastern and southern
parts of the U.S. and increases occur in some northern areas and areas associated with new
ethanol plants. Figure III-18 indicates that changes in absolute ambient concentrations of
acrolein are between ± 0.001 |ig/m3 with the exception of the increases associated with new
ethanol plants. These increases can be up to and above 0.005 |ig/m3 with percent changes above
50% and are due to increases in emissions of acrolein from the new plants. As discussed in
Section 3.4.1.3 of theRFS RIA, uncertainties in the placement of new ethanol plants limit the
model's projected location of associated emission increases. Ambient acrolein increases in
upper Michigan, Canada, the Northeast, and the Rocky Mountain region are driven by
wintertime changes (Figures III-19 and 111-20), and occur in the same areas of the country that
have wintertime rather than summertime increases in ambient 1,3-butadiene. 1,3-butadiene is a
precursor to acrolein, and these increases are likely associated with the same emission inventory
issues in areas of high snowmobile usage seen for 1,3-butadiene, as described above.
Figure 111-17. Acrolein Annual Percent Changes Change in Concentration Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022
33
-------
Figure 111-18. Acrolein Annual Absolute Changes in Ambient Concentrations Between the
RFS1 Mandate Reference Case and the RFS2 Control Case in 2022 (ug/m3)
Legend
00410
= 0005
States snow modeled changes t»tv.«*n the RFS1 baseline and EISA control c
Map colors do not indicate the seventy of exposure
Scale range and increments may not be comparable between toxics
Absolute Difference - £tSA_rr minus RFSi, far/lcrotem
Figure 111-19. Summer Changes in Acrolein Ambient Concentrations Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes and (b)
Absolute Changes (ug/m3)
a
b
34
-------
Figure 111-20. Winter Changes in Acrolein Ambient Concentrations Between the RFS1
Mandate Reference Case and the RFS2 Control Case in 2022: (a) Percent Changes and (b)
Absolute Changes (jig/m3)
a
35
-------
Appendix A: 8-Hour Ozone Design Values for RFS-2 Scenarios (units are
ppb)
State Name
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Alabama
Arizona
Arizona
Arizona
Arizona
Arizona
Arizona
Arizona
Arizona
Arizona
Arkansas
Arkansas
Arkansas
Arkansas
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
County Name
Baldwin
Clay
Colbert
Elmore
Etowah
Houston
Jefferson
Lawrence
Madison
Mobile
Montgomery
Morgan
Russell
Shelby
Sumter
Talladega
Tuscaloosa
Cochise
Coconino
Gila
La Paz
Maricopa
Pima
Final
Yavapai
Yuma
Crittenden
Newton
Polk
Pulaski
Alameda
Amador
Butte
Calaveras
Colusa
Contra Costa
El Dorado
Fresno
Glenn
Imperial
Inyo
Kern
Kings
Lake
Los Angeles
Madera
Baseline DV
77.30
74.00
72.00
70.70
71.70
71.00
83.70
72.00
77.30
76.70
69.30
77.30
71.30
85.70
64.00
72.00
73.30
71.30
73.00
80.30
72.00
83.00
76.00
79.30
72.00
75.00
87.30
72.70
75.00
79.70
78.30
83.00
83.70
91.30
67.00
73.30
96.00
98.30
65.50
85.00
82.30
110.00
85.70
60.70
114.00
79.30
2022 RFS1
DV
63.76
56.43
50.47
52.07
54.66
57.36
62.27
55.16
59.03
63.66
49.96
62.50
55.24
63.20
53.65
52.98
53.52
62.66
64.74
62.93
62.13
68.91
63.54
62.93
62.68
63.11
66.66
58.70
62.86
59.98
70.90
71.11
69.84
80.22
57.77
69.50
79.46
86.26
56.24
74.04
71.67
98.45
73.42
53.07
103.39
68.48
2022 AEO
DV
64.20
56.65
50.77
52.38
54.92
57.55
62.55
55.40
59.39
64.07
50.26
62.80
55.47
63.50
53.78
53.18
53.74
62.74
64.77
63.10
62.17
69.06
63.67
63.10
62.74
63.18
66.84
58.90
63.01
60.21
70.90
71.11
69.85
80.23
57.77
69.50
79.47
86.29
56.24
74.05
71.72
98.49
73.46
53.08
103.39
68.51
2022 RFS2
DV
64.40
56.88
51.00
52.63
55.14
57.82
62.73
56.02
59.81
64.24
50.51
64.07
56.26
63.65
54.02
53.40
53.93
62.74
64.80
63.10
62.20
68.99
63.67
63.08
62.77
63.19
67.01
59.46
63.44
60.65
70.89
71.21
70.39
80.36
57.96
69.49
79.51
86.37
56.44
74.04
71.76
98.47
73.55
53.16
103.23
68.61
36
-------
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Connecticut
Connecticut
Connecticut
Connecticut
Connecticut
Connecticut
Connecticut
Delaware
Delaware
Delaware
Marin
Mariposa
Mendocino
Merced
Monterey
Napa
Nevada
Orange
Placer
Riverside
Sacramento
San Benito
San Bernardino
San Diego
San Francisco
San Joaquin
San Luis Obispo
San Mateo
Santa Barbara
Santa Clara
Santa Cruz
Shasta
Siskiyou
Solano
Sonoma
Stanislaus
Sutter
Tehama
Tulare
Tuolumne
Ventura
Yolo
Adams
Arapahoe
Boulder
Denver
Douglas
El Paso
Jefferson
La Plata
Larimer
Montezuma
Weld
Fairfield
Hartford
Litchfield
Middlesex
New Haven
New London
Tolland
Kent
New Castle
Sussex
49.70
86.30
56.70
89.30
61.00
59.30
96.30
84.30
94.00
112.30
97.30
75.00
123.30
87.70
46.00
75.30
70.70
53.70
76.00
75.30
61.30
79.30
63.50
72.70
47.70
84.70
82.00
82.70
103.70
80.00
89.70
78.70
69.00
78.70
77.00
73.00
83.00
73.30
81.70
63.70
76.00
72.00
76.70
92.30
84.30
87.70
90.30
90.30
85.30
88.70
80.30
82.30
82.70
45.74
75.74
48.97
76.57
54.77
52.18
80.00
81.40
78.00
108.72
81.05
65.39
121.22
76.76
45.37
67.86
62.46
51.43
68.26
65.26
55.65
67.23
54.67
63.93
41.27
74.98
70.06
70.26
89.07
70.43
78.66
67.63
62.90
69.39
67.41
66.55
73.88
64.60
74.36
59.00
66.07
66.07
67.37
78.29
67.41
70.05
75.10
76.14
70.51
71.08
64.51
68.12
69.10
45.74
75.77
48.99
76.63
54.77
52.20
80.01
81.35
78.01
108.66
81.05
65.40
121.17
76.75
45.37
67.91
62.48
51.42
68.27
65.26
55.64
67.25
54.72
63.94
41.28
75.05
70.06
70.28
89.34
70.44
78.68
67.64
62.99
69.52
67.52
66.64
73.99
64.71
74.44
59.04
66.19
66.12
67.49
78.30
67.50
70.15
75.15
76.17
70.54
71.16
64.59
68.18
69.16
45.69
75.88
49.03
76.82
54.80
52.23
80.06
80.83
78.05
108.06
81.09
65.46
120.64
76.66
45.34
68.00
62.56
51.39
68.27
65.25
55.60
67.45
54.93
63.98
41.32
75.21
70.20
70.58
89.41
70.56
78.65
67.67
62.91
69.51
67.51
66.56
73.93
64.74
74.33
59.06
66.20
66.15
67.52
78.22
67.67
70.35
75.24
76.16
70.50
71.33
64.71
68.27
69.25
37
-------
D.C.
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Florida
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Idaho
Washington
Alachua
Baker
Bay
Brevard
B reward
Collier
Columbia
Duval
Escambia
Highlands
Hillsborough
Holmes
Lake
Lee
Leon
Manatee
Marion
Miami-Dade
Orange
Osceola
Palm Beach
Pasco
Pinellas
Polk
St Lucie
Santa Rosa
Sarasota
Seminole
Volusia
Wakulla
Bibb
Chatham
Chattooga
Clarke
Cobb
Columbia
Coweta
Dawson
De Kalb
Douglas
Fayette
Fulton
Glynn
Gwinnett
Henry
Murray
Muscogee
Paulding
Richmond
Rockdale
Sumter
Ada
84.70
72.00
68.70
78.70
71.30
65.00
68.30
72.00
77.70
82.70
72.30
80.70
70.30
76.70
70.30
71.00
77.30
73.00
71.30
79.30
72.00
65.00
76.30
72.70
74.70
66.50
80.00
77.30
76.00
68.30
71.30
81.00
68.30
75.00
80.70
82.70
73.00
82.00
76.30
88.70
87.30
85.70
91.70
67.00
88.70
89.70
78.00
75.70
80.30
80.30
90.00
72.30
76.00
68.78
54.16
53.65
61.49
57.01
58.55
53.82
56.97
62.74
67.06
60.69
64.92
56.17
59.32
55.56
53.22
60.49
55.13
65.33
62.40
53.73
57.86
59.87
56.09
57.85
54.79
65.21
58.88
59.45
51.74
57.05
65.09
57.09
55.41
56.49
59.53
56.28
64.12
52.93
70.62
63.96
66.46
73.01
52.25
65.80
66.91
60.28
56.81
56.54
60.76
65.63
57.41
69.69
68.87
54.45
53.90
61.98
57.34
58.56
54.19
57.20
63.05
67.72
60.89
65.19
56.39
59.76
56.30
53.59
60.76
55.46
65.38
62.87
54.18
57.97
60.19
56.39
58.14
55.10
66.25
59.19
59.90
52.04
57.40
65.36
57.38
55.70
56.90
59.97
56.50
64.36
53.38
70.88
64.29
66.73
73.28
52.53
66.20
67.26
60.54
57.09
56.86
61.00
65.99
57.61
69.81
69.00
54.67
54.11
62.23
57.38
58.63
54.28
57.41
63.06
67.86
61.15
65.16
56.67
59.80
56.37
53.90
60.80
55.60
65.39
62.89
54.24
58.28
60.23
56.44
58.13
55.25
66.45
59.24
59.91
52.18
57.69
65.55
57.71
55.92
57.15
60.18
57.48
64.57
53.62
71.01
64.51
66.90
73.42
53.63
66.38
67.45
60.88
58.02
57.06
62.18
66.18
57.88
69.80
38
-------
Idaho
Idaho
Idaho
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Illinois
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Iowa
Canyon
Elmore
Kootenai
Adams
Champaign
Clark
Cook
Du Page
Effingham
Hamilton
Jersey
Kane
Lake
McHenry
McLean
Macon
Macoupin
Madison
Peoria
Randolph
Rock Island
St Clair
Sangamon
Will
Winnebago
Allen
Boone
Carroll
Clark
Delaware
Elkhart
Floyd
Greene
Hamilton
Hancock
Hendricks
Huntington
Jackson
Johnson
Lake
La Porte
Madison
Marion
Morgan
Perry
Porter
Posey
St Joseph
Shelby
Vanderburgh
Vigo
Warrick
Bremer
66.00
63.00
67.00
70.00
68.30
66.00
77.70
69.00
70.00
73.00
78.70
74.30
78.00
73.30
73.00
71.30
73.00
83.00
72.70
72.00
65.30
81.70
70.00
71.70
69.00
79.30
79.70
74.00
80.30
76.30
79.00
77.70
78.30
82.70
78.00
75.30
75.00
74.70
76.70
81.00
78.50
76.70
78.70
77.00
81.00
78.30
71.70
79.30
77.30
77.30
74.00
77.70
66.30
59.37
57.30
57.90
58.60
56.65
54.50
69.45
62.49
57.72
58.96
60.58
63.36
69.37
60.54
59.13
58.95
55.79
67.06
61.99
60.39
55.10
66.77
55.21
60.94
57.39
63.96
63.90
60.27
63.04
60.84
64.27
63.61
62.41
65.75
62.44
61.36
61.28
60.27
63.38
72.55
67.43
60.45
64.00
63.02
63.99
69.47
57.31
64.84
64.89
62.01
61.42
63.17
56.49
59.53
57.42
58.14
58.74
56.79
54.65
69.59
62.60
57.88
59.13
60.69
63.48
69.50
60.68
59.32
59.09
55.89
67.18
62.11
60.54
55.22
66.90
55.34
61.07
57.55
64.25
64.06
60.47
63.19
61.15
64.54
63.75
62.58
65.95
62.63
61.51
61.55
60.42
63.53
72.76
67.61
60.66
64.16
63.21
64.18
69.70
57.46
65.09
65.06
62.16
61.67
63.32
56.71
59.54
57.44
58.20
59.40
57.92
55.15
69.84
62.89
58.43
59.74
61.37
63.87
69.76
61.07
60.09
59.82
56.46
67.76
62.76
61.28
55.90
67.49
55.91
61.47
58.31
65.09
64.52
61.28
63.33
61.86
65.34
63.84
63.52
66.52
63.17
61.85
62.17
61.17
63.98
73.17
68.08
61.36
64.66
63.81
64.53
70.09
58.66
65.83
65.69
63.11
62.23
63.80
57.58
39
-------
Iowa
Iowa
Iowa
Iowa
Iowa
Iowa
Iowa
Iowa
Iowa
Iowa
Kansas
Kansas
Kansas
Kansas
Kansas
Kansas
Kansas
Kansas
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Kentucky
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Clinton
Harrison
Linn
Montgomery
Palo Alto
Polk
Scott
Story
Van Buren
Warren
Douglas
Johnson
Leavenworth
Linn
Sedgwick
Sumner
Trego
Wyandotte
Bell
Boone
Boyd
Bullitt
Campbell
Carter
Christian
Daviess
Edmonson
Fayette
Greenup
Hancock
Hardin
Henderson
Jefferson
Jessamine
Kenton
Livingston
McCracken
McLean
Oldham
Perry
Pike
Pulaski
Simpson
Trigg
Warren
Ascension
Beauregard
Bossier
Caddo
Calcasieu
East Baton Rouge
Grant
Iberville
71.30
74.70
68.30
65.70
61.00
63.00
72.00
61.00
69.00
64.50
73.00
75.30
75.00
73.30
71.30
71.70
70.70
75.30
71.70
75.70
77.30
74.00
75.00
71.00
78.00
75.70
73.70
70.30
76.70
74.00
74.70
75.30
78.30
73.30
78.70
73.70
73.30
73.00
83.00
72.30
66.70
70.30
75.70
70.00
72.00
82.00
75.00
78.00
79.00
82.00
92.00
73.00
85.00
60.46
63.57
58.14
55.56
52.63
51.30
60.00
49.97
58.37
52.00
59.93
62.36
63.88
59.50
59.33
59.82
63.18
64.35
53.68
60.67
63.49
59.93
62.36
56.67
60.18
61.46
59.28
57.37
63.26
58.82
60.02
61.35
64.51
59.35
63.53
61.16
61.71
58.86
63.73
58.50
53.76
58.12
58.21
55.17
58.85
72.39
67.63
62.91
63.73
72.48
81.32
61.93
75.62
60.61
63.71
58.60
55.71
52.80
51.45
60.15
50.12
58.53
52.19
60.12
62.53
64.02
59.72
59.51
60.01
63.33
64.48
53.94
60.79
63.61
60.07
62.52
56.80
60.35
61.63
59.44
57.53
63.38
58.98
60.16
61.49
64.65
59.50
63.72
61.38
61.89
59.01
63.89
58.66
53.91
58.26
58.36
56.03
58.98
72.47
67.70
63.20
63.99
72.63
81.44
62.09
75.70
61.37
64.42
59.40
56.26
53.94
52.21
60.88
50.99
59.31
53.03
60.58
62.96
64.28
60.28
60.09
60.51
63.58
64.68
54.25
60.97
63.74
60.21
62.78
56.96
60.57
61.98
59.79
57.72
63.53
59.29
60.39
61.94
64.77
59.75
64.01
61.66
62.20
59.46
64.08
58.97
54.13
58.48
58.76
56.36
59.24
72.71
67.92
63.40
64.31
72.96
81.71
62.47
75.87
40
-------
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Maine
Maine
Maine
Maine
Maine
Maine
Maine
Maine
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Jefferson
Lafayette
Lafourche
Livingston
Orleans
Ouachita
Pointe Coupee
St Bernard
St Charles
St James
St John The Baptis
St Mary
West Baton Rouge
Cumberland
Hancock
Kennebec
Knox
Oxford
Penobscot
Sagadahoc
York
Anne Arundel
Baltimore
Calvert
Carroll
Cecil
Charles
Frederick
Garrett
Harford
Kent
Montgomery
Prince Georges
Washington
Barnstable
Berkshire
Bristol
Dukes
Essex
Hampden
Hampshire
Middlesex
Norfolk
Suffolk
Worcester
Allegan
Benzie
Berrien
Cass
Clinton
Genesee
Huron
Ingham
83.00
82.00
79.30
78.30
70.00
75.30
83.70
78.00
77.30
76.30
79.00
76.00
84.30
72.00
82.00
69.70
75.30
61.00
67.00
68.50
74.00
89.70
85.30
81.00
83.30
90.70
86.00
80.30
75.50
92.70
82.00
83.00
91.00
78.30
84.70
79.70
82.70
83.00
83.30
87.30
85.00
79.00
84.70
80.30
80.00
90.00
81.70
82.30
80.70
75.70
79.30
75.70
76.00
72.83
70.27
70.35
69.46
61.82
62.20
74.74
67.71
67.58
67.78
69.83
66.29
74.24
58.25
67.38
56.68
61.11
51.06
56.74
55.72
60.68
69.53
73.83
63.80
64.12
70.33
65.14
62.78
60.08
78.87
63.70
66.14
71.16
62.37
69.99
64.49
69.53
71.26
71.29
70.03
67.67
63.92
68.40
65.30
62.82
75.26
68.76
69.63
66.37
60.43
64.31
63.63
61.87
73.15
70.41
70.47
69.54
62.06
62.44
74.80
67.96
67.93
67.88
70.00
66.42
74.34
58.28
67.53
56.75
61.18
51.18
56.83
55.77
60.74
69.64
74.01
64.07
64.22
70.46
65.28
62.85
60.25
79.07
63.79
66.23
71.27
62.48
70.07
64.64
69.58
71.30
71.39
70.12
67.77
64.01
68.56
65.40
62.92
75.49
68.99
69.83
66.63
60.67
64.56
63.84
62.12
73.25
70.81
70.58
69.76
62.13
62.68
75.10
68.09
68.01
68.00
70.13
66.59
74.60
58.36
67.68
56.85
61.29
51.48
56.96
55.85
60.80
69.80
74.01
64.20
64.37
70.61
65.49
62.99
60.57
79.10
63.91
66.33
71.44
62.67
69.91
64.92
69.63
71.34
71.40
70.29
67.96
64.13
68.42
65.40
63.15
76.18
69.68
70.33
67.42
61.16
65.07
64.30
62.72
41
-------
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Minnesota
Minnesota
Mississippi
Mississippi
Mississippi
Mississippi
Mississippi
Mississippi
Mississippi
Mississippi
Mississippi
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Missouri
Montana
Nebraska
Nebraska
Nevada
Nevada
Nevada
Nevada
New Hampshire
New Hampshire
New Hampshire
New Hampshire
New Hampshire
New Hampshire
New Hampshire
Kalamazoo
Kent
Leelanau
Lenawee
Macomb
Mason
Missaukee
Muskegon
Oakland
Ottawa
St Clair
Schoolcraft
Washtenaw
Wayne
Anoka
St Louis
Adams
Bolivar
De Soto
Hancock
Harrison
Hinds
Jackson
Lauderdale
Lee
Cass
Cedar
Clay
Clinton
Greene
Jefferson
Lincoln
Monroe
Perry
Platte
St Charles
Ste Genevieve
St Louis
St Louis City
Yellowstone
Douglas
Lancaster
Clark
Washoe
White Pine
Carson City
Belknap
Cheshire
Coos
Grafton
Hillsborough
Merrimack
Rockingham
75.30
81.00
75.70
78.70
86.00
79.70
73.70
85.00
78.00
81.70
82.30
79.30
78.30
82.00
67.70
65.00
74.70
74.30
82.70
79.00
83.00
71.30
80.30
74.30
73.70
74.70
75.70
84.30
83.00
73.00
82.30
87.00
71.70
77.50
77.00
87.00
79.70
88.00
84.00
59.00
68.70
56.00
83.70
70.70
72.30
65.00
71.30
70.70
77.00
67.00
78.70
71.70
75.00
61.55
65.22
63.85
64.93
70.95
65.67
60.43
71.44
66.14
67.25
67.67
65.40
66.16
69.27
63.38
55.52
64.38
60.09
65.16
66.94
68.47
51.08
67.43
58.34
54.53
61.11
62.38
70.48
68.47
59.73
69.75
71.24
58.85
63.18
65.23
68.85
66.50
73.02
69.19
54.67
59.73
46.74
74.23
60.56
64.93
55.81
56.01
56.59
63.97
55.17
63.46
56.02
61.50
61.83
65.50
64.33
65.19
71.30
65.86
60.67
71.61
66.41
67.45
67.93
65.53
66.38
69.51
63.50
55.67
64.50
60.28
65.32
67.23
68.70
51.42
67.67
58.56
54.83
61.28
62.56
70.64
68.64
59.85
69.91
71.40
59.00
63.34
65.37
68.98
66.63
73.18
69.32
54.70
59.84
46.85
74.33
60.62
64.97
55.82
55.64
56.65
64.14
55.27
63.53
56.07
61.56
62.79
66.13
65.03
65.64
71.69
66.57
61.32
72.28
66.84
68.07
68.51
66.22
66.68
69.86
64.14
56.34
64.82
60.72
65.66
67.38
68.87
51.61
67.81
58.91
55.11
61.72
63.26
70.89
68.90
60.34
70.74
72.26
59.62
63.85
65.57
69.82
67.22
74.19
69.98
54.76
60.48
47.41
74.20
60.67
65.01
55.87
55.73
56.78
64.43
55.44
63.64
56.16
61.62
42
-------
New Hampshire
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Mexico
New Mexico
New Mexico
New Mexico
New Mexico
New Mexico
New Mexico
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
Sullivan
Atlantic
Bergen
Camden
Cumberland
Gloucester
Hudson
Hunterdon
Mercer
Middlesex
Monmouth
Morris
Ocean
Passaic
Bernalillo
Dona Ana
Eddy
Grant
Lea
Sandoval
San Juan
Albany
Bronx
Chautauqua
Chemung
Dutchess
Erie
Essex
Hamilton
Herkimer
Jefferson
Madison
Monroe
Niagara
Oneida
Onondaga
Orange
Oswego
Putnam
Queens
Rensselaer
Richmond
Saratoga
Schenectady
Suffolk
Ulster
Wayne
Westch ester
Alexander
A very
Buncombe
Caldwell
Caswell
70.00
79.30
86.00
89.30
83.30
87.00
85.70
89.00
88.00
88.30
87.30
83.30
93.00
81.00
73.70
75.30
69.00
66.00
69.50
73.30
71.30
73.70
74.70
86.70
68.70
75.70
85.00
77.00
71.70
68.30
78.00
72.00
75.00
82.70
68.30
73.70
82.00
78.00
84.30
80.00
77.30
88.30
79.70
70.00
90.30
77.30
68.00
87.70
77.00
70.00
74.00
74.30
76.30
57.34
65.49
74.46
72.65
66.32
71.61
73.50
70.25
72.37
71.71
74.07
67.01
75.75
67.07
61.89
68.52
64.38
59.96
65.30
61.55
67.01
59.59
67.68
73.83
57.63
61.25
70.86
64.66
60.64
58.08
64.47
59.75
62.57
71.53
57.13
62.92
66.60
66.62
69.71
69.57
62.29
75.53
64.39
56.91
81.86
62.86
57.22
76.61
57.64
56.78
59.31
54.93
58.13
57.43
65.53
74.48
72.73
66.39
71.68
73.49
70.37
72.46
71.78
74.07
67.10
75.80
67.13
62.02
68.58
64.44
60.01
65.34
61.68
67.05
59.72
67.65
74.02
57.75
61.34
71.07
64.81
60.77
58.21
64.72
59.87
62.81
71.75
57.30
63.10
66.67
66.83
69.76
69.56
62.46
75.54
64.56
57.05
81.83
62.98
57.42
76.60
57.72
56.89
59.43
55.00
58.22
57.60
65.60
74.46
72.88
66.54
71.78
73.47
70.59
72.61
71.91
74.07
67.26
75.92
67.24
62.04
68.60
64.52
60.03
65.40
61.70
67.06
59.97
67.53
74.51
58.04
61.48
71.58
65.23
61.12
58.54
64.90
60.21
63.03
71.96
57.68
63.43
66.78
67.11
69.84
69.54
62.73
75.54
64.85
57.34
81.77
63.25
57.64
76.49
58.15
57.23
59.81
55.52
58.74
43
-------
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Carolina
North Dakota
North Dakota
North Dakota
North Dakota
North Dakota
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Chatham
Cumberland
Davie
Durham
Edgecombe
Forsyth
Franklin
Graham
Granville
Guilford
Haywood
Jackson
Johnston
Lenoir
Lincoln
Martin
Mecklenburg
New Hanover
Person
Pitt
Rockingham
Rowan
Swain
Union
Wake
Yancey
Billings
Burke
Cass
McKenzie
Oliver
Allen
Ashtabula
Butler
Clark
Clermont
Clinton
Cuyahoga
Delaware
Franklin
Geauga
Greene
Hamilton
Jefferson
Knox
Lake
Lawrence
Licking
Lorain
Lucas
Madison
Mahoning
Medina
73.30
81.70
81.30
77.00
77.00
80.00
78.70
78.30
82.00
82.00
78.30
76.00
77.30
75.30
81.00
75.00
89.30
72.30
77.30
76.30
77.00
86.70
66.30
79.30
80.30
76.00
61.50
57.50
60.00
61.30
57.70
78.70
89.00
83.30
81.00
81.00
82.30
79.70
78.30
86.30
79.30
80.30
84.70
78.00
77.70
86.30
70.70
78.00
76.70
81.30
79.70
78.70
80.30
56.71
61.89
61.84
57.26
58.76
62.34
60.04
61.41
62.59
61.78
63.42
60.49
57.23
60.34
60.65
62.23
68.17
61.79
59.81
57.93
57.75
65.70
52.49
58.52
60.77
61.17
56.57
52.72
51.18
56.46
52.93
65.15
75.07
67.28
64.18
65.88
63.54
67.43
63.38
69.39
63.42
63.77
67.74
62.64
61.44
71.72
58.31
61.92
64.21
67.70
62.16
61.66
65.36
56.79
61.96
61.93
57.33
58.84
62.43
60.12
61.64
62.69
61.88
63.59
60.65
57.29
60.41
60.73
62.32
68.22
61.85
59.90
58.02
57.83
65.78
52.64
58.58
60.84
61.33
56.73
52.78
51.52
56.59
52.97
65.43
75.37
67.44
64.37
66.04
63.71
67.31
63.60
69.66
63.72
63.94
67.91
62.84
61.69
71.46
58.42
62.15
64.00
67.98
62.40
61.93
65.66
57.21
63.08
62.45
57.88
59.31
63.18
60.63
61.97
63.34
62.55
63.90
61.01
57.82
60.80
61.21
62.66
68.54
62.20
60.55
58.46
58.42
66.31
52.96
58.98
61.34
61.67
56.83
52.96
52.66
56.71
53.04
66.13
75.86
67.76
64.80
66.39
63.93
67.51
64.21
70.35
64.17
64.26
68.23
63.09
62.32
71.55
58.56
62.73
64.13
68.17
62.94
62.54
65.89
44
-------
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oklahoma
Oregon
Oregon
Oregon
Oregon
Oregon
Oregon
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Miami
Montgomery
Portage
Preble
Stark
Summit
Trumbull
Warren
Washington
Wood
Adair
Canadian
Cherokee
Cleveland
Comanche
Creek
Dewey
Kay
McClain
Mayes
Oklahoma
Ottawa
Pittsburg
Tulsa
Clackamas
Columbia
Jackson
Lane
Marion
Multnomah
Adams
Allegheny
Armstrong
Beaver
Berks
Blair
Bucks
Cambria
Centre
Chester
Clearfield
Dauphin
Delaware
Erie
Franklin
Greene
Indiana
Lackawanna
Lancaster
Lawrence
Lehigh
Luzerne
Lycoming
76.70
74.00
83.70
73.00
81.00
83.70
84.30
87.70
82.70
80.00
75.70
76.00
75.70
74.70
77.50
76.70
72.70
78.00
72.00
78.50
80.00
78.00
72.00
79.30
66.30
58.70
68.00
69.30
65.70
56.30
76.30
83.70
83.00
83.00
76.00
74.30
88.00
74.70
78.30
86.00
78.30
79.30
83.30
81.30
72.30
80.00
80.00
75.30
83.30
72.30
83.30
76.30
77.30
60.06
58.37
66.80
57.60
64.54
67.28
66.18
69.70
68.43
65.36
63.78
61.58
66.40
62.14
64.22
62.92
61.02
64.01
59.90
69.61
65.50
65.99
61.32
67.89
62.18
54.33
56.05
59.58
57.84
57.83
60.02
67.83
65.65
68.78
60.39
58.94
73.97
60.00
62.20
66.77
62.17
64.44
68.12
68.93
56.99
63.42
63.12
59.63
66.27
57.98
66.03
60.42
62.99
60.25
58.52
67.14
57.77
64.81
67.64
66.47
69.89
68.63
65.66
63.97
61.92
66.54
62.38
64.39
63.18
61.16
64.21
60.12
69.75
65.85
66.20
61.51
68.12
62.22
54.45
56.26
59.79
58.00
57.67
60.12
67.98
65.83
68.95
60.56
59.11
74.03
60.15
62.42
66.91
62.36
64.57
68.18
69.15
57.07
63.58
63.28
59.87
66.45
58.17
66.20
60.64
63.16
60.64
58.79
67.70
58.14
65.35
68.15
66.97
70.27
68.89
66.26
64.62
62.27
67.53
62.84
65.02
64.18
61.45
64.64
60.44
70.39
66.18
66.89
61.91
68.78
62.26
54.47
56.39
60.21
58.20
57.54
60.29
68.28
66.13
69.26
60.80
59.40
74.13
60.46
62.86
67.07
62.71
64.79
68.24
69.65
57.23
63.93
63.58
60.26
66.67
58.56
66.47
60.96
63.52
45
-------
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Rhode Island
Rhode Island
Rhode Island
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Carolina
South Dakota
South Dakota
South Dakota
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Tennessee
Texas
Texas
Texas
Texas
Mercer
Montgomery
Northampton
Perry
Philadelphia
Tioga
Washington
Westmoreland
York
Kent
Providence
Washington
Abbeville
Aiken
Anderson
Barnwell
Berkeley
Charleston
Cherokee
Chester
Chesterfield
Colleton
Darlington
Edgefield
Oconee
Pickens
Richland
Spartanburg
Union
Williamsburg
York
Custer
Jackson
Minnehaha
Anderson
Blount
Davidson
Hamilton
Jefferson
Knox
Loudon
Meigs
Rutherford
Sevier
Shelby
Sullivan
Sumner
Williamson
Wilson
Bexar
Brazoria
Brewster
Collin
82.00
85.70
84.30
77.00
90.30
77.70
78.30
79.00
82.00
84.30
82.30
86.00
79.00
76.00
76.50
73.00
67.30
74.00
74.00
75.70
75.00
72.30
76.30
70.00
73.00
78.70
82.30
82.30
76.00
69.30
76.70
70.00
67.00
66.00
77.30
85.30
77.70
81.00
82.30
85.00
83.00
80.00
76.30
80.70
80.70
80.30
83.00
75.30
78.70
85.00
94.70
64.00
90.30
64.94
71.24
66.73
60.90
76.12
64.53
63.45
63.69
65.36
69.67
67.79
71.10
60.78
56.79
57.05
54.00
52.82
62.55
57.67
56.52
59.34
56.81
59.45
52.89
55.09
59.16
60.63
60.61
60.12
55.09
57.92
64.60
61.35
56.41
55.86
62.23
58.48
61.14
60.37
61.89
60.51
59.26
57.32
59.07
61.50
67.64
62.79
56.51
60.64
73.28
82.10
56.38
72.47
65.20
71.32
66.89
61.12
76.20
64.68
63.62
63.81
65.46
69.72
67.85
71.15
60.90
57.03
57.14
54.21
52.90
62.65
57.76
56.58
59.43
56.99
59.54
53.12
55.19
59.23
60.70
60.72
60.19
55.20
57.98
64.67
61.43
56.64
56.17
62.52
58.66
61.38
60.65
62.17
61.02
59.56
57.51
59.34
61.70
67.75
62.97
56.66
60.82
73.42
82.18
56.47
72.63
65.75
71.45
67.16
61.35
76.31
65.00
63.93
64.08
65.63
69.80
67.98
71.25
61.39
57.94
57.65
54.90
53.37
63.11
58.27
57.06
59.77
57.46
59.95
54.04
55.71
59.81
61.25
61.90
60.62
55.56
58.48
64.73
61.53
57.41
56.44
62.76
58.83
61.55
60.88
62.41
61.29
59.96
57.77
59.63
61.91
67.90
63.12
56.84
61.02
73.59
82.28
56.61
72.85
46
-------
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Utah
Utah
Utah
Utah
Utah
Utah
Utah
Utah
Utah
Vermont
Vermont
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Dallas
Denton
Ellis
El Paso
Galveston
Gregg
Harris
Harrison
Hidalgo
Hood
Hunt
Jefferson
Johnson
Kaufman
Montgomery
Nueces
Orange
Parker
Rockwall
Smith
Tarrant
Travis
Victoria
Webb
Box Elder
Cache
Davis
Salt Lake
San Juan
Tooele
Utah
Washington
Weber
Bennington
Chittenden
Arlington
Caroline
Charles City
Chesterfield
Fairfax
Fauquier
Frederick
Hanover
Henrico
Loudoun
Madison
Page
Prince William
Roanoke
Rockbridge
Stafford
Wythe
Alexandria City
88.30
94.00
81.70
77.70
80.30
84.30
100.70
79.00
65.70
83.00
78.00
84.70
87.00
74.70
85.00
72.30
78.00
88.70
79.70
81.00
95.30
81.30
72.30
61.30
76.00
68.70
81.30
81.00
70.30
78.00
76.70
78.50
80.30
72.00
69.70
86.70
80.00
80.30
76.70
90.00
72.70
72.30
81.30
82.00
80.70
77.70
74.00
78.70
74.70
69.70
81.70
72.70
81.70
74.61
72.51
64.91
70.26
71.73
73.47
89.17
65.63
57.30
62.70
65.38
74.98
67.73
63.61
72.76
64.40
67.62
66.46
68.23
69.03
73.84
66.57
63.52
54.61
68.29
60.96
71.92
71.69
64.99
67.86
71.42
68.96
70.51
57.74
58.20
71.91
61.41
65.69
62.43
71.76
57.67
57.62
64.59
66.09
62.34
63.21
60.13
61.91
60.78
58.20
63.01
59.33
65.14
74.73
72.69
65.04
70.36
71.82
73.59
89.24
65.84
57.43
62.87
65.54
75.10
67.89
63.76
72.80
64.65
67.76
66.64
68.36
69.18
73.99
66.78
63.62
54.69
68.41
61.09
72.27
71.82
65.03
68.25
71.47
69.04
70.78
57.90
58.37
71.99
61.50
65.75
62.48
71.85
57.76
57.72
64.66
66.14
62.41
63.29
60.23
61.98
60.88
58.30
63.07
59.45
65.22
74.94
72.94
65.25
70.39
71.98
73.80
89.29
66.13
57.59
63.15
65.81
75.36
68.13
63.99
72.80
64.84
68.01
66.91
68.57
69.45
74.24
67.01
63.80
54.84
68.51
61.19
72.37
71.91
65.08
68.41
71.52
69.10
70.90
58.18
58.63
72.12
61.75
66.03
62.71
72.04
57.96
57.93
64.93
66.37
62.59
63.50
60.48
62.13
61.22
58.59
63.32
59.79
65.39
47
-------
Virginia
Virginia
Washington
Washington
Washington
Washington
Washington
Washington
Washington
Washington
West Virginia
West Virginia
West Virginia
West Virginia
West Virginia
West Virginia
West Virginia
West Virginia
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wisconsin
Wyoming
Wyoming
Wyoming
Hampton City
Suffolk City
Clark
King
Klickitat
Pierce
Skagit
Spokane
Thurston
Whatcom
Berkeley
Cabell
Greenbrier
Hancock
Kanawha
Monongalia
Ohio
Wood
Ashland
Brown
Columbia
Dane
Dodge
Door
Florence
Fond Du Lac
Forest
Jefferson
Kenosha
Kewaunee
Manitowoc
Marathon
Milwaukee
Oneida
Outagamie
Ozaukee
Racine
Rock
St Croix
Sauk
Sheboygan
Vernon
Vilas
Walworth
Washington
Waukesha
Campbell
Sublette
Teton
76.70
76.70
59.50
72.30
64.50
68.70
46.00
68.30
65.00
57.00
75.00
78.70
69.70
75.70
77.30
75.30
78.30
79.00
61.50
73.70
72.70
72.00
74.70
88.70
66.30
73.70
69.50
74.30
84.70
82.70
85.00
70.00
82.70
69.00
74.00
83.30
80.30
74.00
69.00
69.70
88.00
69.70
68.70
75.70
72.30
75.00
67.30
70.00
62.70
67.57
70.76
59.23
66.42
59.53
62.11
45.83
58.43
56.82
55.54
60.18
64.75
59.86
61.88
60.83
58.16
62.93
64.31
53.14
62.25
59.48
59.65
61.74
72.97
57.03
61.94
59.77
61.37
76.27
68.76
71.47
60.32
71.63
59.60
61.98
71.67
71.56
61.24
58.71
57.88
74.71
59.00
59.51
62.74
61.14
63.24
64.63
65.10
57.02
67.60
70.77
59.22
66.48
59.67
62.13
45.79
58.72
57.17
55.52
60.29
64.86
59.98
62.05
61.01
58.31
63.12
64.50
53.27
62.40
59.69
59.86
62.02
73.11
57.19
62.17
59.94
61.46
76.39
68.87
71.59
60.55
71.75
59.76
62.20
71.78
71.68
61.41
58.85
58.14
74.84
59.21
59.67
62.91
61.32
63.38
64.68
65.15
57.12
67.76
70.70
59.23
66.37
59.73
62.06
45.78
58.75
57.24
55.50
60.48
65.00
60.36
62.35
61.25
58.70
63.42
64.75
53.94
62.97
60.62
60.83
62.90
73.75
57.74
62.88
60.53
62.05
76.62
69.40
72.11
61.30
72.16
60.42
62.89
72.19
71.91
62.34
59.84
58.97
75.33
59.99
60.31
63.50
61.93
63.88
64.70
65.21
57.18
48
-------
Appendix B: 2002 CMAQ Model Performance
Evaluation for Ozone, Particulate Matter and Air
Toxics
49
-------
A. Introduction
An operational model performance evaluation for ozone, PM2 5 and its related speciated
components, and specific air toxics (i.e., formaldehyde, acetaldehyde, benzene, 1,3-butadiene,
acrolein, and naphthalene) was conducted using 2005 State/local monitoring sites data in order to
estimate the ability of the CMAQ modeling system to replicate the base year concentrations for
the 12-km Eastern and Western United States domain1. This evaluation principally comprises
statistical assessments of model versus observed pairs that were paired in space and time on a
daily or weekly basis, depending on the sampling frequency of each network (measured data).
For certain time periods with missing ozone, PM25 and air toxic observations we excluded the
CMAQ predictions from those time periods in our calculations. It should be noted when pairing
model and observed data that each CMAQ concentration represents a grid-cell volume-averaged
value, while the ambient network measurements are made at specific locations. In conjunction
with the model performance statistics, we also provide spatial plots for individual monitors of the
calculated bias and error statistics (defined below). Statistics were generated for the 12-km
Eastern US domain (EUS), 12-km Western US domain (WUS), and five large subregions2:
Midwest, Northeast, Southeast, Central, and West U.S. The Atmospheric Model Evaluation
Tool (AMET) was used to conduct the evaluation described in this document.3
The ozone evaluation primarily focuses on observed and predicted one-hour daily maximum
ozone concentrations and eight-hour daily maximum ozone concentrations at a threshold of
40ppb. This ozone model performance was limited to the ozone season modeled for the Final
National Renewable Fuel Standard Rule (commonly known as RFS2): May, June, July, August,
and September. Ozone ambient measurements for 2005 were obtained from the Air Quality
System (AQS) Aerometric Information Retrieval System (AIRS). A total of 1194 ozone
measurement sites were included for evaluation. The ozone data were measured and reported on
an hourly basis.
The PM2.s evaluation focuses on PM2 5 total mass and its components including sulfate (SO4),
nitrate (NOs), total nitrate (TNC^NOs+HNOs), ammonium (NH4), elemental carbon (EC), and
organic carbon (OC). The PM2 5 performance statistics were calculated for each month and
season individually and for the entire year, as a whole. Seasons were defined as: winter
(December-January-February), spring (March-April-May), summer (June-July-August), and fall
(September-October-November). PM2 5 ambient measurements for 2002 were obtained from the
'See Air Quality Modeling Technical Support Document, 2010 (EPA 454/R-10-001): Changes to the Renewable
Fuel Standard Program (Figure II-l) for the map of the CMAQ modeling domain.
2 The subregions are defined by States where: Midwest is IL, IN, MI, OH, and WI; Northeast is CT, DE,
MA, MD, ME, NH, NJ, NY, PA, RI, and VT; Southeast is AL, FL, GA, KY, MS, NC, SC, TN, VA, and
WV; Central is AR, IA, KS, LA, MN, MO, ME, OK, and TX; West is AK, CA, OR, WA, AZ, MM, CO, UT, WY,
SD, ND, MT, ID, and NV.
3 Gilliam, R. C., W. Appel, and S. Phillips. The Atmospheric Model Evaluation Tool (AMET): Meteorology Module.
Presented at 4th Annual CMAS Models-3 Users Conference, Chapel Hill, NC, September 26 - 28, 2005.
(http://www.cmascenter.org/)
50
-------
following networks for model evaluation: Speciation Trends Network (STN- total of 260 sites),
Interagency Monitoring of PROtected Visual Environments (IMPROVE- total of 204), Clean
Air Status and Trends Network (CASTNet- total of 93), and National Acid Deposition
Program/National Trends (NADP/NTN- toal of 297). The pollutant species included in the
evaluation for each network are listed in Table A-l. For PM2.5 species that are measured by
more than one network, we calculated separate sets of statistics for each network.
Table A-l. PM2.s monitoring networks and pollutants species included in the CMAQ
performance evaluation.
Ambient
Monitoring
Networks
IMPROVE
CASTNet
STN
NADP
Particulate
Species
PM2.5
Mass
X
X
SO4
X
X
X
N03
X
X
TNO3a
X
EC
X
X
NH4
X
X
X
oc
X
X
Wet
Deposition
Species
SO4
X
N03
X
a TNO3 = (NO3 + HNO3)
The air toxics evaluation focuses on specific species relevant to the RFS2 final rule, i.e.,
formaldehyde, acetaldehyde, benzene, 1,3-butadiene, acrolein, and naphthalene. Similar to the
PM2.5 evaluation, the air toxics performance statistics were calculated for each month and
season individually and for the entire year, as a whole to estimate the ability of the CMAQ
modeling system to replicate the base year concentrations for the 12-km Eastern and Western
United States domains. As mentioned above, seasons were defined as: winter (December-
January-February), spring (March-April-May), summer (June-July-August), and fall (September-
October-November). Toxic measurements for 2005 were obtained from the National Air Toxics
Trends Stations (NATTS). Toxic measurements from 471 sites in the East and 135 sites in the
West were included in the evaluation for the 12km Eastern and Western U.S. grids, respectively.
There are various statistical metrics available and used by the science community for model
performance evaluation. For a robust evaluation, the principal evaluation statistics used to
evaluate CMAQ performance were two bias metrics, normalized mean bias and fractional bias;
and two error metrics, normalized mean error and fractional error.
Normalized mean bias (NMB) is used as a normalization to facilitate a range of concentration
magnitudes. This statistic averages the difference (model - observed) over the sum of observed
values. NMB is a useful model performance indicator because it avoids over inflating the
observed range of values, especially at low concentrations. Normalized mean bias is defined as:
51
-------
*100
NMB =
Normalized mean error (NME) is also similar to NMB, where the performance statistic is used as
a normalization of the mean error. NME calculates the absolute value of the difference (model -
observed) over the sum of observed values. Normalized mean error is defined as:
NME =
•*100
Fractional bias is defined as:
FB= -
n
I
(P+0)
*100, where P = predicted concentrations and O = observed
concentrations. FB is a useful model performance indicator because it has the advantage of
equally weighting positive and negative bias estimates. The single largest disadvantage in this
estimate of model performance is that the estimated concentration (i.e., prediction, P) is found in
both the numerator and denominator. Fractional error (FE) is similar to fractional bias except the
absolute value of the difference is used so that the error is always positive. Fractional error is
defined as:
- o\
i
V i
*100
The "acceptability" of model performance was judged by comparing our CMAQ 2005
performance results to the range of performance found in recent regional ozone, PM2.5, and air
toxic4'5'6 model applications (e.g., Clean Air Interstate Rule7, Final PM NAAQS Rule8, and EPA's
4 Phillips, S., K. Wang, C. Jang, N. Possiel, M. Strum, T. Fox, 2007: Evaluation of 2002 Multi-pollutant Platform:
Air Toxics, Ozone, and Paniculate Matter, 7th Annual CMAS Conference, Chapel Hill, NC, October 6-8, 2008.
5 Strum, M., Wesson, K., Phillips, S., Cook, R., Michaels, H., Brzezinski, D., Pollack, A., Jimenez, M., Shepard, S.
Impact of using lin-level emissions on multi-pollutant air quality model predictions at regional and local scales. 17tt
Annual International Emission Inventory Conference, Portland, Oregon, June 2-5, 2008.
6 Wesson, K., N. Farm, and B. Timin, 2010: Draft Manuscript: Air Quality and Benefits Model Responsiveness to
Varying Horizontal Resolution in the Detroit Urban Area, Atmospheric Pollution Research, Special Issue: Air
Quality Modeling and Analysis.
52
-------
Proposal to Designate an Emissions Control Area for Nitrogen Oxides9). These other modeling
studies represent a wide range of modeling analyses which cover various models, model
configurations, domains, years and/or episodes, chemical mechanisms, and aerosol modules.
Overall, the NMB, NME, FB, and FE statistics shown in Sections B through P below for CMAQ
predicted 2005 ozone, PM2.5, and air toxics concentrations are within the range or close to that
found in recent OAQPS applications. The CMAQ model performance results give us confidence
that our applications of CMAQ using this 2005 modeling platform provide a scientifically
credible approach for assessing ozone and PM2 5 concentrations for the purposes of the RFS2
Final Rule. We discuss in the following sections the bias and error results for the one-hour
maximum ozone concentrations and eight-hour daily maximum ozone concentrations evaluated
at a threshold of 40 ppb, the annual and seasonal PM2.5 and its related speciated components as
well as specific air toxic concentrations.
7 See: U.S. Environmental Protection Agency; Technical Support Document for the Final Clean Air Interstate Rule:
Air Quality Modeling; Office of Air Quality Planning and Standards; RTF, NC; March 2005 (CAIR Docket OAR-
2005-0053-2149).
8 U.S. Environmental Protection Agency, 2006. Technical Support Document for the Final PM NAAQS Rule:
Office of Air Quality Planning and Standards, Research Triangle Park, NC.
9 U.S. Environmental Protection Agency, Proposal to Designate an Emissions Control Area for Nitrogen Oxides,
Sulfur Oxides, and Paniculate Matter: Technical Support Document. EPA-420-R-007, 329pp., 2009.
(http://www.epa.gov/otaq/regs/nonroad/marine/ci/420r09007.pdf)
53
-------
B. One-Hour Daily Maximum Ozone Performance
Ozone Performance: Threshold of 40 ppb
Table B-l provides one-hour daily maximum ozone model performance statistics calculated for a
threshold of 40 ppb of observed and modeled concentrations, restricted to the ozone season
modeled for the 12-km Eastern and Western U.S. domain and the five subregions (Midwest,
Northeast, Southeast, Central and Western U.S.). Spatial plots of the NMB and NME statistics
(units of percent) for individual monitors are also provided as a complement to the tabular
statistical data (Figures B-l - B-24). Overall, one-hour daily maximum ozone model
performance is slightly under-predicted or near negligible in both the 12-km BUS and WUS
when applying a threshold of 40 ppb for the modeled ozone season (May-September). For the
12-km Eastern domain, the bias and error statistics are comparable for the aggregate of the ozone
season and for each individual ozone month modeled, with a NMB range of-1% to -5% and a
FB range of -0.5% to -4%, and a NME and FE range of 11% to 14%. Likewise, for the 12-km
Western domain, the bias and error statistics are similar between the ozone seasonal aggregate
and the individual months, with a NMB and FB approximately -2%, and a NME and FE
approximately 14%. Hourly ozone model performance when compared across the five
subregions shows slightly better performance in the Southeast. In general, the Northeast,
Midwest, Central and West U.S. exhibit similar bias and error statistics for the episodes modeled.
The month of August shows a slightly better bias and error model performance results, although
the results are spatially and temporally comparable across the months modeled.
Table B-l. 2005 CMAQ one-hour daily maximum ozone model performance statistics
calculated for a threshold of 40 ppb.
CMAQ 2005 One-Hour Maximum Ozone:
Threshold of 40 ppb
May
June
July
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
No. of Obs.
21394
9631
4418
4102
6424
4328
8294
19517
9056
4639
4148
4644
4062
7737
19692
9443
4923
4445
4733
NMB (%)
-1.6
-3.4
0.8
5.4
-3.6
-6.4
-3.5
-3.5
-3.7
-4.6
-1.0
-2.7
-6.2
-4.0
1.2
0.4
0.4
4.2
4.2
NME (%)
11.5
12.8
10.0
11.8
11.3
13.4
12.9
12.8
13.0
12.3
14.1
12.5
13.2
13.1
14.2
16.0
12.7
15.2
15.1
FB (%)
-0.8
-2.8
1.0
5.9
-3.0
-5.5
-3.0
-2.8
-3.2
-4.0
-0.1
-2.2
-5.4
-3.6
1.8
1.0
0.9
4.8
4.6
FE (%)
11.6
12.7
10.2
11.7
11.5
13.4
12.8
12.9
13.0
12.4
14.2
12.6
13.3
13.1
14.1
15.8
12.6
14.9
14.8
54
-------
August
September
Seasonal Aggregate
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
3521
8168
19643
9562
4549
4139
5303
3589
8357
18085
8725
4002
3667
5259
3286
7530
98331
46417
22531
20501
26363
18786
40086
-3.8
0.2
0.1
-0.8
0.2
0.2
3.6
-4.1
-1.0
-2.2
-3.6
-3.6
-1.8
-0.1
-6.1
-4.1
-1.2
-2.1
-1.4
1.4
0.1
-5.4
-2.3
14.8
16.2
13.9
15.5
12.2
13.2
14.9
16.2
15.7
12.0
14.1
10.7
11.3
12.1
14.5
14.3
12.9
14.3
11.7
13.3
13.1
14.4
14.5
-3.1
0.7
0.8
-0.6
1.0
1.2
3.9
-2.9
-1.0
-1.3
-3.2
-3.0
-0.7
0.8
-5.1
-3.8
-0.5
-1.7
-0.8
2.3
0.7
-4.4
-2.1
14.9
16.0
13.8
15.5
12.3
13.1
14.5
16.1
15.7
12.0
14.3
10.8
11.3
12.1
14.5
14.4
12.8
14.2
11.7
13.1
13.0
14.4
14.4
55
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for May
- =—= = = = =
CIRCLE=AQS_1hrmax;
Figure B-l. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., May 2005.
O3 NME (%) for run 2005cLtox_05b_alt_bc_v47_N1b_MP_12km for May
/
r
W~\ 5-^-^
\ f
An Atirospneri; Mcde E¥irfyfefcJlMET} Product
CIRCLE=AQS_1hrmax;
Figure B-2. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., May 2005.
56
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for June
I -c^^-- -jp-
CIRCLE=AQS_1hrmax;
Figure B-3. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., June 2005.
O3 NME (%) for run 2005cLtox_05b_alt_bc_v47_N1b_MP_12km for June
^^_^
.
CIRCLE=AQS_1hrmax;
Figure B-4. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., June 2005.
57
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for July
" ----jp—** ~™
->£?*
"
CIRCLE=AQS_1hrmax;
Figure B-5. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., July 2005.
O3 NME (%) for run 2005cLtox_05b_alt_bc_v47_N1b_MP_12krn for ju|y
/ * _LV*'r* "I' *«
* " '^^ *
CIRCLE=AQS_1hrmax;
Figure B-6. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., July 2005.
58
-------
O3NMB(%)forrun2005ci tox 05b alt be v47 N1b MP 12km for August
— ^--' -"--j-—sa-
/ >~ '- ;vv' 7~:, "*—---, u^'
/ '.' "A \ "^r
CIRCLE=AQS_1hrmax;
Figure B-7. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., August 2005.
O3JJME (%) JornjnjggscjJgQgballbcj^TJjIbJjPjauiyorAug^
CIRCLE=AQS_1hrmax;
Figure B-8. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., August 2005.
59
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for September
= "= -- = = =
CIRCLE=AQS_1hrmax;
Figure B-9. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., September 2005.
O3 NME (%) for run 2005ci_tox_05b_alt_bc_v47_N1 b_MP_12km for September
CIRCLE=AQS_1hrmax;
Figure B-10. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., September 2005.
60
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for May to September
-c-- —f-
CIRCLE=AQS_1hrmax;
Figure B-ll. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., seasonal aggregate 2005.
O3 NME (%) for run 200^
CIRCLE=AQS_1hrmax;
Figure B-12. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., seasonal aggregate 2005.
61
-------
O3 NMB (%) tor run 2005ci_tox_05b_alt_bc_v47_JI1b_MP_W12km tor May
An Atmospheric MoSfiKtfttluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-13. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., May 2005.
O3 NME (%) tor run 2005cLtox_05b_alt_bc_v47_N1b_MP_W12km tor May
An Atmospheric MoVefe&luation (AMET) Produ
« i \ V
CIRCLE=AQS_1hrmax;
Figure B-14. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., May 2005.
62
-------
O3 NMB (%) tor run 2005ci_tox_05b_alt_bc_v47_N1b_MP_W12kin (or June
An Atmospheric MoHaFEifeluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-15. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., June 2005.
03 NME (%) for run 2005cLtox_05b_alt_bc_v47_Nlb_MP_W12km for June
CIRCLE=AQS_1hrmax;
Figure B-16. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., June 2005.
63
-------
O3 NMB <%) for run 2005ci_ton_05b_alt_bc_v47_N1b_MP_W12km tor July
An Atmospheric MoHeKfeAlluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-17. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., July 2005.
O3 NME (%) tor run 20Q5ci_tox_05bjaltj>c,jf47jm b_MP_W12km (or July
An Atmospheric MoMeKtfttluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-18. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., July 2005.
64
-------
O3 NMB (%) tor run 2005cLloxJ>5b_altJX!_v47._Nlb_MP_W12km tor August
An Atmospheric MoSfiKtfttluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-19. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., August 2005.
O3 NME (%) tor run 2005cLk>xJ)5b^altJ)cj47J^1bJ>/IPJW12km tor August
An Atmospheric MoSeftftlluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-20. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., August 2005.
65
-------
03 NMB (%) for run 2005ei tox 05b alt be v47 N1b MP W12km for September
An Atmosbheric MoasKEftiluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-21. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., September 2005.
O3 NME (%) tor run 2005ciJox_05b_all_bc_v47^N1b_MP^Wl2lcm tor September
An Atmospheric MoSerefiiluation (AMET) Produ
CIRCLE=AQS_1hrmax;
Figure B-22. Normalized Mean Error (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., September 2005.
66
-------
O3 NMB {%) for run 2005ci tox 05b alt be v47 N1b MP W12km for May to September
An Atmosbheric MoSsKEftlluation (AMET) Prod
CIRCLE=AQS_1hrmax;
Figure B-23. Normalized Mean Bias (%) of one-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., seasonal aggregate 2005.
O3 NME (%) for run 2005ci_toxJ)5b_alt_bc ¥47J41bJ
-------
C. Eight-hour Daily Maximum Ozone Performance
Ozone Performance: Threshold of 40 ppb
Table C-l presents eight-hour daily maximum ozone model performance bias and error statistics
for the entire range of observed and modeled concentrations at a threshold of 40 ppb for the
ozone season modeled for the 12-km Eastern and Western U.S. domain and the corresponding
subregions defined above. Spatial plots of the NMB and NME statistics (units of percent) for
individual monitors based on the aggregate and the individual ozone months modeled
respectively are shown in Figures C-l through C-24. In general, CMAQ slightly under-predicts
eight-hour daily maximum ozone with a threshold of 40 ppb in the months of May, June and
August. Likewise, model predictions in the BUS and WUS are slightly over-predicted in the
months of July and August. For the 12-km Eastern domain, the bias statistics are within the
range of approximately -4% to 7%, while the error statistics range from 11% to 14% for the
aggregate of the ozone season and for most of the months modeled. For the 12-km Western
domain, the bias statistics are within the range of approximately 3% to -3%, while the error
statistics range from 11% to 13% for the aggregate of the ozone season and for the individual
months modeled. The five subregions show relatively similar eight-hour daily maximum ozone
performance.
Table C-l. 2005 CMAQ eight-hour daily maximum ozone model performance statistics
calculated for a threshold of 40 pbb.
CMAQ 2005 Eight-Hour Maximum
Ozone: Threshold of 40 ppb
May
June
July
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-km BUS
12-km WUS
Midwest
Northeast
Southeast
No. of Obs.
19310
8445
3858
3528
6019
3927
7234
17404
8102
4324
3590
3924
3663
6889
17045
8556
4429
3856
3806
NMB (%)
-1.0
-1.6
0.2
5.2
-2.1
-5.8
-1.8
-2.1
-1.9
-3.8
0.3
-0.3
-5.5
-2.2
3.3
3.7
1.8
6.6
7.4
NME (%)
10.9
12.0
10.0
11.4
10.5
12.8
12.1
11.9
11.9
11.6
13.1
11.4
12.1
12.1
13.4
15.0
11.8
14.6
15.0
FB (%)
-0.4
-1.2
0.7
5.4
-1.6
-5.2
-1.5
-1.5
-1.6
-3.4
1.0
0.1
-5.0
-2.0
3.6
3.9
2.3
6.8
7.3
FE (%)
11.0
12.0
10.2
11.2
10.6
13.0
12.0
12.0
11.9
11.8
13.2
11.5
12.3
12.1
13.3
14.7
11.8
14.3
14.5
68
-------
August
September
Seasonal Aggregate
(May - September)
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-kmWUS
Midwest
Northeast
Southeast
Central U.S.
West
3057
7407
16953
8523
4027
3530
4447
3096
7469
15190
7465
3265
2856
4647
2798
6446
85902
41091
19903
17360
22843
16541
35445
-2.3
3.5
1.9
1.6
0.9
1.4
7.4
-3.4
1.4
-1.8
-2.4
-4.2
-2.3
1.5
-6.5
-2.9
0.1
0.0
-0.9
2.4
2.3
-4.8
-0.2
13.2
15.1
12.9
13.9
11.3
12.3
14.7
14.4
14.1
11.2
13.4
10.2
10.6
11.2
13.6
13.7
12.1
13.3
11.1
12.6
12.3
13.2
13.5
-2.1
3.6
2.2
1.5
1.4
2.0
7.2
-3.1
1.2
-1.3
-2.6
-4.0
-1.8
2.1
-6.1
-3.1
0.5
0.1
-0.5
2.9
2.6
-4.4
-0.3
13.5
14.9
12.9
13.9
11.4
12.2
14.1
14.8
14.0
11.3
13.9
10.4
10.7
11.2
14.0
14.1
12.1
13.3
11.2
12.4
12.2
13.4
13.5
69
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for May
' = "3- = = = =
CIRCLE=AQS_8hrmax;
Figure C-l. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., May 2005.
O3 NME (%) for run 2005cLtox_05b_alt_bc_v47_N1b_MP_12km for May
x
Vs/'-x \ ^--«^
CIRCLE=AQS_8hrmax;
Figure C-2. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., May 2005.
70
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for June
— - = ,,-^,- = = = =
?= v -^>r—
CIRCLE=AQS_8hrmax;
Figure C-3. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., June 2005.
O3 NME (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for June
- "
CIRCLE=AQS_8hrmax;
Figure C-4. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., June 2005.
71
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for July
- -= ^p-= = = = = -
CIRCLE=AQS_8hrmax;
Figure C-5. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., July 2005.
O3 NME (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for July
~~~~
x . ^F
•^t^- ^-^
M5
> f " x---^/ ! /. L-,**1" '
*'- '. $>j'~~'-~/r~ '" '''?~~~iv
-" '"'7^1 * ; fhjfe^
CIRCLE=AQS_8hrmax;
Figure C-6. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., July 2005.
72
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for August
• = = "T^"""*^ = "= = ^~7^
Sj? '--—,
CIRCLE=AQS_8hrmax;
Figure C-7. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., August 2005.
O3 NME (%) for run 2005ci_tox_05b_alt_bc_v47_N1 b_MP_1 2km for August
\
c x''
i
\
CIRCLE=AQS_8hrmax;
Figure C-8. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., August 2005.
73
-------
^O3INMB (%)forruniSMOSci...tox..O^ajt^b^v^^^^^akm for September
J*'
CIRCLE=AQS_8hrmax;
Figure C-9. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., September 2005.
O3 NME (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for September
- ^^ ,
CIRCLE=AQS_8hrmax;
Figure C-10. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., September 2005.
74
-------
O3 NMB (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for May to September
'" -c-- —f-
CIRCLE=AQS_8hrmax;
Figure C-ll. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., seasonal aggregate 2005.
O3 NME (%) for run 2005ci tox 05b alt be v47 N1b MP 12km for May to September
> / J
*\ rrf**
v-«i .,.' •
— • f- *~' I
,K. >J
_^y^
*J*f«.:,
If
is
^-"L,'.-' £
>;^%v/^
CIRCLE=AQS_8hrmax;
Figure C-12. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Eastern U.S., seasonal aggregate 2005.
75
-------
03 NMB (%) for run 200Sci_tox_05b_alt_bc_v47_N1b_MP_W12km for May
H v-i } &•&
\ r' Y- ^
/ J-i T' ^-r< i /rx*^
\ V-' Hr ^ : I i •:-
X
V t-jfe*
f^ ^
CIRCLE=AQS_8hrmax;
Figure C-13. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., May 2005.
O3 NME (%) for run 2QOSciJoxJ)5bj>\tJ>cj/47JN1bJllPJN12km for May
An Atmo*heric MddeKEValuation (AMET) ProdM
CIRCLE=AQS_8hrmax;
Figure C-14. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., May 2005.
76
-------
03 NMB (%) tor run ^05ciJox_05bma!t_bc_v47_Nlb_MP_W12km for June
: <•
:\J%
i V*
V-*
1 %
!
^
"-'x
An Atmo^Jheric
CIRCLE=AQS_8hrmax;
Figure C-15. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., June 2005.
O3 NME (%) for run 2005ci^tox 05b alt be, v47 N1b MP W12km for June
^^
CIRCLE=AQS_8hrmax;
Figure C-16. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., June 2005.
77
-------
O3 NMB <%) lot tun 2005cLl<"UI5b_alt_bc_v47_N1b_MP_W12km tor July
An Atmosbheric Motfefe&luation (AMET) Prod
CIRCLE=AQS_8hrmax;
Figure C-17. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., July 2005.
O3 NME (%) for run SOOSci tox 05b all be v47 N1b MP W12km for July
An Atmosbheric MoSsKEftlluation (AMET) Prod
CIRCLE=AQS_8hrmax;
Figure C-18. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., July 2005.
78
-------
O3 NMB (%) tor run 2005ci_lox_05b_alt_l>c_v47_N1b_MP_W12Km for August
An Atmosbheric MoSefefeluation (AMET) Prod
CIRCLE=AQS_8hrmax;
Figure C-19. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., August 2005.
O3 NMi (%) for run 2005ci_tox_05b_alt_bcjf47J41b _MP_W12km lor August
An Atmospheric MoSfiKH&luation (AMET) Produ
CIRCLE=AQS_8hrmax;
Figure C-20. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., August 2005.
79
-------
O3 {%) for run 2005ci tox OSb alt be v47 Nib MP W12km for September
.______
An Atmospheric MoteleNEifeluation (AMET) Prod
units = %
coverage limit = 75%
CIRCLE=AQS_8hrmax;
Figure C-21. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., September 2005.
O3 NME (%) for run 2005ci_tox_05b_alt_bc_v47_N1b_MP_W12km for September
= = = = = =
An Atmosbheric MoafiKEfcluation (AMET) Prod
CIRCLE=AQS_8hrmax;
Figure C-22. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., September 2005.
80
-------
O3 NMB (%) for run 2005ci_tox_05b_alt_bc_v47_N1b_MP_W12km tor May to September
An Atmospheric MoSeKtftlluation (AMET) Prodi
CIRCLE=AQS_8hrmax;
Figure C-23. Normalized Mean Bias (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., seasonal aggregate 2005.
O3 NME (%) tor fun aOQSci tox 05b alt be v47 N1b MP W12km for May to September
An Atmospheric MoapKtftlluation (AMET) Prod
CIRCLE=AQS_8hrmax;
Figure C-24. Normalized Mean Error (%) of eight-hour daily maximum ozone (40 ppb
threshold) by monitor for Western U.S., seasonal aggregate 2005.
81
-------
D. Annual PM2.s Species Evaluation
Table D-l provides annual model performance statistics for PM2.5 and its component species for
the 12-km Eastern domain, 12-km Western domain, and five subregions defined above (Midwest,
Northeast, Southeast, Central, and West U.S.). Spatial plots of the NMB and NME statistics
(units of percent) for individual monitors are also provided as a complement to the tabular
statistical data (Figures D-l - D-28). In the East, annual total PM2.5 mass is under-predicted
when compared at STN and IMPROVE sites in the Southeast and Central U.S. In the West,
annual total PM2.5 mass is under-predicted when evaluated at STN sites and IMPROVE sites,
with better performance at STN network (bias ~ -10). Although not shown here, the mean
observed concentrations of PM2.5 are approximately twice as high at the STN sites (EUS =
~13|ig m'3; WUS = -1 l|ig m'3) as the IMPROVE sites (EUS = ~7|ig m'3; WUS = ~4|ig m'3),
thus illustrating the statistical differences between the urban STN and rural IMPROVE networks.
Sulfate is consistently under-predicted at STN, IMPROVE, and CASTNet sites, with NMB
values ranging from -32% to -5%. Overall, sulfate performance is best in the East at urban STN
sites. Nitrate is over-predicted in the 12-km Eastern domain (NMB in the range of 5% to 66%),
while nitrate is under-predicted in the 12-km Western domain (NMB in the range of -29% to -
47%). Likewise, model performance of total nitrate at CASTNet sites shows an over-prediction
in the East (NMB = 27%) and an under-prediction in the West (NMB = -3%). Ammonium
model performance varies across the STN and CASTNet in the East and West, with a mix of
over and under-predictions in the Eastern domain and also an under-prediction in the West.
Elemental carbon is over-predicted at STN sites in the East and West with a bias of-30% and
error of-60%. Although, EC is under-predicted at IMPROVE sites in the East and West with a
NMB of- -20% and error of-45%. Organic carbon is moderately under-predicted for all
domains in the STN and IMPROVE networks (bias - -35% and error - 60%. Differences in
model predictions between IMPROVE and STN networks could be attributed to both the rural
versus urban characteristics as well as differences in the measurement methodology between the
two networks (e.g. blank correction factors, and filter technology used).
Table D-l. 2005 CMAQ annual PM2.s species model performance statistics.
CMAQ 2005 Annual PM2.5 species
PM25
Total Mass
STN
IMPROVE
12-km EUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-km EUS
12-km WUS
Midwest
Northeast
No. of
Obs.
11797
3440
2318
2977
2960
2523
2826
9321
10411
571
2339
NMB (%)
-0.8
-10.0
8.5
10.8
-13.7
-6.3
-10.9
-6.4
-19.9
1.3
12.1
NME (%)
37.6
45.0
35.2
41.6
34.0
39.8
46.1
41.6
44.6
36.8
47.7
FB (%)
-2.4
-9.5
9.2
9.9
-14.5
-9.8
-10.6
-7.2
-21.9
1.3
8.1
FE (%)
38.7
44.4
33.5
38.8
37.1
44.3
45.0
43.6
48.4
37.2
44.3
82
-------
Sulfate
Nitrate
Total
Nitrate
(NO3 +
HNO3)
STN
IMPROVE
CASTNet
STN
IMPROVE
CASTNet
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
1694
2376
9258
13897
3920
2495
3441
3499
2944
3157
9034
10002
531
2253
1685
2350
8896
3170
1142
615
786
1099
300
1091
12741
3655
2495
3442
3499
1812
31339
9027
9987
531
2248
1685
2350
8881
3170
1142
615
786
1099
300
-17.6
-13.3
-22.9
-12.3
-17.0
-5.2
-7.7
-14.5
-22.5
-15.5
-15.2
-14.4
-12.7
-7.7
-17.8
-23.5
-10.5
-19.3
-24.6
-16.7
-14.2
-21.5
-32.2
-23.7
25.5
-41.7
29.8
37.0
13.1
5.2
-47.3
34.2
-29.7
20.7
66.4
42.5
22.8
-45.7
27.3
-3.3
26.1
40.6
25.7
10.4
37.5
41.7
44.8
33.2
42.3
34.0
32.1
30.9
37.2
45.8
34.5
41.0
32.5
34.3
32.7
36.6
42.3
24.8
33.6
23.4
22.6
24.7
34.3
33.8
70.4
65.3
64.7
73.9
78.2
58.8
65.4
86.9
72.5
69.8
106.9
106.6
72.7
75.6
40.3
34.5
37.4
46.0
41.4
33.9
-16.9
-11.9
-23.5
-9.2
-7.8
0.7
-4.0
-12.2
-19.5
-6.7
-6.2
2.7
-5.0
-0.2
-12.2
-16.1
5.0
-18.5
-16.6
-14.8
-12.3
-23.6
-33.6
-15.8
-8.1
-70.8
17.1
3.2
-27.5
-6.2
-79.1
-31.6
-93.8
-7.3
-1.0
-37.5
-20.5
-102.3
21.9
6.4
26.9
34.4
17.2
7.3
43.1
46.2
48.6
35.9
42.8
34.8
33.9
33.4
41.6
44.0
38.9
44.5
34.6
37.3
36.2
40.7
45.1
27.8
35.2
24.8
24.8
27.7
38.4
35.2
78.1
97.5
63.9
73.6
86.0
71.6
99.9
101.6
124.0
82.1
95.4
105.2
95.0
128.4
38.1
39.9
34.1
42.4
39.2
33.3
83
-------
Ammonium
Elemental
Carbon
Organic
Carbon
STN
CASTNet
STN
IMPROVE
STN
IMPROVE
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
1091
13897
3893
2495
3498
3882
3059
3130
3170
1142
615
786
1099
300
1091
14038
3814
2502
3479
3877
3221
3015
8668
9495
602
2117
1584
2123
8518
12619
3582
2380
o o T3
JJZJ
3802
2259
3060
8662
9495
601
2116
1587
2123
8518
-4.7
6.7
-14.9
14.3
14.7
0.8
-4.3
-20.5
-1.3
-14.5
8.1
6.2
-13.6
-4.2
-20.8
25.9
31.1
18.7
37.7
10.7
48.1
38.7
-25.9
-10.2
-16.8
-4.8
-46.1
-31.3
-9.6
-35.1
-32.1
-37.3
-17.4
-45.7
-38.8
-31.7
-29.9
-24.0
-33.1
-7.7
-37.8
-42.5
-22.3
35.7
42.3
55.3
42.0
44.3
39.1
43.6
59.0
34.6
38.8
34.6
37.2
32.6
35.4
39.1
66.0
77.7
51.7
70.5
59.1
86.3
82.8
49.1
57.2
41.8
48.8
51.4
49.5
58.2
52.6
56.7
51.9
52.7
52.9
53.4
57.6
50.6
57.1
43.6
52.4
46.7
54.2
57.3
6.4
12.6
7.1
22.3
25.0
6.5
-0.2
5.8
0.4
-12.1
12.8
11.3
-13.7
-1.1
-13.9
18.2
19.5
20.1
26.7
8.1
26.5
21.0
-28.3
-17.8
-28.3
-15.5
-51.5
-30.1
-18.2
-32.5
-28.2
-31.0
-13.7
-48.7
-37.3
-27.8
-34.3
-29.4
-39.3
-14.4
-48.0
-46.8
-28.1
40.5
45.4
55.0
42.2
46.3
42.0
50.4
57.2
35.8
40.1
33.3
35.9
36.5
39.6
40.2
54.3
62.5
47.1
54.7
49.4
64.3
65.1
56.0
60.4
50.4
54.1
62.8
56.5
61.5
63.9
61.3
62.5
60.3
66.7
66.9
61.4
59.1
62.8
53.2
53.5
60.2
65.2
62.7
84
-------
PM25NMB (%> for run 200tei_tox_0»_alt_bc_v«7_N1b_MP_iacm for January to December
-
V,£->»7T7
> f
~~! ;«\
C!RCLE=IMPROVE; TRIANGLE=STN;
Figure D-l. Normalized Mean Bias (%) of annual PM2.s by monitor for Eastern U.S., 2005.
PM25JNIMI (%> tof ™n 200fci_tox_05b_alt_bc_v47_N1b_MP_12km for
-------
I -JO/" J7W
CiRCLE=IMPROVE; TRIANGLE=STN;
Figure D-2. Normalized Mean Error (%) of annual PM2.s by monitor for Eastern U.S.,
2005.
85
-------
r^
————- —-_
^ / 1 ^ f X--T ; ' ! X^ V
SQUARE-CASTNet;
Figure D-3. Normalized Mean Bias (%) of annual sulfate by monitor for Eastern U.S.,
2005.
SO4 NME (%) tor run 200ScMox_OSb_att_bc_v47_N1b_MP_12km tot January to December
~ - ^'
Figure D-4. Normalized Mean Error (%) of annual sulfate by monitor for Eastern U.S.,
2005.
86
-------
NO3 NMB (%} lor run 200Sei_to»_OSb_alt_bc_«4r_N1b_MP_12ktn lor January to December
"" "
<4* .
M- X 1 VV x ------ / jr
/ / J / -' ,/ • * x . *»^'
f I (.v^ I _•—** I : ,• *^
~
^^citf
\ jl
%/*
s
<* , if ;
* 90
•: fty
."0
-------
TNO3 NMB (%) JO£njr^2005ci_tox_05b_alt_bc_v47_N1b_MP_12km for Januaryto_December
rs (AMET)
CIRCLE=CASTNet;
Figure D-7. Normalized Mean Bias (%) of annual total nitrate by monitor for Eastern U.S.,
2005.
TNO3 NME (%) for run 2005cMoxJ)5b^altJ)C^v47^N1b,MPJ2knn for January to December
CIRCLE=CASTNet;
Figure D-8. Normalized Mean Error (%) of annual total nitrate by monitor for Eastern
U.S., 2005.
88
-------
NH4 HUB (%)for run 2M5d_tox_OSb_aK_bc_v47_N1b_WP_12kin for Januaff to Decemter
CIRCLE=STN; TRIANeLE=CASTN«;
Figure D-9. Normalized Mean Bias (%) of annual ammonium by monitor for Eastern U.S.,
2005.
NH* {%) for run 2005ci tax 05b alt be ¥47 Nib HP 12km for January to December
- """ "' '
CiRCLE^STN; TRIAN6LE=CASTNet;
Figure D-10. Normalized Mean Error (%) of annual ammonium by monitor for Eastern
U.S., 2005.
89
-------
Jox=05^1»mjioR«ci«Jot»W010yo^00512^
~tJ-> .,*j$$>~"~"T-
v '- •"*«• 2*f^er V
r
N
\
1 1
; r
v\
\ i
Figure D-ll. Normalized Mean Bias (%) of annual elemental carbon by monitor for
Eastern U.S., 2005.
EC NME (%) lor ryn 2005ci Jox_OSb_E12km_noPMcut for 20050101 lo
TRIANGLE=STN,
Figure D-12. Normalized Mean Error (%) of annual elemental carbon by monitor for
Eastern U.S., 2005.
90
-------
OC NUB (%J lor run 200Sci_tox_05b_E12km_noPMcut tor 20050101 to 20051231
90
80
60
50
JO
30
20
10
-10
-20
30
-40
-50
-60
-80
-90
100
«-100
An AMET FratfeGI
TRIANGLi=STN;
Figure D-13. Normalized Mean Bias (%) of annual organic carbon by monitor for Eastern
U.S., 2005.
OC NME (%) (or run 2005cl_to«_(H6_E12lHeut to to
Figure D-14. Normalized Mean Error (%) of annual organic carbon by monitor for
Eastern U.S., 2005.
91
-------
PM25 NMB (%j tor run 2005eLto»_05b_»lt_bc_«47_N1b_MP_W12kin tor January to December
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure D-15. Normalized Mean Bias (%) of annual PM2.s by monitor for Western U.S.,
2005.
PM25 NME (%) tor run 2005clJox_05b_alt_bc_i>47_Nll>_MP_Wl2km (or January to December
An AtmospBeric ModSJ H/4fuation (AMET) Produ
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure D-16. Normalized Mean Error (%) of annual PMi.5 by monitor for Western U.S.,
2005.
92
-------
SO4 NMB (%) tor run 2005cl_tox_05b_alt_bc_v47_N1b_MP_W12km lor January to December
\ K^-
An Atmospheric Mod&J Equation (AMET) Produ
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure D-17. Normalized Mean Bias (%) of annual sulfate by monitor for Western U.S.,
2005.
SO4 NME (%) tor run ZOOSci Jox JSb^aH^bc_v47^Nlb^MP_Wl2km for January to December
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure D-18. Normalized Mean Error (%) of annual sulfate by monitor for Western U.S.,
2005.
93
-------
NO3 NMB (%) tor run 2005cLtox_05b_aU_bc_v47_N1b_MP_W12Km tor January Is December
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure D-19. Normalized Mean Bias (%) of annual nitrate by monitor for Western U.S.,
2005.
NO3 NME (%) for run aoo5cl_tox_OSt>_alt_bc_v47_Nl b_MP_W12km tor January to December
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure D-20. Normalized Mean Error (%) of annual nitrate by monitor for Western U.S.,
2005.
94
-------
TKO3 MMB (%) tor run 2005cl tox OSb all be ¥47 NH>_JH»_W12lcnitor January to December
An Atmospheric Mooentvaluation (AMET) Produ
CIRCLE=CASTNet;
Figure D-21. Normalized Mean Bias (%) of annual total nitrate by monitor for Western
U.S., 2005.
)^
\ K;
u-
An Atmospheric MoSe%*uation (AMET) Produ
CIRCLE=CASTNet;
Figure D-22. Normalized Mean Error (%) of annual total nitrate by monitor for Western
U.S., 2005.
95
-------
NH4 NMP (%) tor run 2005cl_tox_05b_all_bc_v47_N1b_MP_W12km for January io December
CIRCLE=STN; TRIANGLE=CASTNet;
Figure D-23. Normalized Mean Bias (%) of annual ammonium by monitor for Western
U.S., 2005.
NM4 NME (%) for run 2005cl_tox_05b_alt_tac_v47_N1b_MP_W12kin for January to December
CIRCLE=STN; TRIANGLE=CASTNet;
Figure D-24. Normalized Mean Error (%) of annual ammonium by monitor for Western
U.S., 2005.
96
-------
EC NMB (%) tor run 2005ei_tox_05b_MP_W12km_noPMcut for 20050101 to 20051231
V x
?
\
\ ,..
An Product
CIRCLE=iMPROVE, TfiiANGLE=STN;
Figure D-25. Normalized Mean Bias (%) of annual elemental carbon by monitor for
Western U.S., 2005.
EC NME (%) for run 2005ci_tox_05b_MP_W12km_noPMcut for 20050101 to 20051231
CIRCL£=IMPROVE; TRIANGLE=STN;
Figure D-26. Normalized Mean Error (%) of annual elemental carbon by monitor for
Western U.S., 2005.
97
-------
OC NMB (%) for run 200ScUox_05b_El2kni_naPMciit for 20050101 to
30
a
10
-10
-20
30
-10
-50
-60
-70
-80
,- -90
• -100
* .: -103
CIRCLE=iypRD¥E; TRIANGLE=STN;
Figure D-27. Normalized Mean Bias (%) of annual organic carbon by monitor for Western
U.S., 2005.
OC NME (%) for run 2005ci_tox_05b_MP_W12km_noPMcut for 20050101 to 20051231
v v- <
\ ^v>r^^-^_
CIRCLE-IMPROVE, TRIANGLE-STN;
AjiAMETProdycl
Figure D-28. Normalized Mean Error (%) of annual organic carbon by monitor for
Western U.S., 2005.
98
-------
E. Seasonal PM2.5 Total Mass Performance
Seasonal model performance statistics for PM2.5 total mass are shown in Table E-l. Spatial plots
of the NMB and NME statistics (units of percent) for individual monitors are also provided as a
complement to the tabular statistical data (Figures E-l - E-16). Total PM2.5 mass is generally
over-predicted in the cooler seasons (winter and fall) in the 12-km Eastern domain for both STN
and IMPROVE networks. In the fall season, PM2.5 is over-predicted for Eastern STN sites with
NMB values ranging from 0% to 15% whereas PM25 is under-predicted at Eastern IMPROVE
sites. In the winter season, PM2.5 is over-predicted for Eastern STN and IMPROVE networks
with NMB values ranging from 3% to 39% and FB values ranging from 60% to 72%. However,
in the 12-km Western domain, PM2.5 is under-predicted in the winter (NMB in the range of-2%
to -11%) and the fall (NMB in the range of -7% to -26%). Note that for comparison of West
versus East STN sites, the total number of Western sites is usually less than a third of the Eastern
sites. In the spring, CMAQ generally over-predicts PM2.s in the East and West at urban STN and
rural IMPROVE sites. In the summer season, PM2.5 is under-predicted in the East and West for
STN and IMPROVE (NMB = ~ 30% and NME = -40%).
Table E-l. CMAQ 2005 seasonal model performance statistics for PM2.s total mass.
CMAQ 2005 PM2.5 total mass
Winter
Spring
STN
IMPROVE
STN
IMPROVE
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
No. of Obs.
2861
895
716
542
762
635
739
2213
2493
573
143
406
546
2217
3159
964
795
612
798
752
773
2385
NMB (%)
8.8
-6.2
22.8
7.3
-2.6
10.0
-10.9
16.3
-1.9
39.4
7.1
3.6
5.6
-8.8
10.5
0.5
31.1
31.2
-7.3
-6.6
3.2
3.4
NME (%)
37.8
54.7
40.9
32.3
35.7
43.0
55.0
44.7
49.3
54.3
34.3
38.6
43.1
49.3
41.2
43.2
51.4
46.2
33.1
37.1
45.5
42.5
FB (%)
5.5
-3.2
18.1
8.5
-5.0
4.5
-8.7
13.4
-2.2
28.7
4.7
-0.9
5.5
-4.3
6.5
-1.1
24.6
25.1
-7.8
-9.4
-0.2
0.2
FE (%)
36.9
53.0
35.7
31.1
36.6
43.0
54.6
43.3
49.0
44.0
34.3
41.0
44.6
49.4
39.1
41.2
43.3
39.5
34.2
40.8
41.9
41.5
99
-------
Summer
Fall
STN
IMPROVE
STN
IMPROVE
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
2692
630
153
429
601
2389
2950
935
754
558
722
701
758
2277
2433
580
156
421
568
2155
2802
962
755
606
785
435
812
2256
2600
556
119
438
549
2304
-24.3
33.9
20.6
-8.6
-14.6
-27.1
-29.9
-20.9
-27.7
-21.1
-39.1
-29.5
-18.2
-38.1
-27.7
-32.6
-31.1
-46.6
-40.7
-26.0
-0.9
-6.9
15.1
0.1
-17.0
6.9
-9.1
-7.4
-21.6
14.55
-5.3
-23.6
-8.01
-26.5
43.0
55.7
44.4
34.5
39.4
44.4
37.8
35.9
36.1
30.8
42.7
40.4
36.2
43.7
43.5
41.2
63.6
48.4
45.9
43.7
36.8
46.2
44.8
32.2
32.5
42.4
47.2
41.8
46.9
48.2
37.1
36.5
44.3
46.0
-26.6
20.2
16.7
-7.0
-12.5
-28.2
-36.2
-21.0
-30.5
-21.6
-49.9
-38.8
-19.0
-45.2
-31.1
-41.3
-33.8
-63.1
-49.1
-29.7
1.3
-4.9
10.9
4.4
-16.0
11.1
-7.8
-8.2
-27.2
7.1
-2.7
-22.8
-5.4
-30.3
47.0
45.5
41.5
36.3
43.5
47.5
46.5
40.3
41.6
33.8
54.8
53.1
40.3
53.4
48.5
50.5
41.9
67.7
57.7
48.0
37.0
45.3
39.2
31.6
36.0
42.9
45.8
43.3
51.1
43.1
39.2
42.4
47.4
51.1
100
-------
PM25 NMB (%) tor run 2005ci_tox.05b_alt_bc_v47_N1b.MP.12km for December to Fabruary
-v^
ftp
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-l. Normalized Mean Bias (%) of PM2.s by monitor for 12-km Eastern U.S.
domain, Winter 2005.
PM25 NME (%? tor run 2005ci_tox_05b_alt_bc_v47-jM1b JP_12km tor Oeeember to February
K _.--/?
-~- w,=
v ,'^TfS?
% 1* 1 >4
»^:..'2
•*?}
Jr /
i _ ji^—<{ ^
1 ! \ - \ \ ^
Ai AiifDSpMens Mode- t-.-a\-y.cr -^MglpF'^j^
CIRCLE-IMPROVE; TRIANGLE-STN;
Figure E-2. Normalized Mean Error (%) of PM2.s by monitor for 12-km Eastern U.S.
domain, Winter 2005.
101
-------
PM25 NMB (%Hp^^
"cf
CIRCLE=lMPROVE; TRIANGLE=STKI;
Figure E-3. Normalized Mean Bias (%) of PMi.s by monitor for 12-km Eastern U.S.
domain, Spring 2005.
PM25 NME (%) for run 20Q5ci tox 05b alt be v47 N1b MP 12km for March to May
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-4. Normalized Mean Error (%) of PM2.s by monitor for 12-km Eastern U.S.
domain, Spring 2005.
102
-------
PM25NMB (%2JaMw_2005ci tox_05b_alt_bc_v47_N1 b_MP
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-5. Normalized Mean Bias (%) of PMi.s by monitor for 12-km Eastern U.S.
domain, Summer 2005.
PM25 NME (%) for run _200_5cijox_05b,.alt.>e_v47_^^^
C!RCLE=!MPROVE; TR!ANGLE=STN;
Figure E-6. Normalized Mean Error (%) of PM2.s by monitor for 12-km Eastern U.S.
domain., Summer 2005.
103
-------
PM25 NMB {%) for run 2005cLtox_05b_alt_bc_tf47_N1b_MP_12km for September to November
-~ —
CIRCLE=1MPROVE; TRIANGLE=STN;
Figure E-7. Normalized Mean Bias (%) of PM2.s by monitor for 12-km Eastern U.S.
domain, Fall 2005.
PM25 NME (%) for run 2005ci_tox_05b_alt_bc_v47_N1b_MP_12km for September to Ncwember
CIRCLE=iMPROVE; TRIANGLE=STN;
Figure E-8. Normalized Mean Error (%) of PM2.s by monitor for 12-km Eastern U.S.
domain, Fall 2005.
104
-------
PM25 NMB (%) tor run 2005cl.to».05b.alt_bc.v47.N1b.MP.W12Kni tor December to February
An Atmospheric ModfclES^^ation (AMET) Product
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-9. Normalized Mean Bias (%) of PM2.s by monitor for 12-km Western U.S.
domain, Winter 2005.
PM25 NME (%) lor run 2005CI tox OSb att be ¥47 N1b MP W12km for December» February
An Atmospheric Model w^ation (AMET) Producl
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-10. Normalized Mean Error (%) of PM2.5 by monitor for 12-km Western U.S.
domain, Winter 2005.
105
-------
PM25 NMB (%) lor run 2005cl_K>x_05ft_aH_6c_v47_Nlb_MP_Wiakm far March lo May
An Atmospneric Model Equation (AMET) Produ
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-ll. Normalized Mean Bias (%) of PM2.s by monitor for 12-km Western U.S.
domain, Spring 2005.
PM25 NME (%) tor run 2005cl_tQ»_0»b »H_>»_tf47_N1b_MP_W12lun tor March to May
An Atmospheric Model Eyjifb(ation (AMET) Produ
J J
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-12. Normalized Mean Error (%) of PM2.s by monitor for 12-km Western U.S.
domain, Spring 2005.
106
-------
PM25 NMB (%) tor run 2005cl_lox_05b_alt_l>c_v47_NH>_MP_W12Km lor June to Augug)
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-13. Normalized Mean Bias (%) of PM2.5 by monitor for 12-km Western U.S.
domain, Summer 2005.
PM25 NME (%) tor run 2005ci Jox_05b_alt_bc_v47_N1b_MP_W12km lor June to Augysl
CIRCLE=IMPROVE; TRIANGLE=STN;
Figure E-14. Normalized Mean Error (%) of PM2.s by monitor for 12-km Western U.S.
domain, Summer 2005.
107
-------
PM25 NMB (%) tor run 2005cl_lo»_05b.alt_bc_v47_N1b_MP_W12km tof September to November
CIRCLE=IMPROVE; TRlANGLE=STN;
Figure E-15. Normalized Mean Bias (%) of PM2.s by monitor for 12-km Western U.S.
domain, Fall 2005.
PM25 HUE (%) tor ran 2005c! toic 05b aa be v47 Nib MP W12Kmtor September to November
CIRCLE^IMPROVE; TRIANGLE-STN;
Figure E-16. Normalized Mean Error (%) of PM2.s by monitor for 12-km Western U.S.
domain, Fall 2005.
108
-------
F. Seasonal Sulfate Performance
As seen in Table F-l, CMAQ generally under-predicts sulfate in the 12-km Eastern and Western
domains throughout the entire year. Spatial plots of the NMB and NME statistics (units of
percent) for individual monitors are also provided in Figures F-l - F-16. Sulfate predictions
during the winter season show NMB values ranging from -7% to -35% and in the East and with
NMB values range from -1% to -10% in the West. In the fall season, sulfate predictions show
NMB values ranging from -14% to -29%, across STN, IMPROVE, and CASTNet networks in
the East and West. In the spring, sulfate predictions for the most part are under-predicted in the
East and West, with NMB values ranging from -2% to -31%. Sulfate predictions during the
summer season are moderately under-predicted in the East and West across the available
monitoring data (NMB values rage from -17% to -38%.
Table F-l. CMAQ 2005 seasonal model performance statistics for sulfate.
CMAQ 2005 Sulfate
Winter
Spring
STN
IMPROVE
CASTNet
STN
IMPROVE
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km BUS
No. ofObs.
3385
1033
828
598
963
766
830
2018
2370
502
129
386
511
2120
760
267
193
142
264
72
255
3626
1085
894
637
988
875
867
2378
NMB (%)
-15.1
-7.3
-17.3
-7.4
-13.1
-22.0
-10.3
-16.2
-1.3
-16.3
-7.1
-13.7
-27.1
8.5
-20.4
-5.4
-19.7
-16.9
-21.2
-35.7
-1.6
-7.3
-12.3
4.2
11.3
-13.7
-23.0
-6.2
-10.3
NME (%)
38.0
55.4
35.9
39.4
36.7
39.3
58.1
37.1
50.6
31.8
35.6
35.7
42.6
51.9
25.4
33.2
24.4
23.4
3-23.9
36.6
34.1
32.9
35.0
34.8
39.2
28.0
33.9
37.5
34.0
FB (%)
-16.7
-2.6
-21.5
-12.7
-12.4
-21.2
-3.5
-6.5
22.1
-18.9
-12.0
-9.5
-21.6
26.6
-20.2
9.4
-24.3
-20.2
-21.5
-40.2
11.3
-6.2
-3.4
2.9
10.8
-13.7
-19.2
0.3
-4.6
FE (%)
40.7
53.7
37.7
42.2
37.9
44.1
54.9
42.8
53.1
34.7
36.8
37.3
47.2
53.4
30.2
35.5
28.7
29.1
25.3
41.7
35.5
33.6
35.8
34.0
36.1
30.4
35.9
36.7
35.4
109
-------
Summer
Fall
CASTNet
STN
IMPROVE
CASTNet
STN
IMPROVE
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
2642
630
147
436
605
2352
832
287
206
155
292
78
274
3512
1075
874
621
941
847
853
2269
2340
590
158
427
572
2066
792
295
192
161
270
75
282
3349
1095
902
639
990
571
900
2147
2427
531
-9.2
4.3
-2.3
-14.7
-25.8
-5.1
-10.9
-18.1
0.4
-4.9
-15.8
-31.2
-16.9
-28.1
-35.3
-22.9
-22.8
-31.3
-37.9
-35.4
-32.8
-32.2
-26.3
-31.3
-37.8
-37.7
-30.1
-22.4
-38.1
-17.9
-18.6
-24.6
-36.2
-38.2
-21.0
-17.2
-17.0
-21.4
-22.2
-21.0
-14.6
-23.7
-19.3
-17.0
34.8
37.2
36.2
29.7
35.4
35.3
23.6
26.3
24.4
22.2
21.9
34.3
25.9
36.2
43.4
32.1
33.1
37.5
44.4
45.8
40.3
43.8
37.2
37.0
42.3
43.9
44.2
25.7
41.2
21.9
23.0
27.0
38.9
41.6
32.3
44.2
35.8
32.0
30.9
35.0
48.8
35.1
41.4
35.8
2.2
4.2
2.2
-11.9
-18.9
4.6
-9.6
-14.8
1.1
-2.9
-17.1
-27.9
-14.2
-27.3
-28.8
-16.5
-12.9
-34.2
-43.8
-27.3
-28.4
-23.9
-15.4
-20.5
-42.3
-34.8
-22.3
-25.3
-40.2
-14.5
-16.8
-31.9
-42.5
-40.1
-12.6
-7.9
-0.2
-13.3
-17.7
-12.3
-6.9
-11.7
-4.0
-0.2
36.8
38.3
36.0
31.2
36.5
37.1
24.6
26.3
26.2
21.6
23.8
34.1
26.1
42.7
45.1
35.3
34.0
44.8
56.4
45.6
46.7
47.2
42.6
40.5
52.4
50.9
47.2
31.7
45.8
24.1
25.1
35.0
48.3
45.9
33.6
44.5
38.3
32.0
31.9
36.7
46.4
39.1
44.0
38.3
110
-------
CASTNet
Midwest
Southeast
Central
West
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
97
436
548
2135
786
293
195
157
273
75
280
-29.3
-26.6
-26.0
-14.9
-21.8
-22.4
-17.2
-23.0
-22.9
-25.2
-21.5
33.8
34.2
35.0
43.3
24.1
30.6
20.8
24.9
24.5
26.7
31.1
-27.1
-20.8
-17.7
-1.5
-19.4
-18.2
-12.5
-19.6
-24.3
-24.5
-17.6
36.5
38.3
40.2
44.6
24.9
32.9
20.3
23.8
26.8
29.9
33.0
111
-------
S04 NUB {%) for run 2005cl_tox_05b_alt_bc_v47_N1b_MP_12km for December to February
- -
CIRCI_E=NADP_dep;
Figure F-l. Normalized Mean Bias (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Winter 2005.
SO4 MME <%) lor run 2005ei_tox_05b_all_be_v47_N1b_MP_12km lor DBOoraber to February
CIRCLE=NADP_dep;
Figure F-2. Normalized Mean Error (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Winter 2005.
112
-------
SO4 NMB (%) for run 2Q05ei tox 05b alt be v*7 N1b MP 12km for March to May
CIRCLE-IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-3. Normalized Mean Bias (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
SO4 NME (%) for run ZOOSei ton OSb alt bo «47 N1b MP 12km tor March to May
- . = == = = =
/ <
j \
\\
CIRCLE-IMPROVE; TRIANGLE-STN; SQUARE-CASTNet;
Figure F-4. Normalized Mean Error (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
113
-------
SO4 NMBJ%) lor run 2005o_tox_Mb_all_bc_v47_N1b_MP_12km lor June to August
"
C1RCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-5. Normalized Mean Bias (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Summer 2005.
SO4 NME {%) tor run 200Sei_tO)i.OSb_a_H_t>c_v47.N1b.lW_12km tor Juneto August
C1RCLE=1MPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-6. Normalized Mean Error (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Summer 2005.
114
-------
SO4 MMB (%> for run 2005ci_lox_05b_8lt_bc_v47_N1b_MP_12kro for September to November
~~ '
CIRCLE=IMPROVE, TRIANGLE=STN; SQUARE=CASTN8t,
Figure F-7. Normalized Mean Bias (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Fall 2005.
SO4 NME {%) tor run 2B05ci_toJi_05b_aIt_bc_v47_N1b_MP_12lim tor Scpternbar to November
C1RCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-8. Normalized Mean Error (%) of sulfate by monitor for 12-km Eastern U.S.
domain, Fall 2005.
115
-------
5O4 NMB j%) tor run 2005cl_lox_05b_aH_bc_v47_N1b_MP_W12Km for December Is February
An Atmospheric ModMation (AMET) Produi
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-9. Normalized Mean Bias (%) of sulfate by monitor for 12-km Western U.S.
domain, Winter 2005.
SO4 NME (%) tor run 2005cl tex_05h_alt_bc_v47_N11»_MP_W12lan tof Peeenibef to February
An Atmospheric ModH^^ation (AMET) Produ
\ \^
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-10. Normalized Mean Error (%) of sulfate by monitor for 12-km Western U.S.
domain, Winter 2005.
116
-------
SO4 NMB (%) tor run 2005cl_tox_05b_aU_>>c_v47_N1b.MP_W12Km tor March to May
An Atmospheric Model EteBuation (AMET) Produ
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-ll. Normalized Mean Bias (%) of sulfate by monitor for 12-km Western U.S.
domain, Spring 2005.
SO4 NME (%) for run 2005cl_lo»_05b_aH_be_v47_N1b_MP_Wia(in lof March to May
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-12. Normalized Mean Error (%) of sulfate by monitor for 12-km Western U.S.
domain, Spring 2005.
117
-------
SO4 NMB (%)tor run 2005eUox_05b_alt_bc_v47_N1b_MP_W12kmlof June lo August
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-13. Normalized Mean Bias (%) of sulfate by monitor for 12-km Western U.S.
domain, Summer 2005.
SO4 NME (%} for run 2005cl_tox_OSb_aH_bc_v47_N1b_MP_W12km for June to ftugusl
CIRCLE=IMPROVE; TRIANGLE=STN; SQUARE=CASTNet;
Figure F-14. Normalized Mean Error (%) of sulfate by monitor for 12-km Western U.S.
domain, Summer 2005.
118
-------
Figure F-15. Normalized Mean Bias (%) of sulfate by monitor for 12-km Western U.S.
domain, Fall 2005.
Figure F-16. Normalized Mean Error (%) of sulfate by monitor for 12-km Western U.S.
domain, Fall 2005.
119
-------
G. Seasonal Nitrate Performance
Table G-l provides the seasonal model performance statistics for nitrate and total nitrate for the
12-km Eastern and Western domains. Spatial plots of the NMB and NME statistics (units of
percent) for individual monitors are also provided as a complement to the tabular statistical data
(Figures G-l - G-32). Overall, nitrate and total nitrate performance is over-predicted in the BUS
and under-predicted in the WUS for all of the seasonal assessments except in the winter and fall
season, where total nitrate is under-predicted in the EUS and WUS and in the spring where
nitrate is over-predicted in the EUS. Likewise, in the East, nitrate and total nitrate are
moderately over-predicted during the spring and summer seasons (NMB values ranging from
0.4% to 70%). In the East and West, nitrate and total nitrate are moderately under-predicted
during the winter season when nitrate is most abundant (NMB values ranging from -2% to -46%).
Table G-l. CMAQ 2005 seasonal model performance statistics for nitrate.
CMAQ 2005 Nitrate
Nitrate
(Winter)
Total
Nitrate
(Winter)
Nitrate
(Spring)
STN
IMPROVE
CASTNet
STN
12-km EUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km EUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km EUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-km EUS
12-km WUS
Northeast
Midwest
Southeast
Central
No. of Obs.
3099
973
829
598
963
479
831
2018
2368
502
129
386
511
2118
760
267
193
142
264
72
255
3254
987
894
637
988
503
NMB (%)
-10.9
-42.8
9.1
-18.7
-14.7
-16.3
-46.9
-2.0
-33.6
40.0
-29.5
3.4
-12.6
-46.3
10.5
13.1
24.5
-6.0
17.7
11.4
14.4
38.8
-33.5
48.8
47.2
37.2
11.6
NME (%)
45.7
61.8
44.7
38.0
57.2
46.8
64.1
59.2
62.1
71.7
43.2
75.5
50.9
74.2
27.9
40.4
30.9
19.9
31.4
30.3
47.4
72.6
55.5
81.3
69.3
86.8
52.3
FB (%)
-16.1
-53.1
9.1
-15.9
-34.6
-10.8
-60.5
-25.0
-78.4
29.3
-35.7
-36.2
-20.1
-84.5
16.9
26.4
30.0
0.3
14.5
13.3
27.2
16.8
-56.9
39.0
36.4
-3.0
7.9
FE (%)
57.8
82.1
47.5
44.2
72.8
56.7
86.1
81.6
108.5
72.2
64.8
82.1
71.9
113.9
31.9
49.7
34.2
20.4
32.1
31.1
51.4
70.4
81.4
70.5
60.4
81.2
60.3
120
-------
Total
Nitrate
(Spring)
Nitrate
(Summer)
Total
Nitrate
(Summer)
Nitrate
(Fall)
IMPROVE
CASTNet
STN
IMPROVE
CASTNet
STN
IMPROVE
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
859
2378
2636
630
147
436
605
2346
832
287
206
155
292
78
274
3150
992
874
621
941
485
846
2269
2339
590
158
427
572
2065
792
295
192
161
270
75
282
3238
1048
902
639
990
460
896
2140
-38.4
41.5
-26.8
68.9
42.3
54.7
21.1
-39.6
12.0
-12.2
36.9
27.9
16.5
-1.4
-11.8
0.4
-68.2
-8.6
30.3
-28.4
11.1
-70.6
-3.6
-65.9
7.1
16.2
-4.9
0.1
-72.2
27.0
-13.3
40.3
40.6
16.8
-6.3
-14.5
72.2
-42.0
82.7
63.1
76.6
78.8
-48.9
102.3
56.8
84.1
67.0
107.3
74.5
97.9
62.6
76.5
35.2
31.1
40.0
33.5
34.5
31.0
31.9
79.9
74.6
78.9
79.8
75.3
88.1
74.1
94.2
82.3
102.3
86.7
96.3
92.0
82.2
41.9
30.9
49.6
45.1
39.6
26.5
31.5
104.6
72.4
109.2
88.4
126.2
107.8
69.1
144.2
-63.8
-6.8
-81.2
29.9
11.1
-9.6
-14.0
-88.7
15.7
-3.6
31.6
24.3
9.7
-1.1
-3.0
-51.1
-120.1
-49.3
-9.3
-73.0
-41.5
-126.0
-78.4
-135.3
-55.8
-34.0
-68.8
-64.2
-142.3
14.4
-13.0
26.3
33.1
8.2
-12.0
-13.6
4.1
-52.3
2.3
28.5
-12.3
16.1
-63.2
-12.7
84.3
91.3
112.3
91.3
79.3
89.7
86.3
116.9
32.9
32.3
36.9
30.0
33.7
31.3
32.6
92.7
127.1
91.2
74.3
100.4
89.7
130.0
115.2
144.1
104.7
91.2
112.3
106.9
148.6
37.6
34.1
43.3
37.3
39.0
29.4
34.6
83.0
91.6
81.3
69.7
94.1
81.5
92.5
100.2
121
-------
Total
Nitrate
(Fall)
CASTNet
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
2421
526
97
436
548
2129
786
293
195
157
273
75
280
-5.4
112.6
85.9
98.4
119.5
-38.1
60.9
6.0
73.9
65.6
57.6
40.7
2.2
98.3
147.0
123.8
161.4
150.5
83.3
66.6
38.2
76.7
66.2
67.6
51.1
37.0
-65.0
-4.3
16.8
-25.5
6.0
-75.4
40.8
17.6
49.7
47.0
36.9
29.7
16.9
112.1
95.0
92.8
107.4
98.9
114.1
50.1
44.1
55.6
0.5
52.3
41.3
44.4
122
-------
Figure G-l. Normalized Mean Bias (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure G-2. Normalized Mean Error (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Winter 2005.
123
-------
Figure G-3. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure G-4. Normalized Mean Error (%) of total nitrate by monitor for 12-km Eastern
U.S. domain, Winter 2005.
124
-------
Figure G-5. Normalized Mean Bias (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
Figure G-6. Normalized Mean Error (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
125
-------
Figure G-7. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
Figure G-8 Normalized Mean Error (%) of total nitrate by monitor for 12-km Eastern U.S.
domain, Spring 2005.
126
-------
Figure G-9. Normalized Mean Bias (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Summer 2005.
Figure G-10. Normalized Mean Error (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Summer 2005.
127
-------
Figure G-ll. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Eastern U.S.
domain, Summer 2005.
Figure G-12. Normalized Mean Error (%) of total nitrate by monitor for 12-km Eastern
U.S. domain, Summer 2005.
128
-------
Figure G-13. Normalized Mean Bias (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Fall 2005.
Figure G-14. Normalized Mean Error (%) of nitrate by monitor for 12-km Eastern U.S.
domain, Fall 2005.
129
-------
Figure G-15. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Eastern U.S.
domain, Fall 2005.
Figure G-16. Normalized Mean Error (%) of total nitrate by monitor for 12-km Eastern
U.S. domain, Fall 2005.
130
-------
Figure G-17. Normalized Mean Bias (%) of nitrate by monitor for 12-km Western U.S.
domain, Winter 2005.
Figure G-18. Normalized Mean Error (%) of nitrate by monitor for 12-km Western U.S.
domain, Winter 2005.
131
-------
Figure G-19. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Western
U.S. domain, Winter 2005.
Figure G-20. Normalized Mean Error (%) of total nitrate by monitor for 12-km Western
U.S. domain, Winter 2005.
132
-------
Figure G-21. Normalized Mean Bias (%) of nitrate by monitor for 12-km Western U.S.
domain, Spring 2005.
Figure G-22. Normalized Mean Error (%) of nitrate by monitor for 12-km Western U.S.
domain, Spring 2005.
133
-------
Figure G-23. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Western
U.S. domain, Spring 2005.
Figure G-24. Normalized Mean Error (%) of total nitrate by monitor for 12-km Western
U.S. domain, Spring 2005.
134
-------
Figure G-25. Normalized Mean Bias (%) of nitrate by monitor for 12-km Western U.S.
domain, Summer 2005.
Figure G-26. Normalized Mean Error (%) of nitrate by monitor for 12-km Western U.S.
domain, Summer 2005.
135
-------
Figure G-27. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Western
U.S. domain, Summer 2005.
Figure G-28. Normalized Mean Error (%) of total nitrate by monitor for 12-km Western
U.S. domain, Summer 2005.
136
-------
Figure G-29. Normalized Mean Bias (%) of nitrate by monitor for 12-km Western U.S.
domain, Fall 2005.
Figure G-30. Normalized Mean Error (%) of nitrate by monitor for 12-km Western U.S.
domain, Fall 2005.
137
-------
Figure G-31. Normalized Mean Bias (%) of total nitrate by monitor for 12-km Western
U.S. domain, Fall 2005.
Figure G-32. Normalized Mean Error (%) of total nitrate by monitor for 12-km Western
U.S. domain, Fall 2005.
138
-------
H. Seasonal Ammonium Performance
Table H-l lists the performance statistics for ammonium PM at the STN and CASTNet sites.
Spatial plots of the NMB and NME statistics (units of percent) for individual monitors are also
provided in Figures H-l - H-16. In the winter, ammonium performance at STN and CASTNet
networks shows an under-prediction in the EUS and WUS (NMB values range from -1% to -
31%), except in the Northeast (NMB values range from 5% to 20%). Likewise, ammonium
performance for the summer season shows an under-prediction in the East and West. However,
in the spring, model predictions in the East are over-predicted, whereas ammonia predictions are
under-predicted in the West. Ammonium predictions in the summer are moderately under-
predicted for the East and West in both the rural and urban sites (NMB values ranging from -6%
to - 37%).
Table H-l. CMAQ 2002 seasonal model performance statistics for ammonium.
CMAQ 2002 Ammonium
Winter
Spring
Summer
STN
CASTNet
STN
CASTNet
STN
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
No. ofObs.
3385
1032
828
598
963
766
829
760
267
193
142
264
72
255
3626
1077
894
637
988
875
859
832
287
206
155
292
78
274
3512
NMB (%)
-5.3
-24.0
4.7
-9.2
-6.7
-8.9
-31.0
-1.4
-6.8
20.5
-12.7
-7.4
-5.8
-12.1
18.8
-2.0
34.7
39.0
5.9
-4.3
-2.5
24.1
-5.3
42.6
42.2
5.3
3.7
-10.4
-18.2
NME (%)
37.5
58.9
34.0
32.5
41.5
41.9
61.1
29.8
37.0
37.3
24.7
27.7
35.4
42.1
46.4
47.4
54.9
55.0
37.1
39.3
52.2
39.8
33.1
48.3
49.6
29.1
34.1
32.7
38.5
FB (%)
-1.8
-8.0
8.7
-3.7
-4.7
-5.5
-14.5
2.7
1.2
22.9
-7.8
-7.3
-4.9
0.7
18.5
17.5
38.0
35.7
7.1
0.5
20.4
18.8
-3.3
33.9
35.7
4.7
4.9
-4.7
-9.0
FE (%)
41.6
63.4
34.0
33.6
44.3
50.2
66.0
31.9
40.7
34.8
25.1
29.3
42.5
41.4
44.4
48.7
51.7
48.6
37.8
42.6
51.4
34.3
32.0
38.6
40.3
29.4
32.9
32.1
48.0
139
-------
Fall
CASTNet
STN
CASTNet
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
1071
874
621
941
847
849
792
295
192
161
270
75
282
3349
1081
902
639
990
571
886
786
293
195
157
273
75
280
-31.6
-17.5
-8.5
-21.4
-27.7
-34.7
-22.6
-32.9
-23.4
-6.8
-32.8
-20.7
-36.9
4.4
-18.3
15.5
1.9
-2.8
6.9
-24.9
3.0
-3.8
7.4
18.7
-11.3
10.9
-14.5
49.6
36.9
35.8
37.3
45.7
54.6
31.3
41.6
29.7
25.2
35.6
31.4
43.5
43.2
62.5
49.9
37.6
40.1
48.5
64.7
38.7
43.4
38.8
44.3
35.6
42.0
37.9
-8.9
-1.4
9.1
-14.3
-27.8
-7.2
-28.0
-37.3
-25.9
-3.6
-44.8
-24.5
-39.1
18.6
8.6
29.5
17.8
9.7
18.6
5.5
7.2
-7.3
12.3
25.5
-9.0
19.5
-10.8
51.7
45.2
40.7
44.6
60.3
54.2
38.5
47.8
34.5
27.0
48.1
38.1
49.0
46.8
59.8
51.3
41.1
43.2
51.0
61.2
38.5
39.9
35.5
40.2
39.7
45.3
38.3
140
-------
Figure H-l. Normalized Mean Bias (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure H-2. Normalized Mean Error (%) of ammonium by monitor for 12-km Eastern
U.S. domain, Winter 2005.
141
-------
Figure H-3. Normalized Mean Bias (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Spring 2005.
Figure H-4. Normalized Mean Error (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Spring 2005.
142
-------
Figure H-5. Normalized Mean Bias (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Summer 2005.
Figure H-6. Normalized Mean Error (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Summer 2005.
143
-------
Figure H-7. Normalized Mean Bias (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Fall 2005.
Figure H-8. Normalized Mean Error (%) of ammonium by monitor for 12-km Eastern U.S.
domain, Fall 2005.
144
-------
Figure H-9. Normalized Mean Bias (%) of ammonium by monitor for 12-km Western U.S.
domain, Winter 2005.
Figure H-10. Normalized Mean Error (%) of ammonium by monitor for 12-km Western
U.S. domain, Winter 2005.
145
-------
Figure H-ll. Normalized Mean Bias (%) of ammonium by monitor for 12-km Western U.S.
domain, Spring 2005.
Figure H-12. Normalized Mean Error (%) of ammonium by monitor for 12-km Western
U.S. domain, Spring 2005.
146
-------
Figure H-13. Normalized Mean Bias (%) of ammonium by monitor for 12-km Western U.S.
domain, Summer 2005.
Figure H-14. Normalized Mean Error (%) of ammonium by monitor for 12-km Western
U.S. domain, Summer 2005.
147
-------
Figure H-15. Normalized Mean Bias (%) of ammonium by monitor for 12-km Western U.S.
domain, Fall 2005.
Figure H-16. Normalized Mean Error (%) of ammonium by monitor for 12-km Western
U.S. domain, Fall 2005.
148
-------
I. Seasonal Elemental Carbon Performance
Table 1-1 presents the seasonal performance statistics of elemental carbon for the urban and rural
2005 monitoring data. Spatial plots of the NMB and NME statistics (units of percent) for
individual monitors are also provided as a complement to the tabular statistical data (Figures II -
116). In the winter, elemental carbon performance is mixed across the STN and IMPROVE
networks in the BUS and WUS, with a moderate over-prediction at STN sites and a moderate
under-prediction at the IMPROVE sites (except during the summer season in the WUS). These
biases and errors are not unexpected since there are known uncertainties among the scientific
community in carbonaceous emissions/measurements, transport, and deposition processes.
Table 1-1. CMAQ 2005 seasonal model performance statistics for elemental carbon.
CMAQ 2005 Elemental Carbon
Winter
Spring
Summer
STN
IMPROVE
STN
IMPROVE
STN
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
No. ofObs.
3441
657
831
602
964
811
520
2072
2279
522
166
386
474
2066
3672
1064
881
637
985
937
822
2296
2563
565
160
408
578
2289
3529
1030
NMB (%)
36.3
8.1
53.3
53.5
17.5
43.7
10.4
-10.1
-19.4
13.3
11.3
-33.4
-17.3
-23.6
31.7
49.5
52.1
14.7
18.9
44.9
65.2
-23.9
-7.6
1.9
-15.3
-43.9
44.9
-3.6
27.4
62.1
NME (%)
73.2
71.3
74.7
78.1
60.1
90.5
71.0
49.2
57.1
51.3
46.1
45.0
51.4
56.7
68.1
86.4
80.4
49.7
61.2
81.9
97.3
48.5
54.7
48.2
42.0
49.7
81.9
55.1
66.9
91.3
FB (%)
24.7
5.4
36.3
41.2
7.9
31.0
0.4
-16.4
-31.7
6.1
-1.9
-34.2
-13.6
-35.7
22.1
24.2
36.6
19.5
15.5
22.5
28.5
-17.7
-11.4
-3.1
-29.4
-42.2
22.5
-11.4
17.2
34.6
FE (%)
55.5
65.2
53.8
56.9
46.0
65.6
66.1
53.3
68.2
47.2
42.4
52.8
52.5
70.0
54.1
63.8
58.5
45.6
49.1
60.9
67.2
52.3
54.0
52.5
50.0
52.2
60.9
54.2
58.0
62.5
149
-------
Fall
IMPROVE
STN
IMPROVE
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
866
621
940
571
806
2182
2301
512
160
384
561
2055
3396
1063
901
642
988
602
867
2118
2352
518
116
406
510
2108
30.4
3.8
18.6
62.5
78.9
-42.2
1.4
-38.1
-35.5
-58.7
-42.7
5.1
10.8
21.4
18.8
8.8
-5.6
42.4
26.4
-25.0
-16.0
3.6
36.6
-47.8
-28.4
-16.8
66.1
40.1
66.9
98.4
10.7
51.1
60.8
47.8
40.6
59.8
51.7
63.0
56.9
70.3
63.1
43.6
52.2
73.7
74.3
47.4
55.6
47.7
-25.4
60.9
44.1
56.8
24.1
8.7
15.5
21.2
41.0
-46.8
0.8
-54.2
-52.9
-81.2
-51.9
4.7
8.3
9.0
10.6
11.9
-6.2
34.1
7.6
-32.2
-29.2
-12.5
-30.5
-49.4
-29.3
-30.7
55.7
44.2
55.9
72.0
65.8
64.1
56.6
66.6
58.6
87.0
65.5
57.1
49.4
59.6
50.9
42.3
46.9
56.4
61.8
54.5
63.7
50.4
51.1
60.3
50.6
65.5
150
-------
Figure 1-1. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure 1-2. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Eastern U.S. domain, Winter 2005.
151
-------
Figure 1-3. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km Eastern
U.S. domain, Spring 2005.
Figure 1-4. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Eastern U.S. domain, Spring 2005.
152
-------
Figure 1-5. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km Eastern
U.S. domain, Summer 2005.
Figure 1-6. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Eastern U.S. domain, Summer 2005.
153
-------
Figure 1-7. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km Eastern
U.S. domain, Fall 2005.
Figure 1-8. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Eastern U.S. domain, Fall 2005.
154
-------
Figure 1-9. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure 1-10. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Winter 2005.
155
-------
Figure 1-11. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Spring 2005.
Figure 1-12. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Spring 2005.
156
-------
Figure 1-13. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Summer 2005.
Figure 1-14. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Summer 2005.
157
-------
Figure 1-15. Normalized Mean Bias (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Fall 2005.
Figure 1-16. Normalized Mean Error (%) of elemental carbon by monitor for 12-km
Western U.S. domain, Fall 2005.
158
-------
J. Seasonal Organic Carbon Performance
Seasonal organic carbon performance statistics are provided in Table J-l. Spatial plots of the
NMB and NME statistics (units of percent) for individual monitors are also provided as a
complement to the tabular statistical data (Figures J-l - J-16).The model predictions generally
show moderate under-predictions for all Eastern sites located in the urban STN sites (NMB
values range from -12% to -67%) and rural IMPROVE sites (NMB values range from -3% to -
50%). Organic carbon performance in the EUS and WUS shows the largest under estimations
during the summer season. For IMPROVE, organic carbon performance shows a negative bias
in the West (NMB= -3%) and a positive bias in the East (NMB=24%). For STN, organic carbon
is under-predicted in the East (NMB= -12%) and West (NMB= -26%). These biases and errors
reflect sampling artifacts among each monitoring network. In addition, uncertainties exist for
primary organic mass emissions and secondary organic aerosol formation. Research efforts are
ongoing to improve fire emission estimates and understand the formation of semi-volatile
compounds, and the partitioning of SOA between the gas and particulate phases.
Table J-l. CMAQ 2002 seasonal model performance statistics for organic carbon.
CMAQ 2002 Organic Carbon
Winter
Spring
STN
IMPROVE
STN
IMPROVE
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
No. ofObs.
2063
606
284
552
520
568
507
2032
2432
149
512
439
366
2307
2241
656
337
583
590
579
560
1917
2369
147
508
NMB (%)
-12.4
-25.6
-18.1
24.4
-32.4
-22.9
-24.2
24.5
-2.9
4.6
89.2
-12.9
-5.1
-4.8
-19.1
-19.8
-19.9
8.5
-25.9
-38.7
-15.9
-27.2
-23.7
-29.7
15.6
NME (%)
51.2
62.0
58.7
61.6
44.0
45.4
63.0
65.4
59.9
43.1
104.2
42.9
52.9
59.3
53.6
63.4
57.1
55.3
47.5
53.2
65.9
53.0
54.5
43.9
46.4
FB (%)
-2.5
-22.4
-0.7
34.0
-27.8
-13.2
-22.0
14.1
-4.3
16.6
63.5
-22.1
-4.5
-5.1
-15.7
-13.7
-6.6
9.8
-22.7
-45.7
-7.1
-32.0
-26.4
-28.6
8.5
FE (%)
55.2
62.6
62.7
60.4
51.5
48.5
64.3
55.4
58.9
45.0
71.9
49.9
51.4
59.4
60.8
64.5
63.8
55.8
54.0
68.3
65.5
58.5
56.3
52.1
45.0
159
-------
Summer
Fall
STN
IMPROVE
STN
IMPROVE
Southeast
Central
West
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
442
293
2307
2690
832
408
754
683
659
684
2133
2595
153
523
460
411
2435
2732
809
418
752
681
698
657
2205
2686
149
514
459
461
2459
-42.8
-53.9
-21.5
-67.5
-55.4
-69.2
-68.1
-68.7
-63.3
-54.0
-66.7
-50.4
-70.4
-66.2
-73.8
-74.2
-49.4
-37.6
-40.4
-44.1
-14.6
-47.6
-43.7
-41.0
-25.5
-29.4
-41.0
4.1
-43.6
-37.8
-29.9
58.3
59.6
53.6
69.4
60.0
69.8
71.5
69.5
64.9
59.3
69.0
63.7
70.7
68.5
74.8
74.5
63.2
50.2
57.9
51.3
45.4
52.3
50.7
59.9
47.2
56.5
44.6
43.9
50.8
51.8
56.4
-60.6
-71.6
-24.4
-95.0
-76.2
-98.4
-88.4
-100.9
-91.1
-73.3
-95.5
-61.9
-103.1
-73.8
-122.3
-120.0
-58.1
-37.8
-38.7
-42.0
-13.2
-53.1
-49.2
-39.6
-32.4
-30.0
-44.0
0.5
-71.3
-47.2
-29.5
73.0
81.0
55.1
100.3
83.6
101.2
97.7
104.5
95.7
81.8
99.0
76.0
104.0
80.0
123.2
120.7
73.0
60.6
63.4
60.1
51.9
65.9
63.5
66.8
57.6
59.9
53.0
45.9
77.4
64.6
59.5
160
-------
Figure J-l. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure J-2. Normalized Mean Error (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Winter 2005.
161
-------
Figure J-3. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Spring 2005.
Figure J-4. Normalized Mean Error (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Spring 2005.
162
-------
Figure J-5. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Summer 2005.
Figure J-6. Normalized Mean Error (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Summer 2005.
163
-------
Figure J-7. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Fall 2005.
Figure J-8. Normalized Mean Error (%) of organic carbon by monitor for 12-km Eastern
U.S. domain, Fall 2005.
164
-------
Figure J-9. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Western
U.S. domain, Winter 2005.
Figure J-10. Normalized Mean Error (%) of organic carbon by monitor for 12-km
Western U.S. domain, Winter 2005.
165
-------
Figure J-ll. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Western
U.S. domain, Spring 2005.
Figure 3-12. Normalized Mean Error (%) of organic carbon by monitor for 12-km
Western U.S. domain, Spring 2005.
166
-------
Figure J-13. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Western
U.S. domain, Summer 2005.
Figure J-14. Normalized Mean Error (%) of organic carbon by monitor for 12-km
Western U.S. domain, Summer 2005.
167
-------
Figure J-15. Normalized Mean Bias (%) of organic carbon by monitor for 12-km Western
U.S. domain, Fall 2005.
Figure J-16. Normalized Mean Error (%) of organic carbon by monitor for 12-km
Western U.S. domain, Fall 2005.
168
-------
K. Annual Hazardous Air Pollutants Performance
An annual and seasonal operational model performance evaluation for specific hazardous air
pollutants (formaldehyde, acetaldehyde, benzene, acrolein, and 1,3-butadiene) was conducted in
order to estimate the ability of the CMAQ modeling system to replicate the base year
concentrations for the 12-km Eastern and Western United States domains. The annual model
performance results are presented in Table K-l below. Spatial plots of the NMB and NME
statistics (units of percent) for individual monitors are also provided as a complement to the
tabular statistical data (Figures K-l - K-24). The seasonal results follow in Sections L-P. Toxic
measurements from 471 sites in the East and 135 sites in the West were included in the
evaluation and were taken from the 2005 State/local monitoring site data in the National Air
Toxics Trends Stations (NATTS). Similar to PM2 5 and ozone, the evaluation principally
consists of statistical assessments of model versus observed pairs that were paired in time and
space on daily basis.
Model predictions of annual formaldehyde, acetaldehyde and benzene showed relatively small
bias and error percentages when compared to observations. The model yielded larger bias and
error results for 1,3 butadiene and acrolein based on limited monitoring sites. Model
performance for HAPs is not as good as model performance for ozone and PM2.5. Technical
issues in the HAPs data consist of (1) uncertainties in monitoring methods; (2) limited
measurements in time/space to characterize ambient concentrations ("local in nature"); (3)
commensurability issues between measurements and model predictions; (4) emissions and
science uncertainty issues may also affect model performance; and (5) limited data for estimating
intercontinental transport that effects the estimation of boundary conditions (i.e., boundary
estimates for some species are much higher than predicted values inside the domain).
As with the national, annual PM2.5 and ozone CMAQ modeling, the "acceptability" of model
performance was judged by comparing our CMAQ 2005 performance results to the limited
performance found in recent regional multi-pollutant model applications.10'11'12 Overall, the
normalized mean bias and error (NMB and NME), as well as the fractional bias and error (FB
and FE) statistics shown in Table J-l indicate that CMAQ-predicted 2005 toxics (i.e.,
observation vs. model predictions) are within the range of recent regional modeling applications.
10 Phillips, S., K. Wang, C. Jang, N. Possiel, M. Strum, T. Fox, 2007: Evaluation of 2002 Multi-pollutant Platform:
Air Toxics, Ozone, and Paniculate Matter, 7th Annual CMAS Conference, Chapel Hill, NC, October 6-8, 2008.
11 Strum, M., Wesson, K., Phillips, S., Cook, R., Michaels, H., Brzezinski, D., Pollack, A., Jimenez, M., Shepard, S.
Impact of using lin-level emissions on multi-pollutant air quality model predictions at regional and local scales. 17th
Annual International Emission Inventory Conference, Portland, Oregon, June 2-5, 2008.
12 Wesson, K., N. Farm, and B. Timin, 2010: Draft Manuscript: Air Quality and Benefits Model Responsiveness to
Varying Horizontal Resolution in the Detroit Urban Area, Atmospheric Pollution Research, Special Issue: Air
Quality Modeling and Analysis.
169
-------
Table K-l. CMAQ 2005 annual model performance statistics for air toxics.
CMAQ 2005 Annual
Formaldehyde
Acet aldehyde
Benzene
1,3-Butadiene
Acrolein
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
12-kmEUS
12-km WUS
Midwest
Northeast
Southeast
Central U.S.
West
No. of Obs.
6365
1928
771
1982
1246
1815
1746
6094
1892
703
1969
1231
1640
1709
11615
3369
1425
2589
2426
4737
2333
8102
1976
516
1902
1226
4142
1082
1660
783
n/a
850
278
n/a
592
NMB (%)
-55.5
-28.4
-77.1
-30.5
-66.2
-43.5
-25.5
-4.2
-19.2
-12.6
-9.5
0.4
1.8
-20.4
-32.6
-38.4
-8.3
21.6
-41.1
-47.0
-30.5
-74.7
-51.9
-78.7
-41.6
-85.4
-66.5
-40.8
-94.4
-95.7
n/a
-90.4
-97.0
n/a
-95.9
NME (%)
65.3
52.1
85.4
51.3
72.2
51.0
52.3
62.0
53.7
58.0
62.8
63.5
57.0
54.1
66.8
60.8
72.7
53.3
68.6
68.3
61.2
85.6
82.1
86.2
55.5
86.4
85.9
77.5
95.0
95.7
n/a
91.5
97.0
n/a
95.9
FB (%)
-39.2
-30.1
-25.8
-28.5
-51.3
-41.4
-26.0
-8.2
-19.5
-12.1
-9.0
-6.2
-4.3
-20.1
-13.5
-30.4
25.2
18.1
-17.2
-32.7
-19.2
-49.4
-34.5
-48.3
-54.8
-106.2
-20.0
-41.9
-131.3
-168.1
n/a
-120.5
-156.4
n/a
-177.6
FE (%)
65.6
60.7
74.0
61.6
70.4
61.5
59.8
60.3
59.6
60.0
63.7
62.2
51.1
60.6
62.8
63.9
62.4
46.8
59.8
69.4
63.4
91.6
91.7
81.9
71.3
111.5
89.2
85.3
142.2
170.4
n/a
134.2
157.0
n/a
177.6
170
-------
Figure K-l. Normalized Mean Bias (%) of annual formaldehyde by monitor for Eastern
U.S., 2005.
Figure K-2. Normalized Mean Error (%) of annual formaldehyde by monitor for Eastern
U.S., 2005.
171
-------
Figure K-3. Normalized Mean Bias (%) of annual acetaldehyde by monitor for Eastern
U.S., 2005.
Figure K-4. Normalized Mean Error (%) of annual acetaldehyde by monitor for Eastern
U.S., 2005.
172
-------
Figure K-5. Normalized Mean Bias (%) of annual benzene by monitor for Eastern U.S.,
2005.
Figure K-6. Normalized Mean Error (%) of annual benzene by monitor for Eastern U.S.,
2005.
173
-------
Figure K-7. Normalized Mean Bias (%) of annual 1,3-butadiene by monitor for Eastern
U.S., 2005.
Figure K-8. Normalized Mean Error (%) of annual 1,3-butadiene by monitor for Eastern
U.S., 2005.
174
-------
Figure K-9. Normalized Mean Bias (%) of annual acrolein by monitor for Eastern U.S.,
2005.
Figure K-10. Normalized Mean Error (%) of annual acrolein by monitor for Eastern U.S.,
2005.
175
-------
Figure K-15. Normalized Mean Bias (%) of annual formaldehyde by monitor for Western
U.S., 2005.
Figure K-16. Normalized Mean Error (%) of annual formaldehyde by monitor for
Western U.S., 2005.
176
-------
Figure K-17. Normalized Mean Bias (%) of annual acetaldehyde by monitor for Western
U.S., 2005.
Figure K-18. Normalized Mean Error (%) of annual acetaldehyde by monitor for Western
U.S., 2005.
177
-------
Figure K-19. Normalized Mean Bias (%) of annual benzene by monitor for Western U.S.,
2005.
Figure K-20. Normalized Mean Error (%) of annual benzene by monitor for Western U.S.,
2005.
178
-------
Figure K-21. Normalized Mean Bias (%) of annual 1,3-butadiene by monitor for Western
U.S., 2005.
Figure K-22. Normalized Mean Error (%) of annual 1,3-butadiene by monitor for
Western U.S., 2005.
179
-------
Figure K-23. Normalized Mean Bias (%) of annual acrolein by monitor for Western U.S.,
2005.
Figure K-24. Normalized Mean Error (%) of annual acrolein by monitor for Western U.S.,
2005.
180
-------
L. Seasonal Formaldehyde Performance
Seasonal formaldehyde performance statistics are provided in Table L-l. Spatial plots of the
NMB and NME statistics (units of percent) for individual monitors are also provided as a
complement to the tabular statistical data (Figures L-l - L-16). The model predictions generally
show moderate under-predictions (bias and error results) for all seasons in the Eastern and
Western sites (NMB values range from -19% to -84%; NME values range from 44% to 90%).
Formaldehyde performance in the EUS and WUS shows the largest under estimations during the
Midwest and Southeast areas. These biases and errors reflect sampling artifacts mentioned
previously among the NATTS monitoring network.
Table L-l. CMAQ 2005 seasonal model performance statistics for formaldehyde.
CMAQ 2005 Formaldehyde
Winter
Spring
Summer
Fall
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. of Obs.
1532
389
468
171
306
456
347
1480
440
472
181
249
435
396
1693
585
608
244
283
425
538
1660
514
434
175
408
499
465
NMB (%)
-54.2
-25.1
-35.9
-64.2
-66.2
-49.0
-19.7
-55.4
-34.4
-36.7
-59.9
-70.6
-48.8
-29.4
-55.8
-26.7
-27.5
-84.0
-50.6
-43.2
-25.1
-55.8
-28.9
-26.3
-67.4
-72.8
-36.0
-27.0
NME (%)
63.0
70.9
48.9
74.1
68.9
62.0
72.5
66.3
60.3
57.9
71.9
79.1
57.1
60.2
65.6
44.3
50.6
90.6
58.3
47.2
44.3
65.3
48.2
49.0
79.4
77.6
44.4
48.8
FB (%)
-46.9
-40.1
-35.5
-39.7
-68.5
-31.5
-35.1
-37.5
-26.4
-26.9
-26.7
-41.0
-42.2
-19.6
-32.0
-23.6
-20.3
-16.9
-39.1
-51.3
-20.8
-41.0
-33.2
-34.3
-23.5
-53.2
-41.3
-30.7
FE (%)
70.1
79.9
57.0
75.9
77.9
65.9
79.8
69.3
65.9
66.3
67.9
77.8
69.0
63.5
60.8
47.8
60.7
76.4
56.9
57.9
47.1
63.1
56.3
62.8
75.0
69.6
54.1
56.2
181
-------
Figure L-l. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure L-2. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
182
-------
Figure L-3. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Spring 2005.
Figure L-4. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Spring 2005.
183
-------
Figure L-5. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Summer 2005.
Figure L-6. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Summer 2005.
184
-------
Figure L-7. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Fall 2005.
Figure L-8. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Fall 2005.
185
-------
Figure L-9. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure L-10. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Winter 2005.
186
-------
Figure L-ll. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Spring 2005.
Figure L-12. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Spring 2005.
187
-------
Figure L-13. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Summer 2005.
Figure L-14. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Summer 2005.
188
-------
Figure L-15. Normalized Mean Bias (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Fall 2005.
Figure L-16. Normalized Mean Error (%) of formaldehyde by monitor for 12-km Western
U.S. domain, Fall 2005.
189
-------
M. Seasonal Acetaldehyde Performance
Seasonal acetaldehyde performance statistics are provided in Table M-l. Spatial plots of the
NMB and NME statistics (units of percent) for individual monitors are also provided as a
complement to the tabular statistical data (Figures M-l - M-l6). Overall, the model predictions
show moderate under-predictions for all Eastern and Western sites (NMB values range from -3%
to -47%). Although, in the summer, acetaldehyde performance in the BUS and WUS shows a
positive bias with over-predictions ranging from 15% in the Midwest to 84% in the Southeast.
Similar to formaldehyde results the biases and errors reflect technical issues with observational
data (uncertainties in monitoring methods and limited measurements in time and space).
Table M-l. CMAQ 2005 seasonal model performance statistics for acetaldehyde.
CMAQ 2005 Acetaldehyde
Winter
Spring
Summer
Fall
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. of Obs.
1454
388
470
141
304
408
346
1417
429
473
164
245
392
383
1622
572
576
238
278
397
525
1601
503
450
160
404
443
455
NMB (%)
-44.8
-30.2
-47.8
-46.1
-41.6
-34.8
-27.9
-29.9
-29.1
-33.1
-44.9
-12.9
-30.5
-29.7
47.8
-5.6
37.6
15.0
84.8
54.0
-8.9
-8.7
-24.4
-16.0
-3.5
-16.0
2.2
-26.0
NME (%)
52.0
69.9
54.4
51.8
52.6
40.8
72.1
50.8
54.8
57.4
50.7
52.2
43.4
58.3
83.6
47.0
76.1
63.5
103.7
87.9
46.1
55.3
52.2
57.6
60.4
52.9
47.8
51.9
FB (%)
-46.5
-44.4
-48.7
-52.7
-44.0
-34.4
-42.1
-25.6
-26.3
-23.0
-45.5
-4.1
-29.4
-25.7
44.9
5.9
40.9
32.0
51.9
49.2
2.6
-11.9
-23.3
-16.9
-7.6
-19.0
-2.3
-25.0
FE (%)
60.5
79.7
62.2
66.6
63.9
44.3
81.7
56.5
60.8
66.3
55.9
57.0
44.6
63.8
66.9
45.4
65.6
56.9
70.0
69.4
45.0
56.9
59.3
59.8
63.1
58.8
46.6
59.9
190
-------
Figure M-l. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure M-2. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
191
-------
Figure M-3. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Spring 2005.
Figure M-4. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Spring 2005.
192
-------
Figure M-5. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Summer 2005.
Figure M-6. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Summer 2005.
193
-------
Figure M-7. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Fall 2005.
Figure M-8. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Fall 2005.
194
-------
Figure M-9. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure M-10. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Winter 2005.
195
-------
Figure M-ll. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Spring 2005.
Figure M-12. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Spring 2005.
196
-------
Figure M-13. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Summer 2005.
Figure M-14. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Summer 2005.
197
-------
Figure M-15. Normalized Mean Bias (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Fall 2005.
Figure M-16. Normalized Mean Error (%) of acetaldehyde by monitor for 12-km Western
U.S. domain, Fall 2005.
198
-------
N. Seasonal Benzene Performance
Seasonal benzene performance statistics are provided in Table N-l. Spatial plots of the NMB
and NME statistics (units of percent) for individual monitors are also provided as a complement
to the tabular statistical data (Figures N-l - N-l6). The model predictions typically show
moderate bias (under-predictions) and error results for all Eastern and Western NATTS sites
(NMB values range from -3% to -52%; NME values range from 50% to 75%). However,
benzene performance in the Northeast shows over-predictions during all the seasons (NMB
values range from 11% in the Summer and Fall to 30% in the Spring and Winter). Similar to the
other HAPs modeled, these biases and errors reflect sampling artifacts among the NATTS
monitoring network.
Table N-l. CMAQ 2005 seasonal model performance statistics for benzene.
CMAQ 2005 Benzene
Winter
Spring
Summer
Fall
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. ofObs.
2846
781
592
322
591
1217
523
2888
769
632
331
618
1169
516
2955
943
751
396
563
1167
675
2926
876
614
376
654
1184
619
NMB (%)
-28.6
-39.2
31.1
27.2
-30.2
-50.3
-31.4
-35.8
-29.8
28.8
-20.6
-51.9
-42.0
-17.0
-36.8
-28.9
11.4
-3.2
-48.4
-51.7
-16.9
-30.9
-47.9
11.7
-29.2
-34.8
-41.8
-44.0
NME (%)
69.7
59.1
56.7
72.0
67.3
74.1
58.9
65.3
56.8
52.1
67.7
70.9
62.3
57.0
67.2
58.7
52.9
76.8
69.2
68.0
57.6
64.0
65.8
49.8
74.9
66.8
64.9
67.5
FB (%)
-6.0
-39.4
25.2
35.0
-6.4
-23.4
-33.8
-19.9
-30.6
19.5
13.1
-22.2
-37.4
-16.1
-15.8
-19.2
17.2
31.4
-26.3
-41.7
-3.0
-12.3
-34.2
10.9
20.8
-14.3
-28.6
-27.1
FE (%)
60.0
64.9
46.2
55.0
55.8
66.4
67.4
63.2
62.1
45.2
54.4
61.0
69.6
60.2
67.0
61.8
51.0
75.3
62.3
73.8
58.2
60.9
67.0
44.1
62.2
60.2
67.9
68.3
199
-------
Figure N-l. Normalized Mean Bias (%) of benzene by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure N-2. Normalized Mean Error (%) of benzene by monitor for 12-km Eastern U.S.
domain, Winter 2005.
200
-------
Figure N-3. Normalized Mean Bias (%) of benzene by monitor for 12-km Eastern U.S.
domain, Spring 2005.
Figure N-4. Normalized Mean Error (%) of benzene by monitor for 12-km Eastern U.S.
domain, Spring 2005.
201
-------
Figure N-5. Normalized Mean Bias (%) of benzene by monitor for 12-km Eastern U.S.
domain, Summer 2005.
Figure N-6. Normalized Mean Error (%) of benzene by monitor for 12-km Eastern U.S.
domain, Summer 2005.
202
-------
Figure N-7. Normalized Mean Bias (%) of benzene by monitor for 12-km Eastern U.S.
domain, Fall 2005.
Figure N-8. Normalized Mean Error (%) of benzene by monitor for 12-km Eastern U.S.
domain, Fall 2005.
203
-------
Figure N-9. Normalized Mean Bias (%) of benzene by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure N-10. Normalized Mean Error (%) of benzene by monitor for 12-km Western U.S.
domain, Winter 2005.
204
-------
Figure N-ll. Normalized Mean Bias (%) of benzene by monitor for 12-km Western U.S.
domain, Spring 2005.
Figure N-12. Normalized Mean Error (%) of benzene by monitor for 12-km Western U.S.
domain, Spring 2005.
205
-------
Figure N-13. Normalized Mean Bias (%) of benzene by monitor for 12-km Western U.S.
domain, Summer 2005.
Figure N-14. Normalized Mean Error (%) of benzene by monitor for 12-km Western U.S.
domain, Summer 2005.
206
-------
Figure N-15. Normalized Mean Bias (%) of benzene by monitor for 12-km Western U.S.
domain, Fall 2005.
Figure N-16. Normalized Mean Error (%) of benzene by monitor for 12-km Western U.S.
domain, Fall 2005.
207
-------
O. Seasonal 1,3-Butadiene Performance
Table O-l presents the seasonal 1.3-butadiene performance statistics. Spatial plots of the NMB
and NME statistics (units of percent) for individual monitors are also provided as a complement
to the tabular statistical data (Figures O-l - O-l6). In general, the model predictions show
moderate to large under-predictions for all Eastern and Western sites during all the seasons
(NMB values range from -29% to -90%). Performance of 1,3-butadiene shows the largest under
estimations in the areas of the Southeast and Central U.S. Likewise, the error results are large
ranging from approximately 50% to 100%. These biases and errors reveal the underlying issues
in the HAPs data (i.e., uncertainties in monitoring methods; limited measurements in time/space,
proportionality issues between measurements and model predictions, emissions and science
uncertainty issues, as well as boundary condition estimates).
Table O-l. CMAQ 2005 seasonal model performance statistics for 1,3-butadiene.
CMAQ 2005 1,3-Butadiene
Winter
Spring
Summer
Fall
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. of Obs.
2028
501
453
133
313
1045
280
1963
491
446
105
294
1011
281
2003
482
489
122
279
1059
246
2108
502
514
156
340
1027
275
NMB (%)
-69.8
-55.3
-32.0
-34.9
-78.4
-65.2
-39.9
-80.3
-36.5
-42.7
-66.6
-83.8
-67.8
-29.4
-79.1
-47.9
-47.8
-94.5
-90.6
-71.8
-42.1
-69.3
-57.0
-49.1
-62.0
-87.7
-60.3
-47.9
NME (%)
84.9
87.6
53.1
70.5
80.4
89.0
83.2
88.8
79.4
52.6
75.3
84.4
88.1
75.3
86.7
72.6
55.7
95.3
91.0
85.0
65.5
81.1
80.3
60.7
70.3
88.8
80.5
76.7
FB (%)
-26.0
-30.7
-32.5
-2.6
-92.9
6.7
-42.7
-52.1
-26.5
-56.0
-65.3
-108.1
-17.8
-33.8
-63.1
-40.2
-66.4
-73.6
-122.1
-39.5
-46.7
-56.2
-40.6
-62.4
-56.1
-103.7
-29.2
-45.0
FE (%)
88.3
100.7
60.0
64.5
100.1
92.4
95.2
93.6
85.6
69.9
93.2
111.0
88.4
81.0
93.8
86.0
77.0
89.9
125.3
90.0
78.0
90.6
94.3
77.1
82.9
111.1
86.0
86.3
208
-------
Figure O-l. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure O-2. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Winter 2005.
209
-------
Figure O-3. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Spring 2005.
Figure O-4. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Spring 2005.
210
-------
Figure O-5. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Summer 2005.
Figure O-6. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Summer 2005.
211
-------
Figure O-7. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Fall 2005.
Figure O-8. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Fall 2005.
212
-------
Figure O-9. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Eastern
U.S. domain, Winter 2005.
Figure O-10. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Winter 2005.
213
-------
Figure O-ll. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Spring 2005.
Figure O-12. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Spring 2005.
214
-------
Figure O-13. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Summer 2005.
Figure O-14. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Summer 2005.
215
-------
Figure O-15. Normalized Mean Bias (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Fall 2005.
Figure O-16. Normalized Mean Error (%) of 1,3-butadiene by monitor for 12-km Western
U.S. domain, Fall 2005.
216
-------
P. Seasonal Acrolein Performance
Seasonal acrolein performance statistics are provided in Table P-l. Spatial plots of the NMB and
NME statistics (units of percent) for individual monitors are also provided as a complement to
the tabular statistical data (Figures P-l - P-16). The model predictions generally show large
under-predictions for all Eastern and Western sites (NMB values range from -85% to -97%).
Acrolein performance in the EUS and WUS shows the similar under estimations during each
season. These biases and errors reflect sampling artifacts among each monitoring network
mentioned above.
Table P-l. CMAQ 2005 seasonal model performance statistics for acrolein.
CMAQ 2005 Acrolein
Winter
Spring
Summer
Fall
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. of Obs.
423
195
216
n/a
80
n/a
153
298
190
180
n/a
25
n/a
143
447
187
251
n/a
52
n/a
n/a
492
211
203
29
121
n/a
160
NMB (%)
-91.9
-93.7
-85.6
n/a
-96.6
n/a
-94.0
-86.3
-94.7
-87.7
n/a
-76.7
n/a
-95.1
-95.8
-96.4
-93.0
n/a
-93.9
n/a
n/a
-96.3
-96.8
-92.9
-98.3
-98.1
n/a
-97.0
NME (%)
92.7
93.7
87.2
n/a
96.6
n/a
94.0
87.3
94.7
88.7
n/a
76.9
n/a
95.1
96.5
96.4
94.8
n/a
93.9
n/a
n/a
96.4
96.8
93.1
98.5
98.2
n/a
97.0
FB (%)
-123.1
-160.5
-108.4
n/a
-165.9
n/a
-169.1
-118.2
-167.1
-121.3
n/a
-107.4
n/a
-177.3
-132.9
-171.7
-119.1
n/a
-142.1
n/a
n/a
-144.8
-172.8
-134.5
-145.4
-166.5
n/a
-180.9
FE (%)
133.8
162.1
119.7
n/a
165.9
n/a
169.1
128.0
167.9
129.2
n/a
108.1
n/a
177.3
151.5
177.3
147.9
n/a
142.2
n/a
n/a
149.5
174.3
137.3
153.5
167.5
n/a
180.9
217
-------
Figure P-l. Normalized Mean Bias (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure P-2. Normalized Mean Error (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Winter 2005.
218
-------
Figure P-3. Normalized Mean Bias (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Spring 2005.
Figure P-4. Normalized Mean Error (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Spring 2005.
219
-------
Figure P-5. Normalized Mean Bias (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Summer 2005.
Figure P-6. Normalized Mean Error (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Summer 2005.
220
-------
Figure P-7. Normalized Mean Bias (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Fall 2005.
Figure P-8. Normalized Mean Error (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Fall 2005.
221
-------
Figure P-9. Normalized Mean Bias (%) of acrolein by monitor for 12-km Eastern U.S.
domain, Winter 2005.
Figure P-10. Normalized Mean Error (%) of acrolein by monitor for 12-km Western U.S.
domain, Winter 2005.
222
-------
Figure P-ll. Normalized Mean Bias (%) of acrolein by monitor for 12-km Western U.S.
domain, Spring 2005.
Figure P-12. Normalized Mean Error (%) of acrolein by monitor for 12-km Western U.S.
domain, Spring 2005.
223
-------
Figure P-13. Normalized Mean Bias (%) of acrolein by monitor for 12-km Western U.S.
domain, Summer 2005.
Figure P-14. Normalized Mean Error (%) of acrolein by monitor for 12-km Western U.S.
domain, Summer 2005.
224
-------
Figure P-15. Normalized Mean Bias (%) of acrolein by monitor for 12-km Western U.S.
domain, Fall 2005.
Figure P-16. Normalized Mean Error (%) of acrolein by monitor for 12-km Western U.S.
domain, Fall 2005.
225
-------
Q. Annual Nitrate and Sulfate Deposition Performance
Annual nitrate and sulfate deposition performance statistics are provided in Table Q-l. Spatial
plots of the NMB and NME statistics (units of percent) for individual monitors are also provided
as a complement to the tabular statistical data (Figures Q-l - Q-8). The model predictions for
annual nitrate deposition generally show small under-predictions for the Eastern and Western
NADP sites (NMB values range from -3% to -18%). Sulfate deposition performance in the EUS
and WUS shows the similar over predictions (NMB values range from 3% to 14%), except for
predicted under-prediction in the Central US (NMB = -9.9%). The errors for both annual nitrate
and sulfate are relatively moderate with values ranging from 54% to 87% which reflect scatter in
the model predictions o observation comparison.
Table Q-l. CMAQ 2005 seasonal model performance statistics for acrolein.
CMAQ 2005 Total Deposition
Nitrate
Sulfate
12-kmEUS
12-kmWUS
Northeast
Midwest
Southeast
Central
West
12-kmEUS
12-km WUS
Northeast
Midwest
Southeast
Central
West
No. ofObs.
7381
2732
1658
1391
1980
1229
2400
7381
2732
1658
1391
1980
1229
2400
NMB (%)
-8.6
-16.3
1.0
-3.3
-3.7
-18.5
-13.4
6.2
3.7
11.3
13.9
7.4
-9.9
11.6
NME (%)
61.3
68.4
57.4
59.7
63.8
61.4
71.8
66.5
75.9
63.7
61.7
71.0
64.4
80.99
FB (%)
-5.1
-16.1
-3.1
-2.0
-6.5
-17.6
-15.4
4.7
3.2
16.0
21.4
6.0
-3.5
5.4
FE (%)
54.2
83.3
65.7
67.5
70.7
78.6
84.1
75.3
86.5
67.2
69.9
73.8
80.3
87.1
226
-------
Figure Q-l. Normalized Mean Bias (%) of annual nitrate deposition by monitor for
Eastern U.S., 2005.
Figure Q-2. Normalized Mean Error (%) of annual nitrate by monitor for Eastern U.S.,
2005.
227
-------
Figure Q-3. Normalized Mean Bias (%) of annual nitrate deposition by monitor for
Western U.S., 2005.
Figure Q-4. Normalized Mean Error (%) of annual nitrate deposition by monitor for
Western U.S., 2005.
228
-------
Figure Q-5. Normalized Mean Bias (%) of annual sulfate deposition by monitor for
Eastern U.S., 2005.
Figure Q-6. Normalized Mean Error (%) of annual sulfate deposition by monitor for
Eastern U.S., 2005.
229
-------
Figure Q-7. Normalized Mean Bias (%) of annual sulfate deposition by monitor for
Western U.S., 2005.
Figure Q-8. Normalized Mean Error (%) of annual sulfate deposition by monitor for
Western U.S., 2005.
230
------- |