United States Air and Radiation EPA420-D-02-004
Environmental Protection October 2002
Agency
v>EPA Diesel PM
Model-To-Measu rement
Comparison
> Printed on Recycled Paper
-------
EPA420-D-02-004
October 2002
PM
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Prepared for EPA by
ICF Consulting
EPA Contract No. 68-C-01-164
Work Assignment No. 0-5
NOTICE
This technical report does not necessarily represent final EPA decisions or positions.
It is intended to present technical, analysis of issues using data thatC are currently available.
The purpose in the release of such reports is to facilitate the exchange of
technical information and to inform the public of technical developments which
may form the basis for a final EPA decision, position, or regulatory action.
-------
CONSULTING
Final Report
DIESEL PM MODEL-TO-MEASUREMENT
COMPARISON
EPA Contract 68-C-01-164
Work Assignment No. 0-5
September 30, 2002
Prepared for:
James Richard Cook, Chad Bailey, Tesh Rao
USEPA Office of Transportation and Air Quality
2000 Traverwood Drive
Ann Arbor, Michigan, 48105
Prepared by:
Jonathan Cohen, Christine Lee, Seshasai Kanchi
ICF Consulting
101 Lucas Valley Road, Suite 230
San Rafael, CA 94903
Rob Klausmeier
dKC - de la Torre Klausmeier Consulting, Inc
1401 Foxtail Cove, Austin, TX 78704
-------
Diesel PM Model-To-Measurement Comparison
EPA Contract 68-C-01-164, Work Assignment No. 0-5
CHAPTER 1. INTRODUCTION AND SUMMARY
The purpose of this project was to compare estimated diesel paniculate matter (diesel PM or DPM)
concentrations based on elemental carbon (EC) and black carbon (BC) data with modeled ambient
concentrations of DPM from the 1996 National-Scale Air Toxics Assessment (NSATA).
The NSATA used DPM inventory estimates from EPA's final rule promulgating 2007 heavy duty
engine standards. Using the ASPEN dispersion model, NSATA developed estimates of 1996 annual
average concentrations of DPM at census tracts nationwide. The goal of this project was to evaluate
the reasonableness of DPM estimates from dispersion models for this case by comparing the
NSATA DPM concentration estimates with estimates based on measured EC and BC
concentrations.
EC measurements can be obtained from PM2.5 monitoring sites that sample PM2.5 using quartz
fiber media. The EC is measured using thermo-optical analysis of the carbonaceous material. Many
studies have used thermal optimal transmission (TOT), the NIOSH method developed at Sunset
laboratories. Some studies have used thermal optical reflectance (TOR), a method developed by
Desert Research Institute. In addition, some sites measure ambient BC with an Aethalometer.
EPA's Office of Air Quality Planning and Standards is reviewing the measurement of EC
through the Speciation Trends Network, and an Agency statement on the issue is forthcoming.
For now, however, existing values developed using the TOT method are being used.
All these carbon concentration measurements can be used to estimate ambient DPM by using
conversion factors based on 1) source apportionment studies, 2) source-receptor model studies, and
3) studies which examine the fraction of EC in DPM.
Our analysis was carried out as a series of steps that are detailed in this report:
1. A nationwide database was compiled containing elemental carbon (EC), organic carbon
(OC), and black carbon (BC) concentration measurements from PM2.5 monitoring sites from
January 1994 to December 2001. The database includes daily, annual, seasonal,
weekday/weekend, and overall average concentrations and other summary statistics for each
monitoring site.
2. Using results from several source apportionment studies, multiplicative conversion factors
to estimate diesel particulate matter (DPM) from EC, OC, and BC concentrations were
compiled. Average conversion factors were compiled together with lower and upper bound
values.
3. Based on the results of steps 1 and 2, average, minimum, and maximum estimates of the
overall average DPM at each monitoring site were computed.
ICF Consulting
-------
4. The monitored values of the DPM (derived in step 3) were statistically compared with
modeled values from the NSATA at the nearest census tract centroid and at the census tract
centroid with the maximum modeled value within 30 km. The goal of the comparison was
to determine if the modeled DPM concentrations in the NSATA agree reasonably well with
estimates from monitored data.
Data Base
As described in Chapter 2, ICF developed a database of EC and BC measurements from PM2.5
monitoring sites, consisting of EC and BC data collected in the time period from January 1, 1994
to December 31, 2001. The database has been provided to EPA in a .DBF format, although ICF's
statistical analyses were performed in SAS using a SAS database. The data sources with
currently available data included 76 EPA PM speciation sampling sites, the Northern Front
Range Air Quality Study (NFRAQS), the Phoenix EPA PM Supersite, Interagency Monitoring of
Protected Visual Environments (IMPROVE), Clean Air Status and Trends Network
(CASTNET), the California Multiple Air Toxics Exposure Study in the South Coast Air Basin
(MATESii), and the 1995 Integrated Monitoring Study (EVIS95). Data were also obtained from
one of EPA's EMPACT grant recipients, Airbeat.
The EC and BC values "below the detection limit" were replaced by one half of the minimum
detection limit (MDL). Missing data were not used or substituted for, to avoid biasing the
estimated standard deviations.
The database includes the following information, where available:
1. For each site measuring carbon on quartz fiber, the method by which EC and OC
fractions (EC/OC) are determined
2. Latitude and longitude coordinates of the monitor
3. Whether the monitor is in an urban or rural tract, based on NSATA assignments
4. The minimum detection limit of the monitor and analytic method
5. Monitor start and end dates
6. Summary statistics of daily average EC, OC, or BC measurements for all data at the site
and also stratified by 1) year, 2) calendar quarter, 3) year and quarter, 4) weekday and
weekend. The following summary statistics were obtained: mean, median, standard
deviation, geometric mean, geometric standard deviation, minimum, 10th, ... 90th
percentile, maximum.
7. The same set of summary statistics were obtained for ECOCX, an EC concentration value
developed from the EPA PM TOT speciation data to estimate the corresponding TOR
value, and for the various monitored DPM estimates computed by applying correction
factors (described below) to the EC values.
8. At each PM2.5 monitoring site, the fractions of PM2.5 which are elemental carbon,
organic carbon, sulfates, and nitrates.
ICF Consulting
-------
Conversion Factors
As described in Chapter 3, ICF developed multiplicative "conversion factors" (CFs) for
estimating ambient DPM based on the ambient EC, BC or OC measurements. For each site and
carbon type (EC, OC, or BC), a low-end, most likely ("average"), and high-end CF was
assigned, as discussed below. For EC sites, the estimated ambient DPM-high equals ambient EC
multiplied by the high-end CF for EC, the estimated ambient DPM-low equals ambient EC
multiplied by the low-end CF for EC, and the estimated ambient DPM-avg equals ambient EC
multiplied by the most likely CF for EC. BC and OC conversion factors were tabulated but not
applied to the concentration data. Separate sets of conversion factors were applied to EC data
collected by the TOR or TOT method.
The CFs were developed using existing source apportionment studies. Source apportionment
studies for the West US included the Northern Front Range Air Quality Study (Denver,
Colorado), the Los Angeles Study (various analyses based on data collected in 1982 at 4 urban
Southern California sites), and the San Joaquin Valley Study. For the East US, information from
the recent source apportionment study for sites in the South-East US was used. We contacted
several experts and reviewed available literature to obtain information from these studies. In
particular, James Schauer from University of Wisconsin-Madison provided very helpful
information. The conversion factors were developed by dividing the reported DPM concentration
by the reported total EC or OC concentration.
Since there are several source apportionment studies, each giving different estimates of the diesel
contribution at different receptors, and since there are several diesel exhaust source profiles in
the literature, several possible CF values could be applied for each site. For the TOR conversion
factors, we developed rural CFs for rural sites and urban CFs at urban sites. We could not obtain
TOR data to match by region or season, since the available data were all collected in the winter
and in the West US. For the TOT conversion factors we developed separate factors by quarter for
the East US and another set of factors for the West US. We could not obtain TOT data to match
by the urban or rural classification. The minimum, average, and maximum of the possible CFs
will give the low-end, most likely, and high-end CF's for that site.
Model to Monitor Comparisons
In Chapter 4, we will describe the DPM model-to-monitor comparisons. Using the CFs to
convert the monitored EC values to estimated DPM concentrations, we compared differences
between the monitored and modeled DPM values. For the modeled values, the NSATA
predictions for 1996 using ASPEN (and CALPUFF, for the background) were used. We
compared the monitored value to the NSATA prediction at the nearest ASPEN receptor site
(census tract centroid). We also compared the monitored value to the maximum NSATA
prediction within 30 km. For the monitored value we separately analyzed the site means of
DPM-high, DPM-low, and DPM-avg, as described above. The site means were computed by
averaging all the daily averages (from January 1, 1994 to December 31, 2001). There were
insufficient data to restrict the monitored data to the modeling year 1996.
ICF Consulting
-------
Separately for each model-to-monitor comparison, and for all locations, urban locations, and
rural locations, we compared the modeled and monitored results using:
Scatterplots of modeled against monitored values that include fitted regression lines
Tables summarizing the regression fits, average difference, and average percentage
difference for the modeled values against the monitored values
Tables summarizing the proportions of modeled values that were within 10%, 25%, 50%,
and 100% of the monitored values
Using the NSATA estimates of the percentage contributions of onroad and nonroad sources at
each matched receptor site, we also evaluated whether the regions dominated by either onroad or
nonroad sources have better model-to-monitor comparisons. The same statistical comparisons
were applied to the subsets of sites dominated by non-road (at least 75 % of modeled DPM is
non-road) or on-road (at least 50 % of modeled DPM is on-road) emissions.
The inventory estimates from the NONROAD model have been revised since the NSATA
assessment was conducted using the draft 2000 NONROAD model (also used to develop
inventories for EPA's 2007 heavy duty standards). The EPA provided a single nationwide
multiplicative adjustment factor of 0.69 for the nonroad ambient DPM based on the ratio of
NONROAD year 1996 predictions from the 2002 and 2000 draft versions of the NONROAD
model . To determine if this 70 % adjustment lead to improved model-to-monitor comparisons,
we applied the same statistical comparisons after adjusting the NSATA predictions using the
nonroad adjustment factor (the onroad ambient DPM is unchanged).
Findings
The model-to-monitor comparisons for non-EPA TOR data (i.e. excluding the ECOCX estimates
of TOR from the EPA TOT data) were based on 15 monitoring sites. The model-to-monitor
comparisons for TOT data were based on 95 monitoring sites. The model-to-monitor
comparisons for TOR data including the EPA ECOCX values were based on 88 monitoring sites.
The regression model analyses were generally less useful because the R squared values were in
most cases less than 0.3 and the regressions tended to be over-influenced by the more extreme
values. Based on the regression results, the best model performance was for the DPM-minimum
monitored value for TOT data but for the DPM-average value for TOR data. Results were very
similar for the modeled values based on the 2000 and 2002 NONROAD model and were a little
better for the rural sites compared to the urban sites. For the Non-road-dominated subset of TOT
sites, the regression model fitted better than the all sites regression, but the monitored values
were significantly overpredicted. For the Non-road-dominated subset of TOR sites including the
EPA ECOCX sites, the regression model fitted a bit worse than the all sites regression. There
was not enough data to evaluate the On-road-dominated subset. The comparisons between the
maximum modeled value within 30 km and the monitored values all showed that the monitored
value was significantly over-predicted.
A summary table of the differences between the nearest modeled values and the monitored
values is given on the next page. Based on the mean percentage difference and based on the
ICF Consulting
-------
fraction of modeled values within 100 % of the monitored value, the best model performance
was consistently for the DPM-maximum value at the nearest census tract centroid using the
estimates consistent with the 2002 NONROAD model. For the non-EPA TOR, for TOT, and for
the combination of TOR data from TOR sites and from EPA TOT converted to TOR (i.e., TOR
and ECOCX), the mean percentage differences were 26 %, 27 %, and -12 % and the fractions of
modeled values within 100 % of the monitored value were 73 %, 80 %, and 92 %, respectively.
These results compare favorably with the results of the model to monitor comparisons for other
pollutants in the NSATA assessment. For instance, ASPEN typically agrees with monitoring data
within 30% half the time and within a factor of 2 most of the time. The best agreement is for
benzene where the results are within a factor of two for 89 percent of the cases and within 30%
59 percent of the time. The median ratio of the benzene model to monitor comparisons was 0.92.
Agreement for other HAPs varies, with median ratios of model to monitor values varying
between 0.65 for formaldehyde to 0.17 for lead.
We can conclude that the modeled diesel PM concentrations in NSATA agree reasonably well
with monitor values, and the agreement is better than for other pollutants evaluated, except for
benzene.
ICF Consulting
-------
Summary of differences between the nearest modeled concentration and the monitored values.
Modeled
Variable1
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
Monitored
Variable2
ECTOR
ECTOR
ECTORH
ECTORH
ECTORL
ECTORL
ECTOT
ECTOT
ECTOTH
ECTOTH
ECTOTL
ECTOTL
TOR
TOR
TORH
TORH
TORL
TORL
N
15
15
15
15
15
15
95
95
95
95
95
95
88
88
88
88
88
88
Mean
Modeled
Value
1.56
1.20
1.56
1.20
1.56
1.20
2.61
2.05
2.61
2.05
2.61
2.05
2.31
1.81
2.31
1.81
2.31
1.81
Mean
Monitored
Value
0.94
0.94
1.16
1.16
0.64
0.64
1.73
1.73
2.10
2.10
1.52
1.52
1.70
1.70
2.23
2.23
1.19
1.19
Mean
Difference
0.63
0.26
0.40
0.04
0.92
0.55
0.88
0.32
0.52
-0.05
1.09
0.52
0.61
0.11
0.08
-0.42
1.12
0.62
Mean
/o
Difference
100
56
62
26
190
126
80
42
61
27
101
58
47
15
13
-12
110
65
Fraction of Modeled
Values Within
10%
0.07
0.07
0.00
0.00
0.13
0.07
0.12
0.11
0.11
0.11
0.09
0.09
0.10
0.17
0.11
0.08
0.10
0.14
25%
0.13
0.13
0.07
0.07
0.40
0.33
0.21
0.37
0.22
0.35
0.17
0.32
0.30
0.30
0.26
0.22
0.26
0.31
50%
0.53
0.47
0.40
0.33
0.47
0.47
0.45
0.53
0.46
0.53
0.43
0.52
0.59
0.59
0.60
0.52
0.41
0.52
100%
0.53
0.60
0.60
0.73
0.53
0.53
0.68
0.77
0.74
0.80
0.63
0.72
0.78
0.85
0.84
0.92
0.65
0.74
Notes:
1. Modeled variable:
concnear Nearest modeled DPM concentration consistent with the draft 2000
NONROAD Model
concnear2 Nearest modeled DPM concentration consistent with the draft 2002
NONROAD Model
2. Monitored variable:
ECTOR EC value multiplied by TOR average correction factor (missing for EC
measured using TOT).
ECTORH EC value multiplied by TOR maximum correction factor (missing for EC
measured using TOT).
ECTORL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOT).
ECTOT EC value multiplied by TOT average correction factor (missing for EC
measured using TOR).
ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC
measured using TOR).
ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOR).
TOR ECOCX value multiplied by TOT average correction factor for EPA data,
ECTOR for TOR data.
TORH ECOCX value multiplied by TOT maximum correction factor for EPA
data, ECTOR for TOR data.
TORL ECOCX value multiplied by TOR minimum correction factor for EPA
data, ECTOR for TOR data.
ICF Consulting
-------
CHAPTER 2. DIESEL PARTICULATE MATTER EC/OC/BC DATABASE
EPA has been provided with three Dbaselll (DBF) files, comprising the EC/OC/BC
(elemental/organic/black carbon) concentration database. The files were developed by compiling
and processing data from the following studies ("source"):
AIRBEAT
CASTNET
EPA (PM Speciation data)
IMPROVE (six selected sites )*
IMS95
MATESII
NERL (Phoenix Supersite)
NFRAQS
All concentrations are reported as |ig/m3.
The dailyavg.final.dbf file contains daily average concentrations by source, site_id, and date. The
variables are listed in Table 2-1. Note that each daily average is possibly averaged across
multiple time periods during the day (e.g. some black carbon data was reported every five
minutes) and/or multiple measuring instruments at the same location (e.g. EPA daily average
data with multiple POCs on the same date). For MDLs, we either used values reported with the
database or used default values obtained from the literature. Raw values below the MDL were
replaced by one half of the MDL prior to computing the daily averages (this did not happen very
often for the EC/OC/BC concentration data). For black carbon, MDLs were not reported in the
databases and could not be obtained from the data suppliers. According to the "Aethalometer
Book" (Hansen, 2000), written by the company that makes the aethalometer instrument that
measures BC, the black carbon MDL depends upon the filter size, air flow rate, and averaging
period. The filter size and air flow rate were not always available. Furthermore, since we are
only interested in daily averages, the MDLs of the five-minute or hourly values do not represent
the precision of the daily averages. For these reasons, we did not use MDLs for the black carbon
data, effectively assuming an MDL of zero. The method variable lists all methods used for that
day ("Aethalometer" applies to all black carbon data, other methods apply to EC and OC).
EPA's Office of Air Quality Planning and Standards is reviewing the measurement of EC
through the Speciation Trends Network, and an Agency statement on the issue is forthcoming.
For now, however, existing values developed using the TOT method are being used.
Minimum detection limits (MDLs) were obtained from the data suppliers if possible. For the
EPA PM speciation data, specific MDL's for each of the various NIOSH methods were supplied;
each daily measurement had an associated measurement method. The EPA MDLs were mostly
equal to or close to 0.146 |ig/m3 for EC and OC. For Airbeat, the data supplier reported EC
MDLs that were either 0.059 or 0.134 |lg/m3 depending on the method used. For IMPROVE and
1 The data from the two Yellowstone Park sites YELL1 and YELL2 were treated as all coming from the YELL1 site.
The Yellowstone Park monitoring site was moved a short distance in 1996.
ICF Consulting
-------
NFRAQS, the data source contact was unable to give specific MDL values for the TOR method,
because the TOR measurements of EC and OC are both sums of three components, each of
which has its own MDL. However, we were able to find a report by Chow and Watson (1998)
that tabulated MDLs of 0.12 |ig/m3 for EC and OC by TOR. We used these values for the
IMPROVE and NFRAQS EC and OC data. For CASTNET, MATESII, and NERL, we were
unable to obtain specific MDL values for the NIOSH measurements, but Gary Lear (EPA)
suggested a typical MDL value of 0.1 |ig/m3 for EC and OC, which was used for these three
studies. For MATESII, the EC MDL was erroneously entered as 0.01 in the database; this has no
affect on the daily averages or other results since all the MATESII EC measurements were 0.47
|lg/m3 or greater.
Since some site-days had more than one measured EC or OC concentration, a daily average can
be treated as being below the MDL if any of the measurements for that day and carbon species
were below the MDL. Using this definition, across the entire dataset, 9.9 % of the 13,993 EC
daily averages were below the MDL and 3.3 % of the 13,804 OC daily averages were below the
MDL.
In addition to the carbon species EC, OC, and BC, the daily average file also includes daily
average concentrations for the various estimated DPM concentrations defined as follows:
ECOCX The EPA EC and OC data were collected using the NIOSH (Sunset
Laboratory) method of thermal optical transmission (TOT). EPA also
computed a value ECOCX intended to approximate the equivalent EC
concentration based on thermal optical reflectance (TOR, as developed
and applied by Desert Research Institute). This value is missing for non-
EPA data.
ECTOR EC value multiplied by TOR average correction factor (missing for EC
measured using TOT).
ECTORH EC value multiplied by TOR maximum correction factor (missing for EC
measured using TOT).
ECTORL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOT).
ECTOT EC value multiplied by TOT average correction factor (missing for EC
measured using TOR).
ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC
measured using TOR).
ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOR).
ICF Consulting
-------
EPATOR ECOCX value multiplied by TOR average correction factor (missing for
non-EPA data).
EPATORH ECOCX value multiplied by TOR maximum correction factor (missing for
non-EPA data).
EPATORL ECOCX value multiplied by TOR minimum correction factor (missing for
non-EPA data).
In the concentration summary file, EC, OC, BC, and the above DPM estimates are all referred to
as "SPECIES".
The TOR and TOT (NIOSH) minimum, maximum, and average conversion factors are reported
in Table 3-2 of Chapter 3 "Diesel Paniculate Matter Conversion Factors." The converted values
estimate the diesel particulate matter (DPM) concentration. The applicable conversion factors
depend upon the measurement method (TOR or TOT), whether the location is in the East or
West US, whether the site is urban or rural, and the calendar quarter. For this calculation, sites
with (signed) longitude less than -92° were treated as being in the West US. The Mississippi
river roughly lies along the -92° longitude line. Sites were defined as urban or rural based on the
NSATA assignment for the nearest census tract centroid; this assignment is given by the variable
urbannear in the site summary file.
The sitesummary.final.dbf file contains summary information about the individual sites
(identified by the source and site_id variables). The variables are listed in Table 2-2. Information
includes: the city or county; state; latitude and longitude; first and last measurement dates for
EC, BC, or OC (within the time frame starting January 1, 1994); method; ratios of EC, OC,
sulfate and nitrate to PM2.5; location, distance, location type (urban or rural) and modeled
concentration for the nearest modeled diesel PM2.5 concentration from NATA; location and
modeled concentration for the maximum modeled diesel PM2.5 concentration from NATA
within 30 km, if any; and the dominant source. Note that the maximum modeled value within 30
km is missing if there are no census tract centroids within 30 km. The method variable lists all
methods used for that site (" Aethalometer" applies to all black carbon data, other methods apply
to EC and OC).
The first set of analyses used the 1996 NSATA model predictions consistent with the year 2000
draft of the NONROAD model. The second set of analyses used 1996 NSATA model predictions
consistent with the year 2002 draft of the NONROAD model: The earlier model's ambient and
background non-road components were both multiplied by 0.69 for every census tract, (since the
1996 national modeled NONROAD DPM was reduced by 31 % for the 2002 draft model). The
values of the nearest modeled concentration and of the location and modeled value for the
maximum modeled concentration within 30 km are each given separately for each version of the
NONROAD model. The dominant source is defined for the 2000 NONROAD model only. If the
total on-road modeled DPM is 50 % or greater of the total modeled DPM, the dominant source is
"On-road." If the total on-road modeled DPM is less than 25 % of the total modeled DPM, the
dominant source is "Non-road." Otherwise there is no dominant source.
ICF Consulting 10
-------
The concsummary.final.dbf file contains summary statistics for the daily average concentration
data by source and site, species (BC, EC, OC, ECOCX, ECTOR, ECTORL, ECTORH, ECTOT,
ECTOTH, ECTOTL, EPATOR, EPATORL, EPATORH), year, calendar quarter, and
weekday/weekend. The variables are listed in Table 2-3. Possible values of year are 1994, 1995,
1996, 1997, 1998, 1999, 2000, 2001, and "All" (all years combined). Possible values of quarter
are 1, 2, 3, 4, and "All" (all quarters, i.e., the entire year or years). Possible values of dayofweek
are "Weekday" (Monday to Friday), "Weekend" (Saturday or Sunday), and "All." For example,
overall averages for a site are obtained by considering year = quarter = dayofweek = "All."
Summary statistics are available by year (including "All") and/or by quarter (including "All").
For the weekend/weekday split, separate weekend and weekday summary statistics are reported
by year (including "All") or by quarter (including "All") but not for specific year and quarter
combinations.
ICF Consulting 11
-------
Variable
Date
Source
Site_id
Citycounty
State
EC
OC
BC
ECOCX
ECTOR
ECTORH
ECTORL
ECTOT
ECTOTH
ECTOTL
EPATOR
EPATORH
EPATORL
MinECMDL
MaxECMDL
MinOCMDL
MaxOCMDL
Sulfate
Nitrate
PM25
Latitude
Longitude
Method
Table 2-l^D
Description
Date
Data source (study)
Site identifier
City or County
State
Elemental carbon daily average
Organic carbon daily average
Black carbon daily average
Estimated EC by TOR (EPA data)
Average estimated DPM (TOR data)
Maximum estimated DPM (TOR data)
Minimum estimated DPM (TOR data)
Average estimated DPM (TOT data)
Maximum estimated DPM (TOT data)
Minimum estimated DPM (TOT data)
Average estimated DPM (EPA ECOCX data)
Maximum estimated DPM (EPA ECOCX data)
Minimum estimated DPM (EPA ECOCX data)
Minimum EC MDL for date
Maximum EC MDL for date
Minimum OC MDL for date
Maximum OC MDL for date
Sulfate daily average
Nitrate daily average
PM2.s daily average
Latitude (degrees and fractions of a degree)
Longitude (degrees and fractions of a degree)
List of all measurement methods used on date,
ICF Consulting
12
-------
Variable
Description
Source
Site_id
Citycounty
State
Latitude
Longitude
MinECDate
MaxECDate
MinBCDate
MaxBCDate
MinOCDate
MaxOCDate
MinECMDL
MaxECMDL
MinOCMDL
MaxOCMDL
EC_PM25
OC_PM25
Sulf PM25
Nitr_PM25
Method
Fipsmax
Tractmax
Dist
Data source (study)
Site identifier
City or County
State
Latitude (degrees and fractions of a
degree)
Longitude (degrees and fractions of
a degree)
First date with non-missing EC
Last date with non-missing EC
First date with non-missing BC
Last date with non-missing BEC
First date with non-missing OC
Last date with non-missing OC
Minimum EC MDL for site
Maximum EC MDL for site
Minimum OC MDL for site
Maximum OC MDL for site
Mean EC divided by mean PM2.5
(for days when both were reported)
Mean OC divided by mean PM2.s
(for days when both were reported)
Mean sulfate divided by mean
PM2.5 (for days when both were
reported)
Mean nitrate divided by mean PM2.5
(for days when both were reported)
List of all measurement methods
used at site, separated by
semicolons
FIPS code for census tract centroid
with maximum modeled DPM
within 30 km (2000 NONROAD
model)
Tract ID code for census tract
centroid with maximum modeled
DPM within 30 km (2000
NONROAD model)
Distance (km) to nearest census
tract centroid
ICF Consulting
13
-------
Table 2-2. Site Summary FjkJ^ite^ummary.dbf)
Variable Descrigtioii
Maxconc Maximum modeled DPM within 30
km (2000 NONROAD model)
Concnear Modeled DPM at nearest census
tract centroid (2000 NONROAD
model)
Fipsnear FIPS code at nearest census tract
centroid
Tractnear Tract ID code at nearest census tract
centroid
Urbannear NSATA location type (U = urban,
R = rural) at nearest census tract
centroid
Fipsmax2 FIPS code for census tract centroid
with maximum modeled DPM
within 30 km (2002 NONROAD
model)
Tractmax2 Tract ID code for census tract
centroid with maximum modeled
DPM within 30 km (2002
NONROAD model)
Maxconc2 Maximum modeled DPM within 30
km(2002 NONROAD model)
Concnear2 Modeled DPM at nearest census
tract centroid (2002 NONROAD
model)
Dominant Dominant source: "Non-road,"
"On-Road," or " " (blank).
Consistent with the 2000
NONROAD model.
ICF Consulting
14
-------
Table 2-3. Concentration Summary File (concsummary.dbf)
Variable Description
Source
Site_id
Year
Quarter
DayofWeek
Species
N
Mean
Median
Geommean
Stddev
Minimum
Maximum
PerclO
Perc20
PercSO
Perc40
PercSO
Perc60
PercTO
PercSO
Perc90
Data source (study)
Site identifier
Calendar year or "All"
Calendar quarter or "All"
"Weekday," "Weekend," or "All"
EC, OC, BC, ECOCX, ECTOR,
ECTORL, ECTORH, ECTOT,
ECTOTH, ECTOTL, EPATOR,
EPATORL, or EPATORH
Number of days
Arithmetic mean
Median
Geometric mean
Standard deviation
Minimum
Maximum
10thpercentile
20th percentile
30th percentile
40th percentile
50th percentile
60* percentile
70th percentile
80th percentile
90%ercentile
ICF Consulting
15
-------
CHAPTER 3. DIESEL PARTICIPATE MATTER CONVERSION FACTORS
This part of the effort was primarily carried out by the consulting firm dKC. We compiled
"conversion factors" (CFs) for estimating ambient diesel particulate matter (DPM) based on
elemental carbon (EC), organic carbon and PM2.5 measurements. We attempted to collect
information on CFs for black carbon (BC), but found none in the literature. See below for a
discussion of the recommended treatment of the black carbon data. The CFs were collected from
existing source apportionment and source-receptor model studies.
We received assistance from the following researchers to identify data sources: James Schauer
(University of Wisconsin), Philip Hopke (Clarkston University), Alan Gertler (Desert Research
Institute), and Steve Cadle (GM Research). Overall, we identified and compiled data on CFs
from the following sources:
1. Zheng, Cass, Schauer et al (2002).
Apportionments O/PM2.5 (mass) and organic carbon inPM2.5for 8 SE US sites:
4 urban, 3 rural and 1 suburban. Season: All 4 individually. Dr. Schauer also
provided an elemental carbon breakdown by season (but not by site).
2. Ramadan, Song, and Hopke (2000).
Apportionments of PM (mass) in Phoenix, AZ. Season: Annual Average.
3. Schauer etal (1996).
Apportionments of primary fine organic aerosol and fine particulate mass
concentrations for 4 urban sites in Southern California. Season: Annual Average.
4. Schauer and Cass (2000).
Apportionments of primary fine organic aerosol and fine particulate mass
concentrations for 3 sites in the Central Valley of California: 2 urban and 1 rural.
Season: Winter
5. Watson, Fujita, Chow, Zielinska et al (1998).
NFRAQS. This was the most comprehensive and current analysis of sources of
ambient PM. Two techniques were used to apportion ambient PM: Conventional
CMB and Extended Species CMB. Extended Species CMB breaks down gasoline
vehicle emissions into 3 categories: cold start, hot transient, and high PM emitter
(e.g. a vehicle with visible smoke). PMwas apportioned into total carbon, organic
carbon, elemental carbon andPM2.5. A total of 9 sites were evaluated: 3 urban,
4 rural, one suburban, undone to characterize regional transport. Two sites were
used for the Extended Species CMB analysis: one rural and one urban. For the
Extended Species CMB analysis, a temporal apportionment was done. Season:
Winter
6. Air Improvement Resources (1997).
Summary and analysis of available data on contribution of gasoline powered
vehicles to ambient levels of fine particulate matter. Most of the data was covered
ICF Consulting 16
-------
in Reference #3. Included projections of sources of fine carbonaceous PM. for 4
Southern California sites. Season: Annual Average.
7. Cass(1997).
Summary and analysis of available data on contribution of motor vehicles to
ambient levels offineparticulate matter. Most of the data was covered in
Reference #3. Included projections of sources of elemental carbon for 3 Southern
California sites. Season: Annual Average.
The numbering of these source apportionment study references is arbitrary but is retained here
for consistency with the attached Excel spreadsheet.
Dr. Schauer noted that several important source apportionment studies are currently underway,
and results should start being available in the next 6 months.
We compiled data on conversion factors (CFs) into an Excel spreadsheet provided to EPA. The
spreadsheet contains a page for CFs from each reference. The following information was
compiled for each data point:
Year data were collected
Site evaluated
State data were collected
Type of site: urban, rural, suburban (in some cases more specific types were used,
e.g. rural down valley).
Season data were collected
Ambient Measurement Technique: Thermal optical transmission (TOT, also
referred to as the NIOSH method, as developed by Sunset laboratories) or thermal
optical reflectance (TOR, as developed and applied by Desert Research Institute,
which was also used for the IMPROVE database and the Northern Front Range
study).
PM measurement parameter (organic carbon, elemental carbon, PM2.s)
Concentration apportioned to diesel powered engines
Concentration apportioned to gasoline engine exhaust. (One of the data sources,
Northern Front Range Air Quality Study, had a breakdown of gasoline powered
vehicle emissions into LDHV Cold Start, LDGV hot stabilized (warmed-up vehicle
emissions, and LDGV high PM emitter)
Total concentration
ICF Consulting 17
-------
% of PM from diesel, for each measurement parameter This value was calculated
for each site (in some cases, by season) as the ratio of the CMB estimated diesel
fine particulate matter to the site average total fine particulate matter
concentration. Both values were either reported in the source apportionment study
report or were obtained directly from the researcher.
Multiplicative conversion factor to convert total organic carbon (OC) or total
elemental carbon (EC) concentration to diesel PM2.5 concentration. Conversion
factors (CFs) were calculated by dividing the diesel PM2.5 concentration reported
by the study by the total organic carbon or elemental carbon concentration
reported by the study.
Z factor: % of diesel PM2.5 that is OC or EC. This was calculated by dividing the
reported % of OC or EC that is from diesels by the CF calculated above. The % of
diesel PM2.5 that is OC or EC varied significantly for data based on the two
measurement techniques: TOR or NIOSH.
The Z factor equals the OC or EC diesel PM2.5 concentration divided by the total diesel PM2.5
concentration. This factor can be compared with the measured OC or EC fraction in the
associated diesel PM2.5 source profile. For most of the studies, the diesel source profiles were not
easily obtained. For reference 5 (NFRAQS), the calculated Z factor for EC is compared with the
source profile Z factor in the attached spreadsheet. The calculated and source profile Z factors
were quite close except for the Chatfield and Highlands sites. For consistency with the treatment
of other studies, for all the NFRAQS sites (including Chatfield and Highlands) we used the
multiplicative conversion factor defined above and did not correct for any differences between
the calculated and source profile Z factors.
The originally proposed approach for this project was to compute the conversion factors as the
percentages of diesel in PM2.5 OC or EC (from CMB) divided by the source profile percentage of
OC or EC in diesel PM2 5 (i.e., divided by the source profile Z factor). This alternative approach
gives almost the same results, as can be shown in the reference 5 worksheets, which demonstrate
that the two methods give almost the same conversion factors for the NFRAQS sites, except for
the Chatfield and Highlands sites. The method used here does not require the CMB study to
provide a source apportionment of OC or EC (just the source apportionment for PM25) and does
not need the diesel source profile.
The spreadsheet also contains sheets that compile available data on the following:
Organic Carbon (OC) conversion factors (conversion factors to convert total OC
to diesel PM2 5 concentration).
Elemental Carbon (EC) conversion factors (conversion factors to convert total EC
to diesel PM2 5 concentration).
Fraction of fine particulate mass attributed to diesels.
ICF Consulting 18
-------
Table 3-1 presents the minimum, maximum, and average diesel fraction of PM2s as a function
of:
Urban or rural
Season
East or West US
The reported minimum, maximum, and average values in Table 3-1 are the minima, maxima, and
arithmetic means of the "% of PM from diesel" values across all sites (and seasons, where
applicable) in the given site subset.
Table 3-2 presents the minimum, maximum, and average EC conversion factors as a function of:
Measurement technique
East or West US
Season
Urban or rural
The reported minimum, maximum, and average values in Table 3-2 are the minima, maxima, and
arithmetic means of the EC conversion factors across all sites (and seasons, where applicable) in
the given site subset. For the NIOSH (same as TOT) data collected in the East, the minimum,
maximum, and average conversion factors are all equal. This is because these values were based
only on the Zheng, Cass, Schauer, et al (2002) study. For this project, Dr, Schauer provided EC
summary data from this study averaged over sites, by season. Hence only one value is available
for NIOSH data for each season in the Eastern US.
Table 3-3 presents the minimum, maximum, and average OC conversion factors as a function of:
Urban or rural
Season
East or West US
The reported minimum, maximum, and average values in Table 3-3 are the minima, maxima, and
arithmetic means of the OC conversion factors across all sites (and seasons, where applicable) in
the given site subset.
Black Carbon
Black carbon is measured on an "aethalometer," a measuring instrument developed by Magee
Scientific. The following summary is taken from the "Aethalometer Book," by Hansen (2000).
"The Aethalometer is an instrument that provides a real-time readout of the concentration
of'Black' or 'Elemental' carbon aerosol particles. (BC or EC). These particles ("soot")
are emitted from all types of combustion, most notably from diesel exhaust. 'BC' is
defined by blackness, an optical measurement. The Aethalometer uses an optical
ICF Consulting 19
-------
measurement, and gives a continuous readout. The 'EC' definition is more common. It is
based on a thermal-chemical measurement, an analysis of material collected on a filter
sample for several hours and then sent to a laboratory. Research at Harvard showed that
the Aethalometer BC measurement is directly related and equivalent to the filter-based
EC measurement. In fact, an option in the software allows it to read out in EC units."
More details are given in the full document (Hansen, 2000) and in various references, including
Allen et al (1999), Chow et al (1993), Hansen and Me Murry (1990), and Liousse et al (1991).
On this basis, and because none of the source apportionment studies that we found used black
carbon measurements, we recommend using the same conversion factors to convert BC and EC
concentrations to diesel PM2.5.
Recommendations
The final columns in Tables 3-2 and 3-3 give our recommendations for which minimum,
maximum, and average EC and OC conversion factors should be applied to the database. For
BC, the available data are more limited and we did not find any source apportionment studies
based on BC measurements. OC is not as useful a surrogate for diesel PM as EC because diesel
PM source profiles tend to contain much more EC than OC, on average, and because the diesel
fraction in EC is typically estimated to be much higher than the diesel fraction in OC. For
example, theNFRAQS study (Watson, Fujita, Chow, Zielinska, et al, 1998), determined that EC
contains about 60 % diesel and OC contains about 8 % diesel. Therefore, EPA determined that
only EC data be used for the model to monitor comparisons.
For EC, as shown in Table 3-2, available CF data based on the NIOSH (TOT) method in the East
mainly allows a breakdown by season. There is not enough seasonal data to stratify by location
type. The seasonal stratification in the East is based only on reference 1, which had data for
January, April, July, August, and October only. Thus for the data in the East, the seasonal
stratification is equivalent to a quarterly stratification: Winter = Quarter 1, Spring = Quarter 2,
Summer = Quarter 3, Fall = Quarter 4. For the East US, we recommend using the EC conversion
factors for each daily mean according to the calendar quarter (equivalently, the season). For the
West US, the available data were collected at urban sites in Los Angeles and the season was not
reported. Thus we suggest applying the same factors for all EC data collected in the West,
regardless of location type. For observations based on the TOR method, we suggest that
conversion factors be based on location type (urban or rural), regardless of season. This is due to
an absence of TOR data from non-Winter observations. For BC, the approximate equivalence
between EC and BC suggests using the same conversion factors as for EC.
For OC, separate conversion factors for TOR and TOT data were not computed, although they
would be preferred due to the wide differences in the two measurement methods. For OC, as
shown in Table 3-3, the data in the East is stratified by Urban or Rural location and by season,
but the data in the West is only available for the winter season. The seasonal stratification in the
East is based only on reference 1, which had data for January, April, July, August, and October
only. Thus for the data in the East, the seasonal stratification is equivalent to a quarterly
stratification: Winter = Quarter 1, Spring = Quarter 2, Summer = Quarter 3, Fall = Quarter 4. For
ICF Consulting 20
-------
the East US, we recommend using the OC conversion factors for each daily mean according to
the calendar quarter (equivalently, the season), and whether the site is urban or rural. For the
West US at rural sites, only winter data are available for OC conversion factors, but at rural sites
in the winter, the CF distributions for East and West are quite similar (the factors in the West are
a little lower). On this basis, assuming the same applies to all seasons, we recommend applying
the OC CF's for the seasonal/quarterly totals to the daily averages at West US rural sites in the
first three quarters. For West US rural sites in the winter, i.e., Quarter 1, we recommend using
the corresponding OC CF distribution. For urban sites, the winter CF distributions are very
different between the East and West sitesthe averages differ by a factor of about twoso this
approach is not recommended. Instead, for West US urban sites, we recommend using the
"Urban All" OC CF distribution since the uncertainty range is conservatively wide, the mean is
close to the mean for West US urban sites, and the minimum is the same as the minimum for
West US urban sites in the winter.
ICF Consulting 21
-------
Table 3-1
Summary of Percent of Fine PM Apportioned to Diesels
Location
type
Rural
Rural
Total
Urban
Urban
Total
Grand
Total
Season
Fall
Spring
Summer
Winter
East
or
West
East
East
East
East
West
Winter Total
All
Fall
Spring
Summer
Winter
West
East
East
East
East
West
Winter Total
% Contribution From Diesels
Minimum*
8.9%
11.4%
7.5%
10.0%
2.7%
2.7%
2.7%
12.7%
7.5%
10.1%
9.6%
17.0%
5.3%
5.3%
5.3%
2.7%
Maximum*
10.9%
15.2%
10.9%
13.5%
11.4%
13.5%
15.2%
35.7%
32.0%
29.9%
25.5%
24.1%
12.7%
24.1%
35.7%
35.7%
Average*
9.8%
12.9%
8.7%
12.1%
5.9%
7.8%
9.1%
21.2%
20.7%
19.8%
14.8%
21.2%
9.4%
13.3%
16.6%
13.9%
Notes:
* Minimum, maximum, or average value across all sites of the % contribution from diesel, which
is defined as the ratio of the CMB estimate of diesel PM2.5 divided by the total PM2.5
ICF Consulting
22
-------
Table 3-2
Summary of Calculated Elemental Carbon (EC) Conversion Factors
(Conversion factors to convert total EC to diesel PMi.s concentration)
Ambient
Measurement
TechniquerTOR
or NIOSH
NIOSH
East or
West
East
East
East
East
West
Season
Fall (Q4)
Spring (Q2)
Summer
(Q3)
Winter (Ql)
Unknown
NIOSH Total
TOR
TOR Total
Winter
Location
Type
General
Mixed
Mixed
Mixed
Mixed
Urban
Rural
Urban
Winter Total
Grand Total
MIN*
2.3
2.4
2.1
2.2
1.2
1.2
0.6
0.5
0.5
0.5
0.5
MAX*
2.3
2.4
2.1
2.2
2.4
2.4
1.0
1.0
1.0
1.0
2.4
AVERAGE*
2.3
2.4
2.1
2.2
1.6
2.0
0.8
0.7
0.8
0.8
1.3
Recommended
Conversion
Factors
EAST
X
X
X
X
X
X
WEST
X
X
X
Notes:
* Minimum, maximum, or average value across all sites of the estimated conversion factors.
ICF Consulting
23
-------
Table 3
Summary of Calculated OC Conversion Factors
(Conversion factors to convert total OC to diesel PM2.s concentration)
Location
Type
General
Rural
Rural All
Urban
Urban All
All
Season
Fall
East
or
West
East
Fall Total
Spring
East
Spring Total
Summer
East
Summer Total
Winter
East
West
Winter Total
All
All Total
Fall
West
East
Fall Total
Spring
East
Spring Total
Summer
East
Summer Total
Winter
East
West
Winter Total
Calculated OC Conversion Factor
Minimum*
0.5
0.5
0.5
0.5
0.5
0.5
0.4
0.2
0.2
0.2
0.6
0.6
0.9
0.9
0.9
0.9
1.0
1.0
0.4
0.2
0.2
0.2
0.2
Maximum*
0.5
0.5
1.3
1.3
1.9
1.9
0.5
0.5
0.5
1.9
1.3
1.3
1.3
1.3
2.0
2.0
1.9
1.9
0.8
0.5
0.8
2.0
2.0
Maximum*
0.5
0.5
0.8
0.8
1.4
1.4
0.5
0.4
0.4
0.6
0.9
0.9
1.0
1.0
1.3
1.3
1.5
1.5
0.6
0.4
0.5
0.9
0.8
Recommended
Conversion Factors
East
X
X
X
X
X
X
X
X
West
X
X
X
X
X
Notes:
* Minimum, maximum, or average value across all sites of the estimated conversion factors.
ICF Consulting
24
-------
CHAPTER 4. MODEL-TO-MONITOR COMPARISONS
Using the CFs to convert the monitored EC values to estimated DPM concentrations, we
compared differences between the monitored and modeled DPM values. For the modeled values,
the NSATA predictions for 1996 using ASPEN (and CALPUFF, for the background) were used.
We compared the monitored value to the NSATA prediction at the nearest ASPEN receptor site
(census tract centroid). These comparisons were made for the original NSATA model predictions
consistent with the draft 2000 NONROAD model ("concnear") and for the revised NSATA
model predictions consistent with the draft 2002 NONROAD model ("concnear2") with non-
road model predictions reduced to 69 % of their original value. We also compared the monitored
value to the maximum NSATA prediction within 30 km. These comparisons were also made for
the original NSATA model predictions consistent with the draft 2000 NONROAD model
("maxconc") and for the revised NSATA model predictions consistent with the draft 2002
NONROAD model ("maxconc2")
For the monitored value we separately analyzed the site means of DPM-high, DPM-low, and
DPM-avg, as described above. The site means were computed by averaging all the daily
averages (from January 1, 1994 onward), since there were insufficient data to restrict the
monitored data to the modeling year 1996. The first set of comparisons used EC data collected
by the TOR method only (average, minimum, and maximum DPM values ECTOR, ECTORL,
and ECTORH respectively). The second set of comparisons used EC data collected by the TOT
(NIOSH) method only (average, minimum, and maximum DPM values ECTOT, ECTOTL, and
ECTOTH respectively). The third set of comparisons combined the EC data collected using the
TOR method with the EPA ECOCX value (based on estimated TOR values). This set of average,
minimum, and maximum DPM values are denoted by TOR, TORL, and TORH, respectively.
Each of these model-to-monitor comparisons were applied to the subsets of all locations, urban
locations, and rural locations. Additionally, but only for the predictions consistent with the draft
2000 NONROAD model, we considered the subsets of sites with modeled DPM dominated by
Non-road (at least 75 % non-road) or by On-road (at least 50 % on-road).
We compared the modeled and monitored results using:
Scatterplots of modeled against monitored values that include fitted regression lines
Tables summarizing the regression fits, average difference, and average percentage
difference for the modeled values against the monitored values
Tables summarizing the proportions of modeled values that were within 10%, 25%, 50%,
and 100% of the monitored values
Table 4-1 summarizes each of the regressions. The modeled values are regressed against the
average, maximum, and minimum "DPM monitored" values for each given data subset. The
"DPM monitored" value is the EC value multiplied by the applicable conversion factor to
convert it to the estimated DPM. Scatterplots are shown for a few representative cases. In each
scatterplot, the fitted regression lines of the modeled values against the DPM monitored values.
Each plot shows the modeled values plotted against the average, maximum, and minimum DPM
monitored values for a given data subset. The three regression lines are shown together with the
ICF Consulting 25
-------
Y=X line. The Y=X line has zero intercept and slope 1 and represents the ideal case where
modeled and monitored values agree precisely. For easier comparison, the regressions have been
numbered in Table 4-1, and those numbers are included in the Figure footnotes.
In some cases, all three regression lines intersect and have the same intercept (but different
slopes). This occurs in cases where the same set of three conversion factor values apply at each
monitored value in the data subset. In some cases there are some data points where all three
DPM monitored values are identical, or almost identical. This is attributable to the fact that for
NIOSH data in the East, the minimum, maximum, and average estimated CFs were identical for
each season, and the seasonal values were almost identical, as shown in Table 3-2.
Table 4-1 presents the regression lines in tabular format. For each case is presented: the
intercept, its standard error, the slope, its standard error, and the R squared goodness-of-fit
statistic. Ideally the intercept should be close to zero and the slope should be close to 1. R
squared values close to 1 indicate a good fit of the simple linear model and values close to zero
indicate a poor fit.
Tables 4-2 to 4-4, respectively, show the monitored DPM values and nearest modeled value
(consistent with the draft 2000 NONROAD model) for the TOR sites, for the TOT sites, and for
the combination of the TOR sites and the EPA sites with the ECOCX data. These data were used
for the "concnear" regressions (except that the TOR outlier value mentioned below was excluded
from the regressions).
Table 4-5 presents the mean modeled and monitored values, the mean difference, the mean
percentage difference, and the fractions of modeled values within 10 %, 25 %, 50 %, or 100 % of
the monitored values. Ideally the mean differences and mean percentage differences should be
close to zero and the fractions of modeled values within a small percentage of the modeled
values should be close to one.
Review of preliminary scatterplots showed clearly that for the TOR data, there was one site with
an extremely high modeled value that was greater than 12 ug/m3 compared to monitored DPM
values close to 1 ug/m3. Other TOR modeled to monitor ratios were much lower. This obvious
outlier value from the IMPROVE Washington DC site (WASH1) was removed from all these
analyses although it was retained in the database.
Results
TOR Regressions excluding EPA data
The TOR comparisons excluding the EPA ECOCX data are based on only 15 monitoring sites
and therefore are relatively less representative. As shown in Figure 4-1 (regressions 1,19, and
37), for all sites combined, the regression line of the nearest modeled concentration against the
DPM average is closest to the Y=X line but, as shown in Table 1, regression 1, the R squared
value is 0.22, indicating a poor regression fit. The model tends to overpredict as shown by the
fact that more of the values are above the Y=X line. The same analysis for rural sites only
(regressions 2. 20, and 38) shows a better fit for the regression but a greater tendency to
ICF Consulting 26
-------
overpredict, with slopes close to 2. There are too few urban TOR locations to properly evaluate
this subset. There are also too few Non-road dominated locations (and zero on-road dominated
locations) to compare the modeled and monitored data. Comparing Figure 4-2 to Figure 4-1, the
regression model using the revised NONROAD model appears to fit slightly worse but the
numerical differences are relatively small for most locations. The comparisons between the
maximum modeled value within 30 km and the monitored values show that the monitored value
is significantly over-predicted (e.g., see Figure 4-3).
TOT Regressions
The TOT comparisons are primarily based on the EPA PM speciation data producing a total of
95 monitored values for each of the DPM-minimum, DPM-maximum, and DPM-average. Figure
4 uses all these data. The best results are for the DPM-minimum case, with a R squared value of
0.17, an intercept of 1.52 and a slope of 0.72. Figure 4-4 shows that there are some monitored
values that are significantly over-predicted and some monitored values that are significantly
under-predicted. The results for the urban and rural subsets are similar although the model
performance appears to be a little better for the rural sites. For the Non-road dominated sites, the
regression model fits better than the all sites regression, but the monitored values are
significantly overpredicted (the slope is 1.82 for the DPM-average, regression 58). There are not
enough data points to properly evaluate the subset of On-road dominated sites. Comparing
Figure 4-4 to Figure 4-5, the regression model using the revised NONROAD model appears to fit
slightly better but the numerical differences are relatively small for most locations; the DPM-
minimum predictions give the best model performance. Similarly to TOR, the comparisons
between the maximum modeled value within 30 km and the monitored values show that the
monitored value is significantly over-predicted.
TOR Regressions including EPA data
The TOR comparisons including the EPA ECOCX data are based on 88 monitoring sites. Of the
three monitored DPM values, the best model performance is obtained for the DPM-average
value, regression 127, with a slope of 0.91 and an intercept of 0.77 although the R squared value
is only 0.12. (Figure 4-6). The regression results for the rural and urban subsets are very similar,
but the regression fit is a little better for the rural subset. For the 18 Non-road dominated sites,
the model performance is a bit worse than the performance for all sites. There are insufficiently
many sites to evaluate the subset of on-road dominated sites. The numerical differences between
the 2000 and 2002 models are relatively small for most locations and the model performance is
very similar. (Using the revised model the R squared values are slightly higher, the intercept is
closer to zero, but the slope is further from 1). The comparisons between the maximum modeled
value within 30 km and the monitored values show that the monitored value is significantly over-
predicted.
Differences and percentage differences
The following table extracted from Table 4-5 summarizes the differences and percentage
differences between the nearest modeled value and the monitored values. The ECTOR outlier
value for the IMPROVE Washington DC site is excluded from these calculations.
ICF Consulting 27
-------
Summary of differences between the nearest modeled concentration and the monitored values.
Modeled
Variable1
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
concnear
concnear2
Monitored
Variable2
ECTOR
ECTOR
ECTORH
ECTORH
ECTORL
ECTORL
ECTOT
ECTOT
ECTOTH
ECTOTH
ECTOTL
ECTOTL
TOR
TOR
TORH
TORH
TORL
TORL
N
15
15
15
15
15
15
95
95
95
95
95
95
88
88
88
88
88
88
Mean
Modeled
Value
1.56
1.20
1.56
1.20
1.56
1.20
2.61
2.05
2.61
2.05
2.61
2.05
2.31
1.81
2.31
1.81
2.31
1.81
Mean
Monitored
Value
0.94
0.94
1.16
1.16
0.64
0.64
1.73
1.73
2.10
2.10
1.52
1.52
1.70
1.70
2.23
2.23
1.19
1.19
Mean
Difference
0.63
0.26
0.40
0.04
0.92
0.55
0.88
0.32
0.52
-0.05
1.09
0.52
0.61
0.11
0.08
-0.42
1.12
0.62
Mean
/o
Difference
100
56
62
26
190
126
80
42
61
27
101
58
47
15
13
-12
110
65
Fraction of Modeled
Values Within
10%
0.07
0.07
0.00
0.00
0.13
0.07
0.12
0.11
0.11
0.11
0.09
0.09
0.10
0.17
0.11
0.08
0.10
0.14
25%
0.13
0.13
0.07
0.07
0.40
0.33
0.21
0.37
0.22
0.35
0.17
0.32
0.30
0.30
0.26
0.22
0.26
0.31
50%
0.53
0.47
0.40
0.33
0.47
0.47
0.45
0.53
0.46
0.53
0.43
0.52
0.59
0.59
0.60
0.52
0.41
0.52
100%
0.53
0.60
0.60
0.73
0.53
0.53
0.68
0.77
0.74
0.80
0.63
0.72
0.78
0.85
0.84
0.92
0.65
0.74
Notes:
3. Modeled variable:
concnear Nearest modeled DPM concentration consistent with the draft 2000
NONROAD Model
concnear2 Nearest modeled DPM concentration consistent with the draft 2002
NONROAD Model
4. Monitored variable:
ECTOR EC value multiplied by TOR average correction factor (missing for EC
measured using TOT).
ECTORH EC value multiplied by TOR maximum correction factor (missing for EC
measured using TOT).
ECTORL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOT).
ECTOT EC value multiplied by TOT average correction factor (missing for EC
measured using TOR).
ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC
measured using TOR).
ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOR).
TOR ECOCX value multiplied by TOT average correction factor for EPA data,
ECTOR for TOR data.
TORH ECOCX value multiplied by TOT maximum correction factor for EPA
data, ECTOR for TOR data..
TORL ECOCX value multiplied by TOR minimum correction factor for EPA
data, ECTOR for TOR data.
ICF Consulting
28
-------
Tables 4-2 to 4-4 show the monitored and modeled (nearest tract, DPM modeled values
consistent with the draft 2000 NONROAD model) DPM values. In most cases the modeled
values are within at most a factor of 2 of the monitored values.
Table 4-5 summarizes the differences and percentage differences between all the sets of modeled
and monitored values. The ECTOR outlier value for the IMPROVE Washington DC site is
excluded from these calculations.
For the TOR comparisons excluding the EPA ECOCX data, the best model performance based
on the mean percentage difference and based on the fraction of modeled values within 100 % of
the monitored value is for the DPM-maximum value consistent with the 2002 NONROAD
model. For all 15 sites the mean percentage difference is 26 % and the fraction of modeled
values within 100 % of the monitored value is 73 %. This fraction would have been 69 %
including the outlier. For the 10 rural sites the mean percentage difference is 20 % and the
fraction of modeled values within 100 % of the monitored value is 70 %. For the 5 urban sites
the mean percentage difference is 39 % and the fraction of modeled values within 100 % of the
monitored value is 80 %. Interestingly, this finding is different to the regression analyses which
found the model performance to be slightly worse with the modeled value consistent with the
revised NONROAD model and better using the DPM-average monitored values.
For the TOT comparisons, the best model performance based on the mean percentage difference
and based on the fraction of modeled values within 100 % of the monitored value is also for the
DPM-maximum value consistent with the 2002 NONROAD model. (The results for the subset of
On-road dominated sites are even better, but are based on only 6 monitors) For all 95 sites the
mean percentage difference is 27 % and the fraction of modeled values within 100 % of the
monitored value is 80 %. For the 30 rural sites the mean percentage difference is 16 % and the
fraction of modeled values within 100 % of the monitored value is 80 %. For the 65 urban sites
the mean percentage difference is 32 % and the fraction of modeled values within 100 % of the
monitored value is 80 %.
For the TOR comparisons including the EPA ECOCX data, the best model performance based
on the mean percentage difference and based on the fraction of modeled values within 100 % of
the monitored value is also for the DPM-maximum value using the 2002 NONROAD model. For
all 88 sites the mean percentage difference is -12 % and the fraction of modeled values within
100 % of the monitored value is 92 %. For the 30 rural sites the mean percentage difference is 6
% and the fraction of modeled values within 100 % of the monitored value is 87 %. For the 58
urban sites the mean percentage difference is -14 % and the fraction of modeled values within
100 % of the monitored value is 95 %.
Discussion
The model performance evaluation based on the regression models leads to different conclusions
than the model performance evaluation based on the differences. One primary reason is that the
regression analyses are more influenced by the extreme values. For most purposes the analysis of
the differences is more useful, especially since all of the regression models fitted relatively
poorly (except for those with only 2 data points). The best model performance based on the mean
ICF Consulting 29
-------
percentage difference and based on the fraction of modeled values within 100 % of the
monitored value is for the DPM-maximum value consistent with the 2002 NONROAD model.
The corresponding fractions of modeled values within 100 % of the monitored value are 73 %
for all TOR sites excluding the EPA ECOCX data, 80 % for all TOT sites, and 92 % for all TOR
sites including the EPA ECOCX data. As discussed in Chapter 1, this performance compares
favorably with the model to monitor results for the other pollutants assessed in the NSATA,
except for benzene.
ICF Consulting 30
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Modeled Variable1
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
Monitored Variable2
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
Subset3
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
15
10
5
4
2
2
15
10
5
14
9
5
4
2
2
14
9
5
15
10
5
4
2
2
15
10
5
14
9
5
4
2
2
Intercept
0.80
0.04
1.78
1.99
-2.37
3.71
0.67
0.14
1.39
3.40
2.24
4.91
4.45
7.60
2.23
2.49
1.66
3.57
0.89
0.04
1.78
2.28
-2.37
3.71
0.73
0.14
1.39
3.45
2.24
4.91
4.90
7.60
2.23
Intercept
Standard Error
0.49
0.61
0.74
2.22
0.36
0.45
0.56
1.33
1.95
2.16
2.07
0.94
1.38
1.49
0.48
0.61
0.74
2.12
0.35
0.45
0.56
1.27
1.95
2.16
2.12
Slope
0.81
1.57
0.07
0.39
3.20
-0.64
0.56
1.07
0.04
1.41
2.16
0.75
1.32
0.00
1.98
0.98
1.50
0.51
0.58
1.37
0.05
0.17
2.80
-0.46
0.40
0.94
0.03
1.10
1.89
0.54
0.82
0.00
1.42
Slope
Standard Error
0.42
0.59
0.55
1.20
0.31
0.43
0.41
1.12
1.79
1.60
1.12
0.79
1.26
1.10
0.33
0.51
0.39
0.89
0.24
0.38
0.30
0.84
1.56
1.15
0.89
R Squared
0.22
0.47
0.01
0.05
1.00
1.00
0.20
0.44
0.00
0.12
0.17
0.07
0.41
1.00
0.11
0.17
0.07
0.19
0.47
0.01
0.02
1.00
1.00
0.18
0.44
0.00
0.12
0.17
0.07
0.30
1.00
ICF Consulting
31
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
Modeled Variable1
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
Monitored Variable2
ECTORH
ECTORH
ECTORH
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
Subset3
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
14
9
5
15
10
5
4
2
2
15
10
5
14
9
5
4
2
2
14
9
5
95
30
65
19
7
12
6
2
4
95
30
65
Intercept
2.52
1.66
3.57
0.82
0.04
1.78
2.07
-2.37
3.71
0.69
0.14
1.39
3.40
2.24
4.91
4.57
7.60
2.23
2.49
1.66
3.57
1.74
1.26
2.05
1.60
0.69
2.31
1.48
1.43
1.24
1.37
0.96
1.63
Intercept
Standard Error
0.89
1.38
1.49
0.49
0.61
0.74
2.20
0.35
0.45
0.56
1.32
1.95
2.16
2.09
0.93
1.38
1.49
0.32
0.43
0.43
1.31
2.18
1.98
0.45
0.74
0.23
0.31
0.31
Slope
0.77
1.31
0.37
1.15
2.33
0.10
0.49
4.75
-0.90
0.79
1.59
0.06
2.05
3.20
1.05
1.80
0.00
2.79
1.42
2.23
0.72
0.51
0.54
0.45
1.82
2.16
1.61
0.21
0.47
0.25
0.39
0.43
0.34
Slope
Standard Error
0.59
1.10
0.79
0.61
0.87
0.77
1.71
0.45
0.64
0.58
1.60
2.65
2.25
1.63
1.12
1.87
1.56
0.14
0.23
0.18
0.59
1.38
0.79
0.18
0.26
0.10
0.17
0.13
R Squared
0.12
0.17
0.07
0.21
0.47
0.01
0.04
1.00
1.00
0.19
0.44
0.00
0.12
0.17
0.07
0.38
1.00
0.12
0.17
0.07
0.12
0.16
0.09
0.36
0.33
0.29
0.25
1.00
0.32
0.13
0.19
0.10
ICF Consulting
32
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
99
95
96
97
98
99
Modeled Variable1
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
Monitored Variable2
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
Subset3
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
Intercept
4.67
-0.35
8.85
-4.24
-4.39
27.73
3.39
2.11
3.17
3.45
-0.09
6.37
2.01
1.48
2.32
2.07
2.64
2.65
1.56
1.59
1.36
1.57
1.14
1.83
9.11
1.06
14.11
19.34
1.75
53.90
3.66
2.80
3.53
Intercept
Standard Error
4.84
2.22
7.13
26.43
7.71
40.39
1.46
2.51
3.37
1.54
4.96
0.30
0.40
0.40
1.36
2.85
1.81
0.44
0.74
0.22
0.29
0.29
4.55
2.11
6.62
28.55
10.41
39.14
1.41
2.48
Slope
7.91
5.49
7.73
27.11
9.05
19.83
0.74
2.06
0.75
5.52
3.88
5.38
0.29
0.30
0.26
1.35
0.65
1.30
0.12
0.26
0.15
0.23
0.24
0.20
4.41
3.54
4.15
12.94
3.80
7.31
0.44
1.14
0.43
Slope
Standard Error
2.15
1.22
2.95
11.98
4.87
16.17
0.59
0.88
1.50
0.84
2.05
0.10
0.16
0.13
0.53
1.42
0.63
0.12
0.18
0.07
0.12
0.09
1.56
0.84
2.14
11.05
5.19
13.65
0.40
0.61
R Squared
0.13
0.42
0.10
0.23
0.41
0.13
0.2783
1.0000
0.2660
0.1271
0.4317
0.0981
0.0792
0.1058
0.0598
0.2792
0.0398
0.2986
0.2004
1.0000
0.2537
0.0903
0.1300
0.0666
0.0793
0.3864
0.0564
0.0746
0.0970
0.0279
0.2261
1.0000
0.2002
ICF Consulting
33
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
Modeled Variable1
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
Monitored Variable2
ECTOTH
ECTOTH
ECTOTH
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
TOR
TOR
TOR
TOR
TOR
TOR
Subset3
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
88
30
58
18
9
9
Intercept
6.54
0.90
10.03
1.52
1.01
1.85
1.92
0.59
2.74
1.39
1.14
1.11
1.21
0.77
1.48
1.02
-1.52
4.54
-10.13
-3.85
14.93
3.06
0.83
2.73
0.91
-0.93
3.37
0.77
0.34
1.26
2.04
0.60
2.93
Intercept
Standard Error
3.17
1.46
4.61
0.33
0.44
0.45
1.23
1.50
2.03
0.47
0.74
0.24
0.32
0.33
4.99
2.37
7.38
23.05
5.39
37.64
1.49
2.51
3.47
1.64
5.14
0.48
0.66
0.73
1.65
2.48
2.39
Slope
3.08
2.50
2.89
0.72
0.85
0.62
1.83
2.65
1.52
0.33
0.86
0.40
0.55
0.67
0.47
11.38
7.43
11.25
33.26
10.36
27.43
1.15
3.76
1.20
7.94
5.25
7.83
0.91
1.06
0.68
1.06
1.57
0.82
Slope
Standard Error
1.09
0.58
1.49
0.17
0.30
0.22
0.61
1.09
0.87
0.25
0.34
0.13
0.21
0.16
2.61
1.60
3.55
11.34
3.92
16.14
0.80
1.16
1.82
1.11
2.47
0.27
0.37
0.40
0.83
1.24
1.22
R Squared
0.0796
0.3952
0.0563
0.1573
0.2227
0.1171
0.3485
0.5400
0.2333
0.3061
1.0000
0.4095
0.1719
0.2580
0.1250
0.1696
0.4350
0.1373
0.3360
0.5827
0.2242
0.3415
1.0000
0.3499
0.1701
0.4462
0.1372
0.1177
0.2286
0.0483
0.0920
0.1861
0.0606
ICF Consulting
34
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
Modeled Variable1
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
Monitored Variable2
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
Subset3
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
5
2
3
88
30
58
87
29
58
18
9
9
5
2
3
87
29
58
88
30
58
18
9
9
5
2
3
88
30
58
87
29
58
Intercept
1.82
1.20
8.23
0.63
0.29
1.02
-3.38
0.28
-9.75
-0.95
0.93
-0.16
4.40
1.10
26.79
-2.27
0.20
-6.62
0.79
0.34
1.26
1.94
0.60
2.93
2.54
1.20
8.23
0.63
0.29
1.02
-7.19
0.28
-9.75
Intercept
Standard Error
1.67
1.20
0.36
0.47
0.55
9.21
2.85
15.68
38.73
8.62
58.79
5.46
3.66
6.41
1.99
10.92
0.48
0.66
0.73
1.57
2.48
2.39
1.71
1.20
0.36
0.47
0.55
8.94
2.85
15.68
Slope
-0.03
0.40
-3.52
0.69
0.80
0.52
10.46
3.58
16.33
20.74
4.03
36.94
-0.02
1.75
-11.99
7.35
2.60
11.39
0.68
0.92
0.49
0.88
1.37
0.59
-0.31
0.35
-2.53
0.53
0.70
0.37
9.67
3.13
11.73
Slope
Standard Error
0.84
0.63
0.20
0.26
0.30
5.06
1.57
8.62
19.53
4.30
29.99
2.74
1.93
3.53
1.09
6.00
0.20
0.32
0.29
0.62
1.08
0.88
0.67
0.45
0.15
0.23
0.22
3.74
1.37
6.19
R Squared
0.0006
1.0000
0.9688
0.1228
0.2455
0.0497
0.0478
0.1616
0.0603
0.0659
0.1118
0.1781
0.0000
1.0000
0.9748
0.0486
0.1729
0.0604
0.1183
0.2286
0.0483
0.1103
0.1861
0.0606
0.0667
1.0000
0.9688
0.1272
0.2455
0.0497
0.0728
0.1616
0.0603
ICF Consulting
35
-------
Table 4-1 . Regression models for modeled against monitored DPM concentrations.
Number
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
Modeled Variable1
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
Monitored Variable2
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
Subset3
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
18
9
9
5
2
3
87
29
58
88
30
58
18
9
9
5
2
3
88
30
58
87
29
58
18
9
9
5
2
3
87
29
58
Intercept
-20.88
0.93
-0.16
5.68
1.10
26.79
-4.91
0.20
-6.62
0.75
0.34
1.26
1.98
0.60
2.93
2.01
1.20
8.23
0.62
0.29
1.02
-4.70
0.28
-9.75
-7.20
0.93
-0.16
4.73
1.10
26.79
-3.19
0.20
-6.62
Intercept
Standard Error
35.28
8.62
58.79
5.73
3.66
6.23
1.99
10.92
0.48
0.66
0.73
1.64
2.48
2.39
1.73
1.20
0.36
0.47
0.55
9.19
2.85
15.68
38.11
8.62
58.79
5.66
3.66
6.40
1.99
10.92
Slope
24.86
3.52
26.55
-0.52
1.53
-8.62
6.78
2.27
8.18
1.31
1.57
0.95
1.57
2.32
1.15
-0.19
0.59
-4.96
1.01
1.18
0.73
16.08
5.30
22.99
34.83
5.98
52.01
-0.27
2.59
-16.88
11.29
3.86
16.04
Slope
Standard Error
14.00
3.75
21.56
2.24
1.39
2.61
0.96
4.31
0.38
0.54
0.56
1.19
1.83
1.72
1.25
0.89
0.29
0.39
0.43
7.24
2.32
12.13
27.75
6.37
42.24
4.09
2.72
5.04
1.62
8.45
R Squared
0.1648
0.1118
0.1781
0.0179
1.0000
0.9748
0.0737
0.1729
0.0604
0.1194
0.2286
0.0483
0.0982
0.1861
0.0606
0.0073
1.0000
0.9688
0.1256
0.2455
0.0497
0.0549
0.1616
0.0603
0.0896
0.1118
0.1781
0.0014
1.0000
0.9748
0.0557
0.1729
0.0604
ICF Consulting
36
-------
Notes for Table 4-1:
1. Modeled variable:
concnear Nearest modeled DPM concentration consistent with the draft 2000 NONROAD Model
concnear2 Nearest modeled DPM concentration consistent with the draft 2002 NONROAD Model
maxconc Nearby (within 30 km) maximum modeled DPM concentration consistent with the draft 2000 NONROAD
Model
maxconc2 Nearby (within 30 km) maximum modeled DPM concentration consistent with the draft 2002 NONROAD
Model
2. Monitored
ECTOR
ECTORH
ECTORL
ECTOT
ECTOTH
ECTOTL
TOR
TORH
TORL
3. Subset:
All
Non-road
On-road
variable:
EC value multiplied by TOR
EC value multiplied by TOR
EC value multiplied by TOR
EC value multiplied by TOT
EC value multiplied by TOT
EC value multiplied by TOR
ECOCX value multiplied by
ECOCX value multiplied by
ECOCX value multiplied by
average correction factor (missing for EC measured using TOT).
maximum correction factor (missing for EC measured using TOT).
minimum correction factor (missing for EC measured using TOT).
average correction factor (missing for EC measured using TOR).
maximum correction factor (missing for EC measured using TOR).
minimum correction factor (missing for EC measured using TOR).
TOT average correction factor for EPA data, ECTOR for TOR data.
TOT maximum correction factor for EPA data, ECTOR for TOR data.
TOR minimum correction factor for EPA data, ECTOR for TOR data.
All sites
Sites dominated by Non-road source (at least 75 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
Sites dominated by On-road source (at least 50 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
ICF Consulting
37
-------
Table 4-2. Nearest modeled concentration and ECTOR, ECTORL, and ECTORH monitored values for each site.
City / County1
Oceanville
Seattle
Shenandoah National
Park
Lake Tahoe
Washington DC
Yellowstone National
Park
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
State1
NJ
WA
VA
NV
DC
WY
CO
CO
CO
CO
CO
CO
CO
CO
CO
CO
Nearest
Modeled
Concentration2
2.032
3.2264
0.9612
0.6111
12.263
0.0548
0.7626
1.9723
1.5939
0.7304
0.7953
2.5918
1.6963
1.3913
0.4174
4.5809
Urban or
Rural
R
U
R
R
R
R
R
U
U
R
U
R
U
R
R
R
Subset5
Non-
road
Non-
road
Non-
road
Non-
road
Non-
road
ECTOR4
0.435755
0.75129
0.308284
0.942438
1.01273
0.106637
0.978226
2.711257
0.452586
1.330274
0.820571
0.815005
0.593728
1.065238
0.560907
2.16979
ECTORL4
0.294003
0.533525
0.207999
0.635862
0.683288
0.071948
0.660008
1.925386
0.321402
0.897534
0.582724
0.549883
0.421633
0.718715
0.378443
1.463955
ECTORH4
0.498756
1.045273
0.352855
1.078694
1.159149
0.122055
1.119656
3.772184
0.629685
1.522603
1.141664
0.932837
0.826056
1.219248
0.642002
2.483495
Notes for Table 4-2:
1. City / county and State may appear multiple times if there are several different monitoring sites at that general location.
2. Modeled variable = "concnear" Nearest modeled DPM concentration consistent with the draft 2000 NONROAD Model
3. Subset:
Non-road Sites dominated by Non-road source (at least 75 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
On-road Sites dominated by On-road source (at least 50 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
4. Monitored variable:
ICF Consulting
38
-------
ECTOR EC value multiplied by TOR average correction factor (missing for EC measured using TOT).
ECTORH EC value multiplied by TOR maximum correction factor (missing for EC measured using TOT).
ECTORL EC value multiplied by TOR minimum correction factor (missing for EC measured using TOT).
5. Outlier value not used for statistical comparisons
ICF Consulting 39
-------
Table 4-3. Nearest modeled concentration and TOT, TOTL, and TOTH monitored values for each site.
City / County1
Chicopee
Harrison
Adams
Champaign
Trigg
Tompkins
Washington
Mercer
Noble
Winn Parish
Jefferson
Maricopa
Fresno
Kern
Riverside
Sacramento
San Diego
Santa Clara
Ventura
Adams
Kent
New Castle
District of Columbia
Dade
Hillsborough
DeKalb
Cook
Cook
State1
MA
MA
PA
IL
KY
NY
IN
PA
OH
LA
Alabama
Arizona
California
California
California
California
California
California
California
Colorado
Delaware
Delaware
District of
Columbia
Florida
Florida
Georgia
Illinois
Illinois
Nearest
Modeled
Concentration2
1.7019
4.0016
1.501
1.205
0.8134
0.7298
0.8839
0.9414
0.9718
0.3588
1.9958
1.5009
0.6956
0.7167
2.3167
1.2622
1.6952
2.5023
2.2541
1.743
2.3066
3.903
3.3894
2.4764
1.441
3.3168
10.2885
2.2496
Urban or
Rural
U
U
R
R
R
R
R
R
R
R
U
U
U
U
R
U
U
U
R
R
U
U
U
U
U
R
U
U
Subset5
Non-
road
On-road
On-road
On-road
Non-
road
Non-
road
Tor
0.683716
1.889732
0.838376
0.736247
0.674421
0.510466
0.709542
0.980116
1.208977
0.444273
3.014976
1.35271
1.486594
1.720329
2.196692
1.189164
1.342048
1.71742
1.7441
1.92981
0.958036
1.980288
1.654408
4.237155
1.048591
2.377515
2.232918
1.827492
TOIL4
0.683716
1.889732
0.838376
0.736247
0.674421
0.510466
0.709542
0.980116
1.208977
0.329798
3.014976
1.004159
1.103546
1.277054
1.630673
0.882754
0.996244
1.274895
1.2947
1.432558
0.958036
1.980288
1.654408
4.237155
1.048591
2.377515
2.232918
1.827492
TOTH4
0.683716
1.889732
0.838376
0.736247
0.674421
0.510466
0.709542
0.980116
1.208977
0.645967
3.014976
1.966824
2.16149
2.501338
3.193963
1.72903
1.951321
2.497108
2.5359
2.80592
0.958036
1.980288
1.654408
4.237155
1.048591
2.377515
2.232918
1.827492
ICF Consulting
40
-------
Table 4-3. Nearest modeled concentration and TOT, TOTL, and TOTH monitored values for each site.
City / County1
Cook
Marion
Linn
Polk
Scott
Wyandotte
East Baton Rouge Parish
Baltimore
Baltimore
Hampden
Suffolk
Oakland
Wayne
Wayne
Hennepin
Hennepin
Harrison
Jefferson
St. Louis
Missoula
Douglas
Washoe
Camden
Middlesex
Morris
Union
Bronx
Bronx
Bronx
State1
Illinois
Indiana
Iowa
Iowa
Iowa
Kansas
Louisiana
Maryland
Maryland
Massachusetts
Massachusetts
Michigan
Michigan
Michigan
Minnesota
Minnesota
Mississippi
Missouri
Missouri
Montana
Nebraska
Nevada
New Jersey
New Jersey
New Jersey
New Jersey
New York
New York
New York
Nearest
Modeled
Concentration2
2.3421
2.3757
1.0643
1.0352
1.4764
1.7453
6.7431
2.9017
3.7513
1.7019
4.6792
1.9572
2.0105
2.0734
2.2469
2.3247
0.9887
1.2994
2.1799
0.5105
1.358
3.9817
4.1874
3.3877
2.8348
5.2935
8.0006
7.1404
8.5915
Urban or
Rural
U
U
R
R
U
U
R
U
U
U
U
U
U
U
U
U
U
R
U
U
U
R
U
U
R
U
U
U
U
Subset3
Non-
road
Non-
road
Non-
road
Non-
road
Non-
road
Non-
Tor
1.438204
1.364853
0.71563
0.588565
0.764604
1.10273
1.420214
1.548465
1.104987
0.586525
1.347231
1.67521
1.552197
1.248826
0.493075
0.856137
0.943544
1.218458
1.582243
0.734287
0.673697
1.604714
1.40286
1.404459
0.828597
4.198329
1.77503
2.606724
2.394239
TOIL4
1.438204
1.364853
0.71563
0.43691
0.764604
0.818591
1.420214
1.548465
1.104987
0.586525
1.347231
1.67521
1.552197
1.248826
0.366025
0.635537
0.943544
1.218458
1.582243
0.545084
0.500106
1.191229
1.40286
1.404459
0.828597
4.198329
1.77503
2.606724
2.394239
TOTH4
1.438204
1.364853
0.71563
0.855767
0.764604
1.603357
1.420214
1.548465
1.104987
0.586525
1.347231
1.67521
1.552197
1.248826
0.716925
1.244813
0.943544
1.218458
1.582243
1.067644
0.979547
2.333234
1.40286
1.404459
0.828597
4.198329
1.77503
2.606724
2.394239
ICF Consulting
41
-------
Table 4-3. Nearest modeled concentration and TOT, TOTL, and TOTH monitored values for each site.
City / County1
Essex
Monroe
Queens
Steuben
Mecklenburg
Burleigh
Cass
Cuyahoga
Tulsa
Multnomah
Adams
Allegheny
Allegheny
Philadelphia
Washington
Westmoreland
Charleston
Shelby
Dallas
El Paso
Harris
Salt Lake
Utah
Chittenden
Richmond
King
State1
New York
New York
New York
New York
North Carolina
North Dakota
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
South Carolina
Tennessee
Texas
Texas
Texas
Utah
Utah
Vermont
Virginia
Washington
Nearest
Modeled
Concentration2
0.2218
1.7872
6.6119
0.5936
2.2706
0.5387
0.6603
6.9579
1.4889
2.085
1.501
2.7774
1.9699
4.9986
1.1656
1.6285
0.893
3.0845
3.3398
0.8152
2.3903
1.5764
0.606
1.855
1.5432
3.6513
Urban or
Rural
R
R
U
R
U
U
R
R
U
U
R
U
U
U
R
U
U
U
R
U
U
U
R
U
U
R
Subset3
road
On-road
Non-
road
Non-
road
Non-
road
Non-
road
Non-
road
Non-
road
Non-
road
Tor
0.383065
0.751336
1.624842
0.444674
1.352154
0.344482
0.591956
2.434012
0.767198
1.14655
0.571485
1.832849
1.365899
1.769076
0.856576
1.588581
0.853549
2.99937
0.843332
0.850771
0.482679
1.292407
0.886836
0.799992
1.145547
1.493096
TOIL4
0.383065
0.751336
1.624842
0.444674
1.352154
0.25572
0.439428
2.434012
0.569515
0.85112
0.571485
1.832849
1.365899
1.769076
0.856576
1.588581
0.853549
2.99937
0.626032
0.631554
0.358308
0.959394
0.658326
0.799992
1.145547
1.108372
TOTH4
0.383065
0.751336
1.624842
0.444674
1.352154
0.500872
0.860697
2.434012
1.115497
1.66707
0.571485
1.832849
1.365899
1.769076
0.856576
1.588581
0.853549
2.99937
1.226195
1.23701
0.701809
1.879144
1.289449
0.799992
1.145547
2.170944
ICF Consulting
42
-------
Table 4-3. Nearest modeled concentration and TOT, TOTL, and TOTH monitored values for each site.
City / County1
King
Milwaukee
Anaheim
Burbank
Fontana
Huntington Park
Central LA
Long Beach
Pico Rivera
Rubidoux
Phoenix
State1
Washington
Wisconsin
Puerto Rico
CA
CA
CA
CA
CA
CA
CA
CA
AZ
Nearest
Modeled
Concentration2
3.2264
2.3524
1.1811
3.5344
2.6058
2.0988
4.6059
3.0531
11.744
2.4915
2.5544
2.5682
Urban or
Rural
U
U
U
U
U
R
U
R
U
U
U
U
Subset3
Non-
road
On-road
On-road
Non-
road
Non-
road
Tor
1.008721
1.306485
5.036414
3.753411
5.178159
5.047655
7.390285
5.754731
4.134551
7.097502
5.597411
1.779664
TOIL4
0.748805
1.306485
5.036414
2.786274
3.843909
3.747032
5.48604
4.271917
3.069207
5.268698
4.155133
1.3211
TOTH4
1.466668
1.306485
5.036414
5.457413
7.528979
7.339229
10.74538
8.367309
6.011587
10.31968
8.138567
2.58761
Notes for Table 4-3
1. City / county and State may appear multiple times if there are several different monitoring sites at that general location.
2. Modeled variable = "concnear" Nearest modeled DPM concentration consistent with the draft 2000 NONROAD Model
3. Subset:
Non-road Sites dominated by Non-road source (at least 75 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
On-road Sites dominated by On-road source (at least 50 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
4. Monitored variable:
ECTOT EC value multiplied by TOT average correction factor (missing for EC measured using TOR).
ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC measured using TOR).
ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC measured using TOR).
5. Outlier value not used for statistical comparisons
ICF Consulting
43
-------
Table 4-4. Nearest modeled concentration and TOR, TORL, and TORN monitored values for each site.
City / County1
Jefferson
Maricopa
Fresno
Kern
Riverside
Sacramento
San Diego
Santa Clara
Ventura
Adams
Kent
New Castle
District of Columbia
Dade
Hillsborough
DeKalb
Cook
Cook
Marion
Linn
Polk
Scott
Wyandotte
East Baton Rouge Parish
Baltimore
Baltimore
Hampden
Suffolk
Oakland
Wayne
Wayne
Hennepin
State1
Alabama
Arizona
California
California
California
California
California
California
California
Colorado
Delaware
Delaware
District of Columbia
Florida
Florida
Georgia
Illinois
Illinois
Indiana
Iowa
Iowa
Iowa
Kansas
Louisiana
Maryland
Maryland
Massachusetts
Massachusetts
Michigan
Michigan
Michigan
Minnesota
Nearest
Modeled
Concentration2
1.9958
1.5009
0.6956
0.7167
2.3167
1 .2622
1 .6952
2.5023
2.2541
1.743
2.3066
3.903
3.3894
2.4764
1.441
3.3168
2.2496
2.3421
2.3757
1 .0643
1.0352
1 .4764
1.7453
6.7431
2.9017
3.7513
1.7019
4.6792
1.9572
2.0105
2.0734
2.3247
Urban or
Rural
U
U
U
U
R
U
U
U
R
R
U
U
U
U
U
R
U
U
U
R
R
U
U
R
U
U
U
U
U
U
U
U
Subset5
On-road
On-road
On-road
Non-road
Non-road
TOR4
2.87283
1 .656433
2.112493
2.241741
3.220693
1 .805432
1.903438
1 .849625
2.626535
2.508408
1 .823765
2.255988
1.981439
2.273114
1 .564298
2.812688
1.834895
1 .264669
1.8532
1.331275
1.576618
1.253591
2.237625
1 .658044
1 .886666
1.90302
0.898909
1 .873434
1.807817
1.792132
2.017959
1.603413
TORL4
2.040126
1.176308
1.500176
1.591961
2.172998
1.282118
1.351717
1.313502
1.77212
1 .69242
1.295138
1 .602079
1.407109
1.614241
1.110878
1.897717
1.303041
0.898098
1.316041
0.89821
1 .063742
0.890231
1.589038
1.11868
1.339806
1.35142
0.638356
1.33041
1.283812
1 .272673
1 .433043
1.138655
TORN4
3.996981
2.304603
2.93912
3.118944
3.686336
2.511905
2.648261
2.573392
3.006275
2.871069
2.537412
3.138766
2.756785
3.162594
2.176414
3.219341
2.552898
1.75954
2.578366
1.523748
1.804563
1.744127
3.113217
1.897761
2.624927
2.64768
1 .250656
2.606518
2.515224
2.493401
2.807595
2.230835
ICF Consulting
44
-------
Table 4-4. Nearest modeled concentration and TOR, TORL, and TORN monitored values for each site.
City / County1
Harrison
Jefferson
St. Louis
Missoula
Douglas
Washoe
Camden
Middlesex
Morris
Union
Bronx
Bronx
Essex
Monroe
Queens
Steuben
Mecklenburg
Burleigh
Cass
Cuyahoga
Tulsa
Multnomah
Adams
Allegheny
Allegheny
Philadelphia
Washington
Westmoreland
Charleston
Shelby
Dallas
El Paso
Harris
State1
Mississippi
Missouri
Missouri
Montana
Nebraska
Nevada
New Jersey
New Jersey
New Jersey
New Jersey
New York
New York
New York
New York
New York
New York
North Carolina
North Dakota
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
South Carolina
Tennessee
Texas
Texas
Texas
Nearest
Modeled
Concentration2
0.9887
1 .2994
2.1799
0.5105
1.358
3.9817
4.1874
3.3877
2.8348
5.2935
7.1404
8.5915
0.2218
1.7872
6.6119
0.5936
2.2706
0.5387
0.6603
6.9579
1.4889
2.085
1.501
2.7774
1.9699
4.9986
1.1656
1 .6285
0.893
3.0845
3.3398
0.8152
2.3903
Urban or
Rural
U
R
U
U
U
R
U
U
R
U
U
U
R
R
U
R
U
U
R
R
U
U
R
U
U
U
R
U
U
U
R
U
U
Subset3
Non-road
Non-road
Non-road
Non-road
On-road
Non-road
Non-road
Non-road
Non-road
Non-road
Non-road
TOR4
1.573738
2.274006
1 .868836
1 .682974
1.43152
2.809452
1.917837
1.733885
1.798739
2.866781
2.212477
1.832909
1.102289
1 .459265
1.582166
1 .223072
1.98954
0.96715
1.512952
2.637039
1.858217
1.8004
1.987328
2.111918
1 .946273
1.839106
2.085923
1 .963036
1.614335
2.415763
1.44195
1.037519
0.88649
TORL4
1.117582
1 .534269
1.327144
1.195156
1.016587
1.895534
1.361942
1.23131
1.213607
2.03583
1.571179
1.301631
0.743713
0.984564
1.123567
0.825205
1.412862
0.686817
1.020787
1.779207
1.319603
1 .278545
1 .340848
1 .499768
1.382136
1 .306032
1.40737
1 .39404
1.146412
1.715542
0.972882
0.736789
0.629536
TORN4
2.189548
2.602778
2.600119
2.341529
1.99168
3.215638
2.668295
2.412362
2.058798
3.988565
3.078228
2.550134
1.261656
1 .670243
2.201275
1.399902
2.768056
1 .3456
1.731692
3.018298
2.585345
2.504904
2.274653
2.93832
2.707858
2.558756
2.387502
2.73118
2.246032
3.361061
1 .650425
1.443505
1.233377
ICF Consulting
45
-------
Table 4-4. Nearest modeled concentration and TOR, TORL, and TORN monitored values for each site.
City / County1
Salt Lake
Utah
Chittenden
Richmond
King
King
Milwaukee
Oceanville
Seattle
Shenandoah National Park
Lake Tahoe
Washington DC
Yellowstone National Park
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
Denver
State1
Utah
Utah
Vermont
Virginia
Washington
Washington
Wisconsin
Puerto Rico
NJ
WA
VA
NV
DC
WY
CO
CO
CO
CO
CO
CO
CO
CO
CO
CO
Nearest
Modeled
Concentration2
1 .5764
0.606
1.855
1 .5432
3.6513
3.2264
2.3524
1.1811
2.032
3.2264
0.9612
0.6111
12.2635
0.0548
0.7626
1.9723
1.5939
0.7304
0.7953
2.5918
1 .6963
1.3913
0.4174
4.5809
Urban or
Rural
U
R
U
U
R
U
U
U
R
U
R
R
R
R
R
U
U
R
U
R
U
R
R
R
Subset3
Non-road
Non-road
On-road
Non-road
Non-road
Non-road
Non-road
Non-road
TOR4
1 .679742
1.790894
1.397072
2.093134
1.518021
0.958905
1 .646925
2.454399
0.435755
0.75129
0.308284
0.942438
1.01273
0.106637
0.978226
2.711257
0.452586
1 .330274
0.820571
0.815005
0.593728
1 .065238
0.560907
2.16979
TORL4
1.19286
1.208314
0.992123
1 .486429
1 .024207
0.680962
1.169555
1 .742979
0.294003
0.533525
0.207999
0.635862
0.683288
0.071948
0.660008
1 .925386
0.321402
0.897534
0.582724
0.549883
0.421633
0.718715
0.378443
1 .463955
TORN4
2.337033
2.049819
1.943752
2.912187
1.737494
1.334129
2.291374
3.414816
0.498756
1.045273
0.352855
1.078694
1.159149
0.122055
1.119656
3.772184
0.629685
1 .522603
1.141664
0.932837
0.826056
1.219248
0.642002
2.483495
Notes for Table 4-4:
1. City / county and State may appear multiple times if there are several different monitoring sites at that general location.
2. Modeled variable = "concnear" Nearest modeled DPM concentration consistent with the draft 2000 NONROAD Model
3. Subset:
Non-road Sites dominated by Non-road source (at least 75 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
On-road Sites dominated by On-road source (at least 50 % of modeled DPM consistent with the draft 2000 NONROAD
Model)
ICF Consulting
46
-------
4. Monitored variable:
TOR ECOCX value multiplied by TOT average correction factor for EPA data, ECTOR for TOR data.
TORH ECOCX value multiplied by TOT maximum correction factor for EPA data, ECTOR for TOR data.
TORL ECOCX value multiplied by TOR minimum correction factor for EPA data, ECTOR for TOR data.
5. Outlier value not used for statistical comparisons
ICF Consulting 47
-------
Table 4-5. Summary of differences and percentage differences between modeled ac
Modeled Variable1
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
Monitored Variable2
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTOR
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
Subset3
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
N
15
10
5
4
2
2
15
10
5
14
9
5
4
2
2
14
9
5
15
10
5
4
2
2
15
10
5
14
9
5
4
Mean
Modeled
Value
1.56
1.41
1.86
2.64
2.67
2.60
1.20
1.07
1.44
4.80
4.30
5.71
6.63
7.60
5.66
3.46
3.10
4.11
1.56
1.41
1.86
2.64
2.67
2.60
1.20
1.07
1.44
4.80
4.30
5.71
6.63
Mean
Monitored
Value
0.94
0.87
1.07
1.65
1.57
1.73
0.94
0.87
1.07
1.00
0.96
1.07
1.65
1.57
1.73
1.00
0.96
1.07
1.16
1.00
1.48
2.11
1.80
2.41
1.16
1.00
1.48
1.23
1.09
1.48
2.11
ainst monitored DPM concentrations.
Mean
Difference
0.63
0.54
0.79
0.98
1.10
0.87
0.26
0.20
0.37
3.81
3.35
4.64
4.98
6.03
3.93
2.47
2.15
3.05
0.40
0.42
0.37
0.53
0.87
0.19
0.04
0.08
-0.04
3.57
3.21
4.22
4.53
Mean % Difference
100
76
147
98
45
151
56
38
93
516
416
696
376
464
288
346
275
473
62
54
78
53
26
80
26
20
39
394
351
472
286
Fraction of Modeled Values Within
10%
0.07
0.00
0.20
0.00
0.00
0.00
0.07
0.10
0.00
0.07
0.11
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
25%
0.13
0.10
0.20
0.25
0.50
0.00
0.13
0.10
0.20
0.07
0.11
0.00
0.00
0.00
0.00
0.07
0.11
0.00
0.07
0.10
0.00
0.00
0.00
0.00
0.07
0.10
0.00
0.07
0.11
0.00
0.00
50%
0.53
0.60
0.40
0.50
0.50
0.50
0.47
0.50
0.40
0.07
0.11
0.00
0.00
0.00
0.00
0.21
0.33
0.00
0.40
0.40
0.40
0.50
0.50
0.50
0.33
0.40
0.20
0.14
0.22
0.00
0.00
100%
0.53
0.60
0.40
0.50
0.50
0.50
0.60
0.70
0.40
0.21
0.33
0.00
0.00
0.00
0.00
0.36
0.33
0.40
0.60
0.70
0.40
0.75
1.00
0.50
0.73
0.70
0.80
0.29
0.33
0.20
0.00
ICF Consulting
48
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
Monitored Variable2
ECTORH
ECTORH
ECTORH
ECTORH
ECTORH
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTORL
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
Subset3
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
Location
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
2
2
14
9
5
15
10
5
4
2
2
15
10
5
14
9
5
4
2
2
14
9
5
95
30
65
19
7
12
6
2
4
Mean
Modeled
Value
7.60
5.66
3.46
3.10
4.11
1.56
1.41
1.86
2.64
2.67
2.60
1.20
1.07
1.44
4.80
4.30
5.71
6.63
7.60
5.66
3.46
3.10
4.11
2.61
1.99
2.90
5.17
3.87
5.92
1.90
2.02
1.84
Mean I Mean
Monitoredpifference
Value
1.80
2.41
1.23
1.09
1.48
0.64
0.59
0.76
1.15
1.06
1.23
0.64
0.59
0.76
0.69
0.65
0.76
1.15
1.06
1.23
0.69
0.65
0.76
1.73
1.36
1.90
1.96
1.47
2.25
1.97
1.25
2.33
5.80
3.25
2.23
2.01
2.63
0.92
0.83
1.10
1.49
1.61
1.37
0.55
0.49
0.68
4.12
3.66
4.95
5.48
6.54
4.43
2.78
2.46
3.35
0.88
0.63
1.00
3.20
2.39
3.68
-0.07
0.77
-0.49
Mean % Difference
392
179
258
228
312
190
161
248
184
114
254
126
104
171
792
665
1021
591
735
446
545
456
707
80
66
86
171
164
175
28
84
1
Fraction of Modeled Values Within
10%
0.00
0.00
0.07
0.11
0.00
0.13
0.10
0.20
0.25
0.00
0.50
0.07
0.00
0.20
0.00
0.00
0.00
0.00
0.00
0.00
0.07
0.11
0.00
0.12
0.13
0.11
0.11
0.14
0.08
0.00
0.00
0.00
25%
0.00
0.00
0.14
0.22
0.00
0.40
0.50
0.20
0.50
0.50
0.50
0.33
0.30
0.40
0.00
0.00
0.00
0.00
0.00
0.00
0.07
0.11
0.00
0.21
0.30
0.17
0.16
0.29
0.08
0.00
0.00
0.00
50%
0.00
0.00
0.36
0.33
0.40
0.47
0.50
0.40
0.50
0.50
0.50
0.47
0.50
0.40
0.07
0.11
0.00
0.00
0.00
0.00
0.07
0.11
0.00
0.45
0.60
0.38
0.26
0.29
0.25
0.50
0.50
0.50
100%
0.00
0.00
0.36
0.33
0.40
0.53
0.60
0.40
0.50
0.50
0.50
0.53
0.60
0.40
0.07
0.11
0.00
0.00
0.00
0.00
0.14
0.22
0.00
0.68
0.73
0.66
0.32
0.29
0.33
0.83
0.50
1.00
ICF Consulting
49
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
Monitored Variable2
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOT
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
Subset3
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
On-road Urban
All kll
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
N
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
Mean
Modeled
Value
2.05
1.55
2.28
18.36
7.11
23.55
48.98
8.95
72.34
4.84
4.68
4.92
12.99
5.17
16.60
2.61
1.99
2.90
5.17
3.87
5.92
1.90
2.02
1.84
2.05
1.55
2.28
18.36
7.11
23.55
48.98
8.95
Mean I Mean
Monitoredpifference
Value
1.73
1.36
1.90
1.73
1.36
1.90
1.96
1.47
2.25
1.97
1.25
2.33
1.73
1.36
1.90
2.10
1.71
2.28
2.29
1.89
2.52
2.71
1.64
3.24
2.10
1.71
2.28
2.10
1.71
2.28
2.29
1.89
0.32
0.19
0.37
16.63
5.76
21.65
47.02
7.48
70.09
2.87
3.43
2.59
11.26
3.81
14.70
0.52
0.28
0.63
2.88
1.98
3.40
-0.81
0.38
-1.40
-0.05
-0.16
0.00
16.26
5.41
21.27
46.69
7.06
Mean % Difference
42
30
47
918
423
1146
2142
489
3106
214
307
168
626
284
783
61
48
67
138
116
151
10
63
-17
27
16
32
832
369
1046
2051
376
Fraction of Modeled Values Within
10%
0.11
0.10
0.11
0.02
0.03
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.02
0.00
0.03
0.11
0.07
0.12
0.16
0.00
0.25
0.00
0.00
0.00
0.11
0.10
0.11
0.03
0.00
0.05
0.00
0.00
25%
0.37
0.40
0.35
0.04
0.03
0.05
0.00
0.00
0.00
0.17
0.00
0.25
0.09
0.13
0.08
0.22
0.27
0.20
0.21
0.14
0.25
0.33
0.50
0.25
0.35
0.40
0.32
0.05
0.03
0.06
0.05
0.14
50%
0.53
0.63
0.48
0.08
0.10
0.08
0.00
0.00
0.00
0.33
0.00
0.50
0.12
0.13
0.11
0.46
0.57
0.42
0.32
0.29
0.33
0.33
0.50
0.25
0.53
0.60
0.49
0.12
0.17
0.09
0.05
0.14
100%
0.77
0.80
0.75
0.12
0.13
0.11
0.05
0.14
0.00
0.33
0.00
0.50
0.21
0.30
0.17
0.74
0.80
0.71
0.47
0.57
0.42
0.83
0.50
1.00
0.80
0.80
0.80
0.19
0.27
0.15
0.16
0.43
ICF Consulting
50
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
Monitored Variable2
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTH
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
ECTOTL
TOR
Subset3
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
Location
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
On-road Urban
All kll
All Rural
All Urban
All hi
N
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
95
30
65
19
7
12
6
2
4
95
30
65
88
Mean
Modeled
Value
72.34
4.84
4.68
4.92
12.99
5.17
16.60
2.61
1.99
2.90
5.17
3.87
5.92
1.90
2.02
1.84
2.05
1.55
2.28
18.36
7.11
23.55
48.98
8.95
72.34
4.84
4.68
4.92
12.99
5.17
16.60
2.31
Mean I Mean
Monitoredpifference
Value
2.52
2.71
1.64
3.24
2.10
1.71
2.28
1.52
1.16
1.69
1.78
1.24
2.09
1.55
1.02
1.81
1.52
1.16
1.69
1.52
1.16
1.69
1.78
1.24
2.09
1.55
1.02
1.81
1.52
1.16
1.69
1.70
69.81
2.13
3.03
1.68
10.90
3.46
14.33
1.09
0.83
1.21
3.39
2.63
3.83
0.35
1.00
0.02
0.52
0.39
0.59
16.84
5.95
21.86
47.21
7.71
70.25
3.29
3.65
3.10
11.47
4.01
14.91
0.61
Mean % Difference
3027
169
256
125
564
245
712
101
86
108
207
219
201
49
106
20
58
46
64
1014
484
1258
2244
615
3194
265
363
216
694
329
863
47
Fraction of Modeled Values Within
10%
0.00
0.17
0.00
0.25
0.02
0.03
0.02
0.09
0.13
0.08
0.05
0.00
0.08
0.00
0.00
0.00
0.09
0.10
0.09
0.02
0.00
0.03
0.00
0.00
0.00
0.17
0.00
0.25
0.00
0.00
0.00
0.10
25%
0.00
0.17
0.00
0.25
0.08
0.17
0.05
0.17
0.27
0.12
0.11
0.14
0.08
0.00
0.00
0.00
0.32
0.43
0.26
0.02
0.00
0.03
0.00
0.00
0.00
0.17
0.00
0.25
0.07
0.10
0.06
0.30
50%
0.00
0.33
0.00
0.50
0.18
0.27
0.14
0.43
0.57
0.37
0.16
0.14
0.17
0.33
0.00
0.50
0.52
0.63
0.46
0.05
0.10
0.03
0.00
0.00
0.00
0.17
0.00
0.25
0.11
0.10
0.11
0.59
100%
0.00
0.33
0.00
0.50
0.27
0.37
0.23
0.63
0.70
0.60
0.26
0.29
0.25
0.83
0.50
1.00
0.72
0.73
0.71
0.11
0.10
0.11
0.00
0.00
0.00
0.33
0.00
0.50
0.16
0.17
0.15
0.78
ICF Consulting
51
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
Monitored Variable2
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TOR
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
Subset3
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
Location
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
N
30
58
18
9
9
5
2
3
88
30
58
87
29
58
18
9
9
5
2
3
87
29
58
88
30
58
18
9
9
5
2
3
Mean
Modeled
Value
2.04
2.44
4.01
3.60
4.41
1.76
2.02
1.58
1.81
1.57
1.93
14.57
6.20
18.75
37.69
8.65
66.73
4.36
4.68
4.16
10.34
4.51
13.26
2.31
2.04
2.44
4.01
3.60
4.41
1.76
2.02
1.58
Mean I Mean
Monitoredpifference
Value
1.60
1.75
1.86
1.91
1.81
1.95
2.04
1.89
1.70
1.60
1.75
1.72
1.65
1.75
1.86
1.91
1.81
1.95
2.04
1.89
1.72
1.65
1.75
2.23
1.83
2.43
2.36
2.19
2.52
2.51
2.34
2.63
0.44
0.70
2.14
1.69
2.60
-0.19
-0.02
-0.31
0.11
-0.03
0.19
12.85
4.55
17.01
35.83
6.74
64.92
2.41
2.63
2.27
8.63
2.85
11.51
0.08
0.21
0.01
1.65
1.41
1.89
-0.75
-0.32
-1.05
Mean % Difference
38
51
128
87
168
-5
4
-12
15
7
19
721
296
933
1802
365
3239
134
134
135
484
189
632
13
21
8
78
64
93
-26
-9
-37
Fraction of Modeled Values Within
10%
0.00
0.16
0.00
0.00
0.00
0.00
0.00
0.00
0.17
0.13
0.19
0.05
0.07
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.06
0.07
0.05
0.11
0.07
0.14
0.06
0.00
0.11
0.40
0.50
0.33
25%
0.20
0.34
0.06
0.11
0.00
0.60
1.00
0.33
0.30
0.17
0.36
0.07
0.10
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.16
0.28
0.10
0.26
0.13
0.33
0.11
0.11
0.11
0.40
0.50
0.33
50%
0.57
0.60
0.33
0.33
0.33
0.80
1.00
0.67
0.59
0.50
0.64
0.17
0.24
0.14
0.11
0.22
0.00
0.20
0.00
0.33
0.23
0.38
0.16
0.60
0.47
0.67
0.33
0.33
0.33
0.80
1.00
0.67
100%
0.73
0.81
0.44
0.44
0.44
1.00
1.00
1.00
0.85
0.87
0.84
0.25
0.38
0.19
0.11
0.22
0.00
0.20
0.00
0.33
0.40
0.55
0.33
0.84
0.77
0.88
0.50
0.56
0.44
1.00
1.00
1.00
ICF Consulting
52
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear
concnear2
concnear2
concnear2
maxconc
maxconc
maxconc
maxconc
maxconc
Monitored Variable2
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORN
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
TORL
Subset3
All
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
Location
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
On-road Urban
All Ull
All
All
All
All
All
Non-road
Non-road
Non-road
On-road
On-road
On-road
All
All
All
All
All
All
Non-road
Non-road
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
Urban
All
Rural
N
88
30
58
87
29
58
18
9
9
5
2
3
87
29
58
88
30
58
18
9
9
5
2
3
88
30
58
87
29
58
18
9
Mean
Modeled
Value
1.81
1.57
1.93
14.57
6.20
18.75
37.69
8.65
66.73
4.36
4.68
4.16
10.34
4.51
13.26
2.31
2.04
2.44
4.01
3.60
4.41
1.76
2.02
1.58
1.81
1.57
1.93
14.57
6.20
18.75
37.69
8.65
Mean I Mean
Monitoredpifference
Value
2.23
1.83
2.43
2.25
1.89
2.43
2.36
2.19
2.52
2.51
2.34
2.63
2.25
1.89
2.43
1.19
1.08
1.24
1.29
1.29
1.29
1.36
1.38
1.34
1.19
1.08
1.24
1.20
1.12
1.24
1.29
1.29
-0.42
-0.27
-0.50
12.32
4.31
16.32
35.33
6.46
64.21
1.85
2.34
1.53
8.09
2.61
10.83
1.12
0.96
1.20
2.72
2.31
3.12
0.40
0.64
0.24
0.62
0.49
0.69
13.37
5.08
17.51
36.40
7.36
Mean % Difference
-12
-6
-14
510
246
643
1303
306
2300
83
104
69
335
152
426
110
105
112
228
178
278
36
54
24
65
59
68
1066
487
1355
2596
589
Fraction of Modeled Values Within
10%
0.08
0.07
0.09
0.06
0.07
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.08
0.17
0.03
0.10
0.13
0.09
0.11
0.11
0.11
0.00
0.00
0.00
0.14
0.10
0.16
0.01
0.03
0.00
0.06
0.11
25%
0.22
0.17
0.24
0.14
0.17
0.12
0.06
0.11
0.00
0.00
0.00
0.00
0.18
0.28
0.14
0.26
0.40
0.19
0.22
0.22
0.22
0.00
0.00
0.00
0.31
0.30
0.31
0.03
0.03
0.03
0.06
0.11
50%
0.52
0.40
0.59
0.22
0.34
0.16
0.11
0.22
0.00
0.00
0.00
0.00
0.32
0.41
0.28
0.41
0.53
0.34
0.28
0.33
0.22
0.40
0.50
0.33
0.52
0.50
0.53
0.07
0.07
0.07
0.06
0.11
100%
0.92
0.87
0.95
0.36
0.45
0.31
0.17
0.22
0.11
0.60
0.50
0.67
0.51
0.62
0.45
0.65
0.67
0.64
0.33
0.33
0.33
0.80
1.00
0.67
0.74
0.73
0.74
0.13
0.14
0.12
0.06
0.11
ICF Consulting
-------
Table 4-5. Summary of differences and percentage differences between modeled against monitored DPM concentrations.
Modeled Variable1
maxconc
maxconc
maxconc
maxconc
maxconc2
maxconc2
maxconc2
Monitored Variable2
TORL
TORL
TORL
TORL
TORL
TORL
TORL
Subset3
Non-road
On-road
On-road
On-road
All
All
All
Location
Urban
All
Rural
Urban
All
Rural
Urban
N
9
5
2
3
87
29
58
Mean
Modeled
Value
66.73
4.36
4.68
4.16
10.34
4.51
13.26
Mean I Mean
Monitoredpifference
Value
1.29
1.36
1.38
1.34
1.20
1.12
1.24
65.44
3.01
3.30
2.81
9.14
3.39
12.02
Mean % Difference
4602
237
246
231
730
328
931
Fraction of Modeled Values Within
10%
0.00
0.00
0.00
0.00
0.02
0.03
0.02
25%
0.00
0.20
0.00
0.33
0.03
0.03
0.03
50%
0.00
0.20
0.00
0.33
0.11
0.14
0.10
100%
0.00
0.20
0.00
0.33
0.24
0.34
0.19
Notes on next page.
ICF Consulting
54
-------
Notes for Table 4-5:
5. Modeled variable:
concnear Nearest modeled DPM concentration consistent with the draft 2000 NONROAD Model
concnear2 Nearest modeled DPM concentration consistent with the draft 2002 NONROAD Model
maxconc Nearby (within 30 km) maximum modeled DPM concentration consistent with the draft 2000 NONROAD
Model
maxconc2 Nearby (within 30 km) maximum modeled DPM concentration consistent with the draft 2002 NONROAD
Model
6. Monitored
ECTOR
ECTORH
ECTORL
ECTOT
ECTOTH
ECTOTL
TOR
TORH
TORL
7. Subset:
All
Non-road
On-road
variable:
EC value multiplied by TOR
EC value multiplied by TOR
EC value multiplied by TOR
EC value multiplied by TOT
EC value multiplied by TOT
EC value multiplied by TOR
ECOCX value multiplied by
ECOCX value multiplied by
ECOCX value multiplied by
average correction factor (missing for EC measured using TOT).
maximum correction factor (missing for EC measured using TOT).
minimum correction factor (missing for EC measured using TOT).
average correction factor (missing for EC measured using TOR).
maximum correction factor (missing for EC measured using TOR).
minimum correction factor (missing for EC measured using TOR).
TOT average correction factor for EPA data, ECTOR for TOR data.
TOT maximum correction factor for EPA data, ECTOR for TOR data.
TOR minimum correction factor for EPA data, ECTOR for TOR data.
All sites
Sites dominated by Non-road source (at least 75 % of modeled DPM consistent with the draft 2000 NONROAD
Model
Sites dominated by On-road source (at least 50 % of modeled DPM consistent with the draft 2000 NONROAD
Model
ICF Consulting
55
-------
Figure 4 1
Scatter plot of nearest modeled DPM concentration against corrected monitored EC concentration
based on TOR. EPA data excluded. All sites. Draft NONROAD 2OOO Model.
Average-
Low
^ t- High
=+= CD
o
1 2 3 A-
Monitored concentration (micrograms / cubic meter)
Regression numbers: 1 19 37
ICF Consulting
56
-------
Figure 4 2
Scatter plot of nearest modeled DPM concentration against corrected monitored EC concentration
based on TOR. EPA data excluded. All sites. Draft NONROAD 2OO2 Model.
Average-
Low
^ t- High
o
123
Monitored concentration (micrograms / cubic meter)
Regression numbers: 7 25 44
ICF Consulting
57
-------
Figure 4 3
Scatter plot of nearby maximum modeled DPM concentration against corrected monitored EC concentration
based on TOR. EPA data excluded. All sites. Draft IM ON ROAD 2OOO Model.
Average
/
>|e »k-/O + O
High
>K3-
O
23456
Monitored conoentration (micrograms / cubic meter)
Hepression numbers: 1O as 4-~7
ICF Consulting
58
-------
Figure 4 4-
Scatter plot of nearest modeled DPM concentration against corrected monitored EC
concentration based on TOT. All sites. Draft NONROAD 2OOO Model.
1O
e>©0 Average
=4=
Low
O
High
456789
concentration (micrograms / cubic meter)
Regression numbers: 55 79 1O3
1O
ICF Consulting
59
-------
Figure 4 5
Scatter plot of nearest modeled DPM concentration against corrected monitored EC
concentration based on TOT. All sites. Draft NONROAD 2OO2 Model.
O
&&& Average
23456789
Monitored concentration (micrograms / cubic meter)
Regression numbers: 64 BB 125
1O
ICF Consulting
60
-------
Figure 4 S
Scatter plot of nearest modeled DPM concentration against corrected monitored EC concentration
based on TOR. EPA data included. All sites. Draft NONROAD 2OOO Model.
&&& Average
* o t
o
* * ox
* *OH-° /
|t ^@0^ 4
. 4. .
_4 ,^f*M4
-h
O
O
»
LOW
High
Y=X
2 3 4- S 6 ~7
Monitored concentration (micrograms / cubic meter)
Regression numbers: 127 151 175
ICF Consulting
61
-------
REFERENCES
Air Improvement Resources. (1997). Contribution of Gasoline Powered Vehicles to Ambient
Levels of Fine P articulate Matter. CRC Project A-18.
Allen, G. A., Lawrence, J., and Koutrakis, P. (1999). Field Validation of a Semi-Continuous
Method for Aerosol Black Carbon (Aethalometer) and temporal Patterns of Summertime Hourly
Black Carbon Measurements in Southwestern P. A. Atmospheric Environment. 33 (5), pp. 817-
823.
Cass, G. R. (1997). Contribution of Vehicle Emissions to Ambient Carbonaceous P articulate
Matter: A Review and Synthesis of the Available Data in the South Coast Air Basin. CRC Project
A-18.
Chow, J. C. and Watson, J. G. (1998). Guideline on SpeciatedParticulate Monitoring. Desert
Research Institute.
Chow, J. C., Watson, J. G., Pritchett, L. C., Pierson, W. R., Frazier, C. A., and Purcell, R. G.
(1993). The DRI Thermal/Optical Reflectance Carbon Analysis System: Description, Evaluation,
and Applications in U.S. Air Quality Studies. Atmospheric Environment. Vol 27A, No. 8, pp.
1185-1201.
Hansen, A. D. A. (2000). The Aethalometer. Magee Scientific Company, Berkeley, California,
USA.
Hansen, A. D. A, and McMurry, P. H. (1990). An intercomparison of measurements of aerosol
elemental carbon during the 1986 carbonaceous species method comparison study. J. Air &
Waste Manage. Assoc. 40, pp. 394-395.
Liousse, C., Cashier, H., and Jennings, S. G. (1993). Optical and Thermal Measurements of
Black Carbon Aerosol Content in Different Environments: Variation of the Specific Attenuation
Cross Section, Sigma (G).Atmospheric Environment. Vol 27A, No. 8, pp. 1203 -1211.
Ramadan, Z., Song, X-H, and Hopke, P. K. (2000). Identification of Sources of Phoenix Aerosol
by Positive Matrix Factorization. J. Air & Waste Manage. Assoc. 50, pp. 1308-1320.
Schauer, J. J., and Cass, G. R. (2000). Source Apportionment of Wintertime Gas-Phase and
Particle Phase Air Pollutants Using Organic Compounds as Tracers. Environmenal Science and
Technology. Vol 34, No. 9, pp. 1821 -1832.
Schauer, J. J., Rogge, W. F., Hildemann, L. M., Mazurek, M. A., Cass, G. R., and Simoneit, B.
R. T. (1996). Source Apportionment of Airborne PM Using Organic Compounds as Tracers.
Atmospheric Environment. Vol 30, No. 22, pp. 3837 -3855.
ICF Consulting 62
-------
Watson, J. G., Fujita, E., Chow, J. G., Zielinska, B., Richards, L. W., Neff, W., and Dietrich, D.
(1998). Northern Front Range Air Quality Study Final Report. Desert Research Institute. 6580-
685-8750.1F2.
Zheng, M., Cass, G. R., Schauer, J. J., and Edgerton, E. S. (2002). Source Apportionment of
PM2.5 in the Southeastern United States Using Solvent-Extractable Organic Compounds as
Tracers. Environmental Science and Technology. To appear.
ICF Consulting 63
------- |