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 ------- |