Impacts of Changes in Source Activities on Air Pollution in Rochester

7383 words (30 pages) Dissertation

16th Dec 2019 Dissertation Reference this

Tags: Environmental StudiesClimate Change

Disclaimer: This work has been submitted by a student. This is not an example of the work produced by our Dissertation Writing Service. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of

Assessing the impacts of changes in source activities on air pollution in Rochester, NY from 2008 to 2013


Significant changes in emissions in the northeastern United States have been achieved in recent years. For example, the 260 MW Russell coal-fired power plant in Rochester, NY closed in 2008, emissions from Kodak Park have been reduced dramatically, and new, heavy-duty diesel trucks are equipped with particle traps and use ultralow sulfur fuel. Particle size distributions have been measured in Rochester, NY since the end of 2001. Prior studies have examined the sources of particles and other air pollutants through 2007. This study investigates whether such changes in emissions has resulted in observable changes in air quality. Data measured between January 2008 and December 2013 at New York State Department of Environmental Conservation (NYS DEC) site in Rochester, NY include ambient submicron particle number size distributions (11–470 nm), gaseous pollutants, black carbon (BC), UV black carbon (UVBC), Delta-C (UVBC–BC) and meteorological variables. Two periods, 2008-2010 and 2011-2013, each divided into three different seasons, were analyzed separately. EPA PMF V5 applied to these data yielded 8 different source profiles: two traffic factors, nucleation, residential/commercial heating, secondary nitrate, secondary sulfate, ozone-rich secondary aerosol, and regionally transported aerosol.   Secondary sulfate and regional transport were not found in the winter samples while secondary nitrate and residential/commercial heating were not resolved in the summer sample data. The profiles were generally characterized by similar number modes for the winter, summer or transition periods. Similar sources were normally resolved in both periods with lower number concentrations during 2011-2013 compared with 2008-2010. Comparison with the 2001-2007 results shows that the previously identified industrial emissions and mixed nucleation/traffic sources were no longer detected in the current analyses.

Keywords: Air pollution, Source apportionment, PMF, Size Distribution, Ultrafine particles

1. Introduction

Associations between exposure to ambient airborne particles and a variety of adverse health effects have been found in many epidemiological studies. Ultrafine particles have been suggested as a potential health threat since they may contain potentially toxic species and can penetrate into the respiratory system (Oberdörster et al., 2005). Ambient particles may include primary particles emitted from anthropogenic activities (e.g. industries, transportation), and natural sources such as forest burning, or secondary particles obtained by gas-to-particle change processes, including nucleation and condensation. They are in a wide scale of particle diameters (Dp) from a few nanometers up to 100 µm (Harrison et al., 2000). Particle number size distributions (PNSD) can exhibit at least four size modes, with distinctive sources and chemical compositions; the nucleation mode (Dp < 30 nm), the Aitken mode (30 nm < Dp < 100 nm), the accumulation mode (100 nm < Dp < 1 µm) and the coarse mode (Dp > 1 µm) (Vu et al., 2015). Research into the sources and contributions of PNSD and its components is required to better understand its impact on health and the environment and ultimately inform targeted regulations to curb these adverse effects.

Multivariate receptor models are now plently used for investigating patterns in environmental data.  Positive matrix factorization (PMF) is currently the most widely used approach to apportion particulate matter sources generally by analyzing PM composition data.  However, it is also been applied to apportion the source contributions of PNSD in a number of locations (Kim et al., 2004; Zhou et al., 2005a, b; Ogulei et al., 2006; 2007a; Kasumba et al., 2009; Thimmaiah et al, 2009, Cuccia et al., 2010; Gu et al., 2011; Friend et al., 2012; Liu et al., 2014; Sowlat et al., 2016 and Liao 2017).  Ogulei et al. (2007) made the initial assessment of the particle sources in Rochester based on measured particle number size distributions and pollutant gases measured from late 2004 to near the end of 2005.  They separated the data into 4 seasons that were analyzed independently using the program PMF2.  Ten sources were identified, including two traffic factors, two nucleation factors, industrial emissions, residential/commercial heating, secondary nitrate, secondary sulfate, ozone-rich secondary aerosol, and regionally transported aerosol with generally similar characteristics in each season.

Kasumba et al. (2009) extended the study to include data from December 2001 to March 2004 at a different site in downtown Rochester that was closed in 2004 in favor of the site used in the Ogulei et al. study.  Kasumba et al. extended the analysis to the end of 2007.  A similar suite of sources was identified with some differences in the relative importance of some of the source types, particularly an indication that industrial emissions were diminishing over time. The industrial emissions were dominated by coal-fired power plants located in the northwest of the sampling site including a 260 MW electricity generating plant and a coal-fired cogeneration system at a major industrial complex, Kodak Park.  The effect of the shutdown of the power plant was observed in subsequent work (Wang et al., 2011a).

The present study updates this earlier work by utilizing EPA PMF V5 to identify and apportion the emission sources using ambient particle number concentrations, gaseous and PM species, and meteorological variables measured in Rochester, NY from 2008 to 2013. The measured species and meteorological variables were utilized in the identification of the suspected sources. The quantification of pollutant source contributions is important because it provides exposure estimates that will be used in epidemiological studies of observed health effects. Furthermore, this information is useful in examining the changes in source emissions during the different time periods and their contributions over time to assess the effectiveness of air quality management strategies in improving air quality.

2. Data Description

Number concentrations of particles in the size range of 11–470 nm were measured at the New York State Department of Environmental Conservation (NYS DEC) site in Rochester, NY (43o08’46” N, 77o32’54” W) (Figure 1) from January 2008 to December 2013 . For more explanation, see Kasumba’s work (Kasumba et al., 2009). Hourly CO, PM2.5, SO2, O3, NO, NO2, NOy, black carbon measured at 880 nm (BC) and at 370 nm (UVBC) and Delta-C (Delta-C = UVBC–BC) (Wang et al., 2011b) concentrations were available.  However, CO, NO, NO2, and NOy measurements were only made beginning in 2010.  Meteorological parameters including wind speed and wind direction were measured at the Rochester International Airport that is about 10 km from the monitoring site.

3. Positive Matrix Factorization

PMF decomposes a data matrix to factor contributions and factor profiles interpretable in physical and chemical terms. This model identify the number of factors, the species profile of each factor, and the amount of mass contributed by each factor.  Uncertainty analysis of PMF modeling estimates a range of plausible values for each element of matrix F with a high probability that include the true value of F. The main uncertainty in PMF analyses arises from rotational ambiguity specific to factor analytic models, meaning there are multiple solutions (G, F) with the same value of value of the objective function, Q  (Henry, 1987). Calculation of feasible bands can be achieved by the displacement (DISP) algorithm (Paatero et al., 2014) that is included in EPA PMF V5.  The present work uses the DISP method for estimating the rotation ambiguity for each dataset. The extent of possible rotations is limited by both non-negativity constraints imposed on the solution and by the number of zero values present in the fitted G and Fmatrices.

The hourly average values were combined with the hourly gas, BC, Delta-C, and PM2.5 mass measurements in the PMF model. All missing concentrations were replaced by median within the given size bin or species and their uncertainties were set as four times the value. The particle number concentration uncertainties were estimated from the equation below (Ogulei et al., 2006):



σis the calculated (estimated) measurement error;


Nis the observed number concentration; and

N̅is the arithmetic mean of the reported values

N. The ultimate uncertainties inputted into EPA PMF were computed based on the measurement errors with the expression:



C3is a constant chosen based on the prior studies to be 0.1 except for the summer data from 2011 to 2013 where it was set at 0.2 because there was more variability observed in these data.  Species measured below detection limits (DL) were replaced by 1/2 DL and assigned an uncertainty equal to 5/6 DL.


The conditional bivariate probability function (CBPF) couples ordinary CPF (Ashbaugh et al., 1985; Kim et al., 2003) with wind speed as a third variable, allocating the apportioned pollutant source contributions to cells defined by ranges of wind direction and wind speed (Carslaw et al., 2006; Uria-Tellaetxe et al., 2014) as:




m∆θ,∆uis the number of samples in the wind sector

∆θwith wind speed interval

∆uhaving contribution C greater than a threshold value x (upper 20% of the fractional contribution,

gk, from each source for this study),

n∆θ,∆uis the total number of samples in that wind direction-speed interval. The extension to the bivariate case provides more information on the nature of the sources because different source types can have different wind speed dependencies.

5. Results and discussion

The 2008-2013 data sets for each season (i.e. winter, transition and summer seasons) were split into two periods, 2008-2010 and 2011-2013, and analyzed independently.  For 2008-2010, particle size distributions, PM2.5, BC, Delta-C, SO2, and O3 were analyzed.  For 2011-2013, CO, NO, NO2, and NOy were also available and added to the analysis.

5.1. Obtaining the optimum number of factors

The number of factors (Table S1) was determined by examining the distributions of scaled residuals and the extent of the rotational ambiguity calculated by the displacement procedure (DISP) (Paatero et al., 2014; Emami and Hopke, 2017).  When DISP intervals are small, there is less rotational ambiguity in the solutions.  The presence of additional species in the datasets not only helps to identify the sources but also decreases rotational ambiguity because of the increased numbers of edge points (Emami and Hopke, 2017).

5.2. Factor identification

5.2.1. Winters, 2008-2010 and 2011-2013

The PMF results for both periods appear in Figures 2 to 7.  The profiles including the number size distributions, gaseous pollutants, PM2.5, BC, and Delta-C are presented in Figures 2 and 3 for 2008-2010 and 2011-2013, respectively.  For the portion of the profile showing number size distributions, these values have been normalized so that they present the fraction of total particles in that bin associated with the given profile.   The other variables have not been normalized and represent the average concentration contributed by that factor to the total concentration of the measured variable. Figures 4 and 5 show the diel patterns of the particle number concentrations, and Figures 6 and 7 present the CBPF plots for the resolved sources for the two periods, respectively.  To ascertain differences in contributions to particle number concentrations between weekdays and weekend days, a Mann-Kendall rank sum test was applied to the resolved contributions of each source factor.  The median values for weekdays and weekend days and the probability of the two sets of values having come from the different distribution are presented in Table 1.  For most sources in all seasons and time periods, weekend days are statistically significantly different from weekdays.

A very sharp mode in the number distribution ranging from 11 to 30 nm in diameter represents nucleation mode particles (Jeong et al., 2004). During the winter, they are most likely generated during dilution and cooling of vehicle exhaust emissions.  This factor contributes the most particles to the smallest size bin.  The nucleation factor shows a strong association with morning rush hour traffic and winds from the southern sector consistent with the traffic-generated pollutants from I-490 to south of the site with lower to higher wind speeds.  There was a statistically significant increase in nucleation derived particles on weekends relative to weekdays for 2008 to 2010, but a reversal of that pattern in 2011 to 2013.   The number concentrations of distributions obtained for 2011-2013 period shows an ~30% reduction from the 2008-2010 period (Table 2).

Two traffic factors, traffic 1 and traffic 2, were identified during both time periods having major number modes around 30 and 54 nm, respectively. Depending upon the type of engine, the size distributions of these factors have a whether unimodal (traffic 1) or bimodal (traffic 2) lognormal distribution, which is in coincidence with the conclusion of Vu’ study (Vu et al., 2015). Similar source profiles and designations have been observed in prior studies (Zhou et al., 2005; Kasumba et al., 2009; Sowlat et al., 2016).

The diel patterns of the two traffic factors also help to identify them. Traffic 1 shows a more pronounced morning peak during the winter periods, and a less pronounced afternoon peak. This pattern suggests that this factor is related to local traffic.

Traffic 1 shows sizes more associated with spark-ignition (Kittelson et al., 2006a), while traffic 2 shows sizes more often associated with diesel emissions (Kittelson et al., 2006b; Ogulei et al., 2007b; Kasumba et al., 2009). Most particles from diesel engines are in the size range about 20-130 nm, while most particles from gasoline engines are in the size range ∼20–60 nm. The contributions of traffic1 and traffic 2 had decreases of 42% and 48% between the two multiyear periods, respectively. Traffic 2 shows nighttime highs supporting the assignment of diesel emissions since nocturnal commercial heavy-duty diesel traffic is more likely on the major highways near the DEC site. The weekday concentrations for traffic 1 and exceeded the weekend concentrations (Table 2).

The CBPF plots for traffic 1 point mainly in the directions of I-490 and I-590 and slightly in the direction of routes 96 and 104 at various wind speeds. The directionality of traffic 2 for both time periods corresponds to the directions of I-490 and I-590 suggesting that this factor may also characterize emissions from the highway traffic.  The CBPF values are all relatively low (<0.4) reflecting the presence of traffic sources surrounding the sampling site.  Thus, although I-490 and NY 96 are south of the monitor, particles are transported to the site from all directions.

Residential/commercial heating has two modes in the number profiles with the modes between 100 and 200 nm. Most residences and commercial buildings in Rochester are heated with natural gas or No. 2 oil (Ogulei et al., 2007).  However, there is a significant amount of recreational wood combustion in the winter (Wang et al., 2012a, b). The diel profiles (Figures 4 and 5) indicate that the nighttime contributions from this factor are higher than the daytime contributions for both time periods. The directionality of this factor (Figures 6 and 7) indicates northeast as well as southeast and southwest of the sampling site, the directions of suburbs of Rochester with primarily single family houses.  The figures suggest the impact of high wind speeds that convey these emissions to the site. The CBPF probabilities for 2008-2010 are about twice those for 2011-2013. Also, the contributions obtained for 2011-2013 are ~89% lower in number concentrations compared to 2008-2011.  The reasons for the shift in direction and decreases in CBPF values are not known.  Particle concentrations may be lower as a result of generally milder winters in the latter time period.

Secondary nitrate has three modes in the number profiles with the major number mode between 200 and 300 nm. The other modes were in the ultrafine and Aitken mode size ranges suggesting that both locally produced and transported particles may contribute to this factor. Nucleation and Aitken mode particles may be formed from local NOx emissions (Zhou et al., 2005). The accumulation mode particles are likely due to transport of NH3 and NOx from more distant sources. Ammonium nitrate can be formed from its precursor gases via the oxidation of NO2 reacting with locally emitted NH3 to form NH4NO3 (Zhou et al., 2005b: Zhao et al., 2007).  Zhou et al. (2005b) were able to resolve 2 nitrate factors because they had hourly nitrate concentrations.  One factor had its primary mode in the accumulation mode range while the other showed an Aitken mode peak.  Since the measurements reported in that paper were collected in warmer periods, a nucleation mode was not observed, likely due to rapid evaporation of the smallest sized particles. Cooler temperatures at night and morning rush hour contribute to its formation. The diel patterns from both periods exhibit morning and evening peaks, which is in agreement with the apportioned secondary aerosol source in Sowlat’s work (Sowlat et al., 2016). The weekday average contributions are more than the average weekend contributions, again supporting the traffic influences on this factor. The directionality of this factor in both time periods is consistent with the locations of the major traffic highways surrounding the sampling site for various wind speeds.

Finally, a factor that was named ozone-rich secondary aerosol was resolved for all the seasons considered in this study, and for both periods. Ozone-rich secondary aerosol has multiple modes (both in the Aitken and accumulation modes), but no single size range consistently dominates the particle number profiles. A similar source has been seen consistently in the prior analyses of Rochester PNSD data (Ogulei et al., 2007a; Kasumba et al., 2009).  The diel patterns show daytime highs suggesting the influence of photochemistry in ozone formation. Significant weekday–weekend variability was observed in both time periods, but in opposite directions.  Thus, it appears that this source represents secondary aerosol that is transported into the Rochester area.

Table 2 summarizes the annual average source contributions (#/cm3) from all the sources resolved by PMF. With the exception of the O3-rich secondary aerosol source, the other winter sources showed significant reductions in 2011-2013 compared to 2008-2010.

5.2.2. Transition Seasons, 2008-2010 and 2011-2013

In reviewing the data, it was found that for spring 2009, there was a period of about 3 weeks with particle number concentrations approximately ten times what had been typically observed. Further examination showed that the high values occurred from March 19 to April 9.  There were regular patterns of formation and growth (Figure S7) suggesting that these values were not the result of instrumental problems.  According to information provided by the New York State Department of Transportation (DOT), DOT was installing an intelligent transportation system (ITS) traffic monitor along I-590 north and south of the monitoring site during this period. They would have been southeast of the site for some of the time and northeast of the site the rest of the time. These roads are relatively close to the site, so there could have been an impact.  The CBPF plots for the relevant four weeks (Figure S8) include this period for the sources with elevated contributions and the total particle number concentrations. There was a shift in direction from southeast to north in the CBPF plots. Northeasterly winds are rare, so that probability is very low or zero at this direction (Figure S8). In addition, the average concentration of gaseous elemental mercury (GEM) during this specific time, 1.54 ng/m3, was significantly (one-way ANOVA p<0.01) higher than that in the other months of this year, 1.39 ng/m3. The correlation of GEM with particle number concentrations is 0.42 at this period. Thus, this installation work with diesel off-road equipment in use may have resulted in the observed high particle number concentrations. The data were analyzed with these three weeks of data excluded.

Seven factors were identified for both the 2008-2010 and 2011-2013 periods, i.e., one nucleation factor, two traffic factors (traffic 1 and traffic 2), residential/commercial heating, secondary nitrate, secondary sulfate, and O3-rich secondary aerosol. The transition season factor profiles are shown in Supplemental Figures S1 and S2. The diel variations in the contributions from the identified sources are presented in Supplemental Figures S3 and S4. The CBPF plots for 2008 to 2010 and 2011 to 2013 are presented in Supplemental Figures S5 and S6, respectively.

The size distribution for nucleation was again dominated by particles close to 20 nm in diameter (Figures S1 and S2). The nucleation factor diel pattern correlates well with traffic rush hours for Rochester suggesting that tail-pipe emissions may have produced the nucleation activity in the vicinity of the highways (Figure 1). During the transition seasons, afternoon nucleation episodes were also observed.  These nucleation events are likely to arise from the formation of lower vapor pressure compounds arising from the higher concentrations of oxidizing species that develop by the late morning hours (Jeong et al., 2004).  The CBPF plots for the nucleation factors point to the intersection of the interstate highways I-490 and I-590 as well as NY route 96 over a range of low to high wind speeds (Figures S5 and S6).

Traffic 1 shows a major number mode at about 30 nm, particles emitted from gasoline vehicles (Vu et al., 2015), and a minor mode at about 110 nm (Figures S1 and S2). In contrast, traffic 2 is characterized by its major number mode at about 90 nm, an accumulation soot mode. The minor mode of traffic 2 is 23 nm, nucleation mode. Median weekday contributions from traffic 1 and traffic 2 exceeded weekend values (Table 1).  The diel patterns were generally similar to the winter patterns and almost indistinguishable between the two periods. The CBPF plots (Figure S5 and S6) indicate winds from the northeasterly direction corresponding to route 286, which carries traffic traveling to and from the downtown area, and I-590 highway, with significant probabilities at high speeds suggesting that these factors may also represent emissions from local highway traffic. The directionality of traffic 2 in 2011-2013 points toward Highways I-490 and I-590 from low to elevated wind speeds although with low probabilities. The small number of contributions greater than the threshold value for this factor could be the result of a greater fraction of the diesel trucks being equipped with catalytic regenerator traps and using ultralow sulfur fuel.

Residential/commercial heating identified for transition (Figures S5 and S6) shows strong associations with both northeasterly and southwesterly wind directions at high wind speeds, similar to winter seasons, suggesting that it has little if any association with traffic on I-490 and I-590. Its diel variation suggests a source emitting predominantly in the evening. The size distribution with a small mode in the number distribution at just over 20 nm suggests a higher temperature combustion source such as natural gas or No. 2 oil.

The secondary nitrate factor for the transition season shows the same modes with winter season (Figures S1 and S2). The directions shown by the CBPF plots for this factor (Figures S5 and S6) coincide with the directions of major highways.

A factor representing secondary sulfate was resolved for summer and transitional periods. This factor was characterized by number modes around 50 and 300 nm. It exhibits daytime maximums that may be due to the atmospheric processing of more local SO2 emissions.  Sulfate driven nucleation events had been observed in the past prior to the closure of the 260 MW coal-fired power plant (Jeong et al., 2004; Wang et al., 2011a). The accumulation mode particles mean that the particles have grown to larger sizes through coagulation. The directions shown in Figures S5 and S6 correspond to the places of local sources of combustion-derived particles mainly from traffic with lower probabilities in 2011 to 2013.

The last identified source for transition season was O3-rich secondary aerosol. Its characterizes are very similar to the corresponding winter season factor.  The profiles, diel patterns, and CBPF plots are found in Figures S7-S12, respectively. Table 2 shows the annual average number concentration reduction of the resolved sources during the second multiple year period.

5.2.3. Summer 2008-2010 and 2011-2013

During the review of the summer season data, two additional periods of very high particle number concentrations were observed between June 6 to 13, 2008 and July 9 and 10, 2008. It has not been possible to determine any likely cause for these high values during these periods. Again the particle number concentrations higher by several orders of magnitude, but the distributions appear reasonable and the data prior to and following these periods appear normal.  Thus, the instrument appears to be functioning normally, and it is assumed that there was some type of vehicular or off-road generator activity in the immediate vicinity of the DEC site that produced these high values.  Again, these periods were excluded from the data set.

For the summer 2008-2010 size distribution data, the signal-to-noise ratios for the 12-16, 16-18.4, 284-305, and 328-470 nm intervals were below 2. Therefore, they were set as weak variables and downweighted in the PMF analysis.  These source types included nucleation, two traffic factors, regionally transported aerosol, secondary sulfate and ozone-rich secondary aerosol. Figures S9 and S10 present the resolved profiles, Figures S11 and S12 give the diel patterns and CBPF plots are shown in Figures 13 and 14 for the 2008-2010 and 2011-2013 periods, respectively.   For the summer data from each of the two periods, only 6 factors were needed to adequately fit the data and provide physically interpretable profiles.  The sources that have disappeared are residential/commercial heating and secondary nitrate.

Similar to the other seasons, a nucleation factor was found to contain particles around 30 nm (Figures S9 and S10). Nucleation incidents were frequently observed in both time periods, with mostly afternoon nucleation events occurring in the summer months (Figure S11 and S12). The CBPF plots point in the directions of the interstate highways for low and high wind speeds for 2011-2013 time (Figure S14). For the 2008-2010 period, the CBPF plot (Figure S13) points to the northwest with wind speeds in the 2 to 4 m s-1 range.  There were several SO2 sources in that direction including the Kodak Park and the Russell coal-fired power plant that closed in April 2008.  Data on annual emissions of SO2 from Kodak Park suggests that it was relatively constant over the entire period of this study so there is no obvious reason for the significant difference in direction for the nucleation factor between the two time intervals.

The major modes for 2008-2010 traffic 1 and 2011-2013 traffic1 were 47 and 38 nm, respectively. Traffic 2 is characterized by major number modes about 89 nm (2008-2010), and 67 nm (2011-2013) with the minor modes at 22 and 16 nm, respectively. Figures S11 and S12 show similar diel patterns for the factors in both periods. The directionality of traffic 1 corresponds to the locations of I-490, I-590 and downtown Rochester as well as the directions of NY routes 104 and 96 for low to high wind speeds. The directionality of traffic 2 coincides to the highways recognized for winter and transition seasons below 2 m/s wind speeds.

Regional transport was only resolved for the summer periods.  It was mostly identified by accumulation mode particles, and was mainly characterized by major number modes between 100 and 300 nm. No significant diel variations were observed for 2011-2013 period suggesting that it is not from local emissions (Figure S12). It is likely that this factor largely represents secondary inorganic aerosol primarily sulfate resulting from the homogeneous processing of SO2 emitted from coal fired power plants located in the western New York and the Upper Ohio River Valley. This source is consistent with long-range transported aerosols explained in Vu’s review (Vu et al., 2015).

The secondary sulfate factor that was resolved for the transition and summer aerosol was characterized by major number modes at about 30 nm and 140 nm in summer. As mentioned previously, these sulfate particles are probably formed through atmospheric procedure of local SO2 emissions.  The diel pattern peaks during the day when the rate of homogenous oxidation would be the highest.   A similar diel pattern was observed in New York City by

Drewnick et al., (2004).  The CBPF plot for this factor in 2011-2013 suggests a direction of SO2 emissions from the Kodak Park complex at wind speeds below 2 m/s (Figure S14). In case of 2008-2010, southwest of sampling site is suggested as the main source location. This observation suggests that it is primarily generated by the urban background and traffic.

The O3-rich secondary aerosol source was found here for both periods and is similar to those for the other two seasons.  Table 2 provides the annually average source contributions (#/cm3) for all sources resolved by PMF showing the declines from 2008 to 2013.

5.3. Comparison with Prior Studies in Rochester

Previous work on the data measured between August 2001 and December 2007 on Rochester by Kasumba et al. (Kasumba et al., 2009) identified sources similar to those obtained in the present study with some differences. In this study, the observed properties of nucleation, traffic 1, traffic 2, residential/commercial heating, secondary nitrate, secondary sulfate, regional transport and O3-rich secondary aerosol were all in good agreement with 2001-2004 and 2005-2007 results. However, the previously identified industrial emissions and mixed source of nucleation and traffic were the sources that were not detected in the 2008 to 2013 data. Also, none of the resolved factors particularly traffic sources show a noticeable association with SO2 concentrations. The closing of the 260 MW Russell coal-fired power plant, located to the northwest of the sampling site starting in February 2008 and completed at the end of April 2008 and the reduction in diesel fuel sulfur to <15 ppm beginning on October 1, 2006 led to a decline of -0.82 ppb/y SO2 concentrations during 2008-2010. In addition, an increasing fractions of the heavy-duty diesel fleet have particle controls.  Beginning in 2004, Tier 4 rules were promulgated to produce less polluting railroad locomotives (USEPA, 2008), reductions in off-road engine emissions as required by the 2004 nonroad emission rules (USEPA, 2004) and improvements in nonroad fuels.  EPA mandated the sulfur content to not exceed 500 ppm effective June 2007 for nonroad, locomotive and marine (NRLM) diesel fuels and 15 ppm (ultra-low sulfur diesel) effective June 2010 for nonroad fuel, and June 2012 for locomotive and marine fuels (Nonroad Diesel Engines, 2017).  A main east-west railroad line runs adjacent to the property on which the DEC site is located and thus, mandated changes in fuels and remanufactured engines may be reflected in the decline in SO2 and sulfate observed at this location.

6. Conclusions

DISP-PMF was successfully employed in comparing the sources contributing to the number concentrations of the measured submicron particles on two different periods of time in Rochester, NY. Except 2008-2010 summer data, low rotational ambiguity associated with the apportioned profiles, indicating more accurate results, have been acquired for the studied data. Most of the resolved sources were found to be anthropogenic, and no significant season-to-season variability in particle number modes emitted from any given source was observed at both periods. The main sources contributing to the submicron particle number concentrations of 2008-2013 period in Rochester include nucleation, local and distant traffic, residential/commercial heating, secondary nitrate, secondary sulfate, regionally transported aerosol and ozone-rich secondary aerosol. Nucleation was accountable for the smallest particles, while traffic was mostly accountable for particles ranging from 20 to 80 nm likely resulting from  the proximity of these sources to the sampling sites. Distant sources were found to be largely accountable for the secondary sulfate and accumulation mode particles. The significant reductions in SO2 emitters including the coal-fired power plant, the coal-fired cogeneration plant, and controls on on-road and nonroad engine fuel sulfur resulted in the elimination of two factors observed in prior studies: mixed traffic and nucleation and industrial emissions. The properties of 2008-2010 resolved sources were comparable to 2011-2013 factors. Noticeable lower number concentrations were observed compared to prior analyses that could be the result to changes in industrial activities and fuel quality over the observation period.


Ashbaugh, L.L., Malm, W.C., Sadeh, W.Z., 1985. A residence time probability analysis of sulfur concentrations at Grand Canyon National Park. Atmos. Environ. 19, 1263–1270.

Carslaw, D.C., Beevers, S.D., Ropkins, K. and Bell, M.C., 2006. Detecting and quantifying aircraft and other on-airport contributions to ambient nitrogen oxides in the vicinity of a large international airport. Atmos. Environ. 40, 5424-5434.

Cuccia, E., Bernardoni, V., Massabò, D., Prati, P., Valli, G., Vecchi, R., 2010. An alternative way to determine the size distribution of airborne particulate matter. Atmos. Environ. 44, 3304-3313.

Drewnick, F., Schwab, J.J., Jayne, J.T., Canagaratna, M., Worsnop, D.R., Demerjian, K.L., 2004. Measurement of Ambient Aerosol Composition During the PMTACS-NY 2001 Using an Aerosol Mass Spectrometer. Part I: Mass Concentrations, Aerosol Science and Technology 38(S1), 92–103.

Emami, F., Hopke, P.K., 2016. Effect of Adding Variables on Rotational Ambiguity in Positive Matrix Factorization Solutions, Chemom. Intell. Lab. Syst. 62, 198–202

Friend, A.J., Ayoko, G.A., Jayaratne, E.R., Jamriska, M., Hopke, P.K., Morawska, L., 2012. Source apportionment of ultrafine and fine particle concentrations in Brisbane, Australia. Environ. Sci. Pollut. Res. 19, 2942-2950.

Gu, J., Pitz, M., Schnelle-Kreis, J., Diemer, J., Reller, A., Zimmermann, R., Soentgen, J., Stoelzel, M., Wichmann, H.E., Peters, A., Cyrys, J., 2011. Source apportionment of ambient particles: comparison of positive matrix factorization analysis applied to particle size distribution and chemical composition data. Atmos. Environ. 45,1849-1857.

Henry, R.C., 1987. Current factor analysis receptor models are ill-posed. Atmos. Environ. 1967 21, 1815–1820.

Harrison, R.M., Shi, J.P., Xi, S., Khan, A., Mark, D., Kinnersley, R., Yin, J., 2000. Measurement of number, mass and size distribution of particles in the atmosphere. Philos. Trans. R. Soc. Lond. Math. Phys. Eng. Sci. 358, 2567–2580.

Hasheminassab, S., Daher, N., Saffari, A., Wang, D., Ostro, B. D., and Sioutas, C., 2014. Spatial and temporal variability of sources of ambient fine particulate matter (PM2.5) in California, Atmos. Chem. Phys., 14, 12085–12097.

Jeong, C.-H., Hopke, P. K., Chalupa, D., Utell, M., 2004. Characteristics of Nucleation and Growth Events of Ultrafine Particles in Rochester, NY, Environ. Sci. Technol. 38, 1933–1940

Jeong, C.H., Herod, D., Zlotorzynska, E.D., Ding, L., McGuire, M. L., Evans, G., 2013. Identification of the sources and geographic origins of black carbon using factor analysis at paired rural and urban sites. Environ. Sci. Technol. 47, 8462–8470.

Jeong, C.-H., Evans, G.J., Hopke, P.K., Chalupa, D.J., Utell, M., 2006. Influence of atmospheric dispersion and new particle formation events on ambient particle number concentration in Rochester, United States, and Toronto, Canada. J Air Waste Manag Assoc 56, 431-443.

Kasumba, J., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2009. Comparison of sources of submicron particle number concentrations measured at two sites in Rochester, NY. Sci. Total Environ. 407, 5071–5084.

Kim, E., Hopke, P.K., Larson, T.V., Covert, D.S., 2004. Analysis of ambient particle size distributions using UNMIX and positive matrix factorization. Environ. Sci. Tech, 38, 202–209.

Kittelson, D.B., Watts, W.F., Johnson, J.P., Schauer, J.J., Lawson, D.R., 2006a.  On-road and laboratory evaluation of combustion aerosols—Part 2: Summary of spark ignition engine results, J. Aerosol Science 37, 931–949.

Kittelson, D.B., Watts, W.F., Johnson, J.P., 2006b.  On-road and laboratory evaluation of combustion aerosols—Part1: Summary of diesel engine results, J. Aerosol Science 37, 913–930.

Liu, Z.R., Hu, B., Liu, Q., Sun, Y., Wang, Y.S., 2014, Source apportionment of urban fine particle number concentration during summertime in Beijing. Atmos. Environ, 96, 359-369.

Liao, H.T., Chou, C.C.K., Huang, S.H., Luc, C.J., Chen, C.C., Hopke, P.K., Wu, C.F., 2017, Source apportionment of PM2.5 size distribution and composition data from multiple stationary sites using a mobile platform. Atmos. Res, 190, 21-28.

Oberdörster, G., Oberdörster, E., Oberdörster, J., 2005. Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles. Environ. Health Perspect. 113: 823–839.

Ogulei, D., Hopke, P.K., Wallace, L.A., 2006. Analysis of indoor particle size distributions in an occupied townhouse using positive matrix factorization. Indoor Air 16, 204–215.

Ogulei, D., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2007a. Modeling source contributions to submicron particle number concentrations measured in Rochester, New York. Aerosol Sci. Technol. 41, 179–201.

Ogulei, D., Hopke, P.K., Ferro, A.R., Jaques, P.A., 2007b. Factor Analysis of Submicron Particle Size Distributions near a Major United States–Canada Trade Bridge, J. Air & Waste Manage. Assoc. 57, 190–203

Paatero, P., Eberly, S., Brown, S. G., Norris, G. A., 2014. Methods for estimating uncertainty in factor analytic solutions. Atmos. Meas. Tech. 7, 781–797.

Paatero, P., 1997. Least squares formulation of robust non-negative factor analysis. Chemom. Intell. Lab. Syst. 37, 23–35.

Sowlat, M.H., Hasheminassab, S. and Sioutas, C., 2016. Source apportionment of ambient particle number concentrations in central Los Angeles using positive matrix factorization (PMF). Atmos. Chem. Phys. 16, 4849-4866.

Tauler, R., Smilde, A., Kowalski, B., 1995. Selectivity, local rank, three-way data analysis and ambiguity in multivariate curve resolution. J. Chemom. 9, 31–58.

Thimmaiah, D., Hovorka, J. and Hopke, P.K., 2009. Source apportionment of winter submicron Prague aerosols from combined particle number size distribution and gaseous composition data. Aerosol Air Qual. Res, 9.209-236.

Uria-Tellaetxe, I. and Carslaw, D.C., 2014. Conditional bivariate probability function for source identification. Environ. Model. Softw., 59, 1-9.

U.S. Environmental Protection Agency (USEPA), 2004., Control of Emissions of Air Pollution From Nonroad Diesel Engines and Fuel; Final  Rule, Federal Register Vol. 69, No. 124,38958-39273.

U.S. Environmental Protection Agency (USEPA), 2008. Control of Emissions of Air Pollution from Locomotive Engines and Marine Compression-Ignition Engines Less Than 30 Liters per Cylinder; Republication; Final Rule, Federal Register Vol. 73, No. 126,  37096-37350.

Vu, T.V., Delgado-Saborit, J.M., Harrison, R.M., 2015. Review: Particle number size distributions from seven major sources and implications for source apportionment studies. Atmos. Environ. 122, 114–132.

Xie, Y.L., Hopke, P.K., Paatero, P., Barrie, L.A., Li, S.M., 1999. Identification of Source Nature and Seasonal Variations of Arctic Aerosol by positive matrix factorization. J. Atmos. Sci. 56, 249–260.

Wang, Y., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2011a. Effect of the Shutdown of a Coal-Fired Power Plant on Urban Ultrafine Particles and Other Pollutants, Aerosol Sci. Technol. 45, 1245–1249.

Wang, Y., Hopke, P.K., Rattigan, O.V., Xia, X., 2011b. Characterization of residential wood combustion particles using the two-wavelength aethalometer. Environ. Sci. Technol. 45, 7387-7393.

Wang, Y., Hopke, P.K., Chalupa, D.C., Utell, M.J., 2011c. Long-term study of urban ultrafine particles and other pollutants. Atmos. Environ. 45, 7672-7680.

Wang, Y., Hopke, P.K., Rattigan, O.B., Chalupa, D.C., Utell, M.J. (2012a) Multiple-year black carbon measurements and source apportionment using Delta-C in Rochester, New York, J. Air Waste Manage. Assoc., 62, 880-887.

Wang, Y., Hopke, P. K., Xia, X., Chalupa, D. C., Zhang, Y., Utell, M. J. (2012b) Source Apportionment Using Positive Matrix Factorization on Daily Measurements of Organic and Inorganic Speciated PM2.5. Atmospheric Environment. 55, 525-532.

Zhao, W., Hopke, P.K., Zhou, L., 2007. Spatial distribution of source locations for particulate nitrate and sulfate in the upper-midwestern United States, Atmos. Environ. 41, 1831–1847

Zhou, L., Hopke, P.K., Stanier, C., Pandis, S.N., 2005a, Mining airborne particulate size distribution data by positive matrix factorization. J Geophys. Res. 110, D07S19.

Zhou, L., Hopke, P.K., Stanier, C., Pandis, S.N., Ondov, J.M., Pancras, J.P., 2005b, Investigation of the relationship between chemical composition and size distribution of airborne particles by partial least squares (PLS) and positive matrix factorization (PMF). J. Geophys. Res. 110, D07S18.

Nonroad Diesel Engines, 2017,

[*] Author to whom correspondences should be addressed.  Email: [email protected]

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this dissertation and no longer wish to have your work published on the website then please: