Back to Contents
by Rudolf B. Husar , rhusar@mecf.wustl.edu and Stefan R. Falke,stefan@mecf.wustl.edu
Center for Air Pollution Impact and Trend Analysis (CAPITA)
February, 1996
The ambient air quality standard for particulate matter is currently under review. In particular, consideration is given to the introduction of an additional standard for fine particles. This is in recognition of the fact that the sources, effects, as well as the atmospheric behavior of fine particles are different from PM10.
The potential monitoring techniques for PM10 and fine particles are also different. High volume filter samples for PM10 followed by gravimetric weighing of filters before and after exposure is a well established technique derived from high volume TSP monitoring technology. Monitoring the fine particle concentration, on the other hand, is more elaborate than for PM10 or TSP because inertial size separation reduces the available air flow.
The aerosol population is a mixture of different particle sizes and each size class is composed of an internal and/or external mixture of chemically diverse particles. Hence, it is not possible to express the aerosol concentration as a single number, as is the case for gaseous pollutants. On the other hand, practical considerations dictate that the number of aerosol parameters to be monitored has to be limited. Thus, routine monitoring of aerosol chemical composition in many size classes does not appear to be practical for enforcement purposes. Rather, the aerosol size - chemical composition distribution function needs to be monitored using integral measures such as PM2.5 and/or total (or size segregated) light scattering coefficient. PM2.5 is the integral of the aerosol mass - size distribution up to about 2.5 µm. The total light scattering is also an integral of the aerosol mass size distribution but also weighed by the size dependent scattering efficiency factor.
Recently, an 'in situ' fine particle monitoring standard has been suggested using integrating nephelometers. Such devices are being considered either as a substitute or in conjunction with size segregated filter and/or impactor samplers.
The purpose of this report is to present a comparative study of the aerosol light scattering - PM2.5 relationship using existing monitoring data. Light scattering - PM2.5 comparisons were conducted using standard correlation statistics, as well as temporal pattern analysis on yearly, monthly, and daily scales. This report also contains a brief discussion of the criteria by which the alternative monitoring techniques (PM2.5 mass concentration and fine particle light scattering) are to be evaluated as a suitable standard.
The different aerosol monitoring methods as a potential fine particle standard can be evaluated from at least four different points of view: relevance to the aerosol effects on health and environment, the relationship to aerosol sources, suitability for enforcement, and suitability for monitoring. It is certain that a full evaluation needs to incorporate additional criteria not listed here.
The aerosol parameter to be monitored has to be a suitable causal measure of health effects as well as the effects on visibility, acid rain, climate, etc. It can be presumed that, for health effects, the penetration into the lung and the health potency of the aerosol chemical species are relevant. On the other hand, visibility effects are determined by the light extinction under atmospheric (humidity) conditions. Acid deposition is primarily influenced by the aerosol acidity and it is not directly effected by particle size. The direct aerosol effect on climate is due to scattering and absorption of sunlight while the indirect aerosol effect on climate is due to the aerosol interaction with cloud processes.
The main point of the above discussion is that each of the aerosol effects is associated with a specific size and/or chemical composition. Therefore, it is not likely that the single monitoring variable would be equally suitable as a surrogate for all of the effects. Thus, a choice in the measurement technique would require a value judgment as to which effects exposure should to be matched most closely.
Ideally, the monitoring technique would allow the clear identification and apportionment of the primary and secondary aerosol sources of each monitoring site. This criteria is relevant for the introduction of effective control measures. The most common method of aerosol source type attribution is the receptor oriented chemical source apportionment which requires reasonably detailed aerosol chemical composition data. Apportionment of light scattering data among the potential source types is also accomplished using the aerosol chemical composition data in conjunction with chemically dependent mass extinction efficiencies. In case of aerosol light scattering, special attention needs to be focused on the influence of ambient water vapor through hydroscopicity. In general, the mass extinction efficiencies are inferred from statistical considerations rather than from direct measurements.
The need for source identification for ambient aerosols dictates that aerosol chemical composition be monitored at least at areas of high exposure, in the vicinity of unique sources, and other strategically important locations.
The monitored aerosol parameter needs to be suitable for documenting unambiguously man-induced influences that are responsible for exceedances. For instance, uncontrollable high humidity events may influence the readings without influencing the ambient concentrations or effects.
We are not fully aware of all the legal and technical issues associated with enforcement but it stands to reason that the monitoring technique has to withstand legal scrutiny.
The monitoring of the aerosol parameter should be accurate, precise, nearly continuous, and inexpensive. Sampling of PM2.5 mass (as is usual for PM10) is intermittent (e.g. sixth day) averaged over 24 hours and it is manual. Monitoring of the light scattering can be done continuously, with high precision with automatic, self calibrating instruments. However, the accuracy needs to be valued in the context of criteria 1) and 2); is the light scattering measurement the relevant parameter?
From the point of view of monitoring ease and data coverage, light scattering instruments appear to have advantages. On the other hand, from the point of view of data validation and source apportionment, mass measurement is better because it permits compositional analysis.
In the current analysis, data from three different sources were utilized. The AIRS data base contains light scattering data for about 100 of sites in the time period 1985-1995. PM2.5 data are also available from AIRS for about 70 locations. Available light scattering and combined PM2.5 data were examined to find locations that monitored for both fine mass and light scattering. In some cases, both parameters were not measured at the same location but were measured at two separate sites very near one another, i.e. PM2.5 observations from Rubidoux, CA were compared with light scattering data from San Bernadino, CA. An overlap of the light scattering and PM2.5 AIRS variables was limited to less than ten locations and short time periods. For this reason, the AIRS PM2.5 data sets were augmented by fine particle monitoring data supplied by the state of California Air Resources Board, ARB. The additional ARB data provided PM2.5 observations in California for about 25 sites during 1989 - 1995. Available light scattering and combined PM2.5 data were examined to find locations that monitored for both fine mass and light scattering. In some cases, both parameters were not measured at the same location but were measured at two separate sites very near one another, i.e. PM2.5 observations from Rubidoux, CA were compared with light scattering data from San Bernadino, CA.
Light scattering and fine particle mass data were also available from the SCENES data base. These were samples for seven sites surrounding the Grand Canyon National Park, operated between 1984 - 1989.
For all the sites, the hourly light scattering data were aggregated into daily averages so that they were temporally compatible with the filter samples.
It is well established that the fine mass concentration (PM2.5)
measured by size segregated filter sampling has a strong statistical
correlation with total aerosol scattering. The main reason for
this relationship is that both the fine particle mass as well
as the light scattering efficiency factor have a peak in the size
range 0.3 - 0.6 µm 
. Exception to this relationship
occur when the characteristic aerosol size is either smaller (e.g.
primary automobile exhaust) or larger (wind blown dust) than the
above size range. The most detailed discussion of the topic is
contained in the NAPAP State of Science Report 24. (Acid Deposition:
State of Science and Technology, Volume III, National Acid Precipitation
Assessment Program, Washington, D.C., 1991)
A comparison of the light scattering coefficient and PM2.5 is shown for thirteen different sites.
| Relationship between daily average light scattering and PM2.5 | ||||||
|---|---|---|---|---|---|---|
| Canby, OR | Eugene, OR | 80 Portland, OR | 15 Portland, OR | Medford, OR | Clayton, MO | St. Ann, MO |
 |
 |
 |
 |
 |
 |
 |
| San Bernadino, CA | Azusa, CA | Sacramento, CA | Bakersfield, CA | Stockton, CA | Modesto, CA | |
 |
 |
 |
 |
 |
 |
|
The goodness of the correlation for these data is comparable to numerous other studies conducted over the past twenty years. However, it is remarkable that the light scattering to PM2.5 ratios are about a factor of two higher than corresponding literature values. A recent review of the existing data by White (NAPAP Tables 24-13 and 24-15) shows that this ratio is between 2.0 and 5.0 m2/g for about 30 eastern and western, urban and rural sites. White also recorded two of the eastern sites having a mass extinction efficiencies of about 10 m2/g.
A good example of light scattering data with low mass extinction efficiency is obtained during the SCENES project (Annual; Quarterly). The slope for the entire data set is 2.78 m2/g and it ranges seasonally between 2.5 m2/g in the summer and 3.4 m2/g in the winter. The R2 for the entire SCENES data set is 0.42, which is substantially lower than the correlations found in this study.
The compilation of light scattering efficiencies by White is reproduced in a table which also includes the results of this study. A summary of fine particle mass ratios from different data sets is also shown in by a scatter plot. The open rectangles show White's compilation while the dark triangles show the results of this analysis. It is clear that the results of this cursory analysis tend to indicate substantially higher extinction efficiencies than the bulk of the literature values. There is also a significant variation of the light scattering ratio between the monitoring sites. It was beyond the scope of this examination to identify the causes of these discrepancies. Hence it is not clear what specific light scattering instruments (nephelometers) were used, what were the sampling (humidity) conditions, and what is the accuracy of these light scattering data sets.
A difference between fine mass and light scattering data sets is the temporal coverage. The hourly light scattering measurements reveal the aerosol fine structure in time that is not available from filter samples collected every sixth day.
The nature of the high time resolution aerosol signal is illustrated through time series for six selected stations, Canby (Portland), OR, Eugene, OR, Medford, OR, Stockton, CA, Azusa, CA, and Clayton (St. Louis), MO.
| Temporal pattern of hourly light scattering, daily average light scattering, and PM2.5 mass concentration | ||||||
|---|---|---|---|---|---|---|
| Canby, OR | Medford, OR | Eugene, OR | Stockton, CA | Azusa, CA | Clayton, MO | |
| Year |  |
 |
 |
 |
 |
 |
| January | 
| 
| 
|  | 
| 
|
| July | 
| 
| 
|  | 
| 
|
Canby, OR is a near urban site adjacent to Portland, OR. The light scattering signal indicates that the highest light scattering average and the high excursions occur during the winter season, December - March. The monthly chart for January shows the occurrence of short term, daily aerosol events as well as aerosol accumulation for about a week (January 23-29). Throughout July, the aerosol signal is low and uneventful.
Eugene, OR displays temporal trends similar to those of Canby, OR. The greatest light scattering occurs during the winter months. The month of January shows high peak values of light scattering and an order of magnitude variation from one day to another. July is less variable with low values.
Medford, OR is a valley site in southern Oregon that showed frequent fine mass concentrations well in excess of 100 µm/m3. The site is characterized by a strong winter peak when both the absolute concentrations and the variability is high. The monthly chart for January clearly shows the existence of long term episodes that may last for a week or more (the squares that represent PM2.5 in the plot are difficult to discern due to their overlap with the light scattering data). The summer concentrations at Medford are low and uneventful.
Stockton, CA also exhibits a strong winter peak. The fluctuations between aerosol events exceed a factor of ten. The time chart for January shows the occurrence of a long episode at about seventy µm/m3 that lasts for about two weeks. Subsequently, the concentrations drop to below 5 µm/m3. The summer concentrations at Stockton are below 10 µm/m3 and relatively constant.
Azusa, CA is in the Los Angeles basin and it is exposed to the Los Angeles smog. (Evidently the light scattering coefficient data at Azusa are quantized) The seasonal pattern at Azusa is rather uniform high concentration aerosol peaks occur throughout the year. The monthly charts for January and July also indicate that aerosol events lasting two to four days occur during both winter and summer.
Clayton, MO is a suburb of St. Louis and it is exposed to the regional haze which is characteristic for much of the eastern U.S. The highest aerosol concentration and light scattering occur during the summer season between June and September, distinctly different from the Western sites. The winter concentrations are relatively low between 5 and 15 µm/m3. The concentrations for July show the occurrence of week long haze episodes that alternate with cleaner air.
The high time resolution light scattering data clearly indicate that the aerosol variation is significant in both season and monthly time scales. Western monitoring sites (except in Los Angeles) show a winter peak while the eastern U.S. exhibits a summer peak in fine aerosol concentration and light scattering. Qualitative inspection of the time charts also indicates that the western sites, e.g. Canby, Medford, Stockton, exhibit somewhat more temporal texture in the winter than the peak summer episodes in St. Louis. However in this analysis, we have not pursued a quantification of the temporal signal variation. The only additional temporal analysis pertains to the diurnal cycle of light scattering discussed next.
Data on the diurnal cycle can only be obtained from high time resolution (hourly) continuous monitoring, e.g. light scattering. The diurnal cycle charts were obtained by averaging all of the available light scattering data for a specific hour of the day and season. The examination of the light scattering data for the above six "characteristic" sites reveals a somewhat varied diurnal pattern that differs from site to site as well as seasonally for each site.
| Diurnal pattern of hourly light scattering for January, April, July and October | ||||||
|---|---|---|---|---|---|---|
| Canby, OR | Medford, OR | Eugene, OR | Stockton, CA | Azusa, CA | Clayton, MO | |
 |
 |
 |
 |
 |
 |
|
Eugene, OR exhibits little diurnal variation during the spring and summer while during the fall and winter it, unlike Canby, OR, displays high diurnal modulation. The light scattering peaks between 6 and 10PM with the lowest values occurring around 4PM.
Medford, OR shows one of the highest diurnal modulations of about 50% of the daily mean. Furthermore, the diurnal cycle at Medford changes shape from one season to another. In January, the peak concentration is at midnight while the lowest light scattering occurs at 6 AM. In April and July, there is a strong peak at about 8 AM which shifts to a 9 AM peak in October. The lowest light scattering in the fall occurs in the afternoon. The diurnal cycle of light scattering in Medford is a clear indication of local sources in the air basin where the sampler is located. The shape and magnitude of the diurnal cycle is influenced by the emission pattern, the atmospheric ventilation, and possibly by the aerosol formation rate during fog conditions.
Stockton, CA also exhibits a moderate diurnal pattern during the cold season, January and October. However, the peaks are shifted from 10 PM in January to 6 AM in October. During July, the light scattering is constant throughout the day.
Azusa, CA also shows a diurnal cycle; a mid day peak 8-12 AM during April - October and an early night peak at 10 PM during January. Again, this diurnal cycle of light scattering at Azusa is determined by the smog emission, transport within the L.A. basin, as well as by the diurnal cycle of the smog formation.
Clayton, MO shows minimal diurnal variation of light scattering. The values are somewhat elevated during the night and early morning and less during the midday hours. It is quite remarkable that during the summer months when the light scattering is the highest, the average Clayton values are constant throughout the day. This tends to suggest that local sources and aerosol accumulation during peak hours is not significant in determining the local light scattering coefficient.
The above analysis indicates that there is a measurable diurnal modulation of up to 50% of the daily average of Western Valley sites where primary particle emissions are significant. However, at other sites, particularly over the east (St. Louis) the diurnal fine particle variation is remarkably low, consequently at those locations, the information content of hourly data is similar to the daily average.
The relationship between fine particle mass and light scattering can be obscured by many physical factors and sampling errors. This cursory analysis prevents us from identifying the causal factors for outliers. Therefore, in the brief discussion below, we merely highlight and expand on the conditions when outliers in the scattergram occurred. For this reason, the statements below are to be taken with a great deal of caution.
A closer analysis of the correlation plots in Figure 2 revealed the presence of days with unusually high light scattering values or PM2.5 concentrations. Evaluating these "outlying" points may provide better insight into the behavior of the BSCAT-PM2.5 relationship. In the following event analysis, hourly and daily light scattering, PM2.5, PM10 and TSP data are examined for three locations: Eugene, OR, Azusa, CA, and Rubidoux/San Bernadino, CA. Two aerosol events with extreme values are plotted for each location.
| Aerosol time series for observations that are outliers on the Bscat-PM2.5 scatter chart | ||||||
|---|---|---|---|---|---|---|
| Eugene, OR (Feb 11-25, 1985) |
Eugene, OR (Jan 19-Feb 1, 1986) |
Azusa, CA (Jul 7-20, 1995) |
Azusa, CA (Jan 13-27, 1994) |
San Bernadino, CA (Mar 9-23, 1995) |
San Bernadino, CA (Oct 22-Nov 4, 1994) |
|
 |
 |
 |
 |
 |
 |
|
The Azusa, CA correlation chart in Figure 8b shows two days with a light scattering of approximately 600 (Mm)-1 with corresponding PM2.5 concentrations of about 25 and 90 µg / m3. The PM2.5 concentration of 25 µg / m3 occurred on July 14, 1995. The figure shows a light scattering peak on the fourteenth. PM2.5 and PM10 concentrations remained relatively constant over the two week period while the TSP concentration exhibited a peak on the fourteenth. The ozone data did not show any unusually high concentrations on or around the 14th. The other high light scattering day in Azusa occurred on January 20, 1994. Daily average light scattering, PM2.5 and PM10 all peaked on this day and resulted in the highest PM2.5 concentration for the site as seen in the scatter plot.
Rubidoux/San Bernadino, CA shows high Bscat to PM2.5 ratios on March 17, 1995 and October 29, 1994. The rise in daily average light scattering on the March date has a corresponding increase in aerosol concentrations (PM2.5, PM10, TSP). The actual peak of the light scattering event occurs on the 19th on a day with aerosol concentration data. The October plot displays a steady decrease in all aerosol concentrations. The daily average light scattering shows a relatively constant value until the 29th, when it peaks slightly and then decreases.
| Submit your comments, feedback, questions, and ideas pertaining this page. Your input will be automatically added to the existing annotations. In order to add a new comment, you must be registered with the CAPITA People's Page. |
