Developments in Environmental Science, Volume 9 Allan H. Legge (Editor) Copyright r 2009 Elsevier Ltd. All rights reserved. ISSN: 1474-8177/DOI:10.1016/S1474-8177(08)00202-7
35
Chapter 2 Use of Trace Metals as Source Fingerprints Richard L. Poirot Abstract This chapter introduces and updates the Fast Aerosol-Sensing Tools for Natural Event Tracking (FASTNET) and Combined Aerosol Trajectory (CATT) tools, and gives examples of applications for analyzing aerosol pollution events and space/time patterns and trends in selected aerosol ionic, carbonaceous, and trace element species. The goal is to encourage use of and feedback on these tools from a broad, international community of air pollution researchers and analysts, so that a growing herd of ‘‘fast cats’’ may soon enhance the rate at which our collective knowledge of the causes and effects of air pollution evolves.
2.1. Introduction
In an ideal world, the pace at which we are able to find, acquire, understand, merge, and analyze air pollution-related data would be limited only by the speed of our mouse clicks, as we sit sipping latte in a wireless Starbucks Cafe´, and all the information we needed would appear magically on our laptop computer screens. In the real world, such information rarely lies at our fingertips, we face obstacles such as those articulated in the 1989 National Academy of Science (NAS, 1989) report on ‘‘Information technology and the conduct of research: The users view.’’ The researcher is not aware of all the relevant data; if he is aware of the data, he cannot get access to them; if he can access the data, he cannot read them; if he can read the data, he does not know how good they are; and if he finds the data to be good, he cannot merge them with other data. Corresponding author: E-mail:
[email protected]
36
Richard L. Poirot
There have been considerable advances in the field of information technology in the 15 years since this rather gloomy NAS assessment, and there is currently a multitude of air quality-related data accessible in various online data archival and retrieval systems. At the same time, the air pollution researcher’s ‘‘data problem’’ has taken on new dimensions, as the sheer number, size, and complexity of the ‘‘relevant’’ data have expanded exponentially. This ‘‘data deluge’’ problem is especially acute for those with an interest in aerosol pollution, as aerosols are so inherently complex and as there are so many different kinds of relevant data—from extensive, new, surface-based monitoring networks, meteorological and aerosol forecast models, satellite imagery and associated data products, etc. FASTNET (Fast Aerosol-Sensing Tools for Natural Event Tracking; Poirot et al., 2004) and CATT (Combined Aerosol Trajectory Tools; Husar et al., 2004) are two recently developed, online data acquisition and analysis tools that will help improve our efficiency as air quality analysts. Both tools were developed by the Center for Air Pollution Impact and Trends Analysis (CAPITA) at Washington University, with funding support from the five U.S. Regional Planning Organizations (RPOs), and are maintained by CAPITA under the larger domain and infrastructure of the DataFed Federation (http://datafed.net/, last accessed on July 15, 2008), developed with initial funding support from the National Science Foundation (Husar, 2001) and the National Aeronautics and Space Administration (NASA) (Falke & Husar, 2003). This chapter introduces and updates the FASTNET and CATT tools, and gives examples of applications for analyzing aerosol pollution events and space/time patterns and trends in selected aerosol ionic, carbonaceous, and trace element species. The goal is to encourage use of and feedback on these tools from a broad, international community of air pollution researchers and analysts, so that a growing herd of ‘‘fast cats’’ may soon enhance the rate at which our collective knowledge of the causes and effects of air pollution evolves.
2.2. Methods
FASTNET and CATT are two of several ‘‘projects’’ developed and maintained within—and dependent on the fundamental data and architecture developed for—the DataFed Federation (http://datafed.net/; see Fig. 2.1, last accessed on August 19, 2008). Both projects use a common data viewer (Fig. 2.2) that allows users to access, explore, screen, aggregate, layer, display, and acquire air quality-related data
Figure 2.1.
Datafed Federation (http://Datafed.net, last accessed on July 17, 2008).
Use of Trace Metals as Source Fingerprints 37
38
Richard L. Poirot
Figure 2.2. Data Viewer (http://webapps.datafed.net/datafed.aspx, last accessed on July 17, 2008).
from roughly 100 different sources. These various data can be selected from an extensive data catalog (‘‘data catalog’’) through which some data are directly accessed from their native data repositories, maintained by the data generators, whereas others are automatically cached by DataFed.net to ensure archiving, facilitate navigation and merging of relational space/time features, or generate ‘‘value-added’’ data products. For the FASTNET project, 14 specific datasets, summarized in Fig. 2.3, are emphasized in the DataFed catalog. These include various surface-based aerosol, meteorology, and visibility data; aerosol forecast model results, and satellite data and images. Many of these data sets are available in near-real time, but have only become available (or archived) quite recently, while others, such as the filter-based aerosol chemistry data from the IMPROVE network (Interagency Monitoring of Protected
NOAA/NCDC
DOD/NRL
USDA/USFS
NASA/USFS
GOES
NASA/GSFC
NASA/CAPITA
NPS/CIRA
NPS/CIRA
MWH
NPS
NEXTRAD
NAAPS
HMS FIRE
MODIS FIRE
GASP
TOMS AI
SEAWiFS
VIEWS CHEM
ATAD
MWH webcam
NPS webcam
FASTNET datasets and characteristics.
5/1/2003
NOAA/STI
ASOS STI
Figure 2.3.
7/30/1998
NOAA/PSWC
4/5/2004
7/19/2004
1/1/1988
3/1/1988
1/1/2000
7/25/1996
7/16/2004
8/1/2003
10/1/2003
1/1/2001
4/19/1997
7/1/2002
91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 Start Date
SURF MET
90
EPA/STI
89
AIRNOW
88
Provider
Dataset/ Time
aerosol aerosol
hour 6 hour day day .5 hour day day 3 hour
now now now now now now
aerosol archived archived
day day
now now
aerosol
meteorol
aerosol
fire
fire
aerosol
meteorol
aerosol
6 hour
cache
archived
cache
5/31/2002
1/1/2003
12/31/2002
aerosol
hour
meteorol
now
cache
hour
aerosol
now
cache
Type
hour
Time Resolution
now
End Date
Use of Trace Metals as Source Fingerprints 39
40
Richard L. Poirot
Visual Environments) are available with a time lag of approximately 1 year, but have a longer historical record. The CATT project currently uses ‘‘only’’ the IMPROVE aerosol data and 5-day backward trajectories calculated for IMPROVE sites and sample dates by the National Park Service, using the National Oceanic and Atmospheric Administration (NOAA) Atmospheric Transport and Dispersion (ATAD; Heffter, 1980) model. CATT links the aerosol chemistry and trajectory data, allowing detailed exploration of space, time, and composition patterns. Additional details on FASTNET data tools, techniques, and sample applications are available in Poirot et al. (2004). The FASTNET homepage, illustrated in Fig. 2.4, can be accessed at http://datafed.net/ projects/FASTNET/FASTNET_Links.htm (last accessed on July 15, 2008). Additional details on CATT data, tools, techniques, and sample applications are available in Husar et al. (2004). The CATT homepage, illustrated in Fig. 2.5, can be accessed at http://datafed.net/projects/catt/ CATT_Links.htm (last accessed on July 15, 2008).
2.3. Sample applications
‘‘SURF_MET’’ is one of the many useful FASTNET datasets, and includes hourly global surface meteorological observations from more than 10,000 sites, with especially dense coverage over North America. These data are provided initially and in near-real time by the NOAA World Data Center (WDC) for Meteorology (http://www.ncdc.noaa.gov/ oa/wmo/wdcamet.html, last accessed on July 15, 2008), in Asheville, NC, but the historical archive is not accessible online. Starting in mid-1998, the Plymouth State Weather Center (PSWC; http://vortex.plymouth.edu/, last accessed on July 15, 2008) at Plymouth State College, New Hampshire began routinely downloading and archiving these data, and has made them accessible to CAPITA. At CAPITA, the WDC/PSWC data are routinely downloaded and reformatted in a relational database to facilitate exploration of both space and time patterns in the data. CAPITA also uses the visibility (visual range), temperature, and dew point data to calculate several ‘‘processed data variables,’’ including light extinction (Bext—in inverse megameters, calculated in this case as 3000/ Visual Range in kilometer), humidity-screened light extinction (FBext— with hours with relative humidity W90% filtered out), and humidityscreened and adjusted light extinction (RHBext, which first filters out RH W90%, and then reduces the remaining data values by an inverse hygroscopic growth function). This RHBext variable relates very closely to concentrations of fine particle PM2.5 mass, but includes much more
Figure 2.4.
FASTNET website (http://datafedwiki.wustl.edu/index.php/FASTNET, last accessed on July 17, 2008).
Use of Trace Metals as Source Fingerprints 41
42
Richard L. Poirot
Figure 2.5. CATT website (http://datafed.net/projects/catt/CATT_Links.htm, last accessed on July 17, 2008).
dense spatial and temporal coverage than current PM2.5 networks. CAPITA also adds a wind vector display option (SURF_MET_WIND) based on the wind speed and direction data. These processed data variables, along with the raw data are then redistributed via FASTNET, providing universal access to highly ‘‘usable’’ forms of the data— generated initially by 10,000 individual weather observation stations, collected and posted by the WDC, downloaded and archived by the PSWC, accessed, processed, and redistributed by CAPITA, and subsequently employed by a broader FASTNET ‘‘user community’’ for analysis of air pollution events. The raw data were not collected for this purpose, but take on added value as they pass through many helping hands. Figure 2.6 shows both the RHBext and wind vector spatial displays over the U.S. for February 19, 2004 at 17:00 UTC. Note the two separate
-125
-120
-115
Figure 2.6.
-110
-105
-100
-95
RHBext and wind vectors on February 19, 2004, 17:00 UTC.
Data Provider: Plymouth State Weather Center
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
SURF_MET.RHBext: Surface Extinction Coefficient & Surface Wind Vectors
-90
-85
-80
-75
Delivery: DataFed.Net
-70
800 450 100
RHBext
KPTT 2004-02-19T17:00:00 190
Use of Trace Metals as Source Fingerprints 43
44
Richard L. Poirot
areas of high (dried) extinction—one along the eastern half of the U.S.– Canada border, and another near the Texas–Mexico border. The northeastern event, associated with very low wind speeds, was subsequently identified by the FASTNET user community as being caused primarily by high ammonium nitrate concentrations under low mixing heights (Poirot & Husar, 2004b). The southwestern event, associated with very high wind speeds, was subsequently identified by the FASTNET user community as being caused primarily by high windblown dust concentrations (Husar & Poirot, 2004). Figure 2.7 shows other FASTNET options and data, illustrating the smoke impacts from the July 2002 Quebec forest fires with a spectral reflectance (350 nm) image from the SEAWIFS satellite, underlying the concurrent 1-h SURF_MET RHBext data on the left, and the 24-h IMPROVE organic carbon (OC; red) and sulfate (SO4; yellow) on the right side of the figure. The satellite view integrates across the vertical column, and the surface data indicate where the highest smoke concentrations mixed down to the surface. Surface impacts were highest in the eastern Mid-Atlantic region, with generally lower concentrations closer to the source area, as much of the smoke passed by aloft (Poirot & Husar, 2004a; Taubman et al., 2004). Zooming out (FASTNET zoom tool) to cover a slightly larger area and adding the IMPROVE chemical data on the right, it can also be observed that there were also concurrently moderately high SO4 concentrations, but ‘‘displaced’’ to the west and south of the Ohio River Valley region, which most typically experiences the highest summer SO4 concentrations. For readers who may have an interest in further exploration of the combined SEAWIFS images and surface meteorology data, the view (and the underlying data that can be browsed) used for the left side of Fig. 2.7 can be found by opening the FASTNET Viewer and then selecting ‘‘File,’’ ‘‘Open Page,’’ ‘‘RichP,’’ and ‘‘Qfire1.page.’’ From here, you can explore other time periods of interest, zoom or pan to other regions, add or delete other datasets, change data scales and rendering options, download (some) data in ASCII csv format, and save your page by ‘‘File,’’ ‘‘Save Page As,’’ and then create your own directory and file name. One of the various and most ‘‘simple’’ options provided by the CATT tool is to select a single sample day for the IMPROVE network and plot the back trajectories (4 per day, 5 days back) for all the sites with data on that date. Relative concentrations of different chemical species of interest can also be plotted, and the trajectories can be ‘‘color weighted’’ to emphasize the flows associated with highest (and lowest) species concentrations. Note that the scales in Fig. 2.8 are different from each other for the color-weighted trajectories, ranging from W20 mg/m3 for OC, to 10 mg/m3
-86
-84
-82
-80
-78
-76
-74
-72
-68
-66
Delivery: DataFed.Net
-70
RP11 2002-07-07 116.00.00 NaN
Delivery: DataFed.Net
-100 -98 -96 -94 -92 -90 -88 -86 -84 -82 -80 -78 -76 -74 -72 -70 -68 Data Provider: Coop Inst for Research Atmosphere, Colo State
24
26
28
30
32
34
36
38
40
42
44
46
48
50
Figure 2.7. July 7, 2002, SEAWIFS Spectral Reflectance (350 nm) and Surface RHBext (left panel) with fine organic carbon (red) and sulfate (yellow) from IMPROVE sites (right).
Data Provider: Plymouth State Weather Center
-88
32
34
36
38
40
42
44
46
48
50
SURF_MEI.FBext: Surface Extiction Coefficient
Use of Trace Metals as Source Fingerprints 45
Figure 2.8. CATT color-weighted back trajectories for IMPROVE sites on July 7, 2002, unweighted (top left) and weighted by high (red) OC (top right), SO4 (bottom right), and chloride Cl (bottom left).
46 Richard L. Poirot
Use of Trace Metals as Source Fingerprints
47
for SO4 and 1 mg/m3 for chloride (Cl). The trajectories for highest OC concentrations on July 7, 2002 clearly implicate the central Quebec origin of the smoke, whereas those weighted by highest SO4 reside primarily over the Ohio River Valley—even as the highest observed SO4 concentrations were displaced to the south and west of that high emission region. On the same day, the highest (but not very high) Cl concentrations were observed at far western sites and associated with marine flows. A similar set of color-weighted trajectories are displayed in Fig. 2.9 for August 12, 2002, when a major SO4 event was impacting the Northeast (upper left) and northwestern forest fires were contributing to high OC at several western sites (upper right). A major carbon study was being conducted at Yosemite National Park during this time period (McMeeking et al., 2004) and carbon-14 dating, soluble potassium, and a number of molecular organic wood smoke tracers, including levoglucosan, retene, vanillian, dehydroabietic acid, and 7-oxo-dehydroabietic acid, all confirmed the strong influence of wood smoke—traced, in turn, primarily to fires in Oregon, based on OC concentrations during this time period (Engling et al., 2004). The origins of smoke plumes from large, isolated forest fires, such as those impacting the Northeast on July 7, 2002 and the Northwest on August 12, 2002, are relatively easy to trace, in part because the emissions modulations are so extreme in space and time. The origins of large SO4 events, however, are much more difficult to trace to individual sources, because there are so many contributing sources, and because SO4 is predominantly a secondary pollutant, resulting from highly variable atmospheric transformation of gaseous SO2. The trajectories in the lower two panels of Fig. 2.9 are color weighted by highest concentrations of selenium (Se; left) and nickel (Ni; right). For this event, the highest selenium trajectories correspond to the highest SO4 trajectories, passing over the Ohio River Valley, whereas the highest nickel trajectories are much more constrained to the Northeast. Expanding from individual events to multi-year patterns, several of the CATT tools allow analyses of long-term patterns based on various gridded ensemble trajectory metrics (see bottom row of CATT Links page in Fig. 2.5). One of these optional gridded trajectory metrics is referred to as ‘‘incremental probability,’’ which is described in detail in the CATT Illustrated Instruction Manual. Briefly, a grid, for which domain and grid-square size can be user defined, is employed to aggregate trajectory endpoints from large numbers of trajectories for each grid square. For a given site and time domain (also user defined), an ‘‘all day probability field’’ is calculated for all IMPROVE sample days. A grid square’s probability is expressed as the fraction of trajectory endpoints in that
Figure 2.9. CATT color-weighted back trajectories for IMPROVE sites on August 12, 2004 emphasizing highest concentrations of SO4 (top left), OC (top right), Se (bottom left), and Ni (bottom right).
48 Richard L. Poirot
Use of Trace Metals as Source Fingerprints
49
square divided by the total in all squares. For a selected pollutant species, a ‘‘high day probability field’’ is calculated in the same way but limited to a high subset of pollutant concentration days, where ‘‘high’’ (or any other part of the distribution) can be defined by the user either as an absolute value or as a percentile range of the distribution. The incremental probability is the high day probability minus the all day probability. How much more likely is a location to be upwind if the pollutant was higher than it is on an everyday basis? Other optional ensemble trajectory metrics include the potential source contribution function, upwind average concentration, and concentrationweighted probability. Each can be calculated for individual sites, or aggregated for groups of sites. The gridded results can be displayed in CATT, with user-defined rendering options, and/or exported in ASCII csv format, which includes the latitude and longitude of grid-square locations and the calculated metric. These csv files can then be entered directly into spatial analysis programs such as ArcViews for additional analysis and more refined plotting. Figure 2.10 shows the CATT incremental probability fields for highest selenium and nickel aggregated for all IMPROVE sites from 2000 through 2003. If selenium is high, the air has most likely resided over the ‘‘Midwest,’’ and if nickel is high, the air has most likely resided over the ‘‘East Coast,’’ especially the Northeast and Southeast coastal areas. In Figure 2.11, these gridded incremental probability data have been exported to ArcView, interpolated, and are compared with (1998 EPA EGRID) U.S. SO2 emissions from coal (left) and oil (right) utility sources. Selenium appears to be a good tracer for influence from coalburning emissions, whereas nickel (like vanadium) is a good tracer for residual oil combustion. These observations are consistent with recent modeling results from the Positive Matrix Factorization (PMF) and Unmix receptor models for a number of northeastern IMPROVE sites (Coutant et al., 2002; Lee et al., 2003; Poirot et al., 2001; Polissar et al., 2001; Song et al. 2001), which identified coal and oil combustion sources with strong selenium and nickel contents, respectively. For those receptor-modeling studies that also included trajectory interpretation of the sources (Poirot et al., 2001; Polissar et al., 2001; Lee et al., 2003), the coal and oil sources were associated with flows from the Midwest and Northeast urban corridors, respectively. The models, which can only identify sources with fixed chemical compositions, also tended to divide the coal source influence into two separate ‘‘source components,’’ one with a high sulfur:selenium ratio and one with a lower sulfur:selenium ratio, interpreted as representing the maximal and
50
Richard L. Poirot VIEWS_OL:SEf LYBR1 2002-07-31 NaN ug/m3
Combined Aerosol Trajectory Tool (CATT) 54 52 50 48 46 44 42 40 38 36 34 32 30 28 SEf ug/m3 26 0.001 0.0008 24 0.0006 0.0004 22 0.0002 0 20 -125 -120
Monitoring Network IMPROVE
Se -115
-110
-105
-100
-95
-90
-85
-80
Data Provider: Coop Inst for Research Atmosphere, Colo State
-70
Delivery: DataFed.Net
VIEWS_OL:NIf LYBR1 2002-07-31 NaN ug/m3
Combined Aerosol Trajectory Tool (CATT) 54 52 50 48 46 44 42 40 38 36 34 32 30 28 NIf ug/m3 26 0.001 0.0008 24 0.0006 0.0004 22 0.0002 0 20 -125 -120
-75
Monitoring Network IMPROVE
Ni -115
-110
-105
-100
Data Provider: Coop Inst for Research Atmosphere, Colo State
-95
-90
-85
-80
-75
-70
Delivery: DataFed.Net
Figure 2.10. Incremental probability fields for top 10% Se (top) and Ni (bottom), all IMPROVE sites 2000–2004.
minimal degrees of secondary aerosol formation—from coal sources— encountered at the receptor sites, respectively. This concept of varying sulfur:selenium ratios can be further explored over a larger domain by applying the CATT ‘‘upwind average concentration’’ calculation. For each grid square, an average pollutant
Use of Trace Metals as Source Fingerprints
51
Figure 2.11. Incremental probabilities for highest 10% Se (top) and Ni (bottom) from all IMPROVE sites 2000–2002, Compared with 1998, SO2 emissions from coal (top) and oil (bottom) utility boilers.
52
Richard L. Poirot
concentration is calculated for a receptor site for all trajectories passing through that grid square en route to the receptor. Like the incremental probability metric, this calculation can also be aggregated across multiple receptor locations, following the approach of Kenski (2004). Figure 2.12 displays the average upwind selenium (top) and sulfate (bottom) for two different 4-year time periods, 1992–1995 (left) and 2000–2003 (right). This analysis was constrained to 42 IMPROVE sites that commenced operation before 1992, so that any apparent changes over time would not be influenced by the addition of new monitoring sites closer to or further from emission source areas. The spatial patterns for both pollutants and time periods are generally quite similar, although the magnitude of the average upwind sulfate has declined substantially in recent years, whereas the magnitude of the average upwind selenium has remained relatively constant. For whatever reason(s), the effective control strategies for SO2—including both stack scrubbers and switching to lower sulfur coals—have had relatively minimal influence on selenium emissions. Figure 2.13 compares sulfate and selenium as measured at individual sites across the IMPROVE network for the 2000–2003 period on the left with the upwind average sulfate and selenium as calculated for individual grid cells during this same time period on the right. The measured species are not well correlated (R2 ¼ 0.36) due to variable secondary transformation rates that result in high sulfate:selenium ratios during periods of rapid transformation (typically during summer) and much lower ratios during colder, less humid periods when transformation rates are much slower. However, both species strongly tend to come from the same upwind locations. Figures 2.14 and 2.15 provide two perspectives on the shift in sulfate:selenium ratios over time and space. Figure 2.14 compares the upwind sulfate:selenium ratios over the 1992–1995 and 2000–2003 time periods, and shows that although the correlations are strong in both time periods, the slope has decreased by approximately 12% in recent years. Per unit of selenium, we are seeing less sulfate. Figure 2.15 shows the shifting spatial pattern of sulfate:selenium ratios over time, with the darker shading emphasizing the highest selenium:sulfate ratios. During the earlier time period, the highest selenium:sulfate ratios occurred primarily in the Northwest, where the sulfate concentrations are lowest. But in recent years, as sulfate has declined, the area of higher selenium:sulfate ratio has shifted further east. Whether this reflects unique features of emissions control equipment or a shift to lower sulfur (Western) coal—which may have slightly higher selenium relative to sulfate—is a question for future study. Selenium remains as a good tracer
Figure 2.12. Average upwind Se (top) and sulfate (bottom) for 42 long-term IMPROVE sites, over two 4-year periods, 1992–1995 (left) and 2000–2003 (right).
Use of Trace Metals as Source Fingerprints 53
0
5
10
15
20
25
0
0.002
0.006
Selenium (ug/m3)
0.004
0.008
0.01
Average Upwind Sulfate (ug/m3) 0
2
4
6
8
0
0.0008
0.0012
0.0016 Average Upwind Selenium (ug/m3)
0.0004
y = 4365x R2 = 0.85
0.002
Figure 2.13. Sulfate vs. selenium at 42 IMPROVE sites, 2000–2002, as measured at the sites (left) and as calculated for upwind average grid locations (right).
Sulfate (ug/m3)
30
54 Richard L. Poirot
Use of Trace Metals as Source Fingerprints
55
Figure 2.14. Shift in upwind SO4:Se ratios over time.
for coal sulfate, but the quantitative relationship is not constant over space and time. The overall change in average upwind sulfate (1992–1995 minus 2000–2003) is plotted in Fig. 2.16. Changes have generally been greatest over areas in the eastern U.S. and Canada, where total SO2 emissions reductions have been greatest over this time period. The area of greatest reduction, over the southern Ohio River Valley, is displaced slightly to the southwest of locations where SO2 reductions have been greatest. Possibly this may reflect more efficient sulfate formation in moist, warm airflows moving northeast off the Gulf of Mexico around the backside of stagnating summer high-pressure systems. No change is evident in western areas, and (slight) increases are indicated over areas in Alberta (and Montana) and Mexico. These apparent increases are very small and well within the uncertainty of the analysis method. It should also be cautioned that these far northern and southern areas are near the edges of
56
Richard L. Poirot
Figure 2.15. Spatial pattern of changing upwind sulfate:selenium ratios, 1992–1995 (top) and 2000–2002 (bottom).
the ATAD trajectory domain and may be further influenced by the relative scarcity of rawindsonde data used to drive the model. Figure 2.17 displays average upwind aerosol nitrate concentrations over the two time periods, and shows a slight decrease over southern
Use of Trace Metals as Source Fingerprints
57
Figure 2.16. Change in upwind average SO4 (mg/m3) between 1992–1995 and 2000–2003.
California, with a slight increase (about 0.1 mg/m3) over the central ‘‘corn belt’’ region, where relatively high agricultural ammonia emissions (rather than NOx) appear to be an important factor for aerosol nitrate formation. This area of elevated upwind nitrate appears to extend into the Canadian plains, although this may also be related to flows of colder and relatively sulfate-free air most conducive to aerosol nitrate formation. The extent to which this increase in nitrate may be due to increases in agricultural ammonium and/or decreases in acidic sulfate concentrations is not clear. This apparent change may have also been influenced by network-wide changes in IMPROVE nitrate sampling methods, although those changes were estimated to result in lower (not higher) post-96 concentrations (Schichtel, 2006). Several ‘‘unusual,’’ large-scale springtime nitrate aerosol events have been observed in the past few years over a region extending east of the earlier-indicated area of highest upwind nitrate, along the U.S.–Canada border, including several in 2004 (Poirot & Husar, 2004b; Husar & Poirot, 2004) and one in 2005 (Kenski et al., 2005). In both cases, relatively large groups of data analysts in this large transboundary region were quickly attracted by common interest to exchange data and work on a collaborative group analysis. FASTNET tools and users made substantial contributions to these analyses, and the CATT can be employed for additional retrospective analyses as soon as the slower
58
Richard L. Poirot
Figure 2.17. Average upwind nitrate for 42 IMPROVE sites, 1992–1995 (top) and 2000– 2002 (bottom).
filter chemistry and trajectory data become available. Quite possibly such events are not so much unusual as they are typical, but have only recently become readily detectable with the recent proliferation of new continuous and speciation measurements, and—equally important—with
Use of Trace Metals as Source Fingerprints
59
the development and more widespread use of fast and powerful new data acquisition and analysis tools. 2.4. Conclusions
The intent here was simply to introduce the new FASTNET and CATT data acquisition and analysis tools and encourage others to try them out (http://datafed.net, last accessed on August 19, 2008). A number of specific sample applications were presented without intent to draw specific conclusions, but rather to provoke future exploration. Like many of the featured data that improve in value by passing through many hands, the analysis tools will also be improved with use and feedback from data analysts.
ACKNOWLEDGMENTS
The many datasets presented here were graciously provided by data generators too numerous to mention individually, but each are referenced in the datafed.net catalog. Special thanks to Serpil Kayin, (MARAMA) and Gary Kleiman (NESCAUM), who provided the management for the CATT and FASTNET projects, and of course to the toolmakers: Rudy Husar, Steffan Falke, and Kari Hoijarvi at CAPITA.
REFERENCES Coutant, B., Kelly, T., Ma, J., Scott, B., Wood, B., and Main, H. 2002. Source apportionment analysis of air quality monitoring data: Phase I final report. Prepared for the Mid-Atlantic/Northeast Visibility Union and Midwest Regional Planning Organization by Battelle Memorial Institute, Columbus, OH and Sonoma Technology, Inc., Petaluma, CA, USA. Engling, G., Herkes, P., Carrillo, J., Kreidenweis, S.M., and Collett, J.L. 2004.Organic aerosol composition in yosemite national park during the 2002 yosemite aerosol characterization study. Paper #66, Air &Waste Management Association International Specialty Conference on: Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Falke, S., and Husar, R. 2003. Application of NASA ESE data and tools to particulate air quality management. Proposal to NASA Earth Science REASoN Solicitation CAN-02OES-01. Heffter J.L. 1980. Air resources laboratories atmospheric transport dispersion model (ARLATAD). Technical Memo ERL ARL-81, NOAA, Rockville, MD, USA. Husar, R. 2001. Monitoring and analysis of large-scale aerosol events: Information technology research on collaboration in virtual workgroups. Proposal to NSF on Information Technology Research, NSF 00-126 ITR/AP-IM(GEO).
60
Richard L. Poirot
Husar, R., and Poirot, R. 2004. Texas-Mexico Dust Event, 19 February 2004. http://capita. wustl.edu/capita/capitareports/040219TexasDust/040219TexMexDust.ppt (last accessed on July 15, 2008). Husar, R., Poirot, R., Gebhart, K., Schichtel, B., and Malm, W. 2004. Combined aerosol trajectory tool (CATT) for the IMPROVE Chemical Dataset. Paper #97, A&WMA international specialty conference on regional and global perspectives on haze: Causes, consequences and controversies, Asheville, NC. USA. Kenski, D. 2004. Quantifying transboundary transport of PM2.5: a GIS analysis. Paper #247, 97th Annual A&WMA Conference, Indianapolis, IN, USA. Kenski, D.M., Husar, R., Harrell, M., Conaster, N., Swinford, B., and Turner, J. 2005. The great midwestern PM2.5 event of February 2005. http://www.inawma.org/files/2005-1011_The_Great_Midwestern_PM2.pdf (last accessed on August 16, 2008). Lee, J., Yoshida, Y., Turpin, B., Hopke, P., Poirot, R., Lioy, P., and Oxley, J. 2003. Identification of sources contributing to the mid-Atlantic regional aerosol. J. Air Waste Manag. Assoc. 52, 1186–1205. McMeeking, G.R., Kreidenweis, S.M., Carrico, C., Lee, T., Carrillo, J., Day, D.E., Hand, J.L., and Malm, W.C. 2004. Dry aerosol size distributions and derived optical properties during the yosemite aerosol characterization study. Paper #27, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, consequences and controversies, Asheville, NC, USA. National Academy of Sciences (NAS). 1989. Information technology and the conduct of research: The user’s view. Report of the Panel on Information Technology and the Conduct of Research, National Academy Press, Washington, DC, USA. Poirot, R., and Husar, R. 2004a. Chemical and physical characteristics of wood smoke in the northeastern US during July 2002: Impacts from Quebec forest fires. Paper #93, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Poirot, R., and Husar, R. 2004b. Winter PM Event over the Northeast and Quebec. http:// capita.wustl.edu/capita/ (last accessed on August 16, 2008). Poirot, R.L., Wishinski, P.R., Hopke, P.K., and Polissar, A.V. 2001. Comparative application of multiple receptor methods to identify aerosol sources in northern Vermont. Environ. Sci. Technol. 35, 4622–4636. Poirot, R., Husar, R., Funk, T., Raffuse, S., Dye, T., Kleiman, G., and Kenski, D. 2004. Aerosol and haze observations with the FASTNET distributed monitoring system. Paper #93, Air & Waste Management Association International Specialty Conference on Regional and Global Perspectives on Haze: Causes, Consequences and Controversies, Asheville, NC, USA. Polissar, A.V., Hopke, P.K., and Poirot, R.L. 2001. Atmospheric aerosol over vermont: Chemical composition and sources. Environ. Sci. Technol. 35, 4604–4621. Schichtel, B. 2006. A Discontinuity in the nitrate ion time series at June 1996. Data Advisory: http://vista.cira.colostate.edu/improve/Data/QA_QC/qa_qc_Branch.htm (last accessed on August 16, 2008). Song, X.H., Polissar, A.V., and Hopke, P.K. 2001. Sources of fine particle composition in the northeastern U.S. Atmos. Environ. 35, 5277–5286. Taubman, B.F., Marufu, L.T., Vant-Hull, B.-L., Piety, C.A., Doddridge, B.G., Dickerson, R.R., and Li, Z. 2004. Smoke over haze: Aircraft observations of chemical and optical properties and the effects on heating rates and stability. J. Geophys. Res. 109, D02206, doi:10.1029/2003JD003898.