IDENTIFICATION OF SOURCES
    CONTRIBUTING TO THE CONTAMINATION
OF THE GREAT WATERS BY TOXIC COMPOUNDS
                     A report prepared for:

                     Mellissa McCuIlough
                 Great Waters Program Coordinator
             Pollution Assessment Branch, ESD (MD-13)
             Office of Air Quality Planning and Standards
               U.S. Environmental Protection Agency
                     Durham, N.C. 27701


                            By
              Gerald J. Keeler and Jozef M. Pacyna
                    Air Quality Laboratory
                   The University of Michigan
                Ann Arbor, Michigan 48109-2029

                     Terry F. Bidleman
                Atmospheric Environment Service
                     4905 Dufferin Street
                 Downs view, Ontario M3H 5T4
                      Jerome O. Nriagu
                     Environment Canada
                National Water Research Institute
                  Burlington, Ontario L7R 4A6
                  Revision Date: 17 March, 1993

-------
                           DISCLAIMER


     This document was prepared by researchers in Great Waters-
related scientific disciplines, and a draft of this report was
reviewed by an expanded group of scientists at a workshop held in
November 1992 in Chapel Hill, North Carolina.  Other workshop
participants included representatives from the U.S. Environmental
Protection Agency, the National Oceanic and Atmospheric
Administration, the International Joint Commission, and the
affected States.

     This report has been reviewed by the Office of Air Quality
Planning and Standards, Pollutant Assessment Branch, U.S.
Environmental Protection Agency, and has been approved for
distribution as received from the team of authors.   Approval does
not signify that the contents reflect the views and policies of
the U.S. Environmental Protection Agency, nor does  mention of
trade names or commercial products constitute endorsement or
recommendation for use.

-------
                          TABLE OF CONTENTS







   EXECUTIVE SUMMARY	i





   ACKNOWLEDGEMENTS	xv





1.  INTRODUCTION	1





2.  ATMOSPHERJC DEPOSITION AS A MAJOR PATHWAY	2





3.  GENERAL PROCEDURES FOR SOURCE IDENTIFICATION	4





      3.1. Source characterization	5





      3.2. Source apportionment techniques	6





      3.3  Applications of source apportionment techniques for metals	8





      3.4  Problems in the application of source apportionment techniques for





          organic compounds: changes in chemical profiles	12





             3.4.1  Alteration of PAHs in the Environment	20





             3.4.2  Alteration of PCDDFs in the Environment	22





      3.5  Applications of source apportionment techniques for orgamcs	24





             3.5.1 Volatile organic compounds (VOCs)	24





             3.5.2 Semivolatile orgamic compounds (SOCs)	28





                   A. Pol\ cyclic Aromatic Hydrocarbons (PAHs)	28





                   B.  PCDDFs	33





                   C.  Pesticides	35





                   D.  PCBs	42

-------
4  IDENTIFICATION OF LOCAL SOURCES	45





       4.1. Major source categories for toxic compound emissions	 o





       4.2. Emission profiles for major source categories	87





       4.3. Emission profiles for diffuse sources of organics	92





              4.3.2 Air-surface exchange processes	92





                    a) Volatilization from Soils	92





                    b) Air-Water Gas Exchange	98





                    c) Air - Plant Exchange	101





       4.4. Evaluation of emission inventories. Comparison with European studies	102





       4.5. Application of source-receptor techniques to study the origin of pollution	104





5. IDENTIFICATION OF DISTANT SOURCES	105





       5.1. Emissions from North America outside the Great Waters Regions	107





       5.2. Emissions from sources outside of North America	109





       5.3. Application of source apportionment techniques for identification of





           impacts of emissions on the Great Waters from distant source regions	111





6.  BENEFITS FROM EMISSION REDUCTION	115





7. CONCLUSIONS	119





8. RECOMMENDATIONS FOR FUTURE RESEARCH ACTIVITY	121





       8.1. Measurement programs	121





       8.2. Modeling estimates	122





9. REFERENCES	124

-------
                             EXECUTIVE SUMMARY

       The Great Waters Program  as  written in  section 112(m) of the Clean  Air  Act
 Amendments (CAAA) specifies that U.S. EPA, in  co-operation with Commerce/NOAA,  shall
 conduct a program to identify and assess the extent of atmospheric deposition of hazardous air
 pollutants  (HAPs) to  the Great Waters  which include the Great  Lakes, Lake  Champlain,
 Chesapeake Bay, and other coastal  waters.   In order to meet the ambitious  goals set forth in
 the 1990 CAAA, EPA identified the need to determine the  state-of-knowledge of atmospheric
 deposition of HAPs as a major pathway for loadings to the Great Waters.

       This report evaluates the available data on the sources of HAPs  within and outside the
 watersheds of the Great Waters (Great Lakes, Lake Champlain,  and  Chesapeake Bay), and
 discusses the difficulties in deriving such quantitative information. An  effort is  made to  reconcile
 the emission data discussed with measured and estimated loading rates.  Detailed inventories of
 the sources and emission intensities of HAPs  has become an  indispensable tool in environmental
 management. The quantitative targets are amenable to legislative controls, and the emission of
 the HAPs is easier to regulate than  the resulting atmospheric deposition or food chain effects.
 Any attempt to  set guidelines on deposition rates for HAPs, in fact, requires that the sources of
 the pollutants be known, and that emission rates be determinable.

       Atmospheric deposition is one of the major sources of lead, arsenic,  cadmium,  mercury.
 PAHs, several organochlorine pesticides, (e.g.,  lindane, DDT, chlordane, dieldrin, toxaphene),
 PCBs, PCDDs.  and PCDFs measured in the Great Waters.  The other inputs include non-point
 sources, e.g., agriculture practices in  the region, urban runoff, leaching from landfills, etc., direct
 industrial  discharges, tributary inputs, and direct dumping  of  wastes.   Specific  examples of
 loadings  estimates are  given  in  the "Relative  Loadings"  section   of the  report.    While
 approximately half of the lead is of atmospheric origin, it is noted that tributaries account for a
 substantial  fraction as  well. However, designating the loadings as being from tributary  inputs
 maybe somewhat misleading as much of the lead and other contaminants  in the tributary waters
 are the results of atmospheric deposition and subsequent runoff.  There are many regions  where
high quality data are needed to fill-in  the mass balance estimates for many HAPs. Even with the
high-quality data in some areas it is  very difficult to get a reasonable mass  balance for a large
majority of the critical pollutants.

       Evidence of atmospheric deposition  as a  source of persistent  semi-volatile organic
compounds (SOCs) to water bodies is provided by their accumulation in soils, sediments,  and

-------
peat bogs  in  the  Great  Waters region and  other  locations.   The  trends  of  contaminant
concentrations with  depth in dated  layers of sediments  or peat cores track '".:ir  known
production/release history. These trends show peak accumulation of PAHs in the 1950s, PCBs
in the mid-1970s, and organochlorine pesticides in the late  1960s to mid-1970s, depending on
the chemical.   Rapid  increases of  PCDDs/PCDFs in Great  Lakes  sediments  after  1940
paralleled the production  of chlorinated aromatic compounds, suggesting that incineration of
chlorine-containing  waste was  the  most significant  contributor.    Other  indicators  of
atmospheric sources include  direct measurements of SOCs in air, rain, and snow  from the
Great Waters and remote regions; and accumulation of persistent organochlorine compounds in
biota from  small inland  lakes in the Northwest Territories, the high Arctic, and Antarctica.

       Identifying the specific sources or source types  emitting the pollutants into the atmosphere
which ultimately are deposited is another matter.   Identification  of the major sources and the
deposition  pathways of-the critical pollutants should be made for the individual  compounds
separately as their sources and behavior in the environment differ substantially.  Volatile organic
compounds (VOCs)  and SOCs are  emitted by both point and area sources.   Examples of the
former are stack and fugitive emissions from industrial  processes and incinerators.   Sources
that emit pollutants over broad areas include vehicle  exhaust and evaporation  of pesticides and
PCBs.

       In general, both local  and distant  sources contribute to  the pollution load at a given
receptor.  There are  various definitions concerning the meaning of local  and distant sources.  In
this work local sources are those in the states adjacent to the  Great Waters. For the Great Lakes
these  states are  Illinois,  Indiana, Michigan,  Minnesota,   New  York,  Ohio, Pennsylvania,
Wisconsin, and Ontario in Canada.  For Lake Champlain the emission sources in the states of
New York, Vermont and the Province of Quebec are considered local while local sources for the
Chesapeake Bay are located in the states of Virginia and Maryland.  Sources outside the above
defined regions are regarded here as distant, regardless of how much they may contribute to the
total loading of the Great Waters.  This  somewhat artificial division can be justified when
considering various policy measures to reduce the pollution load.  Local  and distant sources can
be of either anthropogenic or natural origin.  It is believed that natural sources are more important
when discussing  the impact  of  distant source emissions  on the atmospheric deposition  of
pollutants to the Great Waters.

       Although  a  number  of  source-receptor  techniques are  available for estimating the
contributions,  it  is still premature to conclude what part of  pollution load originates within the
study region and what part results from  long range transport.  The major reason for the present

-------
 uncertainty is the lack of reliable input data for application of these techniques including properly
 reported emission data.  The present lack of monitoring data as well as emissions information is
 also  problematic for  other  regions, eg. the  North  Sea  and  the  Baltic  Sea,  the two  most
 extensively studied regions in Europe with respect to the environmental behavior of hazardous air
 pollutants.

        In general, local sources are characterized on the basis of emission measurements and/or
 emission  estimates.   Representative measurements are considered  the best information for
 accurately describing emissions.  These measured  data can be used  for  source characterization
 directly or  indirectly.   Direct  use is when  stack tests  are  performed  using  both continuous
 monitoring or representative grab sampling during short measurement campaigns. Measured data
 can also be utilized indirectly by transformation of the measurements into an emission factor or
 inclusion in a special calculation procedure.

       Source characterization  through measurements is often very expensive,  and  on some
 occasions extremely difficult to perform. In these cases  other methods are often used, and in most
 circumstances are based on emission factors and /or mass balance calculations.  The transparency
 and comparability of the data used to elaborate or  select an emission factor from a handbook of
 emission factors are of great importance. The transparency of the data refers to the level of detail
 specified for the methods  used to prepare  a set  of emissions  factors,  e.g.,  measurements  of
 emissions  rates  or  concentrations  in  exhaust gases together with  the  technological  and
 meteorological conditions at the time that the measurements were obtained. Comparability of the
 data refers to emission factor verification/validation which includes comparison of the factors for
 a given source category obtained by various estimation methods.

       Material or mass balance calculations can be applied  to characterize emissions sources
through the assessment of their emission quantities.  The input quantities of the raw materials or
fuel and the output rates of specific pollutants are  determined.   These rates are used to assess
what fraction of a given pollutant is released in the gas phase while the balance is made for the
amount of the pollutant associated with particles.

       Emission sources can also be characterized using the concentrations of a given pollutant
measured in ambient air at a receptor site.  Measured concentrations are then compared with
emission source profiles for major source categories likely to contribute to pollution at the site.
This method, referred  to as receptor modeling, is  useful when emissions from a given source
region originate from one dominant source  or group of sources which have well denned emission
profiles.

                                            iii

-------
       An important pan of source characterization is to assess the accuracy of the methods used
to prepare emission profiles and to verify source data. The accuracy of source char:  ;-rization of
emissions depends  on whether the characterization  has  been  made  on  the  basis  of actual
measurements at the specific source or through estimation procedures.  The methods based on
measurements are considered as more accurate than those using assumptions and calculations,
e.g., methods based on emission factors or material balances.  Unfortunately,  most verification
procedures focus on activity data (statistical information) and emission factors.

       Dispersion models have been the traditional work horse  for calculating source-receptor
relationships  for air pollutants.  These models require detailed emissions inventories for  various
sources for the pollutants of interest, e.g., SO2, NOx, etc. Even if the dispersion models were
accurate it  is very unlikely that  the emissions  inventories  would be adequate.   Emissions
inventories for the criteria pollutants have  many  short-comings,  as  discussed earlier  in  this
document, and these inadequacies are even more  severe for HAPs or for pollutants which have
large contributions from fugitive process emissions, natural sources, and dusts.  The limitations of
the dispersion oriented  methods have led to the development  of receptor models.   Receptor
models assess contributions from various sources  based on observations at sampling or receptor
sites.

        Several methods are currently available to assess sources and source regions for  various
air pollutants based on the chemical composition of the air at a given receptor. Both the statistical
methods and modeling  are used together with meteorological  data   in  order to obtain  this
assessment.   The origin of the pollution measured at a given  receptor can be  studied using
information on the che^r^l composition of aerosols and/or mixture of gaseous pollutants.

        Statistical methods  have been developed which use  information  on  the  chemical
composition  of aerosols to study  contribution  of sources  or even  source  regions  to  the
contamination at a given receptor.   The  applicability of multivariate techniques for  resolving
sources and source regions for aerosols measured at several locations remote from major emission
regions has been tested using absolute principal component analysis (APCA) and the chemical
mass balance (CMB)  methods.  APCA methods indicate the composition of major components.
such as pollution, crust, and sea-salt components which contribute to the measured concentrations
at receptors.  In the past; the APCA methods were applied to air concentrations of total (both fine
and  coarse  fractions) aerosols.  Further improvement of this receptor modeling method  was
obtained by applying APCA to aerosol elemental concentration measurements in separate  particle
size fractions.  The results of this application of APCA gives  the basis for interpretation of
                                            IV

-------
 coupled chemical  reactions and  physical  processes in remote  locations,  as well as giving
 information concerning atmospheric aging processes and, therefore, the history of the compounds.

        Distinguishing  sources through  chemical  mass balance (CMB) models  and statistical
 methods (e.g., factor analysis, principal component analysis) is more  problematic  for organic
 compounds than metals.   This  is because VOCs and SOCs are  transformed by chemical
 reactions in the atmosphere, and different rates of reactivity lead to changes in ratios among
 compounds during transport from source to receptci.  In addition, some  SOCs  are associated
 to a greater extent with atmospheric particles than others.   Preferential removal of paniculate
 species by  precipitation and dry deposition can alter the relative proportion of SOCs and other
 pollutants in ambient air.  These changes are reflected in the chemical profiles of atmospheric
 deposition  and sediments.  For example, ambient air contains light and heavy PCBs, PCDDs,
 and  PCDFs that are distributed between the vapor and particle phases.   The heavier, more
 particle-bound compounds predominate in rain and sediments.

       With a total VOC emission rate  of  19.5 megatons (metric) per year, the United States
 leads the world in the release of most VOC types.  Emissions are highest in the eastern third of
 the country.  The relative contribution of various  industries and vehicles to ambient air VOCs
 varies with location. Examples are presented in this report showing the use of CMB modeling
 and factor analysis to estimate the contribution of these point and area sources. The agreement
 between CMB results and emission inventories is encouraging in many cases.

       Less work has  been done  with CMB and statistical  methods for reconciliation of SOC
 sources. This is largely due to: a) difficulties in sampling and analytical techniques  for SOCs,
 and b) alteration of chemical profiles by selective reactivity and physical removal of certain
 SOCs, as mentioned above.  In a few cases CMB models have been applied to estimating the
 contribution of PAH sources in urban air.  Principal  component analysis has  been used to
 distinguish  patterns of PCDDs and PCDFs  from different combustion sources in urban air, to
 examine changes in compound  profiles that  occur as a  result of  selective  atmospheric
 deposition,  and to differentiate PCDDs/PCDFs from combustion and pulp mills.

       Several  "marker"  compounds have  been  proposed  to help distinguish  PAHs from
 different sources:  wood  combustion, spark vs.  diesel engines,  and  unbumed vs. burned
petroleum products (combustion vs. street runoff). The main problem with  these markers is
 that few are unique to a particular source type, and they are best applied in combination with
 other organic and inorganic tracers and with multivariate methods such as  factor  and principal
component  analysis.

-------
       Identification of emission sources in the Great Waters regions and their characterization
with respect to atmospheric emissions has  been  carried  out for some time in botn the United
States  and Canada.  As a result,  major source categories  have been defined  for all of the
combustion-related HAPs,  and include these  sources:  production of electricity and  heat,
combustion of fuels in industrial, commercial, and residential units, including wood  combustion,
manufacturing and use of various  industrial goods, mobile source emissions, incineration of
municipal and industrial wastes, and incineration of sewage sludge.  Figure 1 displays the mercury
emissions from various source categories in the Great Lakes Basin.   There  are, however,
differences in quantitative assessment of the fluxes from the above sources,  reported by various
research groups in the United States and Canada.  These differences should be resolved through
thorough examination of the available  data using verification techniques for emission data and
joint supplementary research programs in both countries.

       A recent emissions inventory for Ontario and eastern North America has  been prepared
under contract for the Ontario Ministry of the Environment (MOE).  Inventories  for PAHs and
PCDDs/PCDFs  were derived after compilation of emission  factors  for  a large number of
source types, including industrial processes, vehicles, residential combustion (oil, gas, wood),
power plants, incinerators, open burning, and forest fires. Estimated annual releases in eastern
North  America were 9397 metric tons PAHs and 414 kg PCDDs/PCDFs.  The  breakdown of
PAH emissions by source type was:  residential wood combustion 31%, other stationary fuel
combustion (including power plants) 17%,  industrial processes  29%, transportation 12%,  other
11%.  In the case of PCDDs/PCDFs, stationary fuel  combustion and solid waste  incineration
each accounted for 46% of total releases.

       The uncertainties in these estimates are major and difficult to quantify.  This is largely
due to the quality of emissions data,  which are often incomplete and highly variable among
sources,   even within  the  same  class.    For  example, reported  emission  factors  for
PCDDs/PCDFs from incinerators span more than an order of magnitude.

       Emissions data currently reported by EPA and the IJC reveal that a  large  portion of the
HAPs  in  the United States are generated outside the Great Lakes region. This is particularly true
for emissions from point sources.   The states neighboring the  Great Lakes Basin, particularly
Missouri, generate large quantities of these emissions in electricity and heat producing power
                                          VI

-------
                             "rcDcervc  Sources cf  Mercury
                              - me  Greet Lakes  Bcsir.
         ossn
 6.
 C
    0
                                                            ^
             IL
IN
                            Ml
MN
NY
OH      PA
       Wl     ONT
   10
 C
 C
 o
Waste Incineration

'9.


1

9, m% MM ™™
              IL
IN
                             Mi
MN
NY
OH
PA
                Wl
ONT
   10

    5
v.
a:


9 ma
$%''\ V/////A

1


'9>,


Wk mm Wfa
             IL
IN
                                   MN
        NY
       OH
        PA
       Wl
                      ONT
c  i J
          ceMcneojs Sources
                    IN
       Mi
                                   MN
       NY
       OH
       PA
       Wl
                      ONT
        Total Emission By State
    5 - With  % of Total Emission

|  :oJ      3.5*     4-

2.   .  I
                                          6.4*    6.6X
                           3.3*
                                  2.3*
                                                         5.6*
                                           2.0*    16.1*
             IL
IN
                            M!
MN
NY
OH
        PA
                                                                 Wi
              ONT
      Figure 1   Contribution of ttmospheric emissions of mercury from various source
                categories to the state emissions in the Great Likes region.
                                    Vll

-------
 plants,  primary and  secondary non-ferrous smelters, steel and iron manufacturing plants, and
waste incinerators.   Sources  in the St.  Louis/Granite City area were found to be  a major
contributor to HAPs transported over Lake Michigan during a summertime pollution episode.

       Emissions in regions outside the Great Lakes, Lake Champlain Basin, and the Chesapeake
Bay, heavy metals and  persistent organic compounds can  and  do reach the surface  of  these
waters.   While entering the atmosphere, these  pollutants are subject to long range transport,
transformations, and deposition processes "en route".  The extent to which these processes occur
depends on  stack  parameters,  temperature  and  velocity  of exhaust  gases, meteorological
conditions, and the physical and  chemical forms of pollutants.  Recent studies provide the  basis
for estimating what fraction of the HAPs emitted from major point sources is deposited in the
vicinity  of the emission sources  (local deposition) and what part  is transported and deposited
outside  the emission region.

       Although  PCBs  are no longer sold in the United States and Canada, large reservoirs
still remain.  Of the 640,000 metric  tons of PCBs produced in the U.S., 85% are estimated to
remain  in service (largely in transformers and capacitors),  buried in landfills, and circulating
in the environment.  Releases of PCBs  in eastern  North America from leaking  transformers
and landfill gases were estimated at a few hundred kilograms per year.  PCBs are also released
during combustion, but emission factors for incineration of various types of waste are highly
variable.

       It has been difficult to  obtain  reliable figures for the production and use of pesticides in
the United States and Canada because of proprietary restrictions which protect  their release.
However a recent  survey  has provided such information for  herbicides on  a  state-by-state
basis.  Annual usage of the top  ten herbicides in the U.S. totaled over  150,000 metric tons in
 1987-89.  Four chemicals - atrazine, alachlor, metolachlor, and  EPTC — accounted for  62%
of this  total.  The bulk of these pesticides (80% or more) were  used  on  corn and soybeans.
The National  Oceanic and Atmospheric  Administration (NOAA) reported application figures
for 35  herbicides,  insecticides, and fungicides in coastal drainage areas in the U.S.,  and
provided details of use by crop and season. Use of the 35 chemicals in  1987 amounted to over
 13,000 metric tons.  Three estuarine drainage basins in the mid-Atlantic  and  southeast states
ranked  highest in pesticide use: Chesapeake Bay, Albemarle/Pamlico Sound,  and Winyah Bay.

        Organochlorine  insecticides  such as  DDT,  aldrin,  endrin,  dieldrin,  chlordane,
 heptachlor, lindane, and toxaphene have been of most concern in  the Great Waters because of
their persistence  and tendency to accumulate in biota.  Most of  these insecticides have  been

                                           viii

-------
 banned or severely restricted in the  United  States and Canada,  but are still used in Mexico.
 Central and  South America, Africa,  and Asia.  Reliable statistics on worldwide use that  are
 needed to estimate the contribution of these foreign sources are largely unavailable.

       In the case of emissions from point sources with a stack height of > 150 meters (e.g. large
 power plants,  primary non-ferrous smelters, cement  kilns,  steel and  iron  plants and  waste
 incinerators), all of which employ high temperature processes, only 15 to 20 % of toxic  emissions
 were deposited  locally.  The majority of the pollutants were transported out  of the emission
 region.    Research  in the  Great  Waters Regions  is  needed  to  quantify  local  deposition.
 Particularly, information is needed on the importance of urban area emissions on deposition to
 nearby water bodies.   It is certain that we must consider emission sources both within the Great
 Waters  region  and outside the watersheds in order to assess the origin of  atmospheric toxic
 compounds deposited  on the water surface in the region.

       The quantity  of emissions for the heavy  metals and  persistent organic compounds  of
 concern in the states around the Great  Lakes, Lake Champlain, and  the  Chesapeake Bay is
 difficult to assess due  to diversity of emission  numbers reported by various research groups. For
 most of the  heavy metals  considered in this work emission  estimates differ by  one  order  of
 magnitude and, therefore, are presently under revision.

       Emissions from other source regions  in North America may also affect the amount  of
 pollution load deposited to the Great Waters although no definitive evidence has been provided  by
 measurements and assessment for heavy  metals and persistent organic  pollutants.  Compelling
 evidence suggests that pesticides applied in the southern portion of the U.S. are subsequently
 deposited  into the Great Lakes and other  bodies of water.  Furthermore, the NAPAP concluded
 that  such an impact exists for deposition of sulfate. The source receptor relationships calculated
 for wet sulfate deposition are shown in Figure 2. This figure suggests that the  contribution from
 sources beyond 1000 km dominates the sulfate deposition to Lake Champlain and the Adirondack
Mountain  region. As the sulfates are transported on particles, as are many of the metals and
 organic compounds discussed in this report, one would hypothesize that the metals and organic
 compounds emitted from  sources outside  the study region would also be deposited to the Great
Waters,  especially those compounds that  are  emitted from the same sources as the sulfur.  At
present, regional models of long range transport of air pollutants are available  and with  some
modifications they can be used to assess the contribution of emissions from outside source regions
to the Great waters.  An emission inventory with the appropriate spatial distribution of the data
needs to be prepared.  The experience  gained through NAPAP emissions inventory development
can be used as a starting point for this purpose. Collaboration with the Mexican authorities on
                                           ix

-------
environment protection is highly recommended as the Mexican emissions of heavy metals  and
persistent organic pollutants are expected to contribute to the contamination of the Great Waters.
           *      '   Source apportionment of >vet sulfate deposition (percent per state or subprovincei at
                        Wnueface Mountain, NY
        Input of organochlorines to the Great Waters continues, as indicated by their presence
 in air and  precipitation.    Likely  sources  are long-range  transport from  countries  where
 organochlorines are  still applied,  and volatilization from North  American  farmland treated
 years ago.  Distinguishing "new" from "old" sources of these compounds is an  essential and
 ongoing area of research.  Tools  for this purpose include air  mass trajectories  which can
 provide the history of the air mass on days with elevated levels of pesticides in the ambient air,
 often revealing a path from heavy use areas to the Great Waters, and examination of ratios
 among isomers and breakdown products of certain pesticides for clues to their source.  Models
 have demonstrated that facile  movement of organochlorine pesticides from the southern states
 to the Great Lakes can take place, and elevated concentrations of these compounds in  Ontario
 air have been traced  to air masses arriving from the south.

-------
        Evaporation  rates of pesticides from treated fields have been  measured and related to
 properties of the chemical, the soil, and the evaporation rate of water.  The factors affecting
 pesticide volatilization  are  known, and models have been  developed to predict fluxes from
 soils.   There is a need to  apply these measurement techniques and  models to "old source"
 situations (fields treated in the past) and on a larger scale to  obtain estimates of loadings to the
 atmosphere.

       In the report  to MOE, estimates of herbicide and insecticide releases to the atmosphere
 in eastern  North America were 30,795 and 3,091 metric tons per year, respectively.  These
 were based on usage information and a soil volatilization  model. The insecticide  figure is
 probably low, because  it does not include a large number of current-use chemicals.  At  the
 present time it is not possible to attach a degree of uncertainty to these estimates, other than to
 say that it is probably large.

       Gaseous  SOCs,  particularly  PCBs  and organochlorine pesticides, exchange  freely
 between the atmosphere and the earth's surface -- water, soil, and vegetation.  Seasonal high
 and  low cycles  of these  compounds in ambient air are observed, which  are  correlated  to
 temperature and thus volatilization rates.  The Great Waters  can act as either a sink or source
 of vapor-phase PCBs and pesticides,  and gas exchange forms a large (and poorly understood)
 part of the contaminant  budget.  Long-range transport of persistent SOCs may not  occur in a
 single  event, but in  steps as the compounds are continually deposited and revolatilized from
 land, water,  and vegetation.  Emissions from new sources must be cast in light of this global
 background of "recycled"  material.

       Pesticides, such  as  hexachlorocyclohexane (HCH), have been  found to be  global  air
 pollutants measured in remote areas around the world. As the major application of this pesticide
 is in Asia and the wind patterns at various altitudes do not exclude the air mass transport form the
 Asian continent to  North America, HCH deposited in the Great Waters regions may originate
from as far away as India,  China,  or the former Soviet Union.  This hypothesis can be tested  by
the application of global models, or at least hemispheric models. At present, such models are used
to study the transport of green-house gases in the atmosphere and the transport of sulfur in the
Northern Hemisphere.

       Several methods  can be applied to reduce emissions of toxic heavy metals and persistent
organic pollutants and eventually reduce the atmospheric deposition of these pollutants to  the
Great Watefs. Technological solutions presented in a form  of Best Available Technology (BAT)
package or Best Practicable Technology (BPT) package as well as non-conventional methods

                                          xi

-------
offer  emission  reduction  possibilities  for  point  sources within all  major  source  categories
contributing to the contamination on the Great Waters. Previous experience gained in past efforts
can be used in  recommending emission reduction scenarios in  the Great Waters region.   Cost
estimates and benefits from the implementation  of the control techniques should be carefully
studied.

       New research initiatives are  necessary in order to meet  the requirements outlined  in the
conclusions of this report as well  as to test the important  hypotheses proposed here.   These
activities would include both measurement programs and modeling estimates.

       Source inventories  for  combustion-related SOCs such as PAHs and  PCDDs/PCDFs
need  to be improved.   The wide  variability in  emission factors from various industrial and
incineration processes will probably be reduced as emission  controls are put into place.

       Great improvements in sampling and analytical methods for SOCs have been made over
the  last decade,  making  the determination  of  source  profiles  more  reliable.    These
improvements should lead to increased use of CMB models  and  statistical methods such  as
factor and principal  component  analysis for source  reconciliation of SOCs.   Additional
"marker" compounds should be sought which can be used as source tracers.

       Restrictions which protect the release of production and use statistics for pesticides in
the United States  and  Canada must  be  removed  to allow  free  exchange  of this essential
information.   Further,  an international effort is needed to identify types  and quantities  of
pesticides used in foreign countries.

       New measurement  programs are needed  in order to  improve the quality of source-
receptor techniques which are used to  assess the magnitude and origin of deposited pollutants.
These measurements are needed at both the receptor, the Great  Waters themselves, as well as at
the sources of the emissions. The following are  recommended:
- emission rates and emission factors for toxic heavy metals and persistent organic pollutants from
large point sources in the study region should be evaluated on the basis of measurements of their
concentrations in exhaust gases;
-soil  volatilization models for  pesticides and  mercury need to be  improved  and to the point
where they can be applied regionally.  These  models  should be validated  by experimental
measurements of pesticide and mercury  fluxes from "old source"  areas (previously treated
land) as well as from freshly treated or contaminated fields and areas.
                                           Xll

-------
   a program  should  be  undertaken  to identify  and  quantify  current  sources of  banned
 organochlorine pesticides to the  Great Waters.   In particular, contributions from old  sources
 within the U.S. and Canada should be assessed, and weighed against long-range transport of
 organochlorines from foreign countries.

  new emissions of persistent  SOCs must be evaluated relative to  the global background of
 recycled material.  To do this,  we must determine the quantities of SOCs currently residing in
 the  air,  land, water,  and vegetation  reservoirs and  the rates of exchange  among these
 reservoirs.

 - physical and chemical forms of the most volatile compounds should be established through
 measurements carried out in major sources in the study area; and

 - emission rates for the most volatile metals, and particularly mercury,  as well as SOCs should be
 derived on the basis of measurements  over  the water surface in the Great Waters and the
 surrounding soils.  The results should be representative for the meteorological conditions as in the
 Great Waters and exemplify seasonal changes.

       Measurements at receptors should  provide  information which is  needed in  order  to
 improve the accuracy of source-receptor relationship analysis. The following is recommended:

 - size-differentiated chemical composition of aerosols should be measured at receptors which can
 represent conditions over the water surface in the study area;  and

 - simultaneous measurements of the gaseous and particle phases of the  studied pollutants with the
 help of newly developed techniques (e.g., denuder methods) should be undertaken in order  to
 provide information on gas-to-particle conversions ( and particle-to-gas conversions) for the most
 volatile pollutants under study.

       Improvement is needed within  the three groups  of  estimates:  emission estimates.
 dispersion modeling, and receptor modeling.  The following is recommended for the improvement
 of emission estimates in order to assure better understanding  of source identification in the Great
 waters region:

 - gridded emission inventory for the studied pollutants should be approached for the whole
 territory of the United States and Canada;

 - seasonal changes of mercury and volatile organic compound emissions need to be quantified and
 techniques developed to estimate these emissions; and

 - an approach should be defined to assess emissions of pesticides in the Northern Hemisphere with
 particular emphasis on Mexico and the Asian  countries.

       Improvements in source identification through the  further development of dispersion
modeling for toxic air pollutants  is needed. The following  are recommended:

                                          xiii

-------
- continue to modify and improve the existing long-range transport models so they can be used to
study the contribution of emissions from sources in North America, both within  ard outside the
study region to the hazardous pollution load deposited to the Great Waters; and

- an approach should  be made to apply the existing global  scale models to  investigate  the
possibility of pesticides, e g., lindane, used in  Asia to be transported within air masses to North
America and deposited  also in the Great Waters region.

       From the above discussion it is clear that there is a considerable amount of uncertainty in
our estimates of the sources of the HAPS measured in the Great Water areas.  The level of
research activity must increase and cooperative programs must be implemented before our
understanding of the sources of the  critical pollutants found in the Great Waters will be complete
enough for policy measures to be developed and implemented.
                                           xiv

-------
                            ACKNOWLEDGMENTS

       The authors of this section of the report would like to acknowledge the assistance of the
many people who supplied many of the data, tables, and figures used in this report.  Much of the
best information on emissions of toxic compounds  is only now becoming available and the
following people shared their work with us:  Bill Benjey (EPA-AREAL),  Terry Clark (EPA-
AREAL), Ann Pope (U.S. E.P.A.), P.K. Mishra (OME), John O'Connor (Radian), Jorg Munch
(Dornier GmBH),, Trevor Scholtz (ORTECH), Bob Stevens (EPA-AREAL), Carmen Benkowitz
(Brookhaven National Laboratory), Chris Veldt (TNO), Eva Voldner (AES).  The authors would
also like to thank Mila Simmons (University of Michigan) for her input and information, and
Marion Hoyer for her assistance on many aspects of the report preparation.  We would like to
thank the reviewers of this report, Tom Holsen and Ken Noll for their valuable insights and
suggestions. Lastly,  we thank the participants of the Great Waters Study Report Workshop for
their helpful suggestions in preparing this revision.
                                        xv

-------
 1.  INTRODUCTION

       The Great  Waters Program  as  written  in  section  112(m)  of the  Clean Air  Act
 Amendments (CAAA) specifies that U.S. EPA, in co-operation with Commerce/NOAA, shall
 conduct a program  to identify and assess the extent of atmospheric deposition of hazardous air
 pollutants  (HAPs)  to  the Great  Waters which  include the  Great Lakes, Lake Champlain,
 Chesapeake Bay, and other coastal  waters.  In order to meet the ambitious goals set forth in
 the 1990 CAAA, EPA identified  the need to determine the state-of-knowledge of atmospheric
 deposition  of HAPs as a major pathway for loadings to the  Great Waters.

       Numerous reports  document the fact that the atmosphere is a major source of toxic
 contaminants found in many aquatic  ecosystems.  This report  includes an  evaluation of the
 available data on the sources of HAPs within and outside the watersheds of the  Great Waters
 (Great Lakes, Lake Champlain, and Chesapeake Bay), and will  discuss the difficulties in deriving
 such quantitative information.   Detailed emissions  inventories of the criteria pollutants, e.g. SO2,
 NO2, etc, as well as HAPs have become an indispensable tool in environmental management. The
 quantitative targets  are amenable to legislative controls  and the  emission of HAPs is easier to
 regulate than atmospheric  deposition or  food chain effects.  Any attempt to set  guidelines on
 deposition rates for  HAPs, in  fact, requires that the sources of the pollutants be known and the
 emission rates be determinable.

       There is another cogent reason for concern about air emissions of toxics in this country.
 Of the 12 billion kilograms of the material flows reported in the Toxic Release Inventory (TRI) of
 1988, about 39% (or 5 billion kg)  went into the atmosphere,  while only 6% and  9%  were
 discharged  directly to the surface waters and soils, respectively (EPA, 1990)1   In  many aquatic
 ecosystems, however, large quantities of toxic contaminants can also be derived from industrial
 and municipal wastewater discharges, storm water  and urban run-off, leachates from landfills and
 dump sites, etc.  Furthermore, the transfer from one compartment to another becomes  quite
 ambiguous  as is in the  case of combined storm-sewer systems in which contaminants are both
 directly discharged into the sewers  as well as being deposited from the atmosphere  with the
 precipitation. The wastewater eventually reaches a wastewater treatment plant which separates
the liquids  and solids, discharges  the processed effluents into surface  waters, and then  often
 incinerates  the solids re'moved in  the  process.   The incineration of the sludge results in air
1 The quantity of toxic pollutants emitted is definitely underestimated in the TRI data as this inventors' includes
 only 13% of the nations manufacturing facilities.

-------
emissions which continues the cycle by dispersing the hazardous pollutants into the atmosphere
where they are eventually deposited in a downwind environment by dry and wet processes.  So,
while the focus of this report is  on air toxics, the need for an integrated or whole ecosystem
approach to managing toxic contaminants in the Great Lakes should be emphasized.

2. ATMOSPHERIC DEPOSITION AS A MAJOR PATHWAY

       Atmospheric deposition is one of the major sources of lead, arsenic, cadmium,  mercury,
PAHs,  several organochlorine pesticides, (e.g.  lindane,  DDT, chlordane, dieldrin, toxaphene),
PCBs,  PCDDs, and PCDFs measured in the Great Waters.  The other inputs include non-point
sources, e.g.  agriculture practices in the region, urban runoff, leaching from landfills, etc., direct
industrial discharges, tributary inputs, and direct dumping of wastes.   Specific examples  of
loadings  estimates  are  given in the  "Relative Loadings"  section of the report.   While
approximately half of the Pb is of atmospheric origin, it is noted that tributaries account for a
substantial fraction as well.  However, designating the loadings as being from  tributary inputs
maybe  somewhat misleading as much of the Pb and other contaminants in the tributary waters are
the results of atmospheric deposition and subsequent runoff.  There are  many areas where high
quality data  are needed to fill-in  the mass balance estimates for many pollutants and it is  very
difficult to get reasonable  numbers for a large majority of the critical pollutants.

       Evidence of  atmospheric deposition  as  a source of persistent semi-volatile organic
compounds (SOCs) to water bodies is provided by their accumulation in soils, sediments, and
peat bogs in  the  Great Waters region  and other locations.   The trends of contaminant
concentrations with  depth in dated  layers  of  sediments  or peat cores track  their known
production/release history.  These trends show peak accumulation of PAHs in the 1950s, PCBs
in the mid-1970s, and organochlorine pesticides  in the late  1960s  to mid-1970s, depending on
the chemical.   Rapid increases  of  PCDDs/PCDFs in Great Lakes   sediments after  1940
paralleled the production of chlorinated aromatic compounds, suggesting  that incineration  of
chlorine-containing  waste was  the  most  significant  contributor.    Other  indicators  of
atmospheric sources include  direct measurements of SOCs in air, rain, and  snow from the
Great Waters and remote regions; and accumulation of persistent organochlorine compounds in
biota from small inland lakes  in the Northwest Territories, the high Arctic, and  Antarctica.

       Compared to the other Great Waters considerably more research  has been performed on
the deposition of HAPs to the Great Lakes.  Using the best "available" data quantitative estimates
have recently been made on the fluxes of many toxic substances into these lakes  (Eisenreich and
Strachan,  1992).   The  relative  importance  of  Great  Waters  contamination by atmospheric

-------
 deposition is the focus of another part of this report.  However, the general conclusion can be
 drawn on the basis of a  recent review (ICF, 1992 for EPA), that between 35 and 50% of the
 annual load of the HAPS  of interest enter the Great Lakes waters through atmospheric deposition
 It was concluded, for example, that the atmospheric lead input was responsible for 35 to 47% of
 all inputs to the Great Lakes during the period from 1977 to 1981.

       The deposition of airborne toxic substances has also been studied in Europe. The results
 of long-term measurements on pollution of the  North Sea and Baltic Sea waters lead to  the
 conclusion that  as  much as 50%  of lead and mercury and between 30 and 50% of arsenic,
 cadmium, chromium, copper, nickel, and zinc enter these waters through atmospheric deposition.
 The picture in Europe for the persistent organic compounds, discussed  in this report, is very
 unclear due to lack of reliable measured data

       The good agreement between the results obtained for the European Studies on the North
 Sea and the Baltic Sea, and the findings reported  for the Great Lakes indicates that atmospheric
 deposition is a very important pathway for  HAPs measured in these waters.   This  indication,
 although regarded as preliminary, due to limited data for most of the organic toxins,  argues for
 research on sources and source regions generating atmospheric emissions of the studied pollutants
 to identify the major  sources of the Great Water's contamination.  Once identified, the major
 sources (often called  "hot  spots") can be targeted  and  emissions reduction programs can  be
 implemented leading to a  decrease in the pollution  load to the Great Waters.

       Identifying the specific sources or source types emitting the pollutants into the atmosphere
 which ultimately are deposited is another matter. Identification of the major sources and the
 deposition pathways of the critical pollutants should  be  made for the individual compounds
 separately as their sources and behavior in the environment differ substantially.  Volatile organic
 compounds (VOCs) and  SOCs are emitted by both  point and area sources.  Examples of the
 former are stack and  fugitive emissions from industrial processes and incinerators.   Sources
 that emit pollutants  over  broad areas include vehicle exhaust and evaporation of pesticides and
 PCBs.

       In  general, both local and distant sources contribute to the  pollution load at a given
receptor.  There are various definitions concerning the meaning of local and distant sources.  In
this work local sources are those in the states  adjacent to the Great Waters.  For the Great Lakes
these states  are Illinois, Indiana,  Michigan, Minnesota,  New  York,   Ohio,  Pennsylvania.
Wisconsin, and Ontario in Canada.  For Lake Champlain the emission sources in states of New
York, Vermont and the Province of Quebec are considered as local while for the Chesapeake Bay,

-------
the states of Virginia and Maryland.  Sources outside the above defined region are regarded here
as distant, regardless of how much they may contribute to the total loading cf the f r\.Mt Waters.
This somewhat artificial division can be justified when introducing various policy measures to
reduce the pollution load.  Local and distant sources can be of either anthropogenic or natural
origin.  It is believed  that natural sources are more important when discussing the impact of
distant source emissions on the atmospheric deposition of pollutants to the Great Waters.

       Although a  number of source-receptor techniques are available for estimating the
contributions, it is still premature to conclude what part of pollution load originates within the
study region and what  part results from long range transport within air masses.  The major reason
for the present uncertainty is  the lack of reliable input data  for application of these techniques
including properly reported emission data.  The present lack of-monitoring and emissions data is
also problematic for other water bodies currently being studied with respect to the environmental
behavior of hazardous  air pollutants, e.g. the North Sea and the Baltic Sea.

3. GENERAL PROCEDURES FOR SOURCE IDENTIFICATION

       There are a number of methods which  can be used to identify sources or source regions
contributing to the pollution load measured at a given receptor.  These can be categorized into
statistical methods,  chemical and isotopic trace methods,  meteorological  methods including
trajectory techniques,  and various  combinations  of these approaches.   Emissions of the
compounds  discussed  in this report to the atmosphere are multitudinous.  Most of the compound
classes discussed here stem from human activities, although natural sources make a contribution
in some  cases. For example, PAHs are released both by  anthropogenic combustion and forest
fires and Hg is emitted  during coal  combustion and is released from mines and soils.  Source
characterization, when performed accurately, is a powerftil tool for the assessment of emissions
from local sources.  To assess atmospheric deposition to a receptor such as the Great Waters and
recommend remedial action, it is necessary to know the relative  ontribution of various sources to
the total mass loading.  These are deduced from a combination of source characterization and
source apportionment  techniques such as emissions inventories, dispersion modeling, multi-variate
(statistical) methods such as factor analysis, and chemical mass balance (CMB) models.  Some of
these methods are briefly described by Zweidinger et al. (1990), and Sweet and Vermette (1992)
in their article on VOCs In Illinois and St. Louis, and  expanded upon in their references, and in
Gordon (1988), Henry et al. (1990) and Hopke (1991).

       Source apportionment methods, often referred to  as  receptor  models are  based  on
statistical techniques using measurements and information on sources for an assessment of the

-------
 local source contributions.  These techniques used together with meteorological information can
 also be utilized for assessing the contributions of distant sources.  Receptor models that implicitly
 incorporate meteorological data are often referred to as hybrid models (Keeler, 1987).

 3.1. Source characterization
       In general, local sources are characterized on the basis of emission measurements and/or
 emission estimates.  Representative measurements are considered the best source of information
 for accurately describing emissions.   Measured data  can be used  for source characterization
 directly or  indirectly.   Direct  use is when stack tests are performed using  both continuous
 monitoring or representative grab sampling during short measurement campaigns.  Examples of
 inventories developed for semi-volatile organic  compounds (SOCs)  are given in  Johnson et al.
 (1992). There are several regulations in force requesting measurements of emissions rather then
 estimation for certain types of sources or source categories, e.g.  the Commission of European
 Communities requests emission measurements for power plants  with capacity  higher than 300
 MW   In the U.S.,  the TRI program also  requests any industry with 10 or more employees to
 report the release of any of the 322 compounds on the TRI list.

       Indirect use  of measured data  for source  characterization is  whenever information
 obtained from measurements is transformed to an emission factor or  included in a special
 calculation procedures.  Emission  factors for PAHs in mass/vehicle-km are developed from auto
 exhaust sampling or measurements in traffic tunnels.  Sampling in stacks yields  emission factors
 for industries, power plants, and incinerators. Often these can  be related to fuel usage; e.g.  mg
 PAHs released per ton coal burned.  Aerial pesticide losses are  estimated from models relating
 spray drift and volatilization to application  rates, soil  properties, and  meteorological factors.
 Emissions are usually apportioned  to  a  grid  network for dispersion  modeling  and impact
 assessment.

       Source characterization through measurements is often  very expensive, however, and on
 some occasions extremely difficult to perform. In these  cases other methods are often used, and
 in most circumstances methods  based on emission factors and  /or mass balance calculations are
employed. The transparency and comparability of the data used to elaborate or select an emission
factor from a handbook of emission factors are of great importance.

       Mass balance calculations  can be applied to characterize emission sources  through the
assessment of their emission quantities.  The input and output  rates of  a given pollutant are
determined and are used to assess what portion of a given pollutant is released  in the gas phase

-------
while the whole balance is made for the amounts of the pollutant on panicles.  For example, to
calculate the mercury mass balance for the combustion of coal, the amount of the mercury in the
coal is compared to the amounts of mercury in the fly ash, bottom ash,  and stack dust.   The
imbalance is explained as the amount of mercury leaving the stack in the gas phase.

       Emission sources can also be characterized using the concentrations of a given pollutant
measured  in ambient air at a receptor site.  Measured concentrations  are then compared with
emission source profiles for major source categories likely to contribute to pollution at the site. It
is clear that this method is useful when emissions from a given source region originate from one
dominant source or group of sources which have well defined emission profiles.  A  large electric
power plant in a region with small industry can be a good example of such a case. In the case of
complex source region more statistical  work is needed to specify the  contribution of various
sources to  a  profile which can be then compared with  a  profile  obtained from  ambient
measurements.   Principal  component analysis (PCA) is  often  used in such case, where the
contributions from several sources can be de-coupled in the ambient data. PCA has  been used to
define source profiles in cases where there are no source profiles available from source sampling
(Tuncele/a/.,  1985).

       An important part of source characterization is to assess how accurate are  the  methods
used  to prepare  source profiles  are and  what are  the means of verification  of  source  data.
Accuracy  of   source  characterization  in  terms  of emissions  depends  on  whether  the
characterization has been made on the basis of measurements or estimates. The methods based on
measurements are considered as more accurate than those  using assumptions and  calculations,
e.g. methods based on emission factors or chemical  mass balances.   Unfortunately, most of the
verification procedures focus on activity data (statistical information) and emission factors.

3.2. Source apportionment techniques
       Historically, dispersion  models have  been the traditional work horse for calculating
source-receptor relationships  for air pollutants.    These  models require  detailed emissions
inventories for various sources for the pollutants of interest, e.g.  TSP, SO2, etc.   Even if the
dispersion  models were accurate it  is very unlikely the source emissions inventories  for the
pollutants of interest would be adequate.  Emissions inventories for  the criteria pollutants have
many short-comings, as discussed earlier in this document, and these inadequacies are even more
severe for hazardous  pollutants or for pollutants  which have large contributions from fugitive
process emissions, natural sources, and dusts.  The limitations of the dispersion oriented  methods
have led to the development of receptor oriented models.  Receptor models assess contributions

-------
 from various sources based on observations at sampling or receptor sites.  Gordon (1980, 1888)
 reviews the development of receptor methods and provides a concise overview of the various
 types of receptor models and their applications.  Here, we will briefly describe a few of these
 techniques and discuss how they can be applied along with dispersion models to define source-
 receptor relationships for hazardous air pollutants.

       Receptor models were developed in the early 1970s in an attempt to identify the source(s)
 of paniculate matter in large urban areas and to quantify the amount of paniculate matter emitted
 from the  source(s)  (Miller et a/.,  1972; Gordon,  1980;  Winchester and Nifong, 1971).  The
 chemical element balance, or chemical mass balance (CMB) method as it is now referred, are
 based upon the premise that  the emissions characteristics, in terms of chemical and elemental
 composition as well as physical size and morphology, of various source types are different enough
 that one can identify their contributions by measuring the characteristics in samples collected at a
 receptor site.  Thus, an important  first step in the application of CMB model to  apportion the
 sources of air  pollutants measured in a specific urban area is to define an emissions inventory of
 the number and source types of the important sources of air pollution.  These models assume that
 the composition of all  contributing source types are  known.  This is  often not the  case either
 because the sources are not easily sampled or because the  source classes have widely varying
 compositions (Henry, 1991). This is an important limitation of the  CMB method in that the lack
 of specific source profile information for the pollutants of interest prevent this approach from
 being applied.   Emissions data have been sparse  in the  past, particularly for  SOCs,  but the
 situation is improving.

       While the CMB method has primarily applied to urban scale data, Rahn  and  Lowenthal
 (1984, 1985) also applied this technique to their "regional signatures" to apportion the sulfate and
 trace metals observed on paniculate.  The application of receptor models to regional and global
 scale problems has been controversial and has yet to be developed,  in some peoples opinions, to
 the level  necessary  for it to be thought of as definitive in  nature.  However,  an independent
verification for the  appropriateness of the  trace  element ratio approach  was  performed and
indicated that this technique can be quite powerful (Keeler,  1987; Keeler and Samson, 1989).

       Several methods are currently available to assess sources and source regions for various
air pollutants on the basis of the data on the chemical composition of the air at a  given receptor.
Both the statistical methods and modeling are  used together with meteorological data  in order to
obtain this assessment.  The origin to the pollution measured at a given receptor can be studied
using information on the chemical composition of aerosols and/or mixture of gaseous pollutants.

-------
3.3 Applications of source apportionment techniques for metals
       Statistical  methods have been  developed  which use  information  on  the chemical
composition  of aerosols  to  study  contribution of sources  or even  source  regions  to the
contamination at  a  given  receptor.   The applicability of multivariate  techniques for resolving
sources and source regions for aerosols measured at several remote locations far from major
emission regions  has  been tested with the use  of the absolute principal component analysis
(APCA) and the  chemical mass  balance (CMB)  methods.  The APCA method determines the
composition of the major source components, such as coal combustion related, crustal, or sea-salt
related which contributed to the measured concentrations at the receptors. In the past, the APCA
method was applied to total suspended paniculate concentrations (particles measured in both the
fine and coarse fractions).  APCA has been utilized to study the origin of the Arctic aerosol. The
results of these studies can be summarized as follows.  The anthropogenic component contained
many toxic compounds, however, some crustal material was also found with this component.  A
second component containing the crustal elements was observed with proportions similar to those
found in the average crustal rock.  However,  most of the elements in the soil component  were
highly  enriched and this fact  suggests that the second component also  contained some material
from anthropogenic sources.  The composition of the third component found in the Arctic aerosol
was fairly similar to that of bulk seawater, indicating that  this component is essentially sea-salt.
The three-component solution for the Arctic winter aerosol was confirmed by several  studies
(Barriee/a/., 1992).

       Further improvement of this  receptor modeling method was obtained by applying APCA
to aerosol elemental concentration measurements in separate particle size fractions (e.g.  Li and
Winchester, 199C)  • - .- -esalts of this application of APCA gives the basis for interpretation of
coupled  chemical reactions  and physical processes in  remote  locations,  as  well  as  giving
information concerning atmospheric aging processes and, therefore, the history of the aerosols.
For example, it was noticed that some crustal particle components were found in all size fractions.
Carbonaceous fuel combustion pollutants were indicated by the presence of Si  and Cl and the
absence of Al in all size fractions, and they were usually rich in sulfur.  The combustion of coal
with high ash content may release volatile SiO from reduction of silicon dioxide by carbon, which
then forms a fine aerosol  after SiO  oxidation back to silicon dioxide.  Since Al is non-volatile
during coal combustion  its absence when Si and other metals were present indicates the aerosol
was generated from burning high  ash content coal.

-------
       Principal component  analysis (PCA) has also been applied to  assess the origin  of the
 remote aerosols on the basis of results from automated microanalysis of individual particles.  For
 example,  Anderson el al.  (1992) employed a modified  scanning electron microscope (SEM)
 analysis of particles to study the origin of the Arctic aerosol.  They concluded that the collected
 aerosol contained  a mixture of altered and unaltered particles types from a variety of sources,
 such as coal combustion, sea salt,  and crustal dust.  The  silicate types  appeared  as relatively
 unaltered  particles, as well as particles with sulfate coatings, and also  as particles that have
 reacted with Br. Most silicate particles are probably crustal in origin.  Many compositional types
 of metal-rich particles were  of anthropogenic  origin, and  most  types had temporal variation
 patterns that are individually  distinct.  Other  major particle  types were  of marine origin, but
 extensive  fractionation and reactions of the marine aerosol components  was suggested.  The
 authors concluded that further research is needed on the nature and timing of these reactions, the
 mechanism for fractionation of the marine aerosol, and the sources of some particle types. It was
 also underlined  that the complexity of the  remote aerosols  during  a  period presumed  to  be
 relatively free of pollutants is striking.  Two  important findings were that  (1) the Arctic aerosol
 has pollution products from human activity even in a normal period  of spring; and  (2) many of the
 apparent pollutant  particles in the fine fraction, S-rich species and perhaps  Br-rich species,  are of
 natural origins.

       Single or Individual Particle Analysis  (IPA) has been applied in aerosol  research  to
 investigate the sources and morphology of the collected atmospheric paniculate matter (Dzubay
 and  Mamane,  1989; Mamane,  1990; Sheridan, 1989).   In  the review  by Sheridan (1989)  he
 observed  that particles emitted  by anthropogenic sources,  such  as  carbon  soot  and  coal
 combustion spheres, occurred simultaneously  with the highest concentrations of H2SO4 droplets.
 Mamane (1990) utilized scanning electron microscopy (SEM) to estimate  the  contribution  of
 refuse incinerators  to Philadelphia. Thus, EPA can be used to estimate the  source apportionment
but  also  physical  and chemical processes occurring  during  the  long  range transport  of air
pollutants.  Single particle analysis  provides critical size distribution information that can be
directly used to calculate the deposition of pollutants as a function of size.

       One important limitation of the PCA methods is, however,  that their results do not  allow
one to obtain a fine resolution of the contributions from  various distant  source regions to the
chemical  composition 'of the remote  aerosol.   To  attempt this resolution,   CMB source
apportionment must be  performed using either a  set  of emission  source profiles or  a set  of
elemental signatures. The emission source profiles will be discussed later in this report. One set
of elemental signatures were developed by Rahn  and Lowenthal (e.g. Lowenthal and Rahn, 1985).

-------
The  CNffl  source  apportionment proved to be a good technique to  assess contributions from
major source regions but was often limited to provide fine resolution of the contributions from
more defined areas due to high collinearity of signatures or source profiles used.  For example, the
use of three European signatures in the Rahn and Lowenthal studies was supposed to provide a
fine  resolution of the contributions from various parts of Europe to the contamination  of the
Arctic  aerosol.   This  attempt failed due  to high  collinearity of the European  signatures.
Therefore,  both the PCA and CMB methods are with  limitations when comes to fine  resolutions
of emission source regions contributing to the remote aerosol.

       The origins of aerosols in remote  regions  and their source apportionment  have been
diagnosed with the help of not only heavy metals but also of isotopes, halogens, graphitic carbon,
and organic compounds. Stable lead isotope ratios have been used to assess the contribution of
emissions from various source categories, anthropogenic vs.  natural sources,  or  to  distinguish
various sources within the same source category, e.g.  combustion of gasoline with lead additives
from various manufacturing plants, to the lead concentration or atmospheric deposition at a given
receptor (Sturges and Barrie, 1989; Graney el a/.,  1992). The ratios between the lead  isotopes in
aerosols  in the United States and Western Canada were found to  be much higher than those in the
Arctic aerosol, while for Europe they were lower.  The ratios in Eastern Canada were similar to
the Arctic data, but meteorology argues against this region being  a major contributor to the  Arctic
air pollution.  The stable isotope analysis was taken a step further by Graney and colleagues in
that they linked sediments from three  Great Lakes to air concentrations in the basin to ascertain
the extent  of regional anthropogenic lead pollutiion, and to investigate the extent to which the
sediment cores could be used as indicators of historical atmospheric deposition.  One limitation of
the above  described method is that it requires a detailed information on the  isotope ratios at
emission sources,  which is not always available.   In addition, the isotope ratios at emission
sources  are often not sufficiently different to permit the use of  multivariate statistical models to
resolve the input ratios.

       Concentrations of chlorine, bromine, and iodine have been used to assess the contribution
of emissions from marine,  automotive, and crustal sources to the contamination of the air at a
given receptor.   Other sources,  such  as coal combustion have been also identified using the
halogen  concentrations and their ratios. However, an important limitation of this method  is that
halogens are quite reactive in the polluted  troposphere and their behavior during long  range
transport may be  affected by the chemistry of the atmosphere.  Therefore, the application of
halogens as source tracers is probably limited to pristine regions, such as the Arctic rather than in
the polluted regions, such as the Great Waters area.

                                            10

-------
        High  concentrations  of graphitic  carbon panicles have  been  used  to estimate the
 contribution of combustion  emissions  at  several receptors  (eg.  Rosen  and Hansen, 1984)
 Further work on this subject has been focused on assessing the vertical distribution of graphitic
 carbon particles  and their associated absorption coefficients.   Although this  research is  very
 important for modeling the effects of air contamination on the solar radiation balance, it seems to
 be less useful in assessing the origin of pollutants at receptors.

        A number of receptor-oriented  transport models have  been developed to identify the
 emission source regions contributing to the contamination of certain areas by trace metals.  In
 general, the models have proven useful in calculating trace element concentrations at various
 remote receptors  and trace element  inputs.   A Lagrangian-type  of model was employed  to
 determine the origin of air pollutants measured during the transport episodes to the Arctic  (e.g.
 Pacyna el al, 1985) as well as to the remote locations in Scandinavia (e.g. Pacyna el a/., 1989).
 It was concluded that concentrations  calculated  by models agreed  with measurements within a
 factor of two. Of course, the model results  depend on the quality of emission inventories used as
 an input data. The model performance is also  sensitive to the dry and wet deposition processes.
 Fixed  values  of the  dry  depostion velocities are often used  in models.   This  is certainly a
 simplification of the problem raising the inaccuracy  of the modeling.  The mixing height  also
 affects the performance of the model, but far less than the emission estimates and the wet and dry
 deposition processes.

       The  variational formulation of Eulerian dispersion models also allows for the application
 of both source-oriented and receptor-oriented modeling as complementary tools in identification
 of sources contributing to the contamination of the environment in a given region. Uliasz and
 Pielke (1990) concluded that applicability  of  the receptor  oriented option is limited to linear
 dispersion models and integrals  describing air quality  at the  receptor.  It is necessary to assume
 that all chemical reactions of pollutants are linear and that pollutants do not affect the atmospheric
 dynamics. Uliasz and Pielke (1990) suggest, however, that their variational approach may be still
 useful to perform sensitivity analysis of dispersion models with nonlinear chemistry.

       An improved climatological-type model on Trace Toxic  Air Concentrations in  Europe
(TRACE) has been developed  by Alcamo  el  al. (1993).   The  model addresses  some of the
drawbacks to typical  climatological-type air pollution models by (1) computing time of travel
from an empirical function of geographic distance; (2) maintaining mass conservation by splitting
the computation of decay coefficients spatially, and deriving concentration equations from mass
 considerations, (3) dividing calculations into two  steps and calculating deposition based on local
meteorological variables, thereby avoiding unreasonably "smooth" spatial deposition patterns; and
                                            11

-------
(4) deriving all parameters objectively, that is, without calibration to observations.  TRACE has
been used to compute levels of heavy metals for 1978 to 1985 throughout Europ?  Calculations
agreed with As and  Pb observations  within a factor of two, and  underestimated Cd and Zn
observations.  Using  the model it was estimated that wet deposition exceeds dry deposition in
most of Central Europe. The mean residence time of the mass of heavy metals in Europe's lower
atmosphere was estimated between 64 and 96 hours.

3.4   Problems  in  the  application of  source apportionment  techniques  for organic
     compounds: changes in chemical profiles from source to receptor
       It might seem that if source chemical profiles were known, the same apportionment could
be applied to  atmospheric deposition.  However the problem is much more complex.  Because of
their  reactivity and  exchange  between  different phases  in  the atmosphere  (gas-to-particle
distribution) organic compounds behave less conservatively than elemental tracers. Thus, changes
in the relative abundance of individual compounds occur in transit from source to receptor due to
differential reactivity  and rates of atmospheric deposition.  The limitations imposed by alteration
of profiles are not well understood.  This section gives an overview of factors responsible for
profile changes; problems with individual  compounds or compound classes are dealt with in the
appropriate section.

       Gaseous and  paniculate organic compounds  can be transformed by  photolysis, and are
also more or less reactive toward a number  of atmospheric species,  including radicals (e.g.
hydroxyl, peroxyl), ozone, and nitrogen oxides (Atkinson, 1990; Ballschmiter, 1991; Bunce and
Nakai,  1989; Bunce et a/.,  1989).  Differential  reactivity is a factor  which  limits  source
identifications based on chemical profiles obtained at a distance.  Calculated atmospheric lifetimes
for gas-phase reactions of several VOCs and SOCs with OH radicals range from a few hours to
nearly half a  year (Table 1), and thus ratios of compounds change  in traveling from source to
receptor.   Scheff and Wadden  (1991) considered the effect of reactivity  on CMB  models for
VOCs and concluded that for most species a 2-3 hour transit time from source to receptor would
result in little change  in composition ratios. Nevertheless, diurnal cycles of several gaseous 2- and
3-ring PAHs  were observed in Glendora, California during a photochemical air pollution episode
(Arey et a/.,  1989).  Daytime concentrations were about 2-3 times lower than at night  due to
reaction with the higher Jevels of OH radicals during the day. Elevated nighttime concentrations
of nitronaphthalenes were found  as a  consequence  of  naphthalene reaction with  ^05.
Transformation of paniculate PAHs may also be a problem, and is discussed in that section.
                                           12

-------
TABLE 1  Estimated Atmospheric Lifetimes Due to Gas-Phase Reaction with OH Radicals.




Compound        Lifetime           Compound         Lifetime
Aliphatics
Alkanes, €4 - C \ 3

Aromatics

Benzene
Toluene
Xylenes
Trimethylbenzenes
Ethylbenzene
Ethyltoluenes
Naphthalene
Biphenyl
Fluorene
Anthracene
Pyrene

Chlorinated VOCs

Methyl chloride
Dichloromethane
Chloroform
1 ,2-Dichloroethene
Trichloroethene
Tetrachloroethene


0.5-6 days



5 - 13 days
1-2.5 days
6-10 hours
4 hours
20 hours
7-11 hours
8 hours
2 - 3 days
1.2 days
1.4 hours
4 hours



1 year
110 days
1 50 days
70 days
6 days
90 days

PCBs
Monochloro
DichJoro
Trichloro
Tetrachloro
Pentachloro

PCDDs & PCDFs
2,3,7,8-TCDD
2,3,7,8-TCDF






Pesticides

DDT, DDE
Dieldrin
Chlordane
Hexachlorobenzene
Hexachlorocyclo-
hexane
EPTC
Cycloate

5-11 days
8- 17 days
14-30 days
25-60 days
60- 120 days


2-3 days
7 days








2-4 days
1 day
8 days
80 days

15 days
5.8 hours
5.2 hours
Sources: Altschuller et al., 1991; Atkinson (1987), Bidleman et al. (1990); Kwok et al. (1992).
                                       13

-------
       SOCs. differ from VOCs in that they are associated  to a greater or lesser extent with
atmospheric particles.  This phase distribution profoundly affects their atmospheric chemistry and
physical removal, and thus deposition into the Great Waters (Ballschmiter, 1991, Bidleman, 1988;
Mackay et a/.,  1986; Schroeder and Lane, 1988; Swackhamer and Eisenreich,  1991).  Factors
influencing the extent  of particle  association  include compound  volatility (vapor  pressure),
temperature, and  particle surface area available for  adsorption.  The Junge-Pankow equation
(Pankow,  1987) is the most common model for estimating  the phase distribution of SOCs in
ambient air:

                            is the fraction of the total atmospheric concentration associated with aerosols,
P  T is the liquid-phase vapor pressure of the compound, and  c is a parameter which depends on
thermodynamic properties of the compound and surface properties of the aerosol.  The parameter
0 is the available particle surface area per unit volume of air (cm2/cm3); typical values for urban,
rural, and clean background air are given by Bidleman (1988).

       Experimental estimates of the  phase  distribution  of  SOCs  in air are usually made by
drawing air through a filter to retain particles followed by an adsorbent trap to collect gaseous
SOCs. Such samplers are prone to a number of artifacts which may cause the measured

paniculate fraction on filters to differ from the phase distribution in the free atmosphere (Pankow
and Bidleman, 1992).  Evaluation of these artifacts (Gotham and Bidleman,  1992; deRaat et a/.,
1990; McDow and Huntzicker, 1990; Zhang and McMurray, 1991) and development of improved
techniques to speciate SOCs (e.g. denuder samplers, Appel et a/.,  1989; Coutant et  a/.,  1992.
1989; Krieger and Hites. 1992; Lane el al., 1988) is an active area of research.

       Paniculate fractions measured with filter-adsorbent samplers and predicted by the Junge-
Pankow model have been made in a few cities and rural areas. These comparisons, summarized
by Bidleman (1991), and Pankow and Bidleman (1992) show good agreement between the model
and experimental  results for PAHs,  whereas the sampling  methods yield O-values that are lower
by about a factor or two for organochlorine pesticides, PCBs, PCDDs, and PCDFs (Figure 1).

       The gas-to-particje distribution of SOCs governs the mechanism and rate of their removal
from  the atmosphere. Deposition of SOCs can occur by wet and dry removal of particles, and gas
exchange with soil, plant foliage, and water bodies (Figure 2). An understanding of gas-particle
relationships in the atmosphere is therefore necessary to predict losses of SOCs during transport
                                           14

-------
                                      PAHs
Figure 1.  Paniculate percentages  of PAHs and organochlorines (PCBs, pesticides.
         PCDDFs) in urban air, as determined by high volume  air sampling (solid
         curves) and-predicted by the Junge-Pankow model (Equation 1) as a function
         of the liquid-phase vapor pressure (p°L, Pa) of the chemical. Letters indicate
         individual investigations. Source: Gotham (1990) and Bidleman (1991).
                                      15

-------
           AERIAL DISTRIBUTION  AND REMOVAL OF SOC
              o *° • o   °
               VAPORS
                            PARTICLES
DRY DEPOSITION
(GAS EXCHANGE)
 RAIN/SNOW

SCAVENGING
 RAIN/SNOW

SCAVENGING
    DRY
DEPOSITION
 Figure 2.  Gas-particle partitioning and aerial removal processes for SOCs.  Source:
        Bidleman (1988, 1991).
                                    16

-------
 and deposition. An equilibrium model describing the relationship between wet scavenging and the
 fraction of particle-bound SOCs is (Ligocki and Pankow, 1985)

                 W = WpO + Wg(l-(J>)                                        (2)

 W is the  overall scavenging ratio (concentration  in rain divided by concentration in air, on a
 volume-volume basis), Wp and Wg are scavenging ratios for paniculate and gaseous SOCs.  Wp
 is obtained from field observations for species only in the  paniculate phase (such as certain trace
 metals).  Wg is the equilibrium water/air partition  coefficient, calculated from RT/H where H is
 the Henry's law constant (Pa-m3/mol) for the particular species and R is the gas constant (8.3 Pa-
 m3/deg-mol).  Thus  the  relative importance of particle and gas scavenging for an individual
 compound depends on  its phase distribution (O) and  the Henry's law constant.  These must be
 known as a function of temperature to estimate atmospheric removal rates.

       The actual situation is more complex, because of the variation in particle size distribution,
 re-equilibration of SOCs between the  gas and particle phases during rain events, emission of
 SOCs  into the atmosphere during  rain events, and  differences in meteorological  conditions.
 Recent kinetic models have been formulated to take these factors into account (Tsai et al., 1991;
 Seinfeld et al., 1991);  however  these require a  much more  extensive list of  input variables,
 including the particle size distribution of the SOCs (which is often not available).

       In  rural air at moderate temperatures the simple Equation 2 model predicts that  gas
 scavenging dominates for 2-3 ring PAHs, hexachlorocyclohexanes (HCHs), and dieldrin.  Particle
 scavenging is  more important  for higher-ring PAHs, PCBs,  n-alkanes, chlordane, and DDT
 (Bidleman, 1988,  1991; Swackhamer  and  Eisenreich, 1991; Figure  3).  Rainfall removal  of
 PCDDs and PCDFs changes from a gas to a particle-dominated process  as molecular weight
 increases (Eitzer and Kites. 1989b; Koester and Kites, 1992).

       Deposition of SOCs is a highly temperature-dependent process.  Lowering of compound
 vapor pressure in winter increases the paniculate fraction according to Equation  1.  This can be
 clearly seen in Figure 4, where proportions of PAHs in the gas and particle  phases during winter
 and summer are compared. Henry's law constants also decrease at lower temperatures, resulting
 in greater gas solubility in rain and cloud droplets.  The overall effect is more efficient deposition
 of SOCs in winter than in summer.

      Differential removal by physical processes is one reason why profiles of SOCs at receptor
 sites differ from those at sources.  The case of PCDDs and PCDFs is a good example.  Profiles of
PCDDs and PCDFs from municipal incinerators and industrial waste effluents (Czuczwa and
                                          17

-------
              1500C;
              iZOOCr
              QOOOh
               800C
               3000r
                              GflS-SCRVENGED  SOC
                                    C PflRTICLE
          
-------
                 A. Winter
            100
  D Gaseous

  B Paniculate
               PHEN WVWFUJR PVR  BePM CcdP  B«A CHPY BNTH B«P  BbF  BkF  BaP  Ban* Bjfvo -23P AN'HN COR
              B: Summer
          100
O  Gaseous

D  Paniculate
             PMEN ANTVW FuuR PVP  BcPM CeaP 6aA
                                                         B"  -je-'. SB-xA 5;-,o -13P A\--.N CCR
Figure 4.  Relative proportions  of PAHs  in the gaseous and paniculate phases during
          winter and summer, as determined by high volume air sampling.  Source:  Back
          eial (1992).
                                            19

-------
Hites, 1986) and ambient air (Broman el a/., 199la; Eitzer and Hites, 1989a,b; Smith et a/., 1990)
show the  presence of compounds containing 4-8 chlorines.   Because paniculate  fractions are
greater for the  more highly chlorinated homologs, rain and dry flux distributions are  weighted
toward compounds containing 7-8 chlorines (Eitzer and Hites, 1989b; Koester and Hites, 1992).
Atmospheric degradation  of gas-phase  PCDDs  and PCDFs  may also  contribute to profile
alteration  (Koester and Hites, 1992).  The result is that PCDDs and PCDFs in  Great Lakes
sediments are dominated by 7- and 8-chlorinated compounds (Czuczwa and Hites,  1986; Eitzer
and Hites, 1989b; Koestc:  .  1 Hites, 1992; Figure 5).

   3.4.1 Alteration of PAHs in the Environment

       The problem of selective physical removal of SOCs from the atmosphere and the effect on
chemical profiles  has been discussed  earlier.   In addition,  PAHs can undergo chemical  and
photochemical reactions in the  atmosphere than lead to changes in the  relative proportion of
compounds between source and receptor.

       Many studies have been done to  determine reactivities of PAHs in the gas and particle
phases, with sometimes contradictory results  Light,  mixtures of oxidant gases, and the particle
composition all influence the rate of PAH degradation  (Daisey et al., 1986; Greenburg et al,
1985). The latter authors found good correspondence between experimental PAH stabilities anc
those predicted from molecular properties.

       Laboratory studies show that PAHs are more stable to photolysis  when adsorbed to fly
ash than on  artificial substrates like alumina, silica gel, or glass (Yokley et a/., 1986).  Among the
fly ashes, those with high carbon content and black or gray were most  effective in stabilizing
adsorbed PAHs (Behymer and Hites, 1988; Dunstan et a/, 1989; Yokley et al., 1986).

       Kamens et al. (1988-90) found rapid degradation of PAHs on wood smoke particles when
exposed to natural sunlight.  A. moderate temperatures and humidities, PAHs decayed within an
hour. At low temperature and humidity and reduced light intensity the time scale for PAH loss
increased to days.  Oxygenated PAHs were stable to sunlight alone, but labile in the presence of
0.2 parts-per-million (ppm) ozone and sunlight.  Sunlight was the most important factor causing
loss of benzo(a)pyrene (BaP) from wood soot, followed by ozone and nitrogen dioxide.  Guo and
Kamens (1991) estimated a BaP  half-life of 80 h on wood smoke particles for reaction with 02
parts-per-million atmospheric NC»2    PAHs on  diesel  paniculate  matter were converted by
exposure  to  part-per-million  levels of  ozone  over  a few  hours  (Van  Vaeck and  Van
Cauwenberghe, 1984).

                                          20

-------
                                        590 ff/
                         AVKKAGE AIR
              F4 F5 F6 F7 F8  D4 D5 D6 D7 D8
                                         54 p,/L
                                               J  i
AVEHAGE RAIN
              F4  F5 F6 F7  F8 D4 D5 D6 D7 D8
                                       1100
                    GREAT UKES SEDIMENT
              F4 F5 F6  F7 F8 D4 D5 D6 D7 D8
Figure 5. Homolog profiles of PCDFs (F4 - F8) and PCDDs (D4 - D8) containing 4-8
       chlorines in air and rain from Bloomington, Indiana,  and in Great Lakes
       sediments. Source: Eitzer and Kites (1989b).
                              21

-------
       Losses of parent compounds does not necessarily represent a gain for the environment.
Indeed, reaction of PAHs with atmospheric oxidants leads to formation of PAH  <..;.:r>ones and
PAHs substituted with hydroxy- and nitro-groups, some of which are highly mutagemc (Arey et
a/., 1987; Kamens et a/.,  1985; Kleindienst et a/.,  1986; Nishioka et  al.,  1988;  Nishioka and
Lewtas, 1992; RamdahJ etal., 1986).

       A different  picture of PAH reactivity is seen from field studies.   Freeman and  Cattell
(1990) found that although diurnal variations in total PAH occurred in urban air, ratios among the
PAHs were fairly constant. Gibson et al. (1986) examined ratios of BaP to lead and elemental
carbon (EC) in coastal Delaware and Bermuda.  The expected reduction in BaP  following long-
range transport did not occur; BaP/Pb  and BaP/EC were approximately the same in the two
locations.  However the BaP/EC ratio in Detroit was much higher than in Delaware or Bermuda.
To explain the difference between the inner city and Delaware/Bermuda ratios, the authors
suggested that BaP is rapidly reduced to a low level as paniculate matter ages, after which  further
losses are slow.  Evidence of atmospheric transformations of other PAHs during transport were
found from elevated proportions of nitro- and hydroxynitropyrenes to inert markers in Bermuda
compared to Delaware. The ratio of reactive/unr.eactive PAHs on aerosols was about the same in
stationary southern Norway air compared to air transported from the United Kingdom and  France
(Bjorseth et a/., 1979).

       A method for correcting PAH profiles for degradation and/or  physical  removal  effects
during transit was suggested by Masclet  (1986).  A "relative decay index" (RDI)  was established
based on the diurnal variation of each PAH concentration in urban air.   This RDI concept was
used in a CMB model for PAHs in Paris air (Pistikopoulos et al, 1990).  PAH profiles in ambient
air were matched against profiles of six  5-6 ring PAHs  from three  sources: spark-ignition and
diesel  engines, and  domestic heating.   Losses  of PAHs due to  differential  reactivity were
accounted for by summer and winter RDIs.  Spark-ignition  vehicles accounted for 40-70%  of
these PAHs,  and diesels for 20-40%.  The contribution from domestic  heating rose from a few
percent in summer to 20-40% in winter (Figure 6).

    3.4.2 Alteration of PCDDFs in the Environment

       Degradation of PCDDFs during  transport has  not been given the attention received by
PAHs.  Gaseous species  are photosensitive (Podoll  et a/., 1986) and reactive toward hydroxyl
radicals (Table 1).  PCDDFs in natural waters (Friesen et a/.,  1990) and  on fly ash (Tysklind and
Rappe, 1991) can be photolyzed in solution.  Penta- and hexa-CDDs were photo-reactive when
exposed to natural sunlight in distilled water - acetonitrile solution, and the photolysis rate was

                                          22

-------
            Domestic heating
I   1 Diesel vehicles
Spark -ignition engines
'OOr-m
80
60
          a E3 C




                                                                  SL.


                                                                                      P
                                                                                      1
                                                                                      1
   :h/b/8b
                                                          M»,
                                                                            M VI 111
                                                                                        . I'.VH '
    Figure  6.   Contributions to total  PAHs  in Paris  air, calculated  by  CMB.
              Pistikopoulos el a/. (1990).
                                                     Source:
                                                23

-------
markedly increased in natural water -acetonitrile.  Projected half-lives in sunlit  surface waters
were 27-81 days.  The sensitized photolysis may have been due to hurruc materials (Friesen et al.,
1990).  Less chlorinated PCDDs are also photolyzed in water (Dunlin et al., 1986).  Koester and
Hites (1992) noted that profiles of PCCDFs in urban and rural areas were different, and suggested
that a combination of phase distribution effects and gas-phase reactivity might be responsible.

3.5 Applications of source apportionment techniques for organics
   3.5.1 Volatile organic compounds (VOCs)

       Comparative ambient air concentrations and profiles of individual VOCs in major  U.S.
cities, taken from a national database for VOCs (Shah and Heyerdahl, 1988), were presented by
Edgerton et al. (1989a).  Similarities between aromatic hydrocarbon profiles in several cities and
their correspondence to an average auto exhaust profile suggested that auto exhaust was probably
the dominant source.   Profiles  of industrially related chlorinated and fluorinated hydrocarbons
showed distinct differences among the cities, depending on the source types.

       Source profiles and CMB models have been used to apportion VOCs in several U.S. cities
and urban areas from around the world (Aronian et al., 1989; Doskey et al., 1992; Kenski et al.,
1991; Scheffe/ al, 1989; Scheff and Wadden, 1991;  Sweet and Vermette, 1992).  These papers
describe in great detail the  characteristic "fingerprints" of different VOCs in  sources  such as
vehicle  exhaust,  gasoline vapors, industrial  emissions (refineries, coke ovens, chemical plants),
architectural  coatings (paints,  thinners, and cleanup  solvents), dry-cleaning,  degreasing,  and
graphic arts (printing).  Examples of CMB apportionment  of total  hydrocarbons  are given in
Table 2 for three cities

       A check on the accuracy of CMB results is to compare them with independent emissions
inventories. This was done by Kenski et al. (1991) and Scheffe/ al. (1989) for Chicago, Detroit,
and Beaumont, Texas (see Table 2).  In Detroit,  CMB  calculations  and emissions inventories
agreed well for vehicles, gasoline vapor, coke ovens, and architectural coating contributions.  The
emission  from coke ovens  was  confirmed  by  close  correlation  between  CMB modeling
coefficients and a source-based score derived from Gaussian plume dispersion  models.  Detroit
CMB results were high for refineries and graphic arts sources.  In Chicago, agreement between
CMB and inventory  results was good for all source categories except refineries, for which the
CMB model again gave high values. The opposite  was the case in Beaumont, where inventoried
emissions from refineries  and gasoline vapor exceeded those modeled by CMB.   Very good
agreement between CMB and inventory was found for a polyethylene manufacturing plant.

                                           24

-------
TABLE 2.  Percent Contributions of Various Sources to Total Hydrocarbons in Three Cities:
          Comparison of Emission Inventories with Chemical Mass Balance Results2

   Source                Detroit            Chicago            Beaumont

                          INV      CMB     INV      CMB      INV      CMB

   Vehicles               32.9      28.2      35.1      41.2      11.9      13.6

   Gasoline Vapor        6.9      9.4      7.6      6.4       4.2      19.5

   Point Sources
   Refineries              0.7      16.5      1.3       13.5      23.1      9.1
   Polyethylene Mfg.                                             7.3       7.0
   Other Chem. Plants                                           35.6

   Rubber Mfg.                                                 13.4
   Wood Pulp & Paper                                          1.3

   Coke Ovens             2.0      3.7
Coatings
Architectural
Industrial
Graphic Arts
Other

3.8 2.5 5.5 6.2
14.0
0.7 4.7 9.8 11.0
38.8 34.5 40.6 16.5

0.6

1.4

a) Source: Kenski et al., 1991, Tables 6 and 7.

      CMB models, factor analysis, and wind trajectories were used to investigate VOC sources
in southeast Chicago and East St. Louis by Sweet and Vermette (1992).  The three techniques, if
applied  appropriately,  are  complimentary  in  nature.   Factor  analysis  requires  no  a priori
knowledge  of sources and is useful for  confirming the importance of known emissions or
suggesting sources not inventoried  (Sweet and Vermette, 1992).  In southeast Chicago, factor

                                         25

-------
analysis indicated that four sources types accounted for 78.4% of the variance in the observed
data:  The first and largest factor was attributed to urban air sources (perhaps a combination of
vehicle emissions, gasoline evaporation, and solvent emissions).  A second factor identified only
with benzene  was assigned to coke  oven combustion.  The third  factor comprising carbon
tetrachloride and chloroform was attributed to a regional source outside the city.  A fourth and
unknown factor was characterized by chlorobenzene.

       CMB results showed that under average conditions most aromatics resulted from vehicle
exhaust.   Trichloroethane,  trichloroethylene,  and tetrachloroethylene  came  from  degreasing
solvents and dry-cleaning. In east St.  Louis, a factor associated with several aromatics and light
chlorinated compounds was identified  with a local chemical plant on the basis of wind trajectory
analysis.

       Sweet and Vermette also showed how the situation in south-eastern Chicago can deviate
from the average during a pollution episode.  During one event, 72% of the benzene and 17% of
the ethylbenzene were released by coke ovens  (Table 3).  A high proportion of benzene in coke
emissions was also found by Kenski el al. (1991). By apportioning total VOCs with  assumed
factors, Blakley and Klevs (1990) assigned highest releases of benzene to counties at the  south
end of the lake which had large coke emissions.

       Pollution sources in the Kanawha River Valley, West Virginia were examined  using factor
analysis of VOC and trace element data (Cohen et al., 1991a,b).  The valley contains numerous
sources of VOCs,  including eleven  large industrial  complexes that  use,  store, produce, and
transport  organic chemicals.    Only  at one site were aromatic VOCs  identified  with  an
"automobile"  factor, which  was defined by lead,  bromine, and paniculate carbon.  At  other
locations  in the valley,  VOCs could  not be  apportioned to any one  factor,  but instead  were
ascribed to a factor called "general VOCs". The "general VOC" factor differed little among sites
and was not associated with total particles or element-speciated paniculate matter.

       The authors (1991b) discussed the difference between  their  source apportionment of
VOCs in  Kanawha River Valley and  the receptor modeling results of Stevens et al. (1989) in
Boise, Idaho.  In the latter study several aromatic  hydrocarbons showed excellent correlations to
fine paniculate lead (a gasoline source that is rapidly decreasing).  These tracers were useful as a
lead replacement for estimating  the contribution  of vehicles to ambient concentrations of fine
paniculate extractable organic matter (Zweidinger et a/., 1990).
                                           26

-------
 TABLE 3.  Contribution of Sources to Individual  VOC Concentrations for a South  Chicago
           Sample During a Pollution Episode3


Benzene
Toluene
Ethyl-benzene
m,p-xylene
o-Xylene
1,1,1 -Tri-chloroethane
TrichJoro-ethylene
Tetrachloro-ethylene
Coke Vehicle
Ovens Exhaust
71.6 26.3
0.8 45.6
17.0 68.3
8.3 71.8
1.4 66.6
0 0
0 0
0 0

Paint
0.6
52.0
9.4
18.4
30.3
0
0
0
Gasoline
Vapor
2.3
1.8
5.3
1.8
1.4
0
0
0
Degreasing
Solvent
0
0
0
0
0
100
100
12.9
Dry
Clean.
0
0
0
0
0
0
0
87.1
a) Source:  Sweet and Vermette, 1992; CMB results.

       According to Cohen, the failure of VOCs to show a clear relationship to lead in Kanawha
Valley might have been because: a) there were similar VOC sources at all sites, b) there was good
mixing of pollutants in the valley, and/or c) the sources of VOCs were regional. These studies
point out the value of using elemental and paniculate information in addition to VOCs to speciate
source types.

       Some of the longer-lived halocarbons (e.g. methylchloroform, tri- and tetrachloroethene)
have been  used  as "tracers  of opportunity"  to document pollutant  transport out of  the Los
Angeles basin to the  Nevada - Arizona desert (Bastable et a/.,  1990; Pryor and Hoffer, 1992;
Miller et a/., 19.90; White et a/., 1990). Methyl-chloroform has a distinctly  weekly cycle, being
high during the week and dropping to near-baseline values on weekends.  This cycle has been
                                          27

-------
attributed to releases of methylchloroform from metal fabrication and electronics industries in the
basin (White el  a!.,  1990).   Haze episodes leading  to reduced visibility were correlated  with
elevated concentrations of methylchloroform (Miller el al., 1990).

       The concentrations of trace gases, such as chlorofluorocarbons, chJorocarbons and carbon
monoxide were  also  applied  in the  statistical  methods  to discuss  the sources  and  their
contributions to the contamination at remote receptors (e.g. Khalil and Rasmussen, 1984). It was
concluded that the ratios between these gases can be used as global tracers due to their distinct
and different application pattern.

       The concentration ratios of light hydrocarbons, such as ethane and propane to chlorinated
ethenes have  been used as signatures of emissions from  natural gas  exploitation regions  as
compared with emissions from regions with extended application of industrial solvents (e.g.  Hov
el a/., 1984).   It was indicated  that the use of chlorinated ethenes is  primarily  confined  to
industrialized countries.

    3.5.2 Semivolatile orgamic compounds (SOCs)

    A. Polycyclic Aromatic Hydrocarbons (PAHs)

       Unlike VOCs, very  little work has been done  to relate PAHs in ambient  air  to  their
sources through CMB or factor analysis modeling.  This is because emission factors have been in
short supply, and even within source types the quantities and ratios of PAHs emitted have been
highly variable.  One can see this from  the examples of emission factors given  in Johnson el al.,
(1992), and also  from a review of PAH profiles by Daisey el al. (1986).  These authors critically
examined literature data from the 1960s through early 1980s for PAHs  in several source types.
All PAH concentrations were expressed relative to benzo(e)pyrene  (BeP) because this compound
is fairly  stable and is  found almost  exclusively  in the paniculate phase.  PAH profiles were
examined from auto exhaust (gasoline and diesel), residential coal and wood combustion,  oil- and
coal-fired power plants, and  industrial coal-fired boilers.

       Interpretation of PAH data is plagued by uncertainties due to  sampling  and analytical
artifacts.  PAHs occur in paniculate and gaseous forms, and atmospheric concentrations will be
inaccurate unless both  species are collected.  This has generally been done for source sampling,
but  until the early 1980s  most  ambient  air PAHs  were collected with  filters  only  and
concentrations of 2-4 ring compounds were seriously underestimated. Both the particle-and  gas-
phase compounds are mutagenic (Tuominen el al.,  1988;  Westerholm el al., 1991; Lewis el al,
 1988), providing another reason for collecting the  two fractions   PAHs can undergo reactions
                                           28

-------
 with  oxidant gases during sampling, thus changing their proportions.  Sample extraction and
 analytical methods vary among laboratories  and quality assurance  information is not always
 provided.    According  to  the  Daisey  review,  the  three  most  critical  factors  limiting
 intercomparability of the organic source emission profiles reviewed by  them were:   a)  The
 incompatibility of source emissions and ambient aerosol sampling methods and intervals, b) The
 general lack of emission profiles which represent an average for a given source  type, and c)
 Differences in organic profiles due to variations among sampling and analytical methods rather
 than in source emissions.

       The conclusions of Daisey were that existing data provided some useful information with
 respect to receptor modeling.  PAH profiles  from two source types  that had been  repeatedly
 sampled and analyzed  by the same investigator appeared to be fairly reproducible: coke ovens and
 coal-fired boilers.  Profiles from wood  combustion varied widely and  depended on combustion
 conditions.  Cyclopenta(cd)pyrene was especially enriched in exhaust from spark-ignition engines
 and was suggested as a useful marker compound.  Coronene and benzo(ghi)perylene were also
 relatively  high in vehicle exhaust compared to other sources. Other organics having potential as
 marker    compounds   were   retene   and   levoglucosan  for   wood  combustion,   and
 benzo(b)naphtho(2,1 -d)-thiophene (BNT) for diesel engines and fuels containing sulfur.

       Since the time of the literature  covered in the Daisey review, sampling and analytical
 methods for PAHs have improved greatly.  Filters followed by adsorbent traps to catch gas-phase
 PAHs are now in routine use for ambient air  sampling.  The availability  of standard reference
 materials for urban air paniculate matter and diesel particles has improved quality assurance (Wise
 el a/., 1986; 1988). PAH emission factors and profiles in a large number of source types have
 been  compiled (Johnson el ai,  1992).   Nevertheless, it  is interesting to note  that many of the
 reservations expressed in the Daisey review about data quality and availability were echoed in this
 1992  report.

       The three PAHs cited as vehicle  exhaust markers by  Daisey et al. (1986) and Baek et al.
 (1991a,b) are hardly unique. High levels of coronene were also produced by burning certain types
of vegetation  (Freeman  and Cattell,  1990).   The Johnson  survey found  a high  ratio of
benzo(ghi)perylene to BeP in wood stove emissions. Proportions of cyclopenta(cd)pyrene to BaP
in vehicle exhaust and industrial coal combustion effluents were similar, and  the former compound
was elevated in wood smoke (Table 4).  The Bghip/BeP ratio was much lower in wood soot than
in gasoline soot for profiles presented by Kang and Kamens (1992), in contrast to the information
given in the Johnson  et al. report.  Thus, ratios among the parent PAH compounds are  not
 foolproof source indicators.
                                          29

-------
       Daisey ei a/. (1986) felt that PAH data were best used in combination with other tracers
for  source apportionment^ g. trace elements)   Some  information on PAHs and er?r organic
compounds that may  serve as "marker compounds" of a particular  source type or process is
discussed below and summarized in Table 4.

       The  proportion  of alkyl  homologs  to  unsubstituted compounds  is  an indicator  of
combustion  source temperature.   Low temperature combustion yields soots that  are more
abundant  in  alkylated  PAHs,  whereas  unsubstituted (parent)  PAHs  predominate  at  high
temperatures (LaFlamme and Hites,  1978).   Initially  it was thought  that the alkyl  homolog
distribution in sediment cores could  be used to distinguish  "natural"  (low-temperature) from
"anthropogenic" (high temperature) combustion sources. However LaFlamme and Hites pointed
out  that  coal  combustion  occurs at  moderate temperatures  and yields  alkylated  PAHs.
Furthermore, they  felt that differential water solubility of alkylated and parent PAHs could alter
the distribution  in sediments from what was initially  deposited.  Tan ei al.  (1992)  recently
reported that while wood burning produces mainly parent PAHs,  low-intensity fires in forest floor
waste    ("duff1)    yields    relatively   high    levels    of   alkylated    phenanthrenes,
cyclopenta(def)phenanthrene, and dodecahydrochrysene.

       The ratio of methylphenamhrenes to phenanthrene (MP/P) is low in combustion effluents,
but high in unburned petroleum products  Takada et al. (1990) found that MP/P in auto exhausts
and asphalt were  high  relative to combustion products from a  steam  generator.   Using this
information,  along with sulfur heterocyclic compounds, the authors concluded  that PAHs  in
Tokyo street dust  were strongly  affected by automobile exhausts.  Dusts from residential areas
had a somewhat greater contribution from stationary source combustion products.  The authors
pointed out that the MP/P ratio in auto exhaust is highly variable,  depending on engine load and
cylinder exhaust temperature, and suggested that extensive collection of auto exhausts should be
conducted.  Runoff samples from urban and coastal South Carolina were depleted in MP relative
to crankcase oils and  diese, fuel, and showed PAH profiles similar to the atmospheric particles.
The conclusion was that PAHs in street dust came mainly from atmospheric deposition and not
from dripping crankcase oils (Ngabe, 1992).

       Alkylphenanthrenes were also high in samples from the Baltimore Harbor Tunnel  (Benner
et a/., 1989).  A comparison of tunnel air  samples to  standard reference diesel  and urban air
paniculate matter showed that tunnel and diesel  particles were enriched in MP   Factor analysis
suggested that contributions from  diesel and gasoline-powered  vehicles might  be  separated by
alkvlated PAH content
                                           30

-------
TABLE 4. Marker Information for Identifying Sources of PAHs and Other Organic Compounds.
  Source

  Spark-source engines

  Diesel engines


  Motor vehicles


  High vs. low temperature
  combustion &
  combustion vs. unbumed
  petroleum products
  Wood combustion

  Wood combustion


  Vegetation fires
  Vegetation fires

  and other biogenic
  sources, biogenic

  vs. anthropogenic
  hydrocarbons unresolved
  complex mixture
 Marker Compounds

 cyclopenta(cd)pyrene

 benzo(b)naphtho(2,1 -d)-
 thiophene

 benzo(ghi)perylene,
 coronene
 parent/alkylated PAH
 ratios
methoxyphenols & other lignin
pyrolysis prod.
retene, levoglucosan
carbon-14, potassium

 alkylated phenanthrenes
cyclopenta(def)phenanthrene
dodecahydrochrysene

 carbon preference index

(CPI: odd for n-alkanes &
 alkanones, even for acids

& alcohols), presence of UCM
for (petroleum products),
biogenic markers:
retene, di- & tri-terpen
phytosterol
  Tire wear
Benzthiazoles
 Examples

 Daisey era/., 1986.

 Daisey era/., 1986;
 Alsbergera/., 1989.

 Daisey era/., 1986
 Alsbergera/., 1989.
 LaFlamme & Hites,1978;
 Takadaera/., 1991;

 Ngabe, 1992.
 Hawthorne et a/., 1988, 1989;
 Edye & Richards, 1991.
 Daisey era/., 1986


 Tan era/.,  1992.
Standley & Simoneit, 1987,
1990.
 Simoneit, 1989;
Mazurek & Simoneit, 1984;
Kawamura & Leuenberger et al.
1988;
Gagosian & Pelzer, 1986;

Farmer & Wade, 1988;
Foreman &Bidleman, 1990
Greaves era/.,1987.
Spies era/., 1987.
                                            31

-------
       As indicated by Daisey et al. (1986) other marker compounds may be used to advantage
in  differentiating PAH source contributions.   BNT was  used  as  a diesel  exh. -st indicator by
Alsberg et al. (1989), who employed principal component analysis to examine profiles of PAHs
and aromatic VOCs collected in Gothenburg, Sweden.   Diesel  and gasoline emissions were
separated by enrichment of BNT and particle-associated light PAHs in the former, and aromatic
VOCs  and heavy PAHs (particularly benzo(ghi)perylene  and coronene) in the latter.  Cautions
about using these two PAHs have been mentioned earlier.

       Methoxylated phenols have been suggested as tracers for wood smoke pollution, which is
a major contributor of PAHs (Hawthorne et a/., 1988, 1989). More than 70 species, arising from
pyrolysis of lignin, have been  identified in  soot from residential wood stoves.  Guaiacols  in
hardwood and softwood soots were nearly the same, whereas syringols were much higher in the
hardwood soots.  It was suggested that guaiacols could be used as  markers for wood combustion
in  general and syringols could serve to differentiate wood type.

       Simoneit (1989)  reviewed the wide variety of biomarker compounds available for use  in
source reconciliation.  These include n-alkanes and similar compounds (acids,  alcohols, ketones),
phytosterols, terpenoids, and  terpenols.  Some of these may be useful for investigating regional
sources in long-range transport.  For example, the signatures of C27-29 phytosterols were different
in  aerosols from the western U.S., Nigeria, and southeastern Australia. Further work is needed to
evaluate the stability of marker compounds in the atmosphere and during sampling.

       Air samples from slash-burns along the Oregon coast were  analyzed for a wide variety of
plant wax components and terpenoids (Standley and Simoneit, 1987). PAHs accounted for only a
minor  part of the hexane-soluble material, most of which  was plant waxes,  resins, and thermally
matured  compounds.  Smoke from the burns contained  straight-chain homologs  with a strong
plant wax signature:  n-alkanes and n-alkanones showed an odd-carbon preference, peaking  at
€27.   Straight-chain alkanoic acids and alcohols showed  an even-carbon preference.  The acids
peaked at C22, C24> or ^30> whereas the maximum for the alcohols occurred at C22  A number of
di- and triterpenoids, as well as retene, were suggested as molecular markers for this type of bum.
In a subsequent study (Standley and Simoneit,  1990), polar  cyclic di- and  triterpenoids were
analyzed in extracts of residential wood combustion aerosols. Distinct signatures were found that
could be used to trace the input from coniferous, alder, and oak combustion products.

       Profiles of organic compounds in atmospheric paniculate  matter collected in Colorado
was examined by Greaves et al.  (1987) using factor analysis.  One factor associated with ozone
and  oxygenated  compounds such  as acids, furans,  aldehydes,  ketones,  and  lactones  was

                                           32

-------
 interpreted as arising mainly from photochemical processes.  Factors 2 and 3 which  contained
 terpenoids and odd-chain hydrocarbons were ascribed to biogenic sources.  This hypothesis was
 strengthened by noting that factor 2 was strongly dependent on wind direction, being largest when
 the wind was from a national forest to the west.  Factor 4, associated with CO and several long-
 chain alkanes, was identified with motor vehicles.

       A proposed method for source characterization was  based on the high-resolution gas
 chromatographic profile of neutral-fraction components of fine organic aerosol (particle diameter
 < 2 jim)  (Hildemann et a/., 199la).  A large number of source types were examined:  wood
 combustion; automobile and truck exhausts; fuel oils; natural gas appliances; vegetation detritus;
 tire, brake, and road dusts; cigarette smoke; roofing tar; and charbroiled meats.  The similarities
 or differences among these aerosol signatures were determined by hierarchical cluster analysis.

       Tire  wear can contribute PAHs to road dust, since carbon black is used  in the rubber
 (Voldner and Smith,  1989). Spies et al.  (1987) suggested using benzthiazoles, found in estuarine
 sediments of San Francisco Bay, as indicators of street runoff.  The benzthiazoles are breakdown
 products of antioxidants  added to tire rubber.

    B.  PCDDFs

       A  considerable amount of work has been done to determine differences in congener
 profiles from various sources and relate them to environmental  matrices.  The use of principal
 component analysis (PCA) and other multivariate techniques seems to be more frequent than in
 PAH investigations.

       Czuczwa and Hites (1986) applied  PCA to differentiate sources  of PCDDFs  in  Great
 Lakes sediments.  Profiles of PCDDF homologs (4-8 chlorines) were compared for sediments,
 paniculate ambient air samples, incinerator combustion products, and different PCP  products.
 Sediments from all lakes  except Ontario formed a tight cluster with ambient air particles that was
 separate and distinct from the incinerator or PCP patterns. Clustering of a Lake Ontario sediment
core with the PCP data pointed to contamination from suspected sources on the Niagara River.

       Changes in the relative proportions of PCDDFs in moving from source to receptor were
investigated  by Eitzer and Hites (1989b) using PCA.  Homolog  profiles from different  urban
combustion  sources  were  widely scattered  and showed  no grouping tendency.   Urban and
industrial air samples clustered in the middle of the sources, indicating that a mixture of sources
produces a fairly consistent homolog profile in ambient air.  Differences could be noted between
urban and suburban air groups.
                                          33

-------
       Of special interest was the comparison of paniculate vs. gaseous homolog profiles in air,
and paniculate vs.  dissolved profiles in rain.  Examination of homolog distributors for these
individual phases by  PCA  clearly showed that panicle-bound PCDDFs  are  preferentially
transported to sediments,  giving rise to the pattern alterations shown in Figure 5.  Relative to
whole (paniculate + gaseous) air samples, the homolog profile of the paniculate fraction moved
toward the cluster for sediments, whereas the gaseous fraction moved away.  The same trend was
noted for paniculate and dissolved components of rain samples.

       PCA revealed differences between urban (Indianapolis, IN) and rural (Trout Lake, WI) air
samples, indicating that changes occur during atmospheric transport. The cluster of wet and dry
deposition samples was shifted away from those  of whole air samples and toward  the sediments
(Koester and Kites,-1992).

       Smith el al.  (1990) were able  to  distinguish three patterns of ambient air PCDDFs in
upstate New  York cities  by PCA,  which they  designated as "source-related",  "common
background", and "enhanced lower-chlorinated compounds"  The latter samples were especially
high in TCDFs. Some air samples from Niagara  Falls, NY showed a "source" profile, typical of
municipal waste incinerators; others showed the "common background"  pattern.  A background
pattern was  also yielded  by air  sample from  a parking garage.   The  "lower chlorinated"
distribution was found on consecutive hot summer days when air  masses were  transported from
New  York  City and  New Jersey  to  Albany, where the samples were taken.   The authors
speculated that enhancement of less chlorinated species might result from selective deposition of
heavier, more panicle-bound compounds, but also mentioned the possibility of other, unidentified
source types.

       Clues to the origin of these IcJwer-chlorinated PCDDFs were provided  by  Bacher el al.
(1992), who examined fingerprints of PCDDFs and their brominated analogs in  chimney deposits
from wood burning and in auto exhaust.  Compounds containing 1-4  chlorines were abundant,
especially for the PCDFs.   Auto exhaust yielded the highest levels of brominated dibenzodioxins
and dibenzofurans,  most  containing  1-2  bromines.   The authors felt that the  full homolog
spectrum (1-8 chlorines)  should be considered  when deducing  sources  from  environmental
patterns. Rappe el al. (1989) also commented on the similarity of the isomeric  pattern for tetra-
CDDFs in air paniculate.samples, car exhaust, and municipal incinerator products.

       Emissions data from incinerators  in different countries were compared using PCA
(Edgerton el al.,  1989). Of twelve municipal incinerators, homolog profiles for seven were tightly
grouped, whereas the other five were widely scattered.  Those  five included three incinerators that

                                          34

-------
 were burning under unsteady conditions and two that had PCDDF levels near the detection limit.
 The seven grouped incinerators  were clearly distinguished from the well-separated clusters of
 three sewage incinerators and three Kraft mill boilers.  Ambient air samples from Columbus,  OH
 which were taken near both a sewage sludge and a municipal incinerator were similar to both
 source types.  Air samples from Akron, OH were grouped  in the overlapping clusters formed by
 municipal incinerators and traffic tunnels.  Based on a survey of 400 publications on PCDDF
 sources,  Pitea et  al. (1989) used PCA  and cluster analysis to differentiate PCDDF patterns in
 various types of incinerators.

       Class separation of PCDDFs in human milk samples from Sweden  was achieved using
 partial least squares analysis (Lindstrom et al., 1989).  One set of milk samples characterized by a
 relative high proportion  of PCDFs (especially  hexachloro-DFs) came  from  an area near a
 magnesium plant which had high emissions of these compounds.

       Multivariate analysis of atmospheric data from Stockholm and Baltic areas indicated a
 change in PCDDF proportions in moving from the city to open  coastal areas  (Broman et ai,
 199la). Three air samples showed similarities to settling paniculate matter collected in sediment
 traps from the Baltic. A fourth air sample more closely resembled particles from filtered water.

       Homolog profiles from combustion  sources and in ambient air are substantially  different
 from those in pulp mill effluents.  Combustion patterns of 4-8 chlorinated compounds, exemplified
 in Figures show a wide range of PCDDFs.  Effluents from Kraft bleaching are enriched  in tetra-
 CDDFs,  especially in  toxic compounds substituted in the 2,3,7,8-positions  (Amendola et  a/.,
 1989; Axegard and Renberg, 1989; Clement et a/., 1989; Fouquet et a/.,  1990;  Swanson et al.,
 1988).  Homolog fingerprints in  sediments from large Swedish lakes showed that mill-related
 compounds were distributed throughout the lake.  These patterns could be distinguished  by PCA
 from those in three small lakes that were atmospherically influenced (Kjeller et al., 1990).

       Kang and Kamens (1992)  used results from laboratory chamber degradation experiments
 to correct PAH profiles in urban air for transformation effects.  Corrected profiles were used in a
 CMB model to differentiate  three source types: residential wood combustion,  gasoline spark
ignition emissions, and  diesel engine emissions.

   C. Pesticides

       The application of source  apportionment techniques  to investigate the  sources  of
pesticides has been meagre. This is largely  because of the physio-chemical  nature of pesticides,
how they are used, and where they are utilized.  The  earth's surface  is  a vast reservoir for
                                          35

-------
pesticides and other organic susbstances as well  as metals such as mercury.  The exchange of
pollutants is a two way process with soils, plants, water bodies, etc. acting as t    • rources and
sinks  of airborne  chemicals.   Understanding the two-way nature of air-surface exchange is
necessary if we are to interpret observed annual cycles  of pesticides in the atmosphere,  and
eventually to  understand  the  sources.   Several investigations  have shown that ambient air
concentrations of OC pesticides and PCBs are higher in summer and lower in winter (Hermanson
and Kites, 1989; Hoff et a/.,  199la; Lane et al, 1992a; Larsson and Okla, 1989; Manchester-
Neesvig and Andren, 1989)(Figures 4, 7).  Differences in summer-winter fluxes of PCBs in Green
Bay (Achman et al., 1992) and HCHs in the Great Lakes (McConnell et al., 1992)  have been
observed.

       The temperature effect on these processes can be simply described by the relationship:

                        LogH(orP) = m/T +  b                               (7)

where H or P  are the Henry's law  constant or vapor pressure of the compound and T is the
ambient temperature (Kelvin).  The seasonal cycle of pesticides in air at Egbert, Ontario are
reasonably well described by Equation 7 (Hoff et al., 1991b, Figure 7).  Similar trends were found
for HCHs in the Great Lakes region  (Lane et al., 1992b) and  chlordane in Columbia,  South
Carolina (Bidleman, unpublished data). The plant-air BCF is related to H through Equation 6.

       The implication is that concentrations of pesticides and PCBs in ambient air are controlled
by take-up and degassing from soil and plant  surfaces.  New emissions  of pesticides are thus
super-imposed  on  this background, the magnitude of which varies seasonally and regionally.  A
significant problem is distinguishing local surface exchange phenomena (Equations 6 and 7) from
transport- and usage-related  events.   For  example,  did  the  high endosulfan concentrations
observed at Egbert (Figure 7) result from volatilization or local summertime use of the chemical?
Were elevated DDT and  chlordane concentrations at Egbert due to temperature controlled local
volatilization,  or the  fact  that warm air masses were  transported from  the  southern  U.S.?
Trajectories shown by Hoff et al.  (1991b) do indeed show that high concentrations of DDT,
chlordane, and other OCs were associated with southerly airflow.

       The answer is that all of these phenomena are related, and at present we cannot decouple
local surface exchange effects from transport. The ability to do this is important to understanding
current sources of OC pesticides and PCBs to the Great Waters.
                                           36

-------
   (CO •
   350-
   SCO-

   JCO-
_^  ISO-
'S  100'
g  so-'
~  0-
   100-
   90-
0  I0i
Q  70'
   to-
   SO-

   30 J
   20-1
   10-.
   0 —
                   I
                             PC*
                                     •Djrttnn
                                                          ZOfll*
                                                     ISO 4

                                                     ioc'-

                                                     M-
                                                                        per
                          »08i	
                          iwoj  (d)
                          iwoj
                          I4M-J
                          IJW'   TBftlB»llll
                          IOMJ
                          KXT
                          HC1
                          400-
                                                               Jl
                             Mv  Hi;  U;  Sc»x.
                                                            Hv.  Ih;
                                                                      ipt NX
                       or
                       s
                       u
                      — IOOC
                      C
                      O
                             lo( C = -3560 T *  13.7
                             cu-Chlordaue
'  loi C " -3890'/T * U.8
.. 4.I-DDT
                           loj C - -3JOO/T + 12.6
                           t-HCH
                                                     lot C « -8980 'T - 33 3
                                               | MO JJOO
                                                1000 T(°K)-'
Figure  7.   Annul  cycles of pesticides at Egbert,  Ontario,  1988-1989 (Hoff  et  aL
          1991a,b).  Top:  Monthly trends of chlorobenzenes (CBz) and several pesticides
          (PCC=toxaphene).   Bottom:   Equation  6 plots of atmospheric concentration
          (log) vs. reciprocal temperature.
                                            37

-------
       Isomeric Ratios of Pesticides

       Many technical pesticides, especially OCs, are mixtures of several isor.. .1.  ai;d related
compounds.  Relationships  among  these species provide clues to sample history and  source,
however ambiguities result  from  the  use  of products having different  isomeric  contents.
Differential reactivity of isomers also results in changes in their relative proportions which are not
well understood. Examples of compound ratios as source indicators are given below.

       Hexachlorocyclohexanes.   Technical HCH  (formerly called "BHC")  consists  of five
isomers:  a, (3, y, £• The isomer in  highest proportion is a-HCH, accounting for 55-70% of the
mixture.  The only insecticidally active species, y-HCH, is present at only about 8-15%. Technical
HCH is heavily used 'in Asia; India  alone accounts for over 20,000 metric tons/y. Mexico also
uses technical HCH (FAO,  1986-89).   Ratios of a-HCH/y-HCH  in various technical HCH
products range from 0.6-15; 4-5 has been found in mixtures from two heavy-use countries, India
and China (Bidleman et al,  1992b). Technical HCH has been banned in Canada (1971) and the
U.S. (1978), and replaced by much smaller quantities of lindane (pure y-HCH). Lindane is also
used exclusively in western Europe.

       Atmospheric samples in the  Northern Hemisphere show a global background of a-HCH
with spikes of y-HCH from regional use of lindane. Air samples from India show a-HCH/y-HCH
in the 2-8  range, whereas the same ratio is 0.2-1 in Europe (references in  Bidleman et al.,
1992a,b). In the Great Lakes region a-HCH/y-HCH  ranges from near unity to 20 (Lane et al.,
1992b).  A springtime minimum in this ratio at Egbert (Hoff et al., 199la)  and  Green Bay
(McConnell,  1992) may  reflect regional  use of lindane as a seed treatment. Lane et al.  (1992b)
observed a  three-fold increase in the concentration  of a-HCH  at  Point Petre accompanying
airflow from the S-SW  As a user of technical HCH, Mexico may be the source.

       A confounding factor in interpreting a-HCH/y-HCH ratios is possible interconversion  or
selective removal of the isomers.  Several studies (referenced in Bidleman et  al., 1992a)  have
demonstrated that y-HCH is slowly transformed to a-HCH in soils and sediments by microbial
action.  Suggestions have been made that this transformation also takes place in the atmosphere
and may account for the exceptionally high a-HCH/y-HCH ratios observed in  some Arctic air
samples (Pacyna and Oehme, 1988).  No experimental  evidence  has been found to confirm  or
deny this hypothesis. Ballschmiter (1991) called into question the isomerization of y-HCH in the
atmosphere by noting that a-HCH/y-HCH  ratios are  typically  much lower  in the Southern
Hemisphere,  where use of lindane is more prevalent.  If atmospheric chemistry is responsible for
                                          38

-------
 the high proportion of a-HCH in the Northern Hemisphere, these processes should take place in
 the Southern Hemisphere as well.

        DDT and Related Compounds. Technical DDT contains about 70-80% of p,p'-DDT,
 the insecticidal isomer, 20-30% o,p'-DDT, and minor percentages of other isomers and impurities.
 In the environment, the DDTs are transformed to a greater or lesser extent into DDEs (p,p'-DDE
 and  o,p'-DDE) and DDDs (p,p'-DDD and o,p'-DDD).  Polar metabolites are also formed, but
 their use as markers has so far not been exploited.

       DDT has a very long lifetime in soils. A  1985  survey of California soils ~ 13 years after
 the ban of DDT ~ revealed detectable residues at every one of the 99 sites tested  (Mischke et al.,
 1985).  During this time substantial changes had  taken place in the relative proportions of DDT
 compounds.  The average ratio of DDTs  (sum of p,p'-DDT and o,p'-DDT) to total DDT residues
 (DDTs + compounds in the above paragraph) in soils averaged 0.49. Thus, about half the DDT
 was broken down to DDE and DDD compounds.

       Since DDE has a higher vapor pressure than either DDT or DDD, residues volatilizing
 from soils treated long ago ("old" residues)  would be  expected to  contain  predominantly DDE.
 The  composition of vapors from technical DDT that is freshly applied or evaporated from  soils
 treated relatively recently should be enriched in the parent compounds, p,p'-DDT and o,p'-DDT.
 This information allows judgments to be made  concerning "old"  and "new"  sources  of the
 chemical.

       Peat  cores  from  the  Great Lakes  region  and  eastern  Canada  showed  peak DDT
 accumulations in the late 1960s, but surface samples representing the 1980s still  contained DDT
 residues. Moreover, the composition of  this freshly deposited DDT was largely p,p'-DDT and
 o,p'-DDT. These findings led Rapaport et al. (1985) to the hypothesis that fresh DDT was being
 transported to the Great Lakes region, possibly from use in Mexico and Central America.

       The composition of DDT residues in peat  cores may not be reflective of what is in the
 atmosphere because of differential deposition by precipitation and fallout. Since  the DDTs have
 lower vapor pressures, they are present to a greater extent on atmospheric particles and more
 amenable to deposition. Examination of DDT/DDE ratios in atmospheric samples from different
 areas of the world shows the effect of age on residue composition (Table 5). High levels of total
 airborne DDT have  been found in India, Congo, and  Irkutsk (Siberia), indicating recent usage
(McConnell,  1992). In all  of these locations  the proportion of DDT/DDE was high.  The ratio
was lower, but still above  unity at Lake Baikal, which  was probably the recipient of long-range

                                         39

-------
atmospheric transport.  In North America, the proportion of DDE has increased with time.  This
is exemplified by the decrease in DDT/DDE in Denver from 2.0 in 1980 to 1.2 in  /SO  A nine-
state survey of airborne pesticides in  1970-71 showed average DDT/DDE = 2.6 (referenced in
Bidleman el ai, 1976). Larsson and Okla (1989) noted that DDT/DDE in atmospheric fallout in
Sweden had decreased significantly, from 4.1 in 1972-73 to 2.6 in 1984-85.
      TABLE 5.  Ratios of DDT/DDE in Ambient Air.
Location
Lake Baikal, Siberia
Irkutsk, Siberia
Brazzaville, Congo
India (Several towns)
Porto Novo, India
Denver, Colorado
Egbert, Ontario
Great Lakes
Alert, N.W.T.
Year
1991
1991
1989
1980-82
1987-89
1980, 1986
1988-1989
1990
1988
DDT/DDE
1.2
2.5
3.2
1.8
"mostly DDT"
2.0, 1.2
0.4
0.8
0.6
Reference
McConnel, 1992
McConnel, 1992
Ngabe & Bidleman, 1992
Kaushikera/., 1987
Rameshefa/., 1989
Foreman & Bidleman, 1990
Hoffetal., 1991a
McConnel, 1992
Panonetal., 1991
       Current  DDT/DDE  ratios  in  air  from  the  Great Lakes  region  are  typically <1.
Nevertheless there are some situations in which DDT/DDE > 1 are observed in North America.
McConnell (1992) gives ratios of DDT/DDE for samples collected in 1989-90 in the Great Lakes
region.  One sample  taken in  Green Bay  during June, 1989  showed DDT/DDE =  1.8.   This
sample was also enriched in toxaphene and  chlordane,  and was associated with airflow from the
S-SW  Another air sample from Lake St. Clair in August, 1990 contained unusually high levels  of
DDTs and DDT/DDE = 2.4.  Trajectory information has  not yet been obtained for this sample.
Although  preliminary,  these  observations suggest that pulses of "new"  DDT  are  being
superimposed on a North American background of atmospheric residues that are largely DDE.
                                         40

-------
        The source of this "new" DDT is unclear.  Usage in Mexico and Central America is often
 assumed  to be responsible, because of the DDT/DDE relationships  for "old" soils discussed
 above.  However a recent report documented long-term persistence of parent DDT -p,p'-DDT --
 in soils from New Mexico and Texas (Hitch and Day, 1992). Some of these soils contained five
 times as much DDT as DDE  when sampled  in 1985.  Investigation of the DDT/DDE ratio in
 combination with air trajectory information should be useful for locating such regional sources.

        Chlordane Isomers.  The ratio of trans-/cis-chlordane (TC/CC) in technical chlordane is
 about 1.3. Because TC is slightly  more volatile, the predicted vapor-phase ratio is 1.7.  Average
 TC/CC ratios in air samples from a source region (Columbia, South Carolina) are close the latter
 value (Bidleman et a/.,  1990).  As  chlordane undergoes atmospheric transport, the TC  is depleted
 and  TC/CC decreases.  Ratios from various locations removed from  North America show this
 effect:  Sweden 1.3, Sable Island 0.9, summer Arctic 0.5-0.6, winter Arctic  1.0, Lake Baikal
 (Siberia, summer) 0.6 (Bidleman et al., 1990, 1992c; McConnell,  1992).  The situation at Sable
 Island was interesting because it suggested differences between regional transport and the global
 background. Air samples showing 80% transport from the S-SW showed TC/CC =1.1-1.3.  By
 comparison, when air  masses  arrived mainly from the N-NW, TC/CC  dropped below unity.
 TC/CC  in air samples at Egbert, Ontario showed an annual  cycle, about 0.8-1.0 in summer  and
 1.2-1.5  in winter (Hoff era/., 1991a).

       Reasons for these isomeric  changes have not been determined. TC is more  labile in soils,
 and thus old residues might be expected  to contain higher proportions of CC.   Changes in  this
 ratio, as well as those of DDT/DDE might occur if these pesticides move by the "grasshopper
 effect",  continually being deposited and  revolatilized from  soils.    It  is unknown if chlordane
 isomers are decomposed in the atmosphere.

       Recently a new technique  has been developed which may  help distinguish "old" from
 "new" sources - the ability to separate optical isomers by gas  chromatography. Like amino acids,
 certain  pesticides  have "right" and "left" handed isomers.  These enantiomers have identical
 chemical and physical properties and are  thus  not distinguishable by normal analytical methods.
 The development of chiral-phase capillary gas chromatography columns allows their separation.
 The technique allowed Muller et al. (1992) to determine that, whereas the two enantiomers of a-
HCH were in a 1:1 ratio-in the technical  pesticide, alterations in this ratio occurred in soil, rain,
 and tissue samples. Because only enzymatic processes can change the  enantiomeric ratio, this is
 evidence for biological alteration of residues which are reflected in atmospheric samples (rain).
                                          41

-------
   D.  PCBs

       Little work has been done to  link PCB residues in  receptors to their sources through
examination of congener profiles.  This is probably because of:  a) Lack of routine use of high-
resolution analytical methods,  b) Poor interlaboratory comparability because of different numbers
of congeners included in the analysis, c) Poor information on PCB profiles from different sources,
and d) Changes in PCB profiles in the environment.

       In the 1960s and most of the 1970s, analysis of environmental samples for PCBs was done
by packed-column  gas chromatography (GC).  This  is a low-resolution technique which allows
the  different Aroclors to be distinguished, but does not separate individual congeners. More
recent  analytical  data has been obtained by high resolution capillary GC, which has far greater
separating power (Figure 8).  GC-MS is also becoming more common for determining PCBs as
homolog groups (Slivon et al., 1985; Alford-Stevens el al,  1986).

       The ability to carry out an analysis for individual PCBs is important. Chemical, physical,
and lexicological properties vary substantially among the PCBs.  Differences in volatility, water
solubility, and reactivity often lead to  PCB profiles in  environmental samples  that are markedly
altered from those of the Aroclor fluids responsible for the contamination.  Improvements in
analytical techniques within the last 10-15 years has been critical to our current understanding of
PCB environmental chemistry.  Still, it is common for laboratories to  include different numbers of
congeners in their analytical scheme, making comparisons among research groups difficult.  As is
the case with PAHs, there has been  a lack  of standard reference materials (SRMs)  for PCBs.
Recently marine sediment SRMs for PCBs have become available through NIST and the National
Research Council of Canada (Schantz et a/., 1990).

       As is the  case of PCDDFs, PCB profiles are greatly altered between source and receptor.
This is largely due to the selective phase partitioning that is a consequence of the wide variation in
physicochemical  properties of individual PCB congeners. An example of the difference between
PCB profiles in air and rain is shown in Figure 9 (Duinker and Bouchertall, 1989).  PCBs in rain
obviously originate from washout of the panicle-bound, rather than the vapor-phase, PCBs in air
Further changes in profiles accompany  sedimentation and bioaccumulation processes.

       Schwartz el al. (1987) discussed the problem  of quantifying PCBs as Aroclor mixture
equivalents vs. as the sum of individual congeners.  The conclusion  was that samples should be
analyzed for individual congeners, and total PCBs should be reported  as their sum.
                                           42

-------
                   GC sepantion of Aroctof 1260

                     I00
                     50
                                                  {») PiMtod cohimn
                                                   I  J  1
                      0  3  6  9 12  15 16  21 24 27  30 33 36 39 42  45 « 51 54

                                      Retention Dm* (mm)
                                                 (b) dpAlary column
                   100
~ 75
I
A
2
B
*
| »
B
1
25












1









J








u








JL
,

Ll







t

U










LJL>
: i i i i
16:40 20:00 23:20 28:40 30:00 33:20
RM0ntoo bme (mn)

Figure  8.   Comparison  of packed and capillary-column  gas chromatography  for
         seperating components of Arolocfor 1260. Source Alford-Stevens (1986).
                                         43

-------
                                          Vapor phase
                                                    136
                             Filter
                            Ram
Figure 9.  Profiles of PCBs in the paniculate (filter) and vapor phases of ambient air.
         and in rain.  Source: Duinker and Bouchertall (1992).
                                        44

-------
        Little is known about the relative proportion of PCBs from different sources.   Aroclor
 fluid compositions have been well established (Schulz et al.,  1989), but seldom have emissions
 profiles been established.  Thus it is difficult to determine the origin of PCBs in an ambient air
 sample.

       Profiles of PCBs  in ambient air have been used to suggest sources in South Chicago.
 Murphy et al. (1985) noted that PCB  patterns from  most incinerators tested were similar to
 Aroclors  1248 or 1254.   However one sewage sludge incinerator showed a predominance of
 heavy PCBs, similar to Aroclor 1260.  Holsen et al. (1991) found that PCBs in two spring air
 samples from Chicago matched Aroclor  1260, while a third was similar to Aroclor 1242, a lighter
 PCB mixture.  A series  of air  samples from South Chicago in February, 1989  showed  great
 differences in absolute concentrations of PCBs and their profiles (Gotham, 1990).  Profiles of
 most samples were skewed toward the low molecular weight  PCBs, as  is typical of ambient air.
 For these, total PCBs ranged from 300-2500 pg/m3.  Two samples contained total PCBs of 4700
 and 10,000 pg/m3, and were dominated by heavy components, as is Aroclor  1260.  Thus  three
 separate observations have documented that certain events  in Chicago lead to distinct profiles of
 heavy PCBs in ambient air.

       Several investigations have shown that PCBs volatilize from the Great Lakes and are a
 significant source  to the atmosphere above the water (Achman and Eisenreich, 1992; Baker and
 Eisenreich, 1990;  Hornbuckle et al., 1992). Differences in PCB  profiles over land and over the
 lake may be valuable in distinguishing freshly transported PCBs  from those being recycled into the
 atmosphere from the lake (Hornbuckle et al., 1992).

 4.  IDENTIFICATION OF LOCAL SOURCES

       The first step in defining the contribution of emissions from a given source or source
 region to the  pollution  load  at  a  given  receptor  is  to  prepare   chemical  and  physical
 characterization of this source/source region.  This  information is  being prepared for the Great
Lakes region through several research projects sponsored by the EPA, the Ontario Ministry of the
Environment, and the International Joint  Commission.  A number of institutions have  been
involved in these projects and as a result several technical  reports are now becoming available
The above described emission inventorying has been done mostly for precursors of atmospheric
acid compounds,  such as sulfur and nitrogen  compounds,  as  well as  some photochemical
oxidants, such as selected volatile organic compounds.  The hazardous pollutants, as defined in
the 1990 Clean Air Act Amendments (CAAA), have been less  well defined with respect to  their
emissions but this trend may be changing.
                                         45

-------
       Two  major conclusions  can be  drawn  on the basis of  current work  on emission
inventorying  for toxic compounds.  First, a list of source  categories and processes  generating
emissions of hazardous air pollutants can  and should be defined for the regions of interest. The
above can be done for point and area sources, separately. The second conclusion is that ongoing
work on inventorying emissions of criteria pollutants and VOCs, particularly those efforts that are
based upon emissions inventories prepared under the National  Acid Precipitation Assessment
Program (NAPAP) should  utilize  point  sources listing to estimate the emissions of toxic
compounds as well as the criteria pollutants, e.g. electric power plants, smelters, incinerators,
cement  kilns, and chemical plants. This information is needed when preparing spatial distribution
of toxic compound emissions within the study area.

4.1.  Major source categories for toxic compound emissions
       As pursuant to the requirements of the CAAA of 1990 an interim toxic emission inventory
has been developed for the  continental United States.  Preliminary results of this work which
includes the  geographical distribution  and source type analysis has recently been presented by
Benjey and Coventry (1992).  Several heavy metals were inventoried including arsenic, cadmium,
lead and mercury which are  of primary  interest in  this study.   Altogether emissions of 28
compounds, both heavy metals and persistent organic pollutants have been inventoried based on
the 1985 NAPAP inventory.

       Toxic emissions  in the Great Lakes region were also inventoried within other programs.
such the EPA Region V Project on Air  Toxics Emission Inventories for the  Lake Michigan
Region,  and  other organizations, such as the International Joint Commission (IJC).  The IJC
report provides data on the production, usage and atmospheric emissions of 14 toxic  chemicals,
including the four heavy metals studied in this work, and other priority compounds: polynuclear
aromatic hydrocarbons (PAH), dioxins, furans, polychlorinated  biphenyls (PCBs), and  pesticides,
with a focus on lindane. More  recently, an emission inventory  for toxic compounds has been
prepared within a project from the Ontario Ministry of the Environment.

       An emission inventory of toxic air contaminants  for the Great Lakes states is now being
developed  at the Michigan Department of Environmental Protection (Vial,  1992).  When this
work is completed it wilj be one of the most important sources of information on  emissions in the
Great Lake Region to  date.

       EPA  is  currently developing the Air  Toxics Emission Inventory Protocol for the Great
Lakes States (e.g. Radian.  1992).  The Emission Inventory Branch in the EPA Office of Air

                                           46

-------
 Quality Planning and  Standards has recently prepared the Air Clearinghouse for Inventories and
 Emission Factors (AIR CHIEF) providing information on estimating air emissions of criteria and
 toxic pollutants from selected sources

       Preliminary  results  of the  emission inventory  for heavy metals in  the United States,
 summarized by Benjey and Coventry (1992) suggest that the toxic metal emission sources are
 heavily influenced by primary and secondary metal production (93% of the arsenic emissions and
 more than 83% of the cadmium  emissions), gasoline combustion (about  59%  of the lead
 emissions)  and  waste incineration  (more  than 57% of the mercury emissions).   The  above
 emissions inventory is presently being  revised and the relative contributions from the various
 source categories may change, particularly the importance of lead from  gasoline  combustion.
 Most of the  emission sources for  the above  compounds are located outside the Great  Lakes
 region according to this inventory, and mainly in Arizona for arsenic, and Missouri for cadmium.
 Emissions of lead and mercury are more evenly distributed. It should be admitted, however, that
 the above suggested source category contribution to the total emissions in the United States is
 somewhat surprising.  Taking into account similarities in production technologies and efficiency
 of control  equipment, as well as the chemical composition  of wastes to  be incinerated in the
United States and Western Europe (e.g. Pacyna and Munch, 1988) one  should modify a list of
major source categories contributing to  the atmospheric emissions of toxic metals in the United
States.

       Major source  categories  for emissions of toxic heavy metals  and  persistent  organic
compounds in the Great Waters regions include:
    - combustion of bituminous coal, mostly in pulverized coal dry boilers, lignite, distillate oil .
   residual oil.  and natural gas to produce electricity (emissions of all heavy metals of concern
   and PAH., dioxins, and furans),
   -  combustion of bituminous coal, distillate and residual oil,  and natural gas in industrial
   boilers ( emissions of pollutants as above),
   - cement production in both dry and wet process kilns ( emissions of heavy metals and PAH),
   - production of chJoro-alkali using Hg-cell ( emissions of mercury),
   -  coke production as by-product in primary iron and steel manufacturing ( emissions  of all
   compounds except lindane),
   -  secondary  non-ferrous metal (and mostly lead) production (emissions of heavy metals and
   PCBs),
   - petroleum refineries (PAH),
   - refineries and chemical industry (and particularly production of chJorine and caustic soda,

                                          47

-------
   production of batteries, production of pigments, use of paint),
   - paper and pulp production ( mostly emissions of PCBs),
   - waste incineration ( emissions of all compounds except lindane),
   - glass industry,
   - production of fertilizers,
   - crematories,
   - combustion of fuels in internal engines ( emissions of lead, PAH, and dioxins),
   - use of lindane in livestock treatment ( emissions of lindane), and
   - use of lindane in wood and seed applications ( emissions of lindane).
       Other source categories which emit large amounts of almost all of the compounds of
interest are so-called diffuse sources.  These include:
   - combustion of gasoline and other fuels in motor vehicles,
   - volatilization of compounds from landfills, both flared and unflared
   - re-emission from terrestrial and  aquatic environments (mostly mercury), and
   - combustion of wood and other fuels to produce heat.

Impurities found in such products as pesticides, rubber tires,  pigments and coatings can  also
become import local sources of toxic substances (Ayers, 1987).

       Emissions of heavy metals from natural sources may also be quite important.  However, as
suggested by Lindqvist and Rodhe (1985) for mercury, it is  perhaps misleading to categorize
present day fluxes of mercury from soils, bodies of water, and biota as being "natural emissions"
For example, past anthropogenic Hg emissions have been  dispersed so  thoroughly through the
environment that this distinction is probably no longer meaningful.  The flux of mercury to and
from land and water surfaces has only recently been studied in any meaningful way (Schroeder et
a/., 1992; Vandal et al., 1991). These recent studies have emphasized the importance of the air-
water  exchange of mercury and  its importance in the behavior  and fate of mercury in the
environment.  In addition, we have a very poor understanding of the forms of mercury emitted
from these "natural  systems", but this is the focus of ongoing research (Fitzgerald et a/., 1991;
Pacyna et a/., 1992).  Natural sources which are of major importance include:
    - re-emission of volatilized heavy metals from soil and surface waters,
    - re-suspension of soil panicles,
    - forest fires, and
    - volcanic eruptions.

                                           48

-------
        Potential sources for the critical organic compounds studied in this work are vast.  Some
 of the major source categories include:
    - application of pesticides,
    - combustion of fossil fuels in electric power, co-generation and heat production plants,
    - combustion of fossil fuels in commercial, industrial, and residential units,
    - mobile sources,
    - manufacturing and use of basic organic chemicals, and
    - waste incineration.
        Emission data  currently  reported by EPA  and  LJC,  reveal that  large  portion of the
 emissions of toxic pollutants to the air in the United States can be  generated outside the Great
 Lakes  region.   This is particularly  true  for emissions from point sources.   However, the
 neighboring states, and particularly Missouri, generate  large quantities  of these emissions in
 electricity and heat producing power plants, primary and secondary non-ferrous smelters, steel
 and iron manufacturing plants, and waste incinerators. The contribution of atmospheric emissions
 of toxic compounds generated in various regions of the United States is presented in Figure 10,
 together with the emissions from  12 regions in Canada. The data  were  prepared on the basis of
 research carried out for IJC by Voldner and Smith (1989).  This report covers two U.S. regions,
 specified in Figure  10 as East North Central (ENC) and Middle Atlantic (MAL) and Ontario (ON)
 in Canada as far as the  Great Lakes are  concerned.  The contributions of toxic air compounds
 studied in this work from sources within ENC and MAL to the total U.S. emissions are varying
 from  about 10% for mercury to 27% for lead.  The emissions in ON contribute about 25% to the
 total  Canadian emissions with bigger contribution for arsenic (more than 40%).  The MAL and
 New  England (NEG) should be considered for the Lake  Champlain.  Emissions in NEG do not
 contribute significantly to the total U.S. emissions of the studied compounds except for benzo-a-
 pyrene (BaP - included in PAHs). However, emissions sources in Quebec generate the largest
 amounts of toxic pollutants in Canada, in addition to Ontario.

       The emissions in SAL, having an  impact on atmospheric deposition  of toxic compounds
onto the Chesapeake Bay waters, contribute between 10% and 15% to the total U.S. emissions of
heavy metals, which is considered to be important.  Even more significant is the contribution of
persistent organic compounds reaching as much as 50% for BaP.

       Although -emitted in regions outside the  Great Lakes, Lake  Champlain,  and  the
Chesapeake  Bay, heavy metals and persistent organic compounds can reach the surface of these
waters.  Once emitted into the atmosphere, these pollutants are subject to long range transport,

                                          49

-------
                                     Umtea S;c;es
                  ~ec orci  Percentcce Distribution  of  5 Compounas
                3	—	
                'C3 'cr.nes Total
                NEC   MAL   ENC  WNC   SAL   ESC   WSC   MTN   PAC   NDF
        25 -!
        20 -i
        15 -i
        10 -I
         5 -i
Mercury
i .668 Tonnes Total
(61* Natural Sources)
                NEC   MAL   ENC   WNC   SAL    ESC   WSC   MTN   PAC   NDF
        7,- ___
      % 2C -•
              21 1 Tifires Total
                NEC   MAL   ENC   WNC   SAL   ESC   WSC   MTN   PAC   NDF
                                                                    45.2
             -rsenic
             :.232 Tonnes  Total
         i " _
         t _
         r
                      "-'.   ENC   WNC   SAL    ESC   WSC   MTN   PAC   NDF
                                        50.5
              c55 Tonnes Total
                 NEC   MAL   ENC   WNC   SAL   ESC  WSC   MTN   PAC   NDF
                                    Region in the U.S.
FigurelOa.  Contribution of atmospheric emissions of selected toxic compounds from
         various regions to the total emissions in the United States.
                                         50

-------
                    Peg:or.3i Pe-centcge Distribution of 5 Compouncs
         25 -
        ,20 -
        !15 -
         10 -
          5 -
     Lead
     1 1,466 Tonnes Total
                 NF   PEI  NS   NB   PQ  ON   MB   SA   AB   BC  YUK  NWT
30
25 -
20 -
15 -
10
 5
              Mercury
              3.530 Tonnes Total
               (99*  Natural Sources)
         30
                 NF   PEl   NS   NB   PQ   ON   MB   SA  AB   BC   YUK  NWT

                                   50.9
         -r J Cadmium
             I 322 Tonnes TC'.QI
         10 -!
          H
          o 4-
                 NF
                 NS   N=   PQ   ON   MB   SA   AB   BC  YUK  NWT
                                   34.0 40.5
             ; Arsenic
            ~ 471 Tonnes
                 NF   PEi   NS   NB   PQ   ON   MB   SA   AB   BC   YUK  NWT
                                  Region in  Canada
FigurelOb.  Contribution of atmospheric emissions  of selected toxic  compounds from
         various regions to the total emissions in Canada.
                                         51

-------
during which transformations and deposition processes are occurring. The extent to which these
processes take place depends upon stack parameters such as temperature and velocity of exhaust
gases,  meteorological conditions, and the physical and chemical forms of pollutants.  The results
from recent studies in Europe provide some basis for estimating what part of the heavy metals and
organic compounds emitted from  major point sources is deposited in the area of their emission
sources (local deposition) and what part is transported and deposited outside the emission region.

       In the case of emissions from point sources with a stack height of  > 150 m (e.g. large
power plants, primary  non-ferrous smelters, cement kilns,  steel and  iron  plants  and waste
incinerators), all of which employ high temperature processes, only 15 to 20% of toxic emissions
were deposited locally.  The rest was transported out of the emission region.  Less information on
this subject is available from research in North America, however development of regional models
in the  area of the Great waters (e.g. Clark,  1992) would require data on local deposition.  It is
certain that we must consider emission sources both within the Great Waters region and outside
the watersheds in order to assess the  origin  of atmospheric toxic compounds deposited on the
water surface in the region.

       The quantity of emissions for  the heavy metals and persistent  organic  compounds of
concern  in the states around  the Great Lakes, Lake Champlain, and the Chesapeake Bay is
difficult to  assess due to diversity of emission numbers reported by various research groups.  The
emissions estimates reviewed for this report differed by one order of magnitude, for most of the
heavy metals considered in this, and are presentlt under revision and modification.

       In  order to revise the  emission data more information is required on emissions within
major source categories.  Most of the work in this respect  has  been done for  mercury.  The
emission quantities of mercury  within major source categories in the Great Lakes region are given
in Figure 11. Emissions of mercury during combustion of coal are clearly the highest, followed by
emissions from waste incineration. Therefore, coal-fired power plants and waste incinerators in
New  York, Ohio, and Pennsylvania dominate emissions in the region. To prove this hypothesis,
emissions of SC»2 and total suspended particles (TSP) have been studied and results are shown in
Figure 12.  It can be noted that the S02 emissions in Ohio alone contribute more than 11% to the
total U.S. emissions of this compound and are followed by emissions in Pennsylvania.  More even
in the contribution of TSP emissions  in the states around the Great Lakes to the  total U.S.
emissions of TSP   It should be noted  that total suspended particulates are the major carrier of
atmospheric  heavy metals.  The  difference between the  distribution pattern of SO2 and TSP
emissions as  shown in Figure 12 is mainly due to larger variety of sources emitting TSP compared
to SO2-  The sources of the latter compound are more homogeneous. Even distribution of TSP
                                           52

-------
                                             Sources  cf  Mercury
                                  •n f.e Great  Lakes Bcsm
FOSS
                        Co-r.D-StiOn
                    IL     IN      Ml     MN    NY     OH    PA     Wl    ONT
           8 -\
              Woste incineration
        £  6 H
        !  * H
           2 H
                         V/'M'M   VMWA
                                                         V//<'/Zi
                          IN     M!     MN     NY     OH     PA    Wl    C\"!
        ,
        v     i
        I  M
        ~  2 -J
                              •L
                                                   '/,y7/'*
                                 Ml
                        MN     NY    OH     PA
                                                                 Wl

            IN     Ml     MN     NY    OH     PA
                                                                 Wl
          20
        I/)
           1CH
               Tc;c  Emission B,, State
               Witn * of Totol Emission

                                      2 3*
                               6.4*   6.6*
                                            5.6X
                                                  20*    16'*
                          IN
                                 Ml     MN     NY     OH     PA    Wl    ONT
Figure 11.  Contribution of atmospheric emissions of mercury  from  various source
         categories to the state emissions in the Great Likes region.
                                          53

-------
      PERCENTAGE CONTRIBUTION  OF THE GREAT  LAKE  STATES TO


                 THE  TOTAL EMISSION  OF  S02 & TSP
   10
                                          11.5
                        30.1
   8 -
   6 -
   4 -
      Total U.S. Emission:  21.0 x 10  tonnes
    0 -r
            !L     IN     Ml    MN
 i

NY
OH    PA     Wl    TOTAL
                                                             13.2
    5 -|
      I TSP
      !                             6
       ictc  'J.S. Emission:  1.83 x 10  tonnes
 /c
                                      I      I
            IL     IN     Ml     MN    NY    OH    PA     Wl    TOTAL
Figure 12.  Contribution of emissions in the Great Lakes states to the total emissions of

        SO2 and total suspended particles (TSP) in the United States.
                                   54

-------
 emissions indicates that sources, such as waste incinerators, ferrous and non-ferrous smelters, and
 cement kilns located all-around the U.S., as well as area sources such as heat production and paint
 application (see even distribution of the Hg emissions from  the latter source in Figure  11) can
 substantially contribute to the atmospheric deposition of the studied pollutants to the Great Lakes,
 Lake Champlain, and the Chesapeake Bay. It  is, however, difficult to substantiate quantitatively
 the above suggestion due to a lack of information on source receptor relationships for the studied
 compounds,  and  particularly for  the Chesapeake  Bay and  Lake Champlain.   Although the
 measurements at receptors are sometime  available,  there is a chronic lack of emission  data to
 study these relationships.

       In the evaluation  of the impact of emissions from local sources on deposition of pollutants
 into the  Great Lakes,  Chesapeake Bay and Lake Champlain, the  locations of the  largest point
 sources should be considered.  If we first look at the location  of the 200 largest power plant
 emitters  of sulfur oxides (presented in Figure  13, EPA, 1991), most of the 50 largest plants are
 found  in  a  belt from  Missouri through  Illinois,  Indiana, Michigan,  Ohio, West Virginia, to
 Pennsylvania. The contribution of the power  plants in these states relative to  the total power
 plant emissions for As, Cd, and Pb  are shown in Figure 14. The plant design, and particularly the
 burner configuration, influences the emissions of trace metals (Pacyna, 1989).  Wet bottom boilers
 generate the highest emissions among the coal-fired utility boilers because of the need to operate
 above the ash-melting temperature.  At a typical peak temperature  of about 1550 °C the volatile
 trace elements in the coal ash evaporate  (Pacyna,  1980).  Later they condense  as submicron
 aerosol particles, or on the surface of ash  particles as the flue gas  cools to 370  - 450 °C in the
 convective heat transfer  sections of the power plant.  The emission rates from other types of
 boilers, such as wall-fired and tangential units seem to be lower due to the lower temperature
 involved.  For mercury,  however, the emissions do  not differ as this element is volatile  at  low
 temperatures and passes  the control equipment of electric-power stations almost entirely in the
 gas form.

       The remainder of the largest power plants in the United States are located mostly in the
 southern states; Texas,  Tennessee. Alabama. Florida, and Georgia, as well as Kentucky. The next
 150 largest power plants  are more evenly located in the eastern and southern parts of the country
with a  higher density of plants in a few regions, including the region surrounding the Chesapeake
Bay area.  A fraction of the arsenic, cadmium, mercury, and lead emissions from the largest  200
power  plants, and  particularly those within the power plant belt as defined above, are probably
deposited to the Great Lakes, the Chesapeake Bay waters, and also to the Lake Champlain basin.
The latter region  is probably also affected by emissions  from combustion  of fossil fuels in the

                                           55

-------
Quebec power plants. The degree of this impact is difficult to quantify on the basis of existing
information  on heavy  metal  emissions.   However,  results given in the N'AKA.   :port by
Venkatram et al. (1990)  indicate that nearly half of the sulfate wet deposition  is uue to  sulfur
emissions along the Ohio River Valley/Midwest region discussed above (See Figure 15).
 Figurel3.  Location of 200 major power plants in the United States.
                                          56

-------
      0)
      c
      c
      o
 100



 80



 60



 40



 20



  0
                   Emission of Toxic Metals  frcm  Power Plants


                                      (1987)
                 As
                   IL
                                                           346
               IN
Ml
       I""1

      MO    OH
                                               PA
                  T

      WV   TOTAL  *TOT
                                              (C
                                              ~j
                                              o


w
0)
c
c
o
E-



80 -
60 -

40 -

20 -

Cd






"ftffff^
ys/7s$A
w 9
u i i i i i i i i i i
                   IL
               IN
Ml
           OH
PA
                 wv   TOTAL  rroT
                                                                            o
      CO
      (D

      C

      C
      O
100



 80 -



 60 -



 40 -



 20 -



  0
                Pb
                   IL
               IN
                                                           220
                                   W////A
Ml
MO    OH
                PA
     WV   TOTAL KTOT
                                             o
Figure  14.   Contribution of the As,  Cd,  and Pb  emissions  from power plants in

         Illinois,Indiana, Michigan, Missouri, Ohio, Pennsylvania, and West Virginia to

         the total emissions from power plants in the United States.
                                       57

-------
       The location  of major primary  and secondary non-ferrous metal smelters  was also
reviewed because  of the large  emissions  of atmospheric  heavy  metals  from those plants,
particularly those employing  pyrometallurgical processes (Metal Bulletin Books - MBB, 1986).
The  major primary smelters are  located outside the  Great Waters  study regions, with the
exception of a copper smelter in the Upper Peninsula of Michigan, and their emissions impact on
the Great waters is discussed later in this report.  The type of technology employed in smelters,
refineries,  and other operations, such  as roasting, has  a major influence on the trace element
emissions  (Pacyna, 1989).   Secondary smelters (International Lead and Zinc Study Group -
ILZSG, 1984; 1985)  are located in many areas in the U.S.  However, their contribution to the
atmospheric deposition of heavy metals is much lower than the impact of primary smelters due to
the difference in the raw materials used.  On the other  hand, secondary smelters are considered
mainly as local emitters as they release exhaust gases through rather small stacks (20 to 50 m) in
comparison with tall  stacks  (over 100 m) in primary smelters.  Several  secondary smelters in
Indiana, Illinois, Pennsylvania, Ohio, Michigan, New York, and Ontario  generate emissions of
heavy metals to the atmosphere which are probably deposited in the same region, the Great Lakes
Basin.  The same would apply to the contamination of the Lake Champlain waters by atmospheric
emissions from  secondary smelters in New York and Quebec, as well as the contamination of the
Chesapeake Bay waters by emissions from smelters located in Maryland and West Virginia.

       Another major point source for many of the compounds discussed in this report is sewage
sludge incinerators.  The density of sewage sludge incinerators is greatest in the eastern United
States  as shown in Figure 16.

       VOCs

       A  great deal  of work has been  put into developing inventories and characterizing the
source profiles of different  VOC emitters.   In the U.S. VOCs are mainly anthropogenic; an
inventory of biogenic non-methane hydrocarbons revealed that quantities released were 20 times
lower than those from human activities (Lamb el a/.,  1987). The 1985 NAPAP survey (Saeger el
ai, 1989) identified over 3000 individual  point and area source types for VOCs and included over
600  individual compounds.   Piccot et  al. (1992)  expanded this to a global inventory of VOCs
from anthropogenic sources by assuming that NAPAP data were representative of similar source
types from around the world. Results  showed that the  U.S.  leads the world in emissions of
paraffins,  aromatics,  formaldehyde and  other a.dehydes, and  marginally reactive compounds;
accounting for  23 - 38% of the global total  in each of these categories (Table  6). Emissions of
most VOCs are highest in the eastern third of the country, including the Great Lakes and mid-
                                           58

-------
                  (a)
                                         01
                                         0.1
Figure 15.   Source apportionment of wet  sulfate  deposition (  per  cent  per state or
         subprovince) at a) Whiteface Mountain, NY,  and b) Mt. Mithchell,  NC after
         Fayetal. (1985).
                                       59

-------
TABLE 6.  Global and United States Emissions of VOCs,

Paraffins3
Olefins
BTX Aromaticsb
Other Aromatics
Formaldehyde
Other Aldehydes
Marginally Reactive0
Total
World
50258
38264
14041
4666
1019
307
910
109465
U.S.A.
11785
2772
2534
1750
304
102
264
19511
X jf World
23.4
7.2
18.0
37.5
29.8
33.2
29.0
17.8
       Source: Piccotetal., 1992.




a)     Includes straight- and branched-chain alkanes, alcohols, esters, and ketones.




b)     Benzene, toluene, xylenes.




c)     Includes most chlorinated and fluorinated compounds.
                                  60

-------
Figure 16.  Location of major sewage sludge incinerators in the United States.
            Source:  EPA Document 450/2-90-009 (1990).
                                      61

-------
Atlantic regions (Figure  17)  The heaviest contributors of olefins are tropical Africa and Central
and South America, the result of continental-scale biomass burning.

       The  Lake  Michigan emissions  survey  by Blakley and  KJevs  (1990) speciated  1985
NEDs/NAPAP total VOCs data  and assigned  releases of individual compounds on a  gridded
network.  Over 100,000 tons of air toxics were emitted from point and mobile sources within a
21-county area surrounding the Lake Michigan Basin.  Approximately 15,000 tons were emitted
from point sources, of which over 90% were located in Illinois and Indiana. Releases from point
sources are  listed by facility and county. Benzene and light chlorinated compounds were major
contributors to total VOCs from point  sources, amounting to  5097 and 5548 tons/y.   Highest
releases of benzene occurred in four counties near the southern end of the lake:  Will and Cook
(Illinois),  Lake and Porter  (Indiana).  The  latter three counties were  also listed as having
substantial coke  oven  emissions.  Following the point source presentation is a list of emissions
from all sources (point + area) by  grid within the 21 counties.

       Prototype VOCs emission inventories were made for ten counties in southeast Virginia
(Emmim et al'., 1989) and later expanded to eleven midwestem and mid-Atlantic states (Wind and
Burke,  1990).  The estimates were made  from NEDS point source and area source  data for
VOCs,  and  in the case of the 11-state study,  also from 1988 mileage data from  the  Federal
Fiighway Administration.  Total VOCs were speciated using apportionment factors provided by
Radian  Corporation.  Atmospheric release of carcinogenic VOCs in the Virginia counties was
estimated to be  1987  tons/y, broken down  into:   gasoline  vapor 38.6%, benzene  21.7%,
formaldehyde 21.0%, chlorinated  solvents 10.0%, and acrylonitrile 8.6%.  Vehicles accounted for
about half the benzene released; the remainder was from  point  sources.   For the eleven states.
VOCs emitted totaled  813,400 tons/y.  Vehicles accounted for  66.3% of the 333,200  tons of
benzene + toluene released.

       PAHs
       For PAHs, the relative  importance of different  sources changes seasonally  and as society
achieves greater control measures (Back et a/.,  1991a,b).  For  example, Harkov and Greenburg
(1985) calculated that 183 kg BaP (98% of total) was released  by motor vehicles  in  New Jersey
during the non-heating season, but 6135 kg BaP (98% of total) reached the atmosphere through
residential wood combustion during the heating season.  A 5-10  fold drop in PAH levels in the
Baltimore Harbor Tunnel  between the mid-1970s and 1985-86 was attributed to installation of
catalytic conveners in the later years (Benner et a/., 1989).  The  drop in aerial BaP concentrations
in the Great Lakes region over  the last two decades has been accompanied by a decline in the BaP
content of surface sediments in the lakes (Eadie,  1984; Eadie et al., 1990).
                                          62

-------
                                                     GRID CF\ I  EMISSION   «606 3
Figure 17. Global distribution of total VOCs, 109 g/y.  Source:  Piccot era/.. 1992.
                                        63

-------
       One  of the  most  detailed  studies on emissions of these pollutants and their  spatial
distribution  around  the Great Lakes has been carried  out  for  the  Ontario  Ministry  of the
Environment (Johnson ei a/., 1990)  Spatial distributions of the annual emissions of PAH, PCDD,
PCDF, and insecticides in Eastern North America within 127 km  x 127 km polar stereographic
grid system are presented in Figures 18-21, respectively.

       A great deal of effort was made to compile or estimate PAH emission factors for a large
number of source types,  including  industrial processes, vehicles, residential combustion (wood,
coal and oil), power plants, incinerators, open prescribed burning, and forest fires.

       One  can note the large difference in emission factors among source types.  For example
those from industrial and utility coal combustion are given in mg/metric ton coal bumed, whereas
ferroalloy and aluminum  production and residential wood stoves are  in g/metric ton metal
produced or fuel burned.  Large  differences in emissions and ratios among PAH compounds can
be seen for vehicles with two types of catalysts.

       The  source-to-source variability within  each of these  categories is  also quite large,
typically an order of magnitude or more.  Even within a particular industry, difference'  :n a
process can greatly change quantities of PAH  released.  For example, Johnson  et al. (19£
total PAHs  released by aluminum reduction facilities from 330 to  2,430 g/ metric ton aluminum
produced, depending  on whether the process is a pre-bake  anode Soderberg or a horizontal
Soderberg.  PAHs released from metallurgical coke production were raised from 450 to 1300
mg/metric ton coal charged when partly contaminated water was used for coke quenching instead
of clean water.

        Annual releases of PAHs  to the atmosphere totaled 272 metric tons in Ontario (ONT) and
9397 metric tons in eastern North America  (ENA).  These  ranged  in molecular weight from
acenaphthylene (152) tr coronene (300)  Of total PAHs,  1.9% wa  estimated to be BaP in ONT
and 3.6% in ENA.  Th. difference was attributed to the  greater variety of sources in ENA and
their relative contribution to the total Within Ontario, the largest emissions took place along the
northern shores of lakes Ontario and Erie.   Over ENA, PAH  releases were largest in a wide
diagonal extending  across the midwestern states from southern Illinois to the mid-Atlantic and
southern New England states.  In this corridor, emission fluxes were 1.6 -  6.4 kg/km2-y,  with an
occasional grid  showing  higher values.  Fluxes throughout the Appalachian region  and the
southeast were typically lower: 0.4 - 3.2 kg/km2-y, with "hot spots" found near southern cities.
                                           64

-------
                                                                      1985 Toxics Emission Inventory ,
                                                                     PAH Combined Fluxes (kg/sqkm/yr)[
                                                                                >64.0           |
                                                                                6.4 to64C
                                                                                3.2 10  6.4
                                                                                1.6 to  3.2
                                                                                0.6 to  1.6
                                                                                0.4 to  0.3       |
                                                                                0.2 tc  0.4       j
                                                                                O.i ;o  0.2
                                                                                O.Ol to  0.1
                                                                                <0.01
Figure  18.   Spatial .distributions of annual emissions of PAH in Eastern North America
          within 127 km x 127 km polar stereographic grid system.
                                             65

-------
                                                                          1985 Toxics Emission inventory
                                                                        PCDD Combined Fluxes (g/sqkm/yr)
                                                                                  >O.S4
                                                                                  0.064 10 0.64
                                                                                  0.0064 to 0.064
                                                                                  0.0032100.0064
                                                                                  0.0016100.0032
                                                                                  0.0008100.0016
                                                                                  0.0004 to 0.0006
                                                                                  O.OOC2 to 0.0004
                                                                                  0.0001 to 0.0002
                                                                               £  
-------
                                                                  1985 Toxics Emission Inventory i
                                                                 PCDF Combined Fluxes (g/sqkm/yr)'
                                                                         0.064 to 0.54
                                                                         O.OOW to 0064
                                                                         0.0032 to 0 0064
                                                                         0 CO16 to 0.0032
                                                                         C.OOCfi to 000^6
                                                                         0.30C4 to 00006
                                                                         0 3002 to C 0004
                                                                         0 OOC- 10 C 0002
                                                                         < C 30C:
Figure 20.  Spatial distributions of annual emissions of PCDF in Eastern North America
          within 127 km x 127 km polar stereographic grid system.
                                             67

-------
                                                           15 Toxics Emission Inventory
                                                         Insecticide Fluxes (kg/sqkm/yr)
                                                                 >256
                                                                 1.2810256
                                                                 0 64 10 1 26
                                                                 03210064
                                                                 016IOC.32
                                                                 0 08 10 C 16
                                                                 004 1C 0.08
                                                                 0.02 :c C 04
                                                                 0 3- :o 0 02
                                                                 < 0 01
Figure 21.   Spatial  distributions of annual emissions of insecticides  in  Eastern North
          America within 127 km x 127 km polar stereographic grid system.
                                            68

-------
       Table 7 gives PAH emissions by source class, and also compares the recent estimates of
 Johnson  el al.  (1992)  with  those from mid-1970s data by Peters (1981) and RamdahJ  el al.
 (1983).  A much more  extensive breakdown of ONT and ENA emissions is given in the original
 Johnson  report pp. 322, 323, and 330-333.  For example, 10 types of production facilities are
 listed under the "Industrial" category.  "Stationary Fuel Combustion" is divided into commercial
 and residential fuel and wood combustion, and electric  power generation.  Contributions of on-
 and off-road gasoline and diesel vehicles, aircraft, railroad, and  marine  vessels are listed under
 "Transportation"  Residential wood combustion dominated emissions,  accounting for 51% of
 total PAHs in ONT and 31%  in ENA.

       TABLE 7. Atmospheric Emissions of  PAHs  by Source Class, Tonnes/y


Industrial
Stationary fuel
combustion
Solid waste
incineration
Transportation
Open Sources
Total
Ontario

82.3
155.0
0.7

24.5
9.2
271.7
Eastern
N. America
2,704
4,545
48.5

1,174
926
9,397
Eastern
U.S.A.
1,831
3,882
<73

1,099
440
7,325
Total
A
640
4,044
56

2,266
4,025
11,031
U.S.A.
B
3,497
1,781
50

2,170
4,100
8.598
       Sources: Johnson et al. (1992, p. 321) for all but A and B. Based on 1985 data.
             A = Peters et al., 1981; B = RamdahJ et a/., 1983, summarized by Baek et al.
             (1991, p.283).  Based on mid-1970s data.

       Emissions of BaP were estimated in 10 counties surrounding Lake Michigan by speciating
1985 NAPAP paniculate matter data Blakley and Klevs (1990). Releases totaled 0.96 ton/y, of
which 0.82 tons/y came from Porter County, Indiana.

       Atmospheric emissions of BaP were estimated at 654 tons/y for the entire U.S. by Wilber
et al. (1992) and Voldner and Smith (1989).  The breakdown by source type was: 85.6% wood
combustion,  6.1% agricultural burning, 31% wildfire,  2.0% vehicles,  1.4% coal combustion,
1.3% coke production, and 0.5% other sources.  By comparison,  the Johnson report estimated
release to the atmosphere of 5.1 metric tons/y BaP in ONT and 340  metric tons/y in ENA.
                                         69

-------
       A review of major power plants and aluminum smelters has been performed TO investigate
the impact of these plants on the contamination of the Great waters by PAHs.  1 hi  ,,\\er plants
within  the  belt, as  described earlier, emit large  amounts of PAHs and their impact  on  the
contamination  of the Great Lakes and Chesapeake Bay waters is  thought to be significant for
heavy metals.

       Major  aluminum smelters  in the U.S. are located in many  locations in the country,
including facilities in Indiana, Pennsylvania, Illinois, Ohio,  and New York (MBB, 1986). Their
impact on the contamination of the Great Lakes waters by PAHs is  probably high. However, the
impact is difficult to quantify due to the  lack of emission data to be used in source-receptor
studies. Smelters in Maryland should also be of concern when discussing the origin of PAHs in
the Chesapeake Bay waters.

       PCDDFs

       Voldner and Smith (1989) identified primary sources of PCDDs and PCDFs in Canada as
being combustion of municipal and industrial waste, accidental fires of treated wood products,
production spills during transportation and aerial spraying of herbicides, wood stoves, and PCB
fires.  Erickson (1989) concluded  that PCB fires produce PCDFs,  but not PCDDs.  Secondary
sources listed by Voldner and Smith were volatilization and  erosion  of dust from landfill sites and
from areas where PCDDFs were present as impurities  in herbicides.  From the recent review by
Johnson et al. (1992) the  most  important contributors of PCDDFs in ENA are incineration  of
industrial and  municipal waste, residential and industrial wood combustion, and electric power
generation.  Secondary copper production  (wire reclaiming) was listed as an important source in
ONT.

       Emission factors for some of the major source types  are given in the Johnson et al.  (1992)
report.  The authors stressed the high degree of variability  and/or lack of emission factors for a
number of processes.

       Differences in PCDDF releases by  incinerators  are large, and depend on many variables.
For example, emission factors for the municipal incinerators surveyed by Johnson ranged from 1 -
36 mg PCDDFs per metric ton of refuse burned. Edgerton et al. (1989b) reported total  PCDD
emissions from municipal incinerators ranging from  1 - 4259 ng per dry standard cubic meter air.
Highest releases were from facilities operating under unsteady burning conditions.

       A  great number of incinerator surveys have been done,  and relationships of PCDDF
output to operating conditions have been evaluated.  Many of these studies  are published in the
                                          70

-------
annual symposia on "Chlorinated Dioxins and Related Compounds", Chemosphere, Vols. 19-23
(1989-91). Some of the important variables are:  a) precursor content of the feed (e.g. presence
of chlorine-containing species), b)  type and extent  of feed processing  (eg.  removal of non-
combustible material, shredding or  pelletizing refuse), c) combustion and  operating  conditions
(temperature, oxygen, residence time), d) type of incinerator and supplementary fuel, and f) type
and efficiency of emission control devices (Johnson el al, 1992).

       PCDDF emissions in ONT and ENA as estimated by Johnson et al. (1992) are listed in
Table 8 for different source categories.  Stationary fuel combustion and solid waste incineration
(municipal and industrial) were by far the dominant emitters. The fact that industrial emissions for
ONT and ENA were nearly  identical  was explained by the fact that emission factors were
available for only two industries -- secondary copper production and Kraft pulping, and that
emissions for  areas  other  than ONT could  be underestimated  for lack of point  sources
identification and base quantity information.

       On the gridded map of ENA,  emission fluxes of PCDDFs were remarkably uniform, falling
between 6.4 - 64 mg/km2-y in most regions. Higher fluxes were generally associated with urban
areas. North of the Great Lakes, fluxes dropped to <0.1 - 0.8 mg/km2-y, with an occasional hot
spot.

       TABLE 8.  Atmospheric Emissions of PCDDs/PCDFs by Source Class, kg/y

Industrial
Stationary fuel
combustion
Solid waste
incineration
Transportation
Open Sources
Total

Ontario
15.6
8.62
10.5

0.1
0.016
34.8
Eastern
N. America
17.6
190
191

8.46
6.57
414
             Sources:  Johnson et al. (1992, p. 321)  Based on 1985 data
                                         71

-------
       Pesticides

       It has been very difficult to obtain reliable figures on the production and use of pesticides
because of proprietary restrictions.  Most OC insecticides have been banned in industrialized
countries  (exceptions in the U.S. and  Canada are lindane,  endo-sulfan,  methoxychlor,  and
dicofol).  Some historical usage data are available, and have been compiled by Voldner and Smith
(1989). These and other figures taken from a U.S. EPA (1987c) report on termiticides are given
in Table 9.

       Many OC pesticides which have been banned in the U.S. and Canada continue to be used
in Mexico,  Central  and South  America,  Africa,  and Asia.   Foreign production and  usage
information  is  extremely important for long-lived OC pesticides.  A portion of these chemicals
entering the Great Waters today could  come from sources outside the U.S., or even North
America.

       Reliable  statistics  are  very difficult to obtain,  however  reports  to  the Food  and
Agricultural Organization  of the United  Nations (FAO, 1986-89) show that Mexico used the
following OCs in recent years (metric tons/y):  DDT = 200-300,  technical HCH  = 180-250,
lindane = 15-45, toxaphene = 600-1200.  Toxaphene is manufactured in Nicaragua, and  local
contamination of human milk and foods has been found (Muller et a/., 1988).  The FAO also
reports heavy usage of OC pesticides in India, in excess of 20,000 metric tons technical HCH and
200-900  metric tons DDT per year.   Information  from the India Dept. of  Chemicals &
Petrochemicals (IDCP)  (Spencer, 1991)  indicates annual production of 25,000 - 28,000 metric
tons of technical HCH  between  1986-90,  in agreement with FAO data.  The IDCP gives far
higher figures for DDT:  6700 - 8600 metric tons.

       In 1989/90 Resources for  the Future, Inc.  (RFF) conducted a mail survey of U.S. Dept. of
Agriculture  Extension Service weed scientists to determine herbicide use for major crops. Replies
were integrated with data from other surveys to form a  national herbicide  use database.  The
report (Gianessi and Puffer, 1990) summarizes applications of 50 chemicals on a state-by-state
basis and  by crop treated for 1987.

       Table 10, taken from the RFF report summaries, shows the top ten herbicides used in the
U.S.   These accounted  for 73% of herbicide use in 1987.  Two-thirds of all herbicide in crop
production were applied to corn and soybeans.  Non-crop usage of these 10 chemicals was minor.
with the exception of 2,4-D. The breakdown of totals herbicides by state and crop is given in

                                           72

-------
        TABLE 9. Production, Sales or Usage of Some Organochlonne Pesticides in the
              U.S.A. and Canada, Metric tons/y.

                                                          Toxaphene. U.S.A.
DDT. L
1950-54
1955-59
1960-64
1965-69
1970-72
'S.A DDT
139,000 1968
160,000 1969
141,000 1970
90,000 1971
11,000
Canada
831
894
287
137

Aldrin + Dieldrin. USA
1950-54
1955-59
1960-64
1965-69
1970-73
1981-85
5,750
17,640
30,100
37,730
20,200
2,270-3,400b






1964-68
1969-73
1974-78
1979-83
1984-85a
80,700
100,700
110,900
24,730
4,000
                                             Aldrin + Dieldrin. Canada
                                                   1968
                                                   1969
                                                   1970
                    86
                    58
                    20
       Chlordane. U S.A.b
         1980    4,300
         1985    1,450
         1986    1,800

       Lindane. U.S.A.C
Heptachlor. U.S.A.b
   1980       910
   1985       340
   1986       340

Lindane. Canada
1964
1966
1971
1974
1975
1976
1977
617
312
269
1,600
2,900
13,100
152,000
                                             1968
                                             1969
                                             1970
                                             1971
                                             1984
             16
             3
             3
             3
             250
Mirex-Dechlorane. U.S.A.
1959-63
1964-68
1969-73
1974-75


187
934
344
61


Mirex-Dechlorane. Canada
1963
1964
1965
1966
1967
1968
9.2
25.3
46.1
37.8
23.2
3.9
Sources:  Voldner and Smith (1988), unless stated otherwise.
a) Production banned in 1982, but remaining stocks used until 1986.
b) U.S. EPA (1987b),  1981-85 aldrin/dieldrin figures are imports.
c) Imports from 1974-77.  Large increase in 1977 probably due to ban of technical HCH in 1976.
                                        73

-------
Table 11.  Com- and wheat-growing states in the Midwest and Plains were the major targets for
herbicides;  Texas and California also  ranked among the top ten users.   Heaviest  ;onnages of
herbicides were applied to corn, followed by soybeans.

       TABLE 10.  Top Ten Herbicides Used in 1987-89.

Active Ingredients
Atrazine
Alachlor
Metolachlo
EPTC
2,4-D
Trifluralin
Cyanazine
Butylate
Pendimethalin
Glyphosate
Total Usage
Metric Tons/Y
29,090
25,000
22,727
17,818
15,000
12,273
10,455
8,636
5,909
5,454
% Share
Corn/Sovbeans
84
91
87
79
12
63
90
99
70
44
       Source: Gianessi and Puffer (1990, p. 7).

       The National Oceanic and Atmospheric Administration (NOAA) maintains a data base for
 35 herbicides, insecticides, and fungicides commonly applied  in coastal  watersheds.  A  report
 which is in the final stages of publication (Pait  el al,  1992)  summarizes pesticide usage in
 estuarine drainage areas for  1987.  This information was obtained from the RFF report and  also
 from state data bases.   The seasonality of pesticide  application  was assessed by surveys of
 representative counties within each coastal state.  The information that follows is taken from the
 pre-publication version of the Pait report.

       Over  13,363 metric tons of the 35 pesticides were applied in estuarine drainage  areas. Of
 this, 69% were herbicides, 24% insecticides, and 7% fungicides. Alachlor and atrazine accounted
 for almost 45% of the total herbicides. By region, total pesticide use was divided as follows:

              North Atlantic                    0.8%
              Middle Atlantic                   27.0%

                                           74

-------
              South Atlantic
              Gulf of Mexico
              Pacific
33.5%
34.3%
4.4%
       The breakdown of total pesticides by drainage area is shown in Figure 22 (Pait et al,
1992). Chesapeake Bay ranked first, followed Albemarle/Pamlico Sound in North Carolina and
Winyah Bay in South Carolina.

       TABLE 11.   States and Crops with Largest Annual Use of Herbicides.

State
Iowa
Illinois
Minnesota
Texas
Indiana
Nebraska
Ohio
Kansas
Missouri
California

Crop
Corn
Soybeans
Pasture
Cotton
Sorghum
Wheat
Rice
Alfalfa
Peanuts
Barley
Total Usage
Metric Tons/Y
20,909
20,455
13,636
12,272
11,364
10,000
8,636
8,182
7,727
6,364
Total Usage
Metric Tons/Y
93,636
37,273
12,727
11,818
10,000
8,182
6,818
3,182
2,273
2,272
% Share
Corn/Soybeans
96
95
86
18
97
84
84
43
71
6












       Source: Gianessi and Puffer (1990, pp. ^8 and 9).
                                         75

-------
       Quantities and relative amounts of various pesticides changed with location.  Figure 23
compares tonnages of 15 pesticides used in four of the above estuarine regions.  '/;:  ppendix of
the Pait report contains quantities of each pesticides applied in the 43 drainage basir.o considered.
Data for Chesapeake Bay are given in Table 12. The seasonality of pesticide use in each region is
also given.  For example, atrazine is applied in April-July in Maryland, April- May in  Delaware.
and March-May in Virginia.

       For a study of pesticide drainage into Lake Erie, Baker and Richards (1990) compiled a
list of the most used herbicides and insecticides in the Lake Erie basin (Table 13).  Herbicides

TABLE 12.  Pesticide Use in the Chesapeake Bay Drainage Basin (Metric Tons/Year).
Herbicides
Alachlor
Atrazine
Metolachlor
Cyanazine
Linuron
Simazine
Butylate
2,4-D
Trifluralin
Vernolate
Actifluorfer
Bensulide



546
483
279
141
114
98.3
60.0
42.6
24.6
2* 1

9.1



Insecticides
Carbofuran
Chlorpyrifbs
Ethoprop
Malathion
Carbaryl
Terbufos
Disolfoton
Phorate
Permethrin
Endosulfan
Methyl
Parathion
Fenvalerate
Diazinon
Parathion
Methamidophos
Fungicides
128 Chlorthalonil 17.3
45.5 Metiram 3.47
38.1
31.2
29.7
19.9
12.0
8.69
7.69
6.97
5.20
2.36
1.75
0.84
0.64
        Source:  Pait ei al., (1992, pp. 96-97).
                                           76

-------
                                          Top Ten Estuarine Drainage Areas
                                                                Inl.ll IVxIlllill- I
                                                                 IIKIII Hi- , ir.irl
Pacific
                                                                                           South Atlantic
                                                    Gulf of Mexico
                                                                                FlncUl
      Figure 22. Agricultural pesticide use in estuarine drainage areas.  Source: Pait et al. (1992).

-------
      Table 13.  Pesticide Use in the Lake Erie Basin for 1986 and Ranking  by Use in
            the State of Ohio for 1982.
Pesticide



Alachlor
Metolachlor
Atrazine
Cyanazine
Metribuzin
Chloramben
Linuron
Terbufos
Trifluralin
Butylate
Dicamba
Pendimethalin
Bentazon
Carbofuran
2,4-D
Chlorpyrifos
EPIC
Phorate
Fonofos
Simazine
Total herbicide use in
herbicides listed above
herbicide use in the Lake
Brand
^^^M^VMBHW


Lasso
Dual
Aatrex
Bladex
Lexone, Sencor
Amiben
Lorox, Linurex
Counter
Treflan
Sutan. Genate plus
Banvel
Prowl
Basagran
Furadan
2,4-D
Dursban
Eradicane, Eptam
Thimet
Dyfonate
Princep
the Lake Erie Basin.
make up 97.3% of
Erie Basin.
Typea



H
H
H
H
H
H
H
I
H
H
H
H
H
I
H
I
H
I
I
H
The 15
the total

Total insecticide use in Lake Erie Basin. The 5 insecticides
Quantity
Used
(metric
tons)
1319
897.2
783.9
273.4
255.5
155.5
133.0
74.67
65.7
58.19
54.39
49.49
49.36
48.23
44.44
36.33
34.67
31.51
25.10
24.82
4315.5


245.6
1986
Rank,
bvUse

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20




1982
Rank
bv Ohio
Useb
1
3
2
4
7
6
8
10
12
5
19
17
15
11
14
MR
13
NR
9
16




listed above make up 87.9% of the total insecticide use in
the Lake Erie Basin.





Source: Baker and Richards, 1990 (p. 246).
                                        78

-------
 accounted for 92% of all pesticide tonnages. Corn and soybeans received 99.7% of herbicide and
 90.1% of insecticide applications. In all, 4561  metric tons of pesticides were used in the basin in
 1986.

       Johnson et al. (1992) estimated usage  of several insecticides, herbicides, fungicides, and
 nematocides in  Ontario, Quebec, and  the  Atlantic Provinces for use in their Toxic Chemicals
 Emissions Inventory.  According to the authors, actual sales figures for individual chemicals were
 very difficult to  obtain because of regulations protecting confidentiality.  In some cases estimates
 were made by speciating reported sales of a pesticide class.  For example, "triazine herbicides"
 were broken down into separate quantities of atrazine, simazine, cyanazine,  and metribuzine
 according to their reported usage on particular crops.  Other inventorying methods  are described
 in their report.

       Pesticide emissions  for  ONT  and  ENA  were  made  by Johnson  et al.  (1992),  by
 considering:  a) volatilization from soils, b) volatilization from vegetation,  c) wind erosion of
 soil, and d) spray drift losses during application.  Pesticide application rates in ONT were  based
 on estimates of usage  in Ontario, Quebec, and the Atlantic provinces.  The tonnages of herbicides
 reported by Gianessi and Puffer (1990) for the U.S.  were used to calculate ENA emissions.

       The Jury model  was used to predict soil volatilization.  Climatological estimates of soil
 water flux  were made by difference between monthly evaporation and  precipitation.  Monthly
 values were combined to provide seasonal averages for the model.  A single set of soil properties
 "typical of agricultural cultivated soils" was used..

       Volatilization from foliage was estimated by assuming that 70% of the pesticide impacting
 vegetation evaporates  in a span of 5-10  days and the other 30% is washed off by rain and returns
 to the soil.  These are based on a study of persistence of pesticides on vegetation by Willis and
 McDowell  (1987).   Because the smallest grid size  considered  was  5 km, spray drift  was
 considered to deposit, then re-evaporate.  Emissions were thus  included in the volatilization
 algorithm.

       Windblown  dust  is most significant during tilling  and in the  early growing season.
Johnson et al. (1992) noted that emissions on dust will only be a problem for persistent pesticides,
most of which have been 'discontinued.  They also felt that there were insufficient data on levels of
these chemicals in soils to make a meaningful estimate of windblown dust releases.

       Atmospheric emissions of pesticides  for ONT and ENA are shown in Table 14. Johnson
et al. (1992) note that approximately 2,700 metric tons of pesticides per year were released in
                                           79

-------
ONT, a quantity far exceeding emissions of PAHs. PCDDFs,  and PCBs.  In ONT the highest
ranking chemicals were the dichloropropenes/propanes -nematocides used on tobnc.o  Atrazine.
alachlor, metolachlor, and 2,4-D were the four most heavily emitted herbicides. The former three
are used on field  crops  and  soybeans, whereas 2,4-D is mainly used on grains.  The picture is
different for  ENA, where the top five herbicides (in  rank) were:  2,4-D, alachlor, atrazine.
metolachlor, and trifluralin.  Because of the usage patterns and properties of the pesticides, this
rank of emissions is slightly different from the list of tonnages used  (Table  10).  The asterisks
shown in Table 14, ENA column, refer to compounds  for which U.S. inventory data were not
available.   For these, only  emissions from the Canadian  portion  of the grid (ONT,  Quebec,
Atlantic Provinces) were used.

       Figures for  linda'ne  and chlordane apply  only  to  estimated  releases from  agricultural
applications and  do  not  reflect  their  use  for  seed treatment (lindane) and  as a termiticide
(chlordane).  In the U.S. chlordane has been banned from agricultural usage for many years, and
its use in termite control was stopped in 1988 (EPA,  1987c, Federal Register, 1988). Before this,
large tonnages of chlordane  and heptachlor were applied as termiticides (Table  9).  Endosulfan
releases are only for ONT and eastern Canada, since Johnson et al  (1992) had no data from the
U.S.  However from the Pait et  al.  (1992) report,  it is clear that very large quantities of this
insecticide  are used in estuarine drainage areas, including Chesapeake Bay (Table  12, Figure 23).
Pait et al.  noted that endosulfan runoff was one of the major causes of pesticide-related fish kills
in estuaries.   Because  of the above difficulties, emissions of OC insecticides  from ENA are
probably grossly underestimated.

       The spatial distribution of herbicide emissions in  ENA shows highest releases in the "corn
and grain  belts"   western  Ohio,  Indiana,  Illinois,  Kansas.  Minnesota,  eastern Dakotas  and
Nebraska (the survey did not include western portions of these states).   Fluxes in these regions
were typically 19-38 kg/km2-y,  with pockets exceeding the upper value.  Fluxes in the south.
southeast, and mid-Atlantic  states were 0.6 - 4.8 kg/km2-y. Occasional areas with fluxes in the
4.8 - 9.6 kg/km2-y were located.  A "hot spot" with fluxes  of 9.6 -  19 kg/km2-y was found near
Chesapeake Bay.  This is in accordance with the high reported usage of herbicides in this drainage
basin (Table 12, Figure 22).  Fluxes in the north Atlantic  states ranged from <0.03 -0.6 kg/km2-y.

       The pattern of insecticide emissions in ENA was different. Highest releases were estimated
for the southeast and mid-Atlantic states, certain areas of the Midwest (Ohio, Michigan, eastern
Illinois, western Pennsylvania, southern Ontario), and western New York. In these regions fluxes
were 0.64  - 2.56 kg/km2-y, with a few areas >2.56 kg/km2-y (central Michigan, southeastern
North Carolina). These correspond largely with fruit and vegetable growing areas. Fluxes
                                           80

-------
 TABLE 14.  Estimated Speciated  Pesticide Emissions in Ontario and Eastern North
       America, Kg/Y.
Estimated Emissions Cke)*

Herbicides
Alachlor
Atrazine
Butylate
Cyanazine
2,4-D
Linuron
Metolachlor
Metrabuzine
MCPA
Trifluralin
Total
Insecticides
Carbaryl
Carbofuran
Chlordane
Diazinon
Endosulfan
Lindane
Methoxychlor
Total
Fungicides
Captan
Nematocides
Chloropicrin
Dichloropropenes/propanes
Methylisothiocyanate
Total
Ontario

195,200
220,700
8,950
60,000
184,800
99,900
156,900
51,500
52,400
9,520
1,039,900

19,650
7,410
354
8,580
7,300
2
1,390
44,680

78,030

58,400
1,296,300
185,800
1,540,500
ENA

7,243,000
4,874,000
13,200**
1,534,000
10,691,000
108,700**
3,646,000
58,900**
568,200**
2,058,000
30,795,000

1,872,000
982,700
853**
219.600
11,300**
2,570**
2,110**
3,091,000

132,300**

59,800**
1,330,100**
185,800**
1,575.700
*     Values after rounding
**    Data not available in the U.S.
Source: Johnson er al. (1992, p. 340).
                                  81

-------
Figure 23.  Application of selected pesticides (1000 pounds/y) b drainage basins of the
         middle Atlantic,  south Atlantic, Gulf of Mexico, and Fudiic rc..io::s.  Source:
         Paite/a/. (1992).
 ''"'ion of Mated PotnJo m tfir Afcttlr Mmnhc. 1JI7
      ofSdstd falidda m On Sou* AOntk. 1ST
                                                 { JOOO-

                                                 § isco-

                                                           ill
                    Illl

      , o/Uta* Potato in tkr Gulf e( Mam. 1317
Afptiatio* efSdtcta] fatidda in Ike Pxifu. 19*7
                                           82

-------
 throughout the "cotton belt", the Appalachians, and southern New England were lower:  0.04 -
 0.64  kg/km2-y.  Values for upper New England  and the Atlantic Provinces were <0.01  - 0-08
 kg/km2-y.  Insecticide figures for ENA do not include fungicides and nematocides, and  as noted
 above, releases of several other chemicals are for ONT and eastern Canada only.  Emissions of
 insecticides are thus underestimated for many chemicals.

       A separate inventory and atmospheric transport/deposition model was constructed for
 toxaphene  by Voldner and Schroeder (1989).  Toxaphene was very heavily used in the U.S.,
 especially in the "cotton belt" during the late 1960s through mid-70s (Table 9). Usage  declined
 thereafter,  and production was banned in  1982.  Remaining stocks were  allowed to be applied
 until  1986.  Voldner and Schroeder speciated total toxaphene use according to state and crop,
 using information from several reports referenced in their paper, surveys of U.S.D. A. Cooperative
 Extension programs at various universities, and an agricultural census of crop types.
       Apportioned toxaphene tonnages in eastern and midwestem North America for 1980 are
 shown in Figure 24.

       PCBs

       Originally PCBs were used for a wide variety of purposes (Voldner and Smith, 1989).
These were: closed system electrical  and heat transfer fluids (approximately 60%), plasticizers
(25%), hydraulic fluids and lubricants (10%), and miscellaneous uses (5%) which included flame
retardants,  additives to paints, inks,  seal-ants, and carbonless copy paper.   After 1971 uses were
restricted almost entirely to closed electrical systems such as transformers and large capacitors.

       According to Voldner and Smith (1989), 640,000 metric tons of PCBs were produced in
the U.S.,  of which  70%  were  sold  to  manufacturers  of transformers  and  capacitors.
Approximately 40,000 metric tons were imported  into Canada and  another 3,000 metric tons
entered  Canada in  manufactured fluorescent light ballasts and high intensity  discharge  lighting
fixtures.  As of 1982, the status of the 640,000  metric tons of PCBs in the U.S. was estimated to
be:

          •   Destroyed                                      3%
          •   Exported                                       11%
          •   Buried in landfills                               21%
          •   Still  in service                                  54%
          •   Circulating in the environment                     11%
                                          83

-------
Source:  Voldner an
                               84

-------
       PCBs are released into the atmosphere from both point and area sources.  The former
 include municipal and  hazardous waste land-fills, which contribute PCBs through volatilization
 and release with landfill gases (Murphy et a/., 1985; Lewis et ai, 1985).  PCBs are also emitted
 by refuse  and sewage sludge  incinerators  (Murphy et al.,  1985).   Occasional leakage of
 transformers and capacitors, and emissions from other in-service equipment (e.g. old fluorescent
 light ballasts) releases PCBs on an area scale.

       Like OC pesticides, PCBs are continually cycling through the environment and air-surface
 exchange is important in controlling their atmospheric  concentrations.   Several studies  have
 shown that PCB  levels in ambient air are temperature-dependent,  much  like the pesticides in
 Figure 16 (Hermanson and Kites,  1989; Hoff et ai,  1991b; Manchester-Neesvig and  Andren,
 1989; Larsson and Okla, 1989). Revolatilization of PCBs from surface water must be considered
 in the PCB balance of the Great Lakes (Achman and Eisenreich,  1992; Baker and Eisenreich,
 1990; HornbuckJe et al., 1992, Strachan and Eisenreich, 1988).

       Because of the sporadic nature of PCB emissions,  it is extremely difficult to predict
 quantities released to the atmosphere. Murphy et al. (1985)  estimated that between 10 and  100
 kg/y were emitted from sanitary landfills in the U.S., based on a survey of six sites in the Great
 Lakes area.  At these landfills 70-500 ng of PCBs were expelled per m3  methane.  Differences
 among hazardous waste disposal facilities are very large.  Lewis et al. (1985) found that PCB
 concentrations  over  "hot  spot"  chemical   waste   landfills  greatly exceeded  background
 concentrations, whereas emissions from a well-controlled facility  were negligible.  Based on other
 reports, Voldner and Smith  (1989) estimated that PCBs released to  the environment through
 accidental spills and municipal waste incineration were 50,000  and 5000 kg/y, respectively.

       Johnson et al. (1992) identified the most important sources of PCBs to the atmosphere as:

              1. Transformer leakage
             2. Electric power generation
             3. Industrial fuel combustion
             4. Landfills
             5. Sewage sludge combustion
             6. Waste oil combustion

      Transformer leakage/spillage estimates were based on  a report for utility industry  closed-
system equipment in the U.S.  These losses as percentages of total PCBs  contained  in  the
equipment were: large transformers (0.027%),  large capacitors (0.42%). A loss rate of 0.24% for

                                          85

-------
the industry was assumed.  Of the spilled PCBs,  0.3% was estimated to be evaporated before
cleanup.  The overall atmospheric emission factor was 7.2 mg for each kg PCBs remained in
transformers and capacitors (0.0024 x 0.003 = 7.2 x 10'6 = 0.00072%).

       The information for (2) and (3) was based  on very limited data on PCB emission factors
from bituminous coal combustion.  These were 0.65 mg/metric ton coal for  utility boilers and 3-
26 mg/metric ton for industrial coal-fired stokers. It is not known why coal combustion should be
a source of PCBs.   Emission factors  (mg/metric ton) for incineration sources were given by
Johnson etal as follows: industrial liquid waste 1.18, commercial waste: 2.5, hospital: 29, waste
oil 390, sewage sludge (0.69-14, 5.4 selected), municipal solid waste 1.3-4.5.

       The PCB release estimates of Johnson et al from open sources (including transformer and
capacitor leakage and landfills),  incineration, and stationary fuel combustion are summarized in
Table 15.  Only 38 kg/y was attributed to solid waste incineration in ENA. This differs markedly
from  a U.S.  municipal waste combustion study (quoted  by Voldner and  Smith,  1989)  which
estimated  about 5000 kg/y from this source.  The atmospheric  release of PCBs to ENA from
transformer and capacitor leakage was 370  kg/y.  Voldner and  Smith quote a  report  which
estimated  that 50,000 kg/y PCBs  were  spilled  in the  Great Lakes region.    Applying the
evaporation factor of Johnson el al. (0.3%) yields 150 kg/y PCBs entering the atmosphere.

       No inventory has considered re-emission of PCBs from soil, plants, and water;  and this
may be a significant  part of the  mass balance.   Also,  as  Voldner and Smith point  out, the
availability and data quality for PCB emissions "leaves a lot to be desired"

       When displayed  on a gridded scale (Johnson et a/., 1992), highest emissions of PCBs
occurred in a broad belt extending from Illinois eastward to the mid-Atlantic and southern New
England states.  In  this region fluxes were generally 0.32 - 3.2 g/km2-y (= ug/m2-y).   This is
remarkably similar to the 1.5-1.8 ng/m2-y precipitation and dry deposition  fluxes out of the
atmosphere estimated for the Great Lakes (Eisenreich and Strachan, 1992).

       Other areas  with atmospheric emissions  of the same  magnitude  were Chicago  -
Milwaukee, southeastern Michigan, a band around  southern and western Lake Ontario, and some
populated areas in the south. A few pockets with  fluxes of 3.2 to >32 g/km2/y were identified.
Elsewhere fluxes were usually in the 0.032 - 0.32 g/km2-y range.  Fluxes north of the Great Lakes
and in eastern Canada were <0.001 -0.032 g/km2-y.
                                          86

-------
 4.2. Emission profiles for major source categories
        Emission source profiles provide  a useful  tool  for  quantifying the  contributions  of
 pollutants from various sources in  a given  region.   This  is particularly true in the cases when
 actual emission measurements are not available.  Chemical mass balance (CMB) analysis is then
 frequently used which requires information about source profiles.  It is desirable to prepare a set
 of emission source profiles for fine and course particles separately.  The CMB receptor modeling
 has been described in the literature (Miller et a/., 1972; Gordon, 1980, 1988; Winchester and
 Nifong,  1971;Watsone/a/., 1984).

       An important step of the CMB  analysis  is construction of emission  source profiles.  A
 detailed  description of emission sources and emission generating processes is needed first.  This
 information has been created for the Great Lakes region as mentioned above.  Then the emission
 source profiles  have been constructed using information on emission factors for  several trace
 elements and persistent organic  compounds, such  as PAH, polychlorinated  di-benzo  dioxins
 (PCDD), polychlorinated di-benzo furans (PCDF), and PCBs.  The profiles were constructed on
 the basis of emission factors  calculated for the global emission  survey  for trace elements by
 Nriagu and  Pacyna (1988) and emission factors from the  Ministry of the Environment  (MOE)
 Toxic Chemical Emission  Inventory for Ontario and Eastern North America by Johnson et al
 (1992).

       The emission source profiles  for the combustion of coal in utility, industrial, and residential
 boilers, the combustion of residual oil in utility and industrial boilers, the production of copper,
 lead, and zinc in primary and secondary plants, the production of iron and steel, the incineration of
 municipal wastes and sewage sludge, the  production of cement as well as phosphate fertilizers, the
 combustion  of wood in stoves, and   fireplaces, and the combustion of gasoline and diesel oil in
 mobile transportation are presented in Figures 25a-c.

       The  emission source profiles as  shown in Figures 25a-c can then be used  to  construct
regional  emission profiles  for the Great  Waters  area by combining information on the  source
profiles with information on contribution of a given emission source category to  a total emission
of a studied  compound  in the region.  Regional emission  profiles are used  to  assess emission
contribution from various regions to  the air concentration or deposition at a given  receptor.
                                           87

-------
TABLE 15.  Atmospheric Emission of PCBs in Ontario and Eastern North America, kg/y.

Ontario
Open sources3 36.9
.., —-—-ration 1.3
Su... ._. -'bustion 12.2
Eastern
North America
453.2
38.2
434.9
      a) Includes transfonv ?- & capacitor leakage (80%) and emissions from lanfills (20%).
      Source: Johnson et il. <:?92, pp. 263, 266, 333).

-------
            Coo: Combustion
            (Power Plant)
               1/5  _  _
               <  O  O
       100

        TO -
            Coal Combustion
            (Industrial Units'
               Combustior
Figure 25a.   Emission source  profiles  for  major  source  categories  estimated  with
         cadmium as a reference element.
                                          89

-------
                                      183
      Primary Cu-Ni Production
c.:::
                a
             as a reference element.
maj°r SOUrce
                                                                            with
                                    90

-------
                                   MoDne  Transportation
          Leoded Gasoline
     10
                   u
       1   Unleaded Gasoline
                                      (SI
I
o
o
o
CL
o
u
CL
     10
     10"
     1C'
                                                <
                                                a.
          a
          a
          o
          a
          o
          u
          a.
                                                          o
                                                          o
                                                          a
                                                          Q.
                   o
                   u
                   CL
Figure  25c.   Emission  source  profiles for major source categories estimated  with

         cadmium as a reference element.
                                        91

-------
4.3.  Emission profiles for diffuse sources of organics
       The  earth's  surface is  a vast  reservoir  for  pesticides and  other  organic  chemicals.
Exchange of gaseous SOCs is a two-way process; and plants, soils, lakes, and oceans act as both
sources and sinks of airborne chemicals.   Examples are volatilization of pesticides from sprayed
foliage vs. uptake of pesticides from the air by pine needles.   Since these exchanges occur,
transport of pesticides is not a simple source-to-sink relationship.  Deposition and re-evaporation
of the chemical may occur  many times during transit,  in what is becoming  known as the
"grasshopper effect"

       Understanding the processes underlying these cycles is critical to developing transport and
deposition models for pesticides and determining sources to-the Great Waters.   Below three
processes are discussed, volatilization from soils, gas exchange with water, and gas exchange with
plants.  The two-way nature is  emphasized for the latter two.  In the case of soil, uptake of
gaseous pesticides in non-target areas is poorly understood. The intent is to present an overview
of the processes and their significance. Mathematical details are given in the references.

   4.3.2  Air -surface exchange processes

   a) Volatilization from Soils

       Although spray drift can have a local impact, post-application  volatilization  is a more
important pathway of pesticide loss from  fields.   Many early studies are referenced in work by
Nash and Hill (1990) and Spencer and  Cliath (1990).  These investigations  have been  done by
following pesticide disappearance from  soils under field conditions, and also through controlled
experiments in agroecosystem chambers.  Major losses can take place within a few days if the
chemical is applied to soil surfaces. Incorporation into the soil layer greatly reduces volatilization;
but even so,  long-lived pesticides continue to evaporate.  Spencer and Cliath (1990) state that:
"Even in areas  where DDT use  has been discontinued, the persistence  of DDT residues is
sufficiently great that they will  continue  to  be redistributed for many years.   In these areas.
volatilization from the soil probably will be the main source of DDT components moving into the
atmosphere ..."

       If not washed off by rain or irrigation, pesticides are quickly evaporated from foliage. In
the absence of precipitation, Willis el al. (1985) found half-lives for toxaphene, methyl parathion.
and fenvalerate on cotton plants of 0.7, 0.1, and 3.3 days.  In another study (Seiber et al., 1979).
                                            92

-------
 toxaphene which was aerially sprayed on a cotton field disappeared rapidly from the foliage, with
 59% lost in 28 days. Loss of toxaphene in the aerated top soil was slower, 51% in 58 days.

       The fate of diazinon sprayed on a dormant peach orchard was followed by accounting for
 the amount of pesticide in air,  soil, and tree parts  (Glotfelty et al.,  1990).  Spray drift and
 volatilization during application  were small,  and were exceeded by  evaporation losses over the
 next two days.  After the first 24 h, diazinon in the soil  dissipated by a first-order process with a
 19-d half life.  The conclusion was that most of the diazinon in the atmosphere of California's
 Central Valley comes from from volatilization.

       A  mass balance for a DCPA-treated  onion field yielded  a half-life of 40.5 days for the
 herbicide.  Flux measurements based on air samples taken above the plot showed that in that time
 29% of the DCPA  was lost  by volatilization (Ross et a/.,  1990).

       In  the tropics OC pesticides are heavily used and  volatilization is the main loss process.  A
 mass balance of HCH in the Vellar River estuary (India) showed that of 42,000 kg HCH applied,
 41,830 kg (99.5%) evaporated and only 170 kg was transported  by the river (Takeoka et al.,
 1991). Yeadon and Perfect (1981) found that DDT applied to soils in Nigeria evaporated with a
 half-life of 9 days.

       Field measurements of pesticide volatilization  are time-consuming and  costly.  A classic
 paper by Parmele et al. (1972) described several micrometeorological techniques and used them
 to determine losses of heptachlor and dieldrin from bare  soil and corn fields.   Common methods
 are briefly described below  For  details, see papers by Parmele et al. (1972) and Majewski et al.
 (1989, 1990). All of these flux methods except the last (TPS) require measurement of the vertical
 gradient in pesticide concentration.

       Typically this is accomplished by collecting air samples at five or more heights above the
 soil up to a few meters.

 •Thornthwaite-Holzman, or Aerodynamic (AD): This is the most frequently used field technique,
 and is based on  accurate measurements of pesticide  and wind speed gradients in the  turbulent
boundary layer.  Requirements are a large, uniformly surfaced area with similar land surrounding
it, and a long fetch  (unobstructed  upwind distance).

 •Lysimeter and Energy Balance (EB):  These methods describe pesticide fluxes (F, mass/area-
time) by the general equation:
                                          93

-------
                F = K2(dC/dz)                                         (3)

where Kz is the vertical diffusivity and dC/dz is the pesticide concentration gradient above the
surface.  Flux estimation methods assume that Kz of pesticide vapor and water vapor or heat are
the same.  Kz is determined by direct measurement of water loss from a weighing Lysimeter, or
(more commonly)  by energy balance.  In the latter technique, vertical gradients of temperature
and humidity above the surface are used with measured radiation and soil heat fluxes to deduce
Kz. Field size requirements are the same as those for the AD method.

The method is simple, but the EB instrumentation is sophisticated.

• Integrated Horizontal Flux (HF):  Fluxes from the field can be calculated from the vertically
integrated pesticide concentration and wind profiles.  This was used by Glotfelty et al. (1990) to
determine transport of volatilized diazinon from a treated orchard:

                F =  (1/R) |0u«dz                                (4)

In Equation (4),  C and u are average pesticide concentrations and wind  speeds  at a particular
height, z.  R is the distance the wind has to travel over the treated surface.  In practice, the
integral is estimated by summing over finite height  intervals.  The method requires a uniform
surface  and  a  uniform  source strength, but  is independent  of assumptions  regarding  the
equivalence of pesticide, water vapor, and heat  Kz values.  Also, fetch requirements are not as
critical.  However, IHF gives no information above the highest and lowest sampling point.

• Eddy Correlation (EC): In the EC method pesticide fluxes are determined by measuring short-
term fluctuations in vertical heat flux.  The technique requires meteorological equipment that can
respond rapidly to small changes in temperature, wind  speed, and wind  direction.  As in the EB
method,  it is assumed  that Kz are the same for heat  and pesticide vapor.   Disadvantages are
complexity  of  instrumentation,  need   for  fast-response  sensors,  and  precise  alignment
requirements.

• Theoretical Profile Shape (TPS): A trajectory simulation model predicts that there is a point
above the center of a circular-shaped source where the gaseous horizontal flux to the atmosphere
can be determined from measurements of wind speed and mean pesticide air concentration at that
point. In contrast to the above methods, TPS requires  measurement of wind speed and pesticide
concentration at only one height. A drawback is  that field plots must be circular.
                                           94

-------
        AD, EHF, EB, EC, and IPS methods were compared in a field test of volatilization from
 fallow  soil  of four pesticides: lindane,  diazinon, nitrapyrin, and  chJorpyrifos (Majewski et a/.,
 1990).   No statistical differences were  found at the 95% confidence level for all methods and
 compounds. However the authors cautioned that it is not clear whether any  of the methods are
 accurate ~  that is, if they describe what is really occurring.  Careful mass balance experiments
 were suggested as a way to calibrate flux estimation methods.

        The expense and time required to do these studies have greatly limited  the number of field
 experiments that have been carried out over the  years.  Moreover, results  are likely to be highly
 dependent on local soil properties, meteorological conditions, and pesticide  application methods.

        Concurrent with the development of methods for flux estimation have been experiments to
 identify the factors that influence pesticide volatilization.  Much work has been done in field plots
 and environmental chambers and to follow pesticide dissipation from  soils.  Recent studies and
 previous work are described by Nash and Hill, 1990; Spencer and Cliath,  1990; Clendening et al.,
 1990; Jenkins et al., 1990; and Woodrow etal., 1990.

        In a series of four papers, Jury  et  al.  (1983, 1984a-c)  described  the development and
 application of a model  for assessing the  behavior  of organic compounds in soil.   Losses  of
 chemicals from soil are the result of leaching, volatilization, and degradation.  The Jury model
 incorporates all three processes. To apply the model, it is necessary to know:

 • Properties of the chemical.  Henry's law constant  (H), soil organic carbon - water partition
 coefficient (Koc),

       Diffusion coefficients in water and air, reaction rate coefficients (parameters of first-order
 loss equations that describe chemical and microbial reactivity).  The latter are difficult to assess.

 • Soil properties,  porosity, bulk density,  volumetric water content.

       In addition, fluxes are affected by the amount of chemical applied, depth of incorporation
 into soil, temperature, and  relative humidity.

       The Jury model views  pesticide evaporation  as being controlled by two resistances, the
 soil layer and the air boundary layer.

       That is, emissions of gaseous pesticides are related to:  a)  the supply of pesticides from
below to the soil surface, and  b) the volatilization rate from the soil surface into the atmosphere.
                                           95

-------
The former process occurs by  vapor-phase diffusion in soil pores and movement of dissolved
chemical upward with water flow.

       For some chemicals  water evaporation from the soil  is critical, because  it causes an
upward flow of water  which trans-ports chemical to  the  surface.   If water  flow carries an
insignificant amount  of chemical relative to vapor diffusion, water evaporation has less effect.
Jury el al.  (1984a) and Spencer and Cliath (1990) show how, based on their physicochemical
properties,  substances can be classified by the effect of water evaporation on their volatility from
soils (Figure 26):

• Class I.   High Henry's law  constants (OCs, EPTC,  triallate, trifluralin).   These chemicals
volatilize from the surface faster than mass transport can replenish them from below.  Control
thus resides in the soil.   Volatilization rates are increased by evaporation of water for some of
these compounds,  but not others.  The influence of water flux is less than for Class II and III
compounds. Volatilization rates decrease with time.

• Class III.  Low Henry's law constants (atrazine, prometryn,  napropamide, 2,4-D).   These move
to the surface in water flow faster than they can be lost  to the atmosphere through the boundary
layer.  Without water evaporation their volatilization rates are very low.  As water evaporates,
pesticide concentrations build up at the soil surface and volatilization rates increase with time.

• Class II:  Intermediate in behavior (methyl parathion, parathion, ethoprophos)

       From Figure 26 and the further description of chemical classifications (Jury et al., 1984a),
it is fair to  say that for most pesticides water evaporation will play an important role in loss from
soils.

       In the final paper of the series (Jury et al., 1984c) the authors compare  results of the
behavior assessment  model to volatilization data  from  laboratory chamber experiments.   Very
good agreement was  obtained, and the effects  of water evaporation for chemicals in the different
volatility classes were correctly predicted.  Despite this, Jury et al.  (1984b) stressed that the
model is to screen chemicals for their behavior in soils, not to simulate field results.  In field mass
balance studies (Clendening et al., 1990) volatilization behavior of five Class I  and III compounds
was in accordance with the Jury model, however deep migration by leaching  was difficult to
predict.

       Several  field  models have been  developed to simulate pesticide volatilization.  A six-
compartment model was formulated by Nash and Hill (1990)  to take into account changing

                                           96

-------
     10'*
  o
  •••*
  o>
  fsi
  O
              CATEGORY I
            	 E :2.5mrr\/
-------
mechanisms of loss over time.  When first applied to a soil surface,  pesticides volatilize quickly.
Later the loss rate slows down, in pan due to adsorption of chemical to the soils.

       Woodrow et al. (1990) investigated volatilization of three herbicides and methyl parathion
from flooded rice fields in California's Sacramento Valley.  Good agreement was found between
measured loss rates and those predicted by EPA's EXAMS model. A Gaussian plume dispersion
model (ALOHA) was used to estimate transport of the chemicals out of target areas.  Measurable
air  concentrations  of  pesticides in  residential and  business  areas  were due  to drift during
application, post-application  volatilization,  and photooxidation of  pesticides  during transport
(conversion of methyl parathion to its oxon).

       A volatility model that takes into  account changing  meteorological  conditions,  soil
temperature and moisture, and water evaporation rates over the period of interest was developed
by  Scholtz and  Voldner  (1992), who then applied it  to  the  evaporation of  three pesticides
(lindane, chlordane, and 2,4-D) from sandy loam soil.  Modeled daily emission rates and the
cumulative fraction of pesticide lost over  a 6.5-month period are shown in Figure 27.  Although
the Henry's law  constant  of the pesticides change with  temperature, including this temperature
dependence had little  effect  on the results.   Differences in water  evaporation rates had the
strongest influence on daily fluxes.

    b)._ Air-Water Gas Exchange

   In the frequently used "two-film" model gases are exchanged by diffusion through thin air and
water films on either side of the air-water interface.  The rate at which molecules diffuse through
these films  is slow compared to convective mixing in the bulk air and water, and thus the interface
acts as a resistance to transfer. The equations for describing flux (F = mass/area-time) of material
have the form:

                    F = KAC                                              (5)

where  AC  is the concentration  gradient (that is, difference in concentration from the top to the
bottom of the air or water film). The overall mass transfer coefficient (K) takes  into account the
resistance of both films.  Algebraic expressions for K include transfer constants for the individual
air and water phases and the Henry's law  constant of the chemical. Often the resistance of either
the air or  water will dominate, and  the  chemical is said to be "gas phase" or "liquid-phase"
controlled. Which situation applies is largely related to the Henry's law constant.
                                           98

-------
                                       DAILY EMISSION RATE. SANDY-LOAM
                                       SPAAY APPUCATON: 1 i-D; Henry's U* coefficient s function of
                                             ), independent of icmpeniurc (dotied).
                                  CUMULATIVE EMISSION: SANDY-LOAM
                                  SPRAYED APPUCA.TlON-.1kg/ha.06h April 15
                                                  2.4-D
                                                 MOTTH
Figure 27.  Modeled daily emission rates and cumulative fractions emitted for pesticides
          sprayed on sandy loam soil.  Top:  lindane;  Bottom: lindane, chlordane, and 2,
          4-D. Source: Scholtzand Voldner (1992).
                                             99

-------
       The detailed equations for two-film gas exchange of pesticides, PCBs, and PAHs have
been presented by many authors, including Baker and Eisenreich (1990), Corlidi?! and Bidleman
(1991), WMO (1989), Hinckley el al. (1991), and McConnell et al. (1992).

       As a way of quantitatively describing water bodies as sources or sinks, the model many
shortcomings. Major unknowns are:
    • Mass transfer coefficients as a function of wind speed.
    • The role of bubbles and breaking waves in the exchange.
    • Henry's law constants of SOCs as functions of temperature and salinity.
    • Physical states of SOCs in air and water: To employ the model, it is necessary to know the
       fraction of gaseous and dissolved compound.  Further method development is needed to
       speciate  SOCs  in air  and  water  into gaseous, paniculate,  colloidal, and dissolved
       components.

       Calculations  based on  the two-film model indicate that gas exchange of SOCs is an
extremely important factor in the Great Lakes budget. Murphy (1983) proposed evasion of PCBs
from the  Great Lakes, and a subsequent mass balance study concluded  that the net flux of PCBs
and DDT was out of the lake on an annual basis (Strachan and Eisenreich, 1988).  At that time
these conclusions were  based on limited air and water data. Mass balance  models of Siskewit
Lake also show the importance of volatilization (McVeety and Kites, 1988;  Swackhamer et al.,
1988; Swackhamer and Kites, 1988).

       Recent  paired air and  water measurements  have provided much more insight to  this
process.  Data from Lake Superior  indicated volatilization of PCBs during late summer (Baker
and Eisenreich,  1990).  Intensive studies in Green Bay showed that PCBs volatilized from the
surface water at rates ranging from 13 - 1300 ng/m2-d.  Moreover, the profile of individual PCB
congeners  was  different in the air  over  Green  Bay  than over land (Achman et a/., 1992;
Hornbucklee/a/., 1992).

       Concurrent sampling of air  and  water in Green Bay  and the lower four  Great Lakes
allowed McConnell et al. (1992) to determine the direction of HCH exchange. Transfer of HCHs
to Green Bay was air-to-water  (deposition) in early June.  In August when surface waters were
warmer,  the flux direction was reversed  (volatilization)  in  Michigan,  Huron, and Erie,  but
remained depositional in Ontario. Based on the annual cycle of HCHs in the atmosphere (Hoff et
al., 199la), the authors predicted that transfer of gaseous HCHs was into the  lakes for most of
the year, with short periods of volatilization in late summer.   The loading of HCHs to lakes
Michigan,  Huron,  Erie, and Ontario by gas deposition was  estimated to  be 370 kg/y.  By

                                         100

-------
 comparison,  970 kg/y  entered  the  four  lakes by  precipitation  and dry  panicle  deposition
 (Eisenreich and Strachan, 1992).

       Gas  exchange contributes significant  quantities of OCs to  the oceans.  The  GESAMP
 report (WMO,  1989) concluded that  air-to-water gas transfer of OCs accounted for 24-72% of
 total  atmospheric deposition.  Gotham and Bidleman (1991)  estimated that about  110 metric
 tons/y of HCHs entered the Arctic,  about two-thirds of which  came from  gas uptake by  the
 surface ocean.  Gas exchange formed 20-60% of the total atmospheric budget for  other OC
 pesticides.

    c) Air - Plant Exchange

       Plants accumulate OCs from the atmosphere, and many investigators have suggested their
 use to survey atmospheric contamination in non-agricultural and remote regions.  The case of pine
 needles as an indicator of DDT transport from eastern Europe was mentioned  earlier (Section  V-
 C-lc).  Examination of different plant species from around the world revealed especially high
 levels of DDT,  HCHs, and PCBs in China.  DDT was also high in plants from several  African
 countries, and in Russia near Moscow. Lowest concentrations of most pesticides and PCBs were
 found in lichens from the Antarctic Peninsula (Bacci et al.,  1986,  1988;  Gaggi et  ai, 1985;
 Villeneuve el ai, 1988).

       Accumulation of pesticides and PCBs by plants is  important  for several reasons.  The
 potential for using plants as a monitoring tool was mentioned above. Plants are also at the  base of
 the terrestrial food chain. Transfer of OCs from lichen to reindeer to man was investigated in
 northern  Sweden (Villeneuve et a/., 1985). The percentage of transfer from lichen to reindeer was
 positively correlated  with the octanol-water partition coefficient (Kow) of the compound.

       Finally, the earth's plant biomass is a large reservoir for persistent organic  compounds.
 Accumulation from  the  atmosphere into plants  is related  to  the  proximity  of sources (i.e.,
 concentration of chemical in air) and temperature. The strong role of temperature was shown by
 Calamari el al. (1991), who surveyed plants from around the world for OC pesticides and PCBs.
HCB  concentrations  in plant  tissue were highest in the  Arctic  and Antarctic and  lowest  in
equatorial regions.    For this compound  the "source factor"  can  be eliminated, because
concentrations of HCB in the troposphere are uniform,  varying by no more than a factor of 2-3.
Levels of HCHs and DDT were far higher in  temperate and tropical zones than at the poles,
showing the strong dependence on regional use of these pesticides.
                                         101

-------
       Recent  work  has focused on the rate and  mechanism  of atmospheric  uptake  and
depuration of OCs by plants.  Like  fish in  water, plants  take up chemical  from a polluted
atmosphere  and release it when placed in clean air (Bacci  et al.,  1990a,b).   The steady-state
bioconcentration factor (BCF = concentration of chemical in the plant divided  by that in air) is
related to two common physicochemical properties, Kow and H:

                     BCF = constant  • K^RT/H                             (6)

Reviews of accumulation processes and  comparison of different models  have been written by
Paterson et al. (1990) and Schonherr and Reiderer (1989).

       The idea that large areas of "green space" on the planet can take up and release pesticides
to the atmosphere has important consequences  for long-range transport and seasonal cycles of
pesticides in the atmosphere.

4.4.  Evaluation of emission inventories.  Comparison with European studies
       In general, emission inventories have only recently been compiled for toxic heavy metals
and persistent organic compounds, with the exception of lead.   Emission inventories were first
used to evaluate the environmental impacts of the emissions from single point sources, thus were
of local importance. In the late 1970's, it was recognized that heavy metals  can be transported for
distances of up to a few thousand kilometers.  This suggested the  need for regional, and even
global, emission inventories to be formulated for these persistent pollutants. When  the parameters
affecting the emission quantities  were defined  for this study,  it became clear that a regional
emission inventory for heavy metals should be prepared using so-called bottom-up approach. In
other words, major emission sources should be identified and emission quantities assessed in sub-
regions,  e.g. administrative units such as county or city, and  then  added in order to  obtain
emissions within a region. In this way a detailed assessment of emission  sources can be collected
at a county or city level.

       Emission measurements are often carried out at specific sources and their results are then
reported at a sub-region level. Indeed, measured data have been used to assess emissions of toxic
pollutants in various states in the study region. This  applies particularly to emissions from single
point  sources,  and  mainly waste incinerators.   The emission data  obtained  through  such
approaches are usually accurate providing that the measurements are representative  with respect
to sample collection and analytical methods used. Thus, the above approach is recommended for
collection of emission data from major point sources, including large power plants (e.g. over 1000
                                           102

-------
 MW electricity), heat boilers (e.g. 200 GJ/h capacity), primary and secondary non-ferrous metal
 smelters, cement kilns and already mentioned waste incinerators.  In Europe, there are already
 regulations requiring emission data to be measured for large power plants within the parties of the
 Commission of the European Communities.  Standardization  of measurement  methods and
 harmonization  of reporting procedures are important factors when the bottom-up approach is
 used to obtained emission  data.   These  factors  need to be employed in order to assure  the
 comparability of the emission data to  be  used to construct  the  emission  inventory for a given
 region.

        There are currently very few measurements, on the emissions of hazardous  air pollutants
 from sources within the Great Waters regions. One reason for this could be that on the national
 scale in the U.S. there is no formal obligation to undertake an emission survey for toxics, the TRJ
 program being voluntary.  Only some states have taken the initiative to prepare emission surveys,
 mostly using emission factors from the  literature and  statistical information on the production of
 industrial goods and the consumption of raw materials. The measured data reported  for the study
 region  are more accurate  than the estimated values.   The measured source profiles provide
 information which may be used when estimating emissions for regions with no measurements.

       The second approach used to prepare an emission inventory is called  a top-down approach
 and is used when the inventory is based on emission estimates.  The emission inventories prepared
 by EPA, IJC, and the Canadian authorities, and used in  this  report, are based  on the top-down
 approach.  Major emission source categories for a larger region, e.g. a state are  defined in the first
 step,  emission factors are selected, and  emission quantities calculated.  Further division of these
 estimates in order to obtain spatial distribution of emissions is carried out using either surrogate
 parameters, such as population density maps for distribution of area source emissions or a list of
 point sources with information on  geographical location and emission data for other compounds,
 such  as sulfur and nitrogen  oxides, volatile organic compounds,  and total suspended panicles.
 The NAPAP database provides aJarge body of information in this respect for the sources in U.S.
 and Canada.  Similar  approach  is used in  Europe to assess  sources and fluxes of atmospheric
 heavy metals and persistent organic pollutants within various activities of the UN ECE Task
Force on Heavy Metals Emissions and  Task Force on Persistent  Organic Pollutants, as well  as
within PARCOM and RELCOM.

       Several parameters affect the accuracy of emission data prepared using  emission factors.
The emission factors need to be representative for a given source category,  sub-category, and
industrial activity.  Therefore, the quality  of emission data  are dependent upon selection and
utilization of emission factors which are  transparent with respect to the conditions, both technical
                                           103

-------
and meteorological, at the time the factors were determined.  Various sets of emission factors
were used to calculate emission quantities for the study area by EPA and the Canadi - i authorities.
Emission factors used in the preliminary version of the EPA emission inventory for neavy metals
in the United States (Benjey and Coventry,  1992) seem to overestimate the emissions of mercury,
arsenic, cadmium, and particularly lead. They are higher than the emission factors used in Canada
(Voldner and  Smith,  1989; Johnson et a/., 1990) and Europe (Axenfeld et al.,  1992; Pacyna,
1986).  The preliminary EPA emission factors are now being revised. They need to be validated
and cross-checked.  For example, methods  proposed recently within the UN ECE Task Force on
Emission  Inventories could be used for verification of emission data for heavy metals  and
persistent organic compounds (Mobley, 1992).

       In summary, the  process of collecting  information on emissions for heavy metals  and
persistent organic pollutants from important sources  in  the Great  Waters  region  has begun.
However, it will take some time before an  accurate emission inventory becomes available.  This
conclusion is made  on the basis of the existing  data for the region and research plans, including
those  within  the  international  organizations  dealing  with the  emission  inventorying  for
atmospheric heavy  metals  and persistent  organic  compounds.   Therefore,  the analyses  and
particularly conclusions based on presently  available emission data for the study region should be
considered with some caution. This situation is similar in Europe in terms of atmospheric input of
heavy metals and persistent organic compounds  to the North Sea and the Baltic Sea. The overall
conclusion from the research in Europe is that the current state of knowledge on emission fluxes
of the above  pollutants needs further improvement in order to satisfy  requirements  posed by
policy makers and the scientific community.

4.5.  Application of source-receptor techniques to study the origin of pollution
       Although the source apportionment  techniques are most widely used to assess the sources
of air contamination at distant locations, chemical mass balance model can be applied  to identify
contributions from  local sources, such as in an  urban area (Scheffand Wadden, 1991) .  There
have been attempts to construct both static and dynamic  mass balances for various pollutants in
the study area in order to quantitatively account for the pollutant loadings and the flows through
the environmental system. The mass balancing method has been applied in the Great Lakes basin
to manage phosphorus load reductions to Lake Erie, and to elucidate behavior in smaller regions,
such as Saginaw Bay, as reported by Mackay (1992) for the Virtual Elimination Task Force.

        The current level of knowledge of loadings, water and sediment concentrations, and biota
concentrations is not satisfactory and several assumptions need to be made when preparing the

                                          104

-------
 chemical mass balance for the study area. We are also limited by the available knowledge on the
 behavior of pollutants in the environment. The physical and chemical processes that are important
 depends upon the pollutant being investigated, and varies drastically from organics to metals, and
 even more for mercury.

 5.  IDENTIFICATION OF DISTANT SOURCES

       Atmospheric deposition has been identified as one of the major, if not dominant, pathways
 for toxic heavy metals and persistent organic pollutants measured in the Great waters. However,
 these is no definitive information available to the authors of this report which indicates the relative
 contributions from local and distant sources.  There is presently a limited amount of information
 on  the transport and deposition of HAPs to the Great Lakes-(Rice et a/., 1986; Voldner and
 Schroeder, 1990) which originated outside the Great Lakes Basin.  However, there has recently
 been a number of studies which investigated  the long-range  transport  and  deposition of
 organochlorines (OCs)  to remote areas such as the Arctic. Regional transport of metals, including
 mercury, and other toxic compounds is also the  subject  of intensive studies at EPA at  present
 (e.g. Clark, 1992). While this compelling evidence gives indirect proof that this phenomenon is
 important to the Great  Waters, there is still a lack of quality data on the hazardous pollutants to
 support a definitive conclusion.  However, the results from two research programs: National Acid
 Precipitation  Assessment Program (NAPAP) and  the  Baltic Sea Environmental  Program in
 Europe can provide clues on the potential contribution  of distant  source emissions  to the
 atmospheric deposition  in the Great waters region.

       NAPAP focused mainly on  the deposition of acidic compounds and their precursors. The
 atmospheric transport and deposition of sulfur is discussed here  as  an example for the transport
 and source-receptor relationships for pollutants that undergo chemical transformation and that are
 on particles including heavy metals and persistent organic pollutants. The transport and chemical
 transformations for organics and  metals, such  as mercury, may be  quite  different  and  direct
 analogies should not be drawn at this time. NAPAP (1990) concluded  that for receptors in the
 eastern United States,  more than 70% of the total deposition of sulfur originates  from sources
 within 500 km of the receptors and dry deposition of sulfur species contributes more than half to
the  total deposition at the average  receptor.  In Canada, about two-thirds of the total deposition
 at average receptors  originate  from source areas at distances greater than 500  km from the
receptors.   Wet  deposition constitutes about two-thirds of the total  deposition (except receptors
close to large point sources of sulfur emissions).  As the main sources of heavy metals and
 persistent organic compounds are south of the Great Lakes Basin and these pollutants are more

                                          105

-------
resistant to wet deposition, one can hypothesize that emissions from regions outside the Great
Waters are very important when performing source apportionment studies for th. Great Waters.
While the evidence  to date strongly suggests that  distant  sources are  contributing  to the
contamination  of the  Great Lakes and  other bodies,  further  research is needed to prove this
hypothesis and to quantify the relative loadings for specific pollutants.

      There has been a great deal of work looking at similar problems facing large bodies of
water in Europe.  The Baltic Sea and the Great Waters can  be compared taking into account
similarities in the level of water contamination by heavy metals and persistent organic pollutants,
and emission source categories contributing to  this  contamination,  although the meteorological
conditions, and particularly precipitation patterns are quite different.  The Baltic Sea Program
concluded that major source regions contributing to  the atmospheric deposition of heavy metals
and persistent  organic compounds except for pesticides are located several hundred kilometers
from the Baltic Sea shore (Pacyna, 1992). Most  of the major contributing point sources identified
within the study were at least 500 km far from the Baltic shore.  Even so,  emissions from these
sources and  source regions were found to contribute the majority of the  pollution load from the
atmosphere.  The above results confirm that the regions with high emissions of toxic metals and
organic compounds  can contribute to the contamination of the  environment at remote receptors.

      When diagnosing the major  sources  of contaminants in the Great  Lakes Basin, for
example, the large emissions of  arsenic, cadmium,  lead, mercury,  and  other heavy metals in
Missouri (Benjey and Coventry,  1992)  can be transported and  deposited  in the basin.   Data
supporting the transport of toxic compounds from  the St. Louis area to Lake Michigan  were
obtained  during the Lake Michigan Urban Air  Toxics  Study  performed during the summer of
1991.  The largest air pollution episode observed during the  study was  associated  with mixed-
layer transport from the general area around St. Louis, MO to the southern Lake Michigan area
(Keeler el ai,  1992). Measurements of all of the toxic air pollutants discussed in this report  were
made at several sites around Lake Michigan.  Preliminary results suggest that the iron-steel and
other metallurgical activities in the St. Louis/Granite City area  were probably the largest sources
of heavy metals measured in Chicago and South Haven, MI.  Additional research is needed to
further support this hypothesis both with  further  measurement and through the application of long
range transport models.  The major difficulty in applying dispersion models to this problem is
obtaining a set of reliable emission data within the specified grid system.
                                           106

-------
 5.1. Emissions from North America outside the Great Waters Regions.
        Contribution of heavy metals and persistent organic pollutants from North American
 sources located outside the Great waters study regions can be assessed  with the help of source-
 receptor techniques.   However,  the results would indicate contribution of either emissions
 generated within major source categories or arriving from major source regions if the source -
 receptor techniques are used together with information on meteorological conditions. Emission
 data for major point and area sources within the U.S. and Canada are needed in order to verify the
 results  of source-receptor techniques  and can  be used for developing strategies to  reduce the
 pollution load to the Great waters. The preliminary results from the EPA and the IJC data (e.g.
 Voldner and Smith, 1989) indicate that four states lead the emissions of heavy metals generated
 from the production of energy and industrial goods.   The  states include: Arizona,  Louisiana,
 Missouri,  and Texas.   Primary copper smelters in  Arizona result in high emissions  of arsenic,
 mercury, and cadmium.  Secondary smelters, refineries, and lead alky] plants emit large quantities
 of lead in Texas and Louisiana.  Combustion of fossil fuels,  and primarily coal in Missouri,
 produce large amounts of all studied heavy metals. Industrial activities in Nevada, Montana, Utah
 and New Mexico also generate substantial amounts of atmospheric heavy metals. Recent EPA
 study on toxics in the community (1990) listed a ranking of states emitting toxic pollutants to the
 air in 1988.   The list was topped with Texas followed by Ohio, Tennessee, Louisiana, Virginia,
 and Utah. The list of the top ten was completed by the states in the study area: Indiana, Illinois,
 Michigan, and New York.  Quebec and Manitoba in addition  to Ontario generate  the  largest
 amounts of studied heavy metals in Canada (e.g.  Voldner and Smith, 1989).

       Primary emission sources of PAHs to the atmosphere include combustion of fuels for heat
 and power generation, transportation, solid waste incineration, industrial  processes such as coal
 and coke processing and petroleum refining. Therefore, their sources are similar to the  sources of
 mercury (except transportation) and the states  of Missouri, Texas, and  Louisiana generate  the
 largest PAH emissions.

       The  ubiquitous past  use  of PCBs as  well  as  current generation led  to widespread
geographical distribution of these pollutants in U.S. and Canada. The leaks and spills from current
use all-over U.S. and Canada appear to be the largest source of PCBs to the environment.

       The  major primary sources  of  PCDDs  and PCDFs to  the  atmosphere include  the
combustion of munitipal and industrial waste in many locations  in U.S. and Canada.  Therefore,
the emissions of these pollutants are quite ubiquitous. In addition, it has been speculated that  the
dominant sources of PCDDs to the total environment could be manufacturing of chlorophenols

                                          107

-------
and their derivatives and the disposal of chemicals containing these wastes (e.g. Voldner and
Smith, 1989)  Although the locations of the companies which have been major  p   ucers and
formulators of these products is known, no data is available to speculate on the likely emission
amounts.

       Information on the use of lindane in U.S. and Canada is very limited.  Its application in
agriculture is banned but is probably used in livestock treatment. The use pattern of lindane in
U.S. and Canada is largely unknown.

       It is postulated that large sources of atmospheric emissions of heavy metals and persistent
organic pollutants are located in Mexico, and that the effects of these sources are being felt in
Great Waters areas (Eisenreich and Strachan, 1988).  Two attempts have  been made within the
Global Emission Inventory Activities (GEIA) program of the International Geosphere Biosphere
Program (IGBP) to assess the sulfur dioxide and nitrogen oxides emissions in Mexico as a part of
the global emission inventory for these pollutants. The assessment has been made in co-operation
with the Mexican experts and it was concluded that the sulfur and nitrogen emissions divided  by
the number of inhabitants was one of the largest in the world. As the combustion of fossil fuels is
by far the most important source of sulfur emissions, it can be expected that substantial amounts
of mercury are also emitted into the atmosphere from the Mexican power plants, particularly due
to the lack of desulfurization installations on these Mexican plants.

       Large quantities of nitrogen oxides are emitted in Mexico from  the combustion  of
gasoline. The Mexican gasoline contains lead in amounts of 0.4 g/liter and results in emissions to
the atmosphere which are also relatively high. Recently, a preliminary study was been carried out
at the Norwegian Institute for Air Research  to assess global emissions of atmospheric lead.  The
study is a continuation of global emission inventory development for which preliminary results
have been published by Nriagu and Pacyna (1988).  It has been estimated that  at least 5000 t of
lead is emitted from gasoline combustion in Mexico each year, compared to about 15000 t in U.S.
in 1985. Current emissions of lead in U.S. are significantly lower than those in 1985 and probably
do not exceed 7000 t per year.

       Mexico is also an important producer of non-ferrous metals and particularly copper, lead.
zinc, cadmium, arsenic, silver, gold, antimony, and bismuth. There are several companies in the
country producing the above metals with the two major ones being Industrial Minera Mexico  SA
de CV, and Industrias Penoles SA de CV  Substantial amounts  of atmospheric emissions of all
heavy  metals of concern for this report are expected from the concentrating  facilities in Santa
Eulalia (Chih.),  Taxco (Gro.),  Rosario (Sin.),  Velardena (Dgo.),  Santa  Barbara (Chih.),  and

                                          108

-------
 smelters and refineries in  Nueva Rosita (Coah.),  San Luis Potosi (SLP), Chihuahua (Chih.),
 Monterey (NL), and Torreon (Coah).  In addition, there are several secondary non-ferrous metal
 plants in the country generating atmospheric emissions of. heavy metals.  Although quantitative
 information was not available to the authors of this report, knowledge  on industrial technologies
 employed in the above facilities and information on emission control  equipment, although very
 limited, indicates that emissions from Mexican non-ferrous metal industry are higher than those in
 U.S..

       There are several aluminum plants in  Mexico operated by major companies  such  as
 Almexa Aluminio SA de CV, Aluminio SA de CV, Laminadora de Aluminio SA de CV - Lasa,
 and Reynolds Aluminio SA.  It is expected that substantial emissions of PAHs are generated in
 these plants, particularly in smelters employing the Soderberg process to produce aluminum, such
 as the Aluminio SA de CV smelter in Veracruz.  An emission factor ranging from 500 to 5000 g
 PAH / ton of Al produced was estimated for this process (e.g. Axenfeld el a/., 1992).

       The emissions from the above described sources in North  America but  outside the study
 region can be transported and deposited to the Great Waters areas.  One of the methods utilized
 to study the long range transport of heavy metals and organic pollutants are regional transport
 models.  There are two major groups of input data needed to employ these  models:  gridded
 emissions data and meteorological input data including wind and precipitation observations from
 surface and upper air meteorological sites. The emissions data are always more  difficult to obtain
 and as such they should be given a priority in future research plans. This task requires not only a
 financial support but also international co-operation.

 5.2.  Emissions from sources outside of North America.
       Measurements in the Great waters region show presence of pollutants which are banned in
 U.S. and Canada entirely or partially.  One of this pollutants is y-hexachlorocyclohexane,  an
 insecticide called lindane. An interesting question is what may be the origin of lindane measured
 in regions where its use is banned.

       There is rather limited information on the use  of lindane  in the literature although this
pesticide is widely used in various parts of the world.  The Food and Agriculture Organization
(FAO) statistics indicate that large amounts of lindane have been used by several countries during
the last two decades,, particularly in Asia.  Although  it is difficult to obtain exact values , the
production and use of lindane in India and China is at a level of tens of thousand of tons per year
in each of these countries (as indicated by Semb and Pacyna, 1988).  Similar amounts a*re believed

                                          109

-------
to be used on the former Soviet Union as reported by Komarov (1980).  Lindane is used in the
above mentioned countries to improve the agricultural crop yield and to reduce d.^.     Lindane
has been also used in Europe and Australia (reported by  Semb and Pacyna,  19Sb after other
authors).

       Recent investigations of the Arctic air pollution have shown that lindane has been present
in appreciable amounts in air samples collected at the ground stations as well as with the use of
the aircraft  (Pacyna and  Oehme,  1988).  The  authors suggested the Asian  continent  as  the
probable source of the lindane.  During the spring storms in the Asian deserts  fractions of dust
with small particles can become airborne and subject to long-range transport within air masses.
Spring is the period with extensive use of lindane and other pesticides in Asia.  Therefore, in it
reasonable to suggest that lindane is taken up with dust particles'and transported  out of the region
(Pacyna and Ortar, 1989). It is suggested here that lindane measured  in the Great waters region
can originate partly in Asia  through  the process described above.  A possibility of long range
transport of air  pollutants from the Asian continent over Hawaii to the western United States and
then further east has been already proposed to study, however no support was obtained.  It should
be added that  our understanding of meteorological processes governing the  movement of  air
masses and  wind  patterns do not exclude the possibility of the air mass transport from Asia to
North America.

       Of course, other source regions can not be excluded when discussing the origin of lindane
in the Great Waters region, e.g. the Latin American countries. It is well established  that certain
organochlorine  compounds, such as DDT and PCBs can be deposited to lakes far removed from
their  source   (Hites   and  Eisenreich,  1987).    The  concentration  ratio   of  a-  to   y-
hexachlorocyclohexane (HCH) in air masses can be a helpful tool to estimate the residence time of
aerosols and then to assess source regions.  The y-HCH is photochemically transformed to the  cc-
isomer. Pacyna and Oehme (1988) concluded that the ratios of 50 and higher would indicate old
air masses.  These values would be expected in the Great waters region if the HCHs were to  be
from the sources in Asia.

       It is rather difficult to expect emissions from parts of the world other than mentioned
above to contribute directly to the deposition  in the study region.   One theoretical possibility
could be an episodic transport of pollutants with the Arctic air masses. Emissions from various
source regions  contribute to the contamination of Arctic air, especially from sources in northern
part of the  former Soviet Union and Europe - so called  Eurasian sources.   One of the major
scientific questions still to be answered by the Arctic researchers is what happens to the pollution
load entering the Arctic region.  A part of this load is deposited to the surface but meteorological
                                           110

-------
 conditions in the Arctic do not favor efficient deposition  of the pollutants.  Thus  one theory
 suggests that a part of the pollution brought to the Arctic is carried-out of the region (e.g. Pacyna
 and Shaw, 1991). Could part of this pollution load reach the Great Lakes and other Great Waters
 under certain meteorological conditions?  If so, what is the frequency of occurrence?  These
 questions need to be  addressed in the future research activities in both the Great Waters regions
 and the Arctic.

 5.3.  Application  of source apportionment techniques  for identification  of impacts of
     emissions on the Great Waters from distant source regions.
       There have been only a few studies to date which attempted to diagnose the  sources of
 toxic air pollutants in the Great Lakes basin. Mamane et al. (1992) carried out a systematic study
 of the sources of pollutants and toxic compounds measured in the air over Green Bay as part of a
 larger study  of toxics deposition to Green Bay.   There  findings suggest that regional source
 influences were the dominant contributor to the paniculate mass measured during the study. They
 also identified incinerator emissions impacting the measurement site as the  concentrations of Cl,
 K, and Pb in  fine particles were 2 to 3 times higher than average in these samples.  Confirmation
 that one or more incinerators contributed to the metals measured on particles in Green Bay was
 found by SEM analysis of individual particles collected on the filters. However, this study did not
 directly quantify the sources of the pollutants deposited to Green Bay.

       More recently, Clark  has  utilized  the RELMAP model to calculate the transport and
 deposition.   The RELMAP  model is a  lagrangian  dispersion model which  relies  upon the
 availability of accurate emissions data for the compounds of interest.  Utilizing dispersion models
 for defining  source-receptor  relationships offers the distinct advantage that one  can  directly
 calculate the  contributions from various sources to the actual deposition to the  lakes.  The total
 deposition (dry + wet) of Pb to Lake Michigan was estimated by Clark (1992) and the source-
 receptor relationships are given in Figure 28. Cells which had sources contributing more than 2%
 of the total annual atmospheric deposition  directly to Lake Michigan are indicted  on the Figure.
 The preliminary modeling suggests that sources of Pb outside of the Great Lakes Basin  are in fact
 the major contributors to  the Pb  deposition to Lake Michigan.  Figure 29 shows the  relative
 contributions  of sources contributing to the deposition of Cd to Lake Michigan. Figure 30 shows
the relative contributions of sources contributing to the deposition of benzo(a) pyrene to Lake
Michigan. The relative contribution pattern for this PAH compound is quite different than those
 seen for Pb and Cd.  The contribution of BaP is primarily from local sources in the Chicago/Gray
 area.
                                          Ill

-------
            40-KM EMISSION/DEPOSITION GRID FOR MODELING TOXICS
     95
 PERCENTAGE CONTRIBUTION TO LAKE MICHIGAN DEPOSITION

                             LEAD
Figure 28.  The relative contribution patterns for total  annual deposition to Lake
       Michigan. Preliminary modeling estimates from Clarke (1992).
                               112

-------
             40-KM  EMISSION/DEPOSITION GRID FOR MODELING TOXICS
     95
  PERCENTAGE CONTRIBUTION TO LAKE MICHIGAN DEPOSITION7

                            CADMIUM
Figure 29.  The relative  contribution patterns for total annual deposition to Lake
       Michigan. Preliminary1 modeling estimates from Clarke (1992).
                               113

-------
            40-KM EMISSION/DEPOSITION GRID FOR MODELING ~'XICS
     95
 PERCENTAGE CONTRIBUTION TO LAKE MICHIGAN DEPOSITION

                      BENZO(A)PYRENE
Figure 30.  The  relative contribution patterns for total annual deposition to  Lake
       Michigan. Preliminary modeling estimates from Clarke (1992).
                              114

-------
       Rahn  et  al. (1989)  applied  their  regional  signature technique to  the  aerosols  and
 precipitation samples collected in Underhill Vermont.  They estimated that approximately 80% of
 the sulfate and selenium originated from sources located in the  Midwest whereas 20% of the
 sulfate and selenium deposited to the Lake Champlain basin were derived from local sources on
 the East Coast.   Keeler and Samson (1989) applied a hybrid receptor model  to investigate the
 utility of using the "regional signature" approach and concluded that variations in elemental  ratios
 may reflect different mixes in emissions of sources in different source regions.  In their analysis
 Keeler and Samson calculated the source contribution fields using  Quantitative Transport Bias
 Analysis (Keeler,  1987) for several heavy metals including arsenic measured in  the eastern North
 America.  The contribution field for As contributions is given in Figure 31. This analysis would
 suggest that the largest contributor to As concentrations measured during the month of August, in
 the Lake Champlain basin for example, was found in the Midwest area around Pittsburgh but that
 significant contributions were also observed from a large  region  extending  up towards the
 Canadian smelters in Sudbury and Noranda.

 6.  BENEFITS FROM EMISSION REDUCTION

       A number of the  abatement techniques are available for the reduction of emissions of
 pollutants studied in this work and emitted from sources in the Great Waters area. Implementation
 of these techniques  shall result in several benefits which can be measured in a local environment,
 e.g. around a given point source of emission, as well as in the whole region of the Great Waters.

       Various alternatives to reduce emissions apply  to anthropogenic  sources. Emissions from
 natural sources are  very difficult  to control, if  impossible. In general,  different alternatives are
 proposed for  stationary (point)  and mobile sources.  It is  not intended here to discuss the
 applicability  of different alternatives to reduce emissions or even their detailed description.  It is.
 however, important to identify various alternatives in a view of benefits resulting  from  their
 implementation.

       Three different  groups of methods  can be identified  for reductions  of emissions from
variety of point  sources:  application  of best  available  technology (so-called BAT,  or best
practicable technology  - BPT),  methods leading to  increased energy efficiency,  and methods
resulting in waste minimization. The BAT concept is based on the application of the state -of-the
art control technology to remove pollutants from exhaust gases. As a majority of heavy metals
and persistent organic pollutants from industrial sources is released on particles, the BAT in this
case is concerned with the equipment to efficiently remove dust from exhaust gases leaving
primary and secondary iron and steel plants, primary and secondary non-ferrous smelters,
                                          115

-------
Figure 31.  The spatially averaged As contribution to ambient  concentrations in the
         northeastern United States during August,  1983.
                                       116

-------
 ferroalloy plants, waste incinerators, and fossil fuel power plants producing electricity and heat.
 In general, the installation of fabric filters or equally efficient control techniques (e g. electrostatic
 precipitators - ESPs) with practical  removal efficiency of better than 98% for fine panicles (with
 diameter smaller than 2 pirn) will assure the reduction of dust concentrations in the exhaust gases
 lower than  10 mg/Nm3   It should be  admitted that  both, primary exhaust  gas and fugitive
 emissions must be controlled. The installation of BAT as described above will result in reduction
 of up to 95% of heavy metals  and  persistent organic pollutants on  particles,  introduced  to the
 industrial processes as impurities of raw materials (Pacyna, 1992).

       Removal of mercury from exhaust gases  is different that the  removal of other  heavy
 metals, as the majority  of the mercury emissions occurs  in a gaseous  phase.  Therefore, BAT for
 mercury removal includes both wet scrubbers and ESPs,  capable-of reducing metal concentrations
 in exhaust gases to at  least 50 [ig/Nm3   A presence of any flue gas desulfurization technique
 results in removal of 40 to 80%  of gaseous mercury in exhaust gases.

       Removal of dioxins from exhaust  gases of waste incinerators  could obtained through the
 application of various flue gas cleaning techniques.  The dry flue gas cleaning with scrubbing on
 fabric filters has proved  to be very efficient to remove dioxins  on an industrial scale in  waste
 incineration plants.

       Reduction of emissions  of heavy metals and persistent  organic  pollutants can  also be
 obtained through application of so-called  pre-treatment methods, and  first of all washing of raw
 materials before use and  switch  of fuels.  In fact both methods are applied primarily to fuels and
 not to other raw  materials.  It  has  been recently concluded that   washing  of coal  prior to
 combustion  results in removal  of 10  to  30%  of  heavy metals contained in coal  (reported by
 Pacyna, 1992).  Fuel switching can be used in some operations which involves replacing some of
 the coal and residual oil fired in a boiler with select natural gas. The degree of emission reduction
 depends on the amount  of fuel to be substituted and the kind of fuel.  In the best cases up to 80%
 reduction can be obtained.

       Appropriate  design  and  management  of the  combustion processes in  new incinerators
would result in  considerable reductions in dioxin formation when compared  with many existing
incinerators and could bring the  emission  back to about 1 ng TEQ (Toxic Equivalents Quantity)/
Mm3

       Reduction of emissions through the application of techniques leading to increase of energy
efficiency is based on the use of so-called primary reduction measures.  An example of such

                                          117

-------
measures  is modification of  combustion processes leading to  higher  combustion  efficiency
(through suitable manipulation of the stoichiometry/temperature profiles within the boiler).  It is
difficult to assess to what extent the emission reduction can be achieved through the combustion
modification. For gaseous pollutants, such as nitrogen oxides a reduction range of 60 to 90% was
found technically possibly (reported by Pacyna, 1992).

       Emission reductions  can also be obtained through minimization of wastes.  Production
technologies resulting in lower amounts  of  exhaust gases  should  be given priority over the
technologies with large quantities of waste gases.  The cost of emission  control installations is
usually lower for the low waste gas technologies to produce a certain  industrial product.

       The use of unleaded gasoline  is the best option to reduce  lead emissions from mobile
sources.  Extended research has been carried out on the cost and benefits from using the unleaded
gasoline (e.g. CONCAWE, 1980; CEC, 1984). A production of unleaded gasoline requires about
5% increase in  total energy  (crude  oil).   A  considerable reduction of  dioxin input to the
atmosphere from vehicles can be achieved through the general use  of catalytic converters and the
use of unleaded gasoline.

       Reduction of emissions of pollutants in the Great Waters  Study areas may  be achieved
through decreases in vehicle miles traveled, fuel reformulation, as well as through the introduction
of alternative vehicles. Tailpipe controls result in lowering emissions from other area sources.

       There are also alternatives to reduce emissions from fugitive and indirect sources although
control of releases from these  sources is much more difficult that  the  control of emissions from
the above discussed sources.  Landfills should be organized  as controlled landfills connected to
waste water treatment plants if possible. Leaching potential of metals, defined as the fraction of
metal present in a solid waste that may become water - soluble under certain chemical conditions
should be controlled at least for volatile metals, such as mercury.

       The  phasing out the  pesticides  which are the  most  persistent,  toxic  and  liable to
bioaccumulate is the best way to deal with the unwanted environmental  effects of these chemicals.

       The  implementation  of the  emission reduction  techniques  as described  above would
improve the quality of the environment through the reduction of  atmospheric deposition  of the
pollutants of interest.  This  is the major environmental benefit.  The reduction of atmospheric
deposition will inevitably decrease the uptake of pollutants by surface waters, soils, and  plants in
the vicinity of major point sources of emissions, and limit migration of these pollutants through
various environmental media.
                                           118

-------
        Installation  of equipment to  reduce  the  dust  concentration  in  exhaust  gases,  or
 improvement of its performance will contribute to the improvement of air visibility in the studv
 region.

        Reduction of atmospheric emissions of heavy metals and persistent organic pollutants will
 result in lowering their intake to human body, mostly due to reducing their ingestion in the study
 region.  This intake has already exceeded the WHO/FAO maximum permissible values for some
 pollutants in some locations within the study area.

        Implementation of emission  control techniques  will  be  a substantial  step  towards
 improving the chemical and to some  extent biological recovery of the  environment in the study
 area.   As a result, an increase of fish population, an important factor of local  economy in the
 region, can be expected.  The most obvious and measurable costs are  those  that  stem  from
 damage to fish and wildlife and  the loss of commercial fisheries represents one of the first and
 most  easily identifiable losses of economic value.  As indicated in the documents for virtually
 eliminating inputs to the Great Lakes,  loss income estimates, in  1990 dollars, due to toxics-related
 closures or market losses could be as high as 8.5  million dollars per year including only losses due
 to mercury in Lake St. Clair and toxics in Lake Ontario.  Since most of the emissions deposited in
 the study region originate outside the region,  limitations of these emissions will reduce deposition
 in a much larger region than the study area.

 7.  CONCLUSIONS

 1. Atmospheric deposition is one of the major sources of lead, arsenic, cadmium, mercury, PAHs.
 lindane,  and possible  PCBs, PCDDs, and  PCDFs measured in the  Great Waters.   The other
 sources include leaching from the landfills, direct  industrial discharges, agriculture practices in the
 region,  transport with  river waters, and direct  dumping of wastes. However, identifying the
 specific sources or  source types emitting the pollutants into the atmosphere which ultimately are
 deposited is another matter. Identification of the dominant pathway and the major sources of the
 critical pollutants should be made for the individual compounds separately as their sources and
behavior in the environment differ substantially.

 2.  Emissions from sources within and outside the Great Waters regions both contribute to the
 load of pollution in atmospheric deposition to the waters in the region.  Although a number of
 source-receptor techniques are available for estimating the contributions, it is still  premature to
conclude what pan of pollution load originates within the study  region and what part results from
long range transport within air masses. The major reason for the present  uncertainty is the lack of

                                           119

-------
reliable input data for application of these techniques including properly reported emission data.
The present lack of monitoring data as well as emissions information is also prob'en.-.:.;, ibr other
regions, e.g. the North Sea and the Baltic Sea, the two most extensively studied regions in Europe
with  respect to  the  environmental  behavior of toxic heavy  metals  and  persistent organic
pollutants.

3.  Identification of emission sources in the Great Waters regions and their characterization with
respect to atmospheric emissions has been carried out for some time in both the United States and
Canada.  As a result, major source categories have been defined for all of the studied pollutants.
The major sources include: production of electricity and heat, combustion of fuels in industrial,
commercial, and residential units, including wood combustion, manufacturing and use of various
industrial goods, and incineration of municipal and industrial wastes, and incineration of sewage
sludge dominate in the group of local sources.  There  are, however, differences in quantitative
assessment of the fluxes from the above sources,  reported by  various research groups in  the
United States and Canada.  These differences should be resolved through thorough examination
of  the  available  data  using various techniques  of verification  of emission  data and joint
supplementary research programs in both countries.

4. Emissions from other source regions in North America may also affect the amount of pollution
load deposited to the Great Waters although no evidence has been provided by measurements and
assessment for heavy metals and persistent organic pollutants.  The NAPAP concluded that such
an impact exists for deposition of sulfates.  As the sulfates are transported within  air masses on
particles, as so do most of the metals and organic compounds discussed in this report, one would
also hypothesize that the metals and organic compounds emitted from sources outside the study
region can be deposited to the Great Waters. At present, regional models of long range  transport
of  air pollutants are  available  and  with some modifications they can de  used  to assess  the
contribution of emissions from outside source regions to the Great waters. An emission inventory
with the appropriate spatial distribution of the data needs to be prepared.  The experience gained
through NAPAP emissions inventory development can be used as a starting point for this purpose
Collaboration with the Mexican authorities on environment protection is highly recommended as
the Mexican  emissions  of heavy metals and  persistent  organic  pollutants are expected  to
contribute to the contamination of the Great Waters.

5.  Lindane was found to be a global air pollutant measured in remote areas around the world.  As
the major application of this pesticide is in Asia and the wind patterns at various  altitudes do  not
exclude the air mass transport form the Asian continent to North America, lindane deposited in
the Great Waters regions may originate from as far away as  India,  China, or the former Soviet
                                           120

-------
 Union. This hypothesis can be tested by the application of global models, or at least hemispheric
 models. At  present, such  models are used to study the transport of green-house gases in the
 atmosphere  and the transport of sulfur in the Northern Hemisphere.   A knowledge of emission
 rates or fluxes of lindane in Asia within 1 degree by 1 degree grid system should be elaborated for
 the use in the models.  This task can be carried out in co-operation with international programs
 involved in preparation of global emission inventories, e.g. the IGBP program on Global Emission
 Inventories Activity (GEIA) or the OECD program on emission of greenhouse gases.

 6.  Several  methods can be  applied to reduce emissions of toxic heavy metals and persistent
 organic pollutants  and eventually reduce the  atmospheric deposition of these pollutants to the
 Great Waters. Technological solutions presented in a form of Best Available Technology (BAT)
 package or Best Practicable Technology (BPT) package offer emission reduction possibilities for
 point sources within all major source categories contributing to the contamination on the Great
 Waters. Experience gained in this respect in North America and Europe, and particularly in the
 Baltic Sea Environmental Program can be useful in recommending emission reduction scenarios  in
 the Great  waters region.   Cost estimates and benefits from the implementation  of the  control
 techniques should be carefully studied. Experience gained during NAPAP may prove very useful.

 8. RECOMMENDATIONS ON FUTURE RESEARCH ACTIVITY TO IMPROVE
   THE ACCURACY OF METHODS FOR SOURCE IDENTIFICATION

       New research initiatives are necessary  in order  to meet the requirements outlined in the
 conclusions  of this report  as well as to test the important hypotheses proposed here.   These
 activities would include both measurement programs and modeling estimates.

 8.1. Measurement programs
       New  measurement  programs are needed  in  order  to  improve the quality of source-
 receptor techniques which  are used to assess the magnitude and origin of deposited pollutants.
These measurements are needed at both the receptor, the Great Waters themselves, as well as at
the sources of the emissions. The following source emissions information are recommended.

- emission rates and emission factors for toxic heavy metals and persistent organic pollutants from
large point sources  in the study region should be evaluated on the basis of measurements of their
concentrations in' exhaust  gases.   Large  point sources  should  include electricity and  heat
producing  plants, ferrous and non-ferrous metal smelters, cement kilns, and waste incinerators.
                                         121

-------
They can be defined as in the CEC directives.  The emission measurements shall be representative
and their reporting transparent;

- physical and chemical forms  of the most volatile  compounds should be established through
measurements carried out in major sources in the study area; and

- emission rates for the most volatile metals, and particularly mercury, as well as lindane and other
pesticides should be derived on the basis  of measurements over the water surface in the  Great
Waters and the  surrounding soils.  The results should be representative for the meteorological
conditions as in the Great Waters and exemplify seasonal changes.

       Measurements at receptors should  provide with the information which is needed in  order
to improve the accuracy of source-receptor relationship analysis. The following is recommended:

- size-differentiated chemical composition of aerosols should be measured at receptors which can
represent conditions over the  water surface in the  study area.   The results  used in various
statistical methods for source-receptor modeling, e.g. principal component analysis can be used to
improve the accuracy of identification of source categories discussed; and

- simultaneous measurements of the gaseous and particle phases of the studied  pollutants with the
help of newly developed techniques (e.g. denuder methods)  should be undertaken in order  to
provide information on gas-to-particle conversions ( and particle-to-gas conversions) for the most
volatile pollutants under study.  The results should be useful to explain the chemical behavior of
these pollutants, particularly during the episodes of their transport within air masses from source
regions to the receptors in the Great Waters areas.

8.2. Modeling estimates

       Improvement  is  needed  within  the  three  groups of  estimates:  emission  estimates.
dispersion modeling, and receptor modeling.

       The following  is  recommended for the improvement  of emission estimates in order  to
assure better understanding of source identification in the Great waters region:

-  a set of emission factors and emission rates  should be prepared for all sources contributing to
the  contamination of the  Great  waters,  and  particularly  for sources or  even  whole source
categories for which measurements are not available.  In general, rather limited information  exists
on emission factors for heavy metals and even  less for persistent organic compounds. In the past

                                            122

-------
 major research  has  been carried out in Europe, particularly within the UN  ECE programs.
 PARCOM,  and  RELCOM.  Recently several projects  has been  carried out in  North America.
 mainly  for EPA, IJC, Ontario  Ministry of the Environment, and Environment Canada.   The
 information on  emission factors,  available  from the above mentioned  programs  should be
 reviewed and a set of emission factors and emission rates selected with the aim of their application
 to prepare emission inventory for atmospheric heavy metals and persistent organic compounds in
 North America with  special emphasis on the sources within the Great waters region. Guidelines
 on  emission estimation  and reporting  should be  elaborated  in  order to assure  the  data
 representativeness, comparability, completeness, consistency, and accuracy. As the  subject  is of
 broad interest, a close co-operation with other programs and international organizations is highly
 recommended, particularly with the  UN ECE task forces on  emission of heavy metals  and
 persistent organic pollutants;

 -  gridded emission  inventory for the studied  pollutants should be approached for the whole
 territory of the United States and Canada. A large body of information on the parameters used to
 prepare  spatial distribution of emission data, e.g. geographical  location of point  sources  and
 surrogate parameters to distribute emissions from area sources has been collected during NAPAP
 This information  should be used to distribute the emission data for pollutants under study here;

 - seasonal changes of mercury and volatile organic compound emissions need to be quantified and
techniques developed to estimate these emissions; and

- an approach should be defined to assess emissions of pesticudes (e.g. lindane) in the Northern
Hemisphere  with particular emphasis on Mexico and the Asian countries.  This  task should be
carried out in co-operation with international organizations, such as IGBP

       Improvements in  source identification through the further development of dispersion
modeling is needed.  The following are recommended:

- continue to modify and improve the existing long-range transport models so they can be used to
study the contribution of emissions from sources in North America, both within and outside the
study region to the hazardous pollution load deposited to the Great Waters. Models developed
during NAPAP as well as in other programs in  North America and Europe ( e.g. the UN ECE
EMEP model) should be taken into account; and

- an approach should be  made to apply the existing  global scale models to  investigate the
possibility of lindane used in  Asia to be transported within air masses to North America  and
deposited also in the Great Waters region.
                                          123

-------
9. REFERENCES

Achman, D.A., HombuckJe, K C. and Eisenreich, S.J. (1992). Volatilization of PCI3.S from Green
Bay, Lake Michigan.  Environ. Sci. Technol. 27, 75-87.

AJcamo,  J., Bartnicki, J., Olendrzynski, K. and Pacyna J.M.  (1993) Computing Heavy Metals in
Europe's Atmosphere. Atmos. Environ, (in press).

Alford-Stevens, A.L.  (1986). Analyzing PCBs.  Basic information about PCBs and how they are
identified and measured.  Environ. Sci. Technol. 20, 1194-1199.

Alford-Stevens, A.L., Bellar, T.A., Eichelberger, J.W. and Budde, W L. (1986).  Characterization
of commercial  Aroclors  by automated  mass spectrometric determination  of polychlorinated
biphenyls by level of chlorination. Anal. Chem. 58, 2014-2022.

Alsberg,  T., Hakansson, S., Strandall, M. and Westerholm, R. (1989). Profile analysis of urban air
pollution. Chemometrics and Intelligent Lab. Sys. 7, 143-152.

Amendola, G., Barna, D., Blosser, R., LaFleur, L., McBride, Al, Thomas, F., Tiernan, T.  and
Whittemore, R. (1989).  The occurrence and fate of PCDDs and PCDFs in five bleached kraft
paper pulp and paper  mills. Chemosphere 18,  1181-1188.

Anderson, J.R., Buseck, P.R., Saucy,  D.A. and  Pacyna, J.M.  (1992).  Compositions,  size
distributions, and principal component analysis of individual fine-fraction panicles from the Arctic
aerosol at Spitsbergen, May - June, 1987. Atmos. Environ. 26A, 1747 - 1762.

Appel, B.R., Cheng,  W. and Salaymeh. F  (1989).  Sampling of carbonaceous particles in the
atmosphere II  Atmos. Environ. 23, 2167-2175.

Arey, J.,  Zielinska, B., Atkinson, R. and Winer, A.M. (1987). PAH and nitroarene concentrations
in ambient air during  a wintertime high NOX episode in the Los Angeles Basin.  Atmos. Environ.
21, 1437-1444.

Arey, J.,  Atkinson, R., Zielinska, B. and McElroy, P. A. (1989).  Diurnal concentrations of volatile
polycyclic aromatic hydrocarbons and nitroarenes during a photochemical air pollution episode in
Glendora, California.  Environ. Sci. Technol. 23, 321-327.

Aronian, P.P.,  Scheff, P.A. and Wadden,  R.A. (1989).   Wintertime  source  reconciliation of
ambient organics. Atmos. Environ. 23, 911-920.

Atkinson, R. (1987).  Estimation of OH radical reaction rate constants and atmospheric lifetimes
for polychlorobiphenyls,  dibenzo-p-dioxins, and dibenzofurans. Environ. Sci. Technol. 21, 305-
307.

Atkinson, R. (1990).  Gas phase tropospheric chemistry of organic compounds: a review.  Atmos.
Environ. 24A, 1-41.
                                          124

-------
 Axelgard. P  and Renberg, L  (1989)  The influence of bleaching chemicals and lignin content on
 the formation of polychlorinated dioxins and dibenzofurans.  Chemosphere 19, 661-668

 Axenfeld.  F,  Munch.  J.,   Pacyna,  J.M..   Duiser,  J.A.   and  Veldt,   C.   (1992).  Test-
 Emissionsdatenbasis  der  Spurenelemente  As, Cd, Hg,  Pb,  Zn,  und der Speziellen Organischen
 Verbindungen y-HCH (Lindan),  HCB.  PCB,  und PAX fur  Modellrechnungen  in  Europa.
 Forschungsbericht 104 02 588, Dornier GmbH, Friedrichshafen, Germany (in German).

 Bacci, E., Calamari, D., Gaggi, C., Fanelli, R., Focardi,  S. and Morosini, M. (1986).  Chlorinated
 hydrocarbons in lichen and moss samples from  the Antarctic Peninsula.  Chemosphere 15, 747-
 754.

 Bacci, E.,  Calamari,  D., Gaggi, C., Focardi,  S. and  Morosini, M.  (1988).   Organochlorine
 pesticide and PCB residues in plant foliage (Magnifera  indicd) from West Africa. Chemosphere
 17,693-702.

 Bacci, E., Calamari, D., Gaggi, C. and Vighi, M. (1990a). Bioconcentration of organic chemical
 vapors in plant leaves: experimental measurements and correlation.  Environ.  Set. Technol.  24,
 885-889.

 Bacci, E.,  Cerejeira,  M.J., Gaggi,  C., Chemello, G., Calamari, D.  and Vighi, M.  (1990b).
 Bioconcentration of organic chemical vapours in plant leaves:  the azalea model.  Chemosphere
 21,525-535.

 Bacher, R., Swerev, M. and Ballschmiter. K. (1992).  Profile and pattern of monochloro- through
 octachlorodibenzodioxins and dibenzofurans in chimney deposits from  wood burning. Environ.
 Sci. Technol.  26, 1649-1658.

 Back, S.O., Field,  R.A., Goldstone, M.E., Kirk, P W., Lester, J.N.  and Perry, R. (1991a).   A
 review of atmospheric polycyclic aromatic hydrocarbons:  sources, fate, and behavior.  Water,
 Air, SoilPollut. 60, 279-300.

 Back,  S.O..  Goldstone.  M.E.,  Kirk,  P W.W.,  Lester,  J.N.  and Perry, R.  (1991b).   Phase
 distribution and particle  size dependency of polycyclic  aromatic  hydrocarbons in the urban
 environment.  Chemosphere 22, 503-520.

 Baker, D.B  and  Richards,  R.P (1990).   Transport  of soluble pesticides through drainage
 networks in large agricultural basins. In. Long-Range Transport of Pesticides. Kurtz, D.A. (ed.),
 pp. 241-270. Lewis Publishers, Chelsea, Michigan.

 Baker, J.E. and Eisenreich, S.J. (1990).   Concentrations  and  fluxes of polycyclic aromatic
 hydrocarbons and polychlorinated biphenyls across the air- water interface of Lake Superior.
Environ. Sci.  Technol. 24. 342-352.
                 *
 Ballschmiter,  K. (1991).   Global distribution of organic compounds. Environ.  Carcino. Ecotox.
Rev. C9,  1-46.
                                          125

-------
Barrie, L.A., Gregor, D.J., Hargrave, BT, Lake. R, Muir, D.C.G., Shearer, R., Tracey, B. and
Bidleman, T.F  (1992).   Arctic contaminants:  sources,  occurrence, and pathos  Sci.  Tot.
Environ.  122, 1-74.

Bastable, H.G., Rogers, D.P and Schorran, D.E. (1990).  Tracers of opportunity and pollutant
transport in southern California.  Atmos. Environ. 24B, 137-151.

Behymer, T.D. and Hites, R.A. (1988). Photolysis of polycyclic aromatic hydrocarbons adsorbed
on fly ash. Environ. Sci.  Techno!. 22, 1311-1319.

Benjey, W.G., and Coventry, D.H. (1992).  Geographical distribution and source type analysis of
toxic metal emissions. Presented at the Int. Symp.  on Measurement of Toxic and  Related Air
Pollutants, Durham, NC, May 3 -8, 1992.

Benner,  Jr., B.A., Gordon, G.E. and  Wise, S.A.  (1989).   Mobile  sources  of  atmospheric
polycyclic aromatic hydrocarbons: a roadway tunnel study.  Environ. Sci.  Technol. 23, 1269-
1278.

Bidleman, T.F. and  Olney, C.E.  (1974).   Chlorinated  hydrocarbons in  the Sargasso  Sea
atmosphere and surface water. Science 183, 516-518.

Bidleman, T.F  and Olney,  C.E. (1975).  Long range transport of toxaphene insecticide in the
atmosphere of the western North Atlantic.  Nature 257, 475-477.

Bidleman, T.F., Rice, C.P and Olney, C.E (1976).  Fiigh molecular weight  hydrocarbons in the
air and sea:   rates  and mechanisms of air-sea transfer. In: Marine Pollutant  Transfer. Windom,
H.L., Duce,  R.A.  (eds.), pp.  323-351   Lexington Books (D.C. Heath and Co.), Lexington,
Massachusetts.

Bidleman, T.F., Wideqvist,  U., Jansson, B. and Soderlund, R. (1987). Organochlorine pesticides
and  polychlorinated biphenyls in the atmosphere  of southern Sweden.  Atmos. Environ. 21.
641-654.

Bidleman, T.F. (1988). Atmospheric processes:  wet and dry deposition of organic compounds
are controlled by their vapor-particle partitioning. Environ. Sci. Technol. 22, 361-367, correction
on 726-727.

Bidieman, T.F., Zaranski, M.T. and Walla, M.D. (1988). Toxaphene: usage,  aerial transport, and
deposition.  In: Toxic Contamination in Large Lakes. Vol. Ill, Schmidtke, N.W. (ed.), pp.  257-
284. Lewis Publishers, Chelsea, Michigan

Bidleman. T.F., Atlas, E.L., Atkinson, R., Bonsang, B., Burns, K.,  Keene, W.C.,  Knap, A.H.,
Miller, J., Rudolph, J. and Tanabe, S. (1990). The long-range transport of organic  compounds.
In:  The Long-Range Atmospheric Transport of Natural and Contaminant Substances. Knap, A.H.
(ed.), pp. 259-301. NATO ASI Series C, Vol. 297, Kluwer Academic Publishers, Boston.
                                          126

-------
 Bidleman,  T.F  (1991).  Gas-particle  distribution and atmospheric  deposition of  semivolatile
 organic compounds.  In: Municipal Waste Incineration Risk Assessment. Travis, C.C.  (ed.), pp.
 65-86. Plenum Publishing Co., New York.

 Bidleman,  T.F.,  Walla, M.D.,  Roura, R.,  Carr, E. and Schmidt, S. (1992a).  OrganochJorine
 pesticides in the atmosphere of the Southern Ocean and Antarctica.  Marine Pollut. Bull, (in
 press).

 Bidleman,   T.F.,  Patton,  G.W.  and  Hinckley,  D.A.   (1992b).   Hexachlorocyclohexanes,
 hexachlorobenzene, and polychlorocamphenes in the Arctic atmosphere and surface ocean.  In:
 Proceedings of the Eighth International Conference of the Comite Arctique International on the
 Global Significance of the Transport and Accumulation of Polychlorinated Hydrocarbons in the
 Arctic, Shearer, R. (ed.), Plenum Publishing Co. (in press).

 Bidleman, T.F., Cotham, W.E., Addison, R.F. and Zinck, M.E. (1992c). Organic contaminants in
 the northwest Atlantic  atmosphere at  Sable Island, Nova  Scotia, 1988-89.   Chemosphere  24,
 1389-1412.

 Bjorseth, A., Lunde, G. and Lindskog, A. (1979).  Long range transport of polycyclic aromatic
 hydrocarbons. Atmos. Environ. 13, 45-53.

 Blakley, P.  and  Klevs, M. (1990).  Air Toxics  Emission Inventories for the Lake Michigan
 Region.  Contract  68-02-4398  from the U.S.  Environmental Protection Agency, Region V, to
 Engineering Science, Inc.,  Fairfax, Virginia,  192 pp.

 Bobet, E., Berand, M.F and Dann, T. (1989).  The measurement of PCDD and PCDF in ambient
 air in southwestern Ontario. Chemosphere 20,  1439-1445.

 Bodo,  B.A.  (1991).  Trend analysis and mass discharge estimates of atrazine in southwestern
 Ontario Great Lakes tributaries, 1981 -89. Environ. Toxicol. Chem.  10,  1105-1121.

 Broman, D.. Naf. C. and Zebuhr,  Y. (199la).  Long-term high- and low-volume air sampling of
 polychlorinated dibenzo-p-dioxins and dibenzofurans and polycyclic aromatic hydrocarbons along
 a transect from urban to remote areas on the Swedish Baltic coast. Environ.  Sci.  Technol. 25.
 1841-1849.

 Bumb, R.R., Crumett, W.B., Cutie, S.S., Gledhill, J.R., Hummel, R.H., Kagel,  R.O., Lamparski.
 L.L., Luoma, E.V., Miller, D.L.,  Nestrick, T.J.,  Shadoff,  L.A.,  Stehl, R.H.   and  Woods, J.S.
 (1983). Trace chemistries of fire. Science 210, 385.

Bunce, N.J.,  Landers, J.P., Langshaw, J-A.  and Nakai, J.S.  (1989).  An assessment of the
importance of direct solar  degradation of some simple chlorinated  benzenes and biphenyls in the
vapor phase. Environ.  Sci.  Technol. 23,  213-218.

Bunce, N.J. and Nakai, J.S. (1989).  Atmospheric chemistry of chlorinated phenols.  J. Air Pollut.
 Cont. Assoc. 39, 820-823.
                                          127

-------
Calamari, D., Bacci, E., Focardi, S., Gaggi, c., Morosini, M. and Vighi, M. (1991).  Role of plant
biomass in the global environmental partitioning of chlorinated  hydrocarbons.  Environ.  Sa.
Technol. 25, 1489-1495.

Christmann,  W.,  Kloppel,  K.D.,  Partscht,  H.  and  Retard, W.  (1989).  Determination of
PCDD/PCDF in ambient air. Chemosphere 19, 521-526.

Clark,  T.L.  (1992). Exposure/risk assessment status report:  regional analysis. Presented at the
Joint Workshop on Mercury and Multimedia Risk Assessment, Durham, NC, October 22, 1992.

Clement, R.E., Suter, S.A., Reiner, E., McCurvin, D. and Hollinger,  D. (1989).  Concentrations
of chlorinated dibenzo~p-dioxins and dibenzofurans in effluents and centrifuged particulates from
Ontario pulp and paper mills. Chemosphere 19, 649-654.

Clendening, L.D., Jury, W.A. and Ernst, F.F. (1990).  A  field mass balance study of pesticide
volatilization, leaching,  and persistence. In: Long-Range Transport  of Pesticides. Kurtz, D.A.
(ed.), pp. 47-60. Lewis Publishers, Chelsea, Michigan.

Cohen, J. and Pinkerton, C. (1966).  Widespread translocation of pesticides by air transport and
rainout, In:   Organic Pesticides in  the Environment. Gould,  R.F. (ed.),  pp.  163-176.  Adv. in
Chem.  Ser., vol. 60, Amer. Chem. Soc., Washington, D.C.

Cohen, M.A., Ryan, P.B., Spengler, J.D., Ozkaynak, H.  and Hayes, C. (199la).  Source-receptor
study of volatile organic compounds and paniculate matter in the Kanawha Valley, West Virginia.
I. methods and descriptive statistics.  Atmos. Environ. 25B, 79-93.

Cohen, M.A., Ryan, P.B., Spengler, J.D., Ozkaynak, HI, Hayes,  C. (1991b).  Source-receptor
study of volatile organic compounds and paniculate matter in the Kanawha Valley, West Virginia.
II. Analysis of factors contributing to VOC and particle exposures.  Atmos. Environ. 25B, 79-93.

Gotham,  W.E. (1990).  Chemical  and Physical Properties Affecting the  Transport  and Fate of
Semivolatile Organic Compounds in the Environment. Ph.D. dissertation, Dept. of Chemistry and
Biochemistry, University of South Carolina, Columbia, South Carolina.

Gotham,  W.E.  and Bidleman,  T.F.  (1991).   Estimating the  atmospheric deposition of
organochlorine contaminants to the Arctic. Chemosphere 22, 165-188.

Gotham,  W.E.  and Bidleman, T.F. (1992).   Laboratory  investigations  of the partitioning of
organochlorine compounds between the gas phase and atmospheric aerosols on glass fiber filters.
Environ.  Sci. Technol. 26, 469-478.

Coutant, R.W., Callahan, P.J., Chuang, J.C. and Lewis, R.G. (1992). Efficiency of silicone grease
coated denuders for collection of   polycyclic aromatic hydrocarbons.   Atmos. Environ. 26A,
2831-2834.

Czuczwa, J.M. and Hites, R.A. (1986).  Airborne dioxins and dibenzofurans:  sources and fates.
Environ.  Sci. Technol. 20, 195-200.

                                          128

-------
 Daisey, J.M., Cheney, J.L. and Lioy, P J  (1986).  Profiles of organic paniculate emissions from
 air pollution sources:   status and  needs  for receptor source  apportionment modeling.  J. Air
 Pollut. Com. Assoc. 36. 17-33.

 Dann, T. (1990). The Detroit Incinerator Monitoring Program, Data Reports 3 and 4, PMD-89-
 8, PMD-90-8, Environment Canada, Conservation and Protection, Ottawa.

 deRaat, W.K., Bakker, G.L. and  Meijere, F.A.  (1990).  Comparison of filter materials for
 sampling of mutagens and  polycyclic aromatic hydrocarbons in ambient airborne particles. Atmos.
 Environ. 24A, 2875-2887.

 Doskey, P V., Porter, J.A. and Scheff, P.A. (1992).  Source fingerprints for volatile non-methane
 hydrocarbons. J. Air Waste Mgmt. Assoc.  (in press).

 Ducret, J. and Cachier,  H. (1992).   Paniculate carbon content in rain  at various temperate and
 tropical locations. J. Atmos. Chem.  15, 55-67.

 Duinker, J.C. and  Bouchertall, R. (1989).  On the distribution of atmospheric  polychlorinated
 biphenyl congeners between the vapor phase, aerosols, and rain. Environ. Sci. Technol. 23, 57-
 62.

 Dunstan, T.D.J., Mauldin,  R.F., Jinxlan, Z., Hipps, A.D., Wehry, E.L. and Mamantov, G. (1989).
 Adsorption and photodegradation of pyrene on magnetic, carbonaceous,  and mineral subtractions
 of coal stack  ash. Environ. Sci. Technol. 23, 303-308.

 Dzubay, T.G. and Mamane, Y. (1989). Use of electron microscopy data in receptor  models for
 PM-10. Atmos. Environ. 23, 467-476.

 Eadie, B.J. (1984).   Distribution of polycyclic aromatic hydrocarbons  in the Great  Lakes. In:
 Toxic Contaminants in the Great Lakes. Nriagu, J.O. and Simmons, M.S.  (eds.), pp. 195-211.
 Adv. Environ. Sci. Technol. Ser. Vol.  14, J. Wiley and Sons, New York.

 Eadie, B.J.,  Robbins,  J.A., Faust,  W.R.  and Landrum, P.F.  (1990).  Polycyclic aromatic
 hydrocarbons in sediments and  pore waters of the  lower Great  Lakes:   reconstruction of a
 regional benzo(a)pyrene source function, pp.  171-189, In:

Edgerton,  S.J., Holdren, M.W. and  Smith, D.L. (1989a).  Interurban  comparisons of ambient
volatile organic compound  concentrations in U.S. cities. J. Air Pollut. Cont. Assoc. 39, 729-732

Edgerton, S.J., Czuczwa. J.M., Rench, J.D.,  Hodanbosi, R.F and Koval, P.J. (1989b). Ambient
air concentrations of polychlorinated dibenzo-p-dioxins and dibenzofurans in Ohio:  sources and
health risk assessment. Chemosphere 18, 1713-1730.

Eduljee, G.H. (198-7).   Comment  on "Airborne dioxins,  sources  and fates"   Environ.  Sci.
 Technol. 21, 922-924.
                                          129

-------
Eisenreich, S.J. and Strachan. W.M.J. (1992).  Estimating Atmospheric Deposition  of Toxic
Substances to the Great Lakes - An Update.  Proceedings of a workshop held at the Canada
Centre for Inland Waters, Burlington, Ontario, Jan. 31- Feb. 2,  1992, sponsored by the Great
Lakes Protection Fund and Environment Canada, 59 pp.

Eitzer, B.D. and Kites, R.A. (1989a).  Polychlorinated dibenzo-p-dioxins and dibenzofurans in the
ambient air of Bloomington, Indiana. Environ. Sci.  Technol. 23, 1389-1395.

Eitzer, B.D. and Kites, R.A. (1989b).  Atmospheric transport and deposition of polychlorinated
dibenzo-p-dioxins and dibenzofurans. Environ. Sci.  Technol. 23, 1396-1401.

Emmim, J., Gregory, C. and Wind, J.J. (1989).  A Prototype Toxics Loading Inventory for Ten
Counties in Southeast Virginia, Prepared for the U.S. Environmental Protection Agency, Office of
Resources Management, by American Management Systems, Inc., Contract GS-OOK-85-AFD-
2777, 47 pp.

Environmental Protection Agency (1987a).  National Dioxin Study ~ Report to Congress.  EP V
530/SW-87-025, Research Triangle Park, North Carolina.

Environmental  Protection Agency  (1987c).  Chlordane,  Heptachlor,   Aldrin,  and  Dieldrin.
Technical  Support Document, Office of Pesticide Programs, Washington, D.C.

Environmental Protection Agency (1988a).  Compilation of Air Pollutant Emission Factors.  Vol.
i.  Stationary Point and Area Sources, Supplement B., 4th ed., U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards. NTIS PB89-128631.

Environmental Protection Agency (1988b).  Air  Emissions Species Manual, Vol.  I.  Volatile
Organic Compound Species Profiles. Report No. EPA-450/2-88-033.

Environmental Protection Agency (1988c).  Air Emissions  Species Manual, Vol. II. Paniculate
Matter Species Profiles. Report No. EPA-450/2-88-033.

Environmental Protection Agency (1990). Toxics in the Community. 1988. National and Local
Perspectives.  U.S. Environmental  Protection Agency,  Washington, DC, Report No. EPA
560/4-90-017.

Environmental Protection Agency  (1991). National Air Quality and Emissions Trends Report,
1990. U.S. Environmental Protection Agency, Research Triangle Park, NC, 1991.

Erickson,  M.D.  (1989).   PCDFs and related compounds  produced from PCB fires: a review.
Chemosphere 19, 161-165.

Federal Register (1988).   Chlordane and heptachlor termiticides.  notice of cancellation and
amendment of existing stocks determination. 52, 11798-11801

Food and Agricultural Organization  of the  United Nations (1986-89).  Production Yearbook.
Vols. 39-42, Rome.

                                         130

-------
 Fouquet,  A.,  Messier, F.,  Cogan, D.  and  Grondin,  A. (1990).  Investigation for  presence of
 chlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs) in bleaching processes from
 pulp and paper industries.  Chemosphere 20, 1707-1714.

 Freeman,  D.J. and Cattell, F.C.R. (1990),  Woodburning as a source of atmospheric polycyclic
 aromatic hydrocarbons. Environ. Sci. Technol. 24, 1581-1585.

 Friesen, K.J.,  Muir, D.C.G.  and Webster, G.R.B. (1990).   Evidence of sensitized photolysis of
 polychlorinated dibenzo-p-dioxins in natural waters under sunlight  conditions.  Environ.  Sci.
 Technol. 24, 1739-1744.

 Gaggi, C., Bacci, E., Calamari, D.  and Fanelli, R. (1985). Chlorinated hydrocarbons in plant
 foliage: an indicator of tropospheric contamination level. Chemosphere 14, 1673-1686.

 Gianessi, L.P.  and Puffer, C. (1990, rev. 1991).  Herbicide Use in  the United States.  Quality of
 the Environment Division, Resources for the Future, Washington, D.C., 128 pp.

 Gibson, T.L.,  Korsog, P.E. and Wolff, G.T.  (1986).   Evidence  for the transformation of
 polycyclic organic matter in the atmosphere. Atmos. Environ. 20, 1575-1578.

 Glotfelty,  D.E., Schomburg, C.J., McChesney, M.M., Sagebiel, J.C. and  Seiber, J.N. (1990).
 Studies of the distribution,  drift,  and volatilization of diazinon residues  resulting  from  spray
 application to a dormant peach orchard.  Chemosphere 21, 1303-1314.

 Graney, J.R., Halliday, A.N., Keeler, G.J., Robbins, J.A. and Nriagu, J.O  (1992) Isotopic tracing
 of lead pollution in the Great  Lakes. Paper presented at  the  1992 Fall AGU Meeting.

 Greaves, R.C., Barkley, R.M., Sievers, R.E. and Meglen, R.R.  (1987).  Covariations in  the
 concentrations of organic compounds associated with  springtime atmospheric aerosols.  Atmos.
Environ. 21,2549-2561.

 Greenburg, A., Darack,  F.,  Harkov,  R., Lioy, P  and  Daisey, J.  (1985). Polycyclic  aromatic
 hydrocarbons in New Jersey,  a comparison of winter and summer concentrations over a two-year
 period. Atmos. Environ. 19,  1325-1339.

 Gregor, D.J. (1991). Trace organic chemicals in the arctic environment:  atmospheric transport
 and deposition. In: Pollution of the Arctic Atmosphere. Sturges, W.T. (ed.), pp. 217-254. Elsevier
Applied Science, New York.

Guo, Z. and Kamens, R.M. (1991).   An experimental technique for studying heterogeneous
reactions of polycyclic aromatic hydrocarbons on particle surfaces.  J. Atmos. Chem. 12, 137-151.

Hansen, A.D.A., Conway, T.J., Steele, L.P., Bodhaine, B.A.,  Thoning, K.W.,  Tans,  Z.  and
Novakov,  T. (1989)., Correlations among combustion effluent species at Barrow, Alaska:  aerosol
black carbon, carbon dioxide, and methane. J. Atmos. Chem. 9, 283-299.
                                          131

-------
Hanson, D.J. (1991).  Dioxin toxicity:  new studies prompt debate, regulatory ac.i^,.  Chemical
and Engineering News, Aug. 12, 7-14

Hargrave, B.T., Vass, W.P., Erickson.  P.E. and Fowler, B.R. (1988).  Atmospheric uansport of
organochlorines to the Arctic Ocean.  Tellus 40B, 480-493.

Harkov, R. and Greenburg,  A.  (1985).  Benzo(a)pyrene in New Jersey -results from a twenty-
seven site study. J. Air Pollut. Cont. Assoc. 35, 238-243.

Hawthorne,  S.B.,  Miller, D.J., Barkley,  R.M.  and Krieger,  M.S.  (1988). Identification of
methoxylated phenols as candidate tracers for atmospheric wood smoke pollution.  Environ. Sci.
Technol.  22, 1191-1196.

Hawthorne,  S.B., Krieger,  M.S., Miller, D.J. and  Mathiason, M.B.  (1989).   Collection and
quantitation  of methoxylated phenol tracers for atmospheric  pollution from residential wood
stoves. Environ. Sci. Technol. 23, 470-475.

Heffter, S.L. (1980). Air Resources Laboratories atmospheric transport and dispersion model.
NOAA  Technical  Memorandum  ERL  ARL-81,  National  Oceanic  and  Atmospheric
Administration Air Resources Laboratory, Silver Spring, Maryland, February 1980.

Henry, R.C., Lewis, C.W., Hopke, P.K. and Williamson, H.J. (1984). Review of receptor model
fundamentals.  Atmos. Environ.  18, 1507-1515.

Henry, R.C.  (1991).   Multivariate receptor models. In: Receptor Modeling  for Air Quality
Management. Hopke, P.K. (ed.), pp. 117-147. Elsevier, New York.

Hermanson,   M.H.  and  Hites,  R.A.  (1989).    Long-term  measurement  of  atmospheric
polychlorinated biphenyls in the vicinity of Superfund sites. Environ.  Sci.  Techno!. 23, 1253-
1258.

Hildemann, L.M., Markowski, G.R. and Cass, G.R.  (199la). Chemical composition of emissions
from urban sources of fine organic aerosol.  Environ. Sci. Technol. 25, 744-759.

Hildemann, L.M.,  Mazurek, M.A.,  Cass, G.R.  and Simoneit, B.R.T.  (1991b).   Quantitative
characterization of urban sources of organic aerosol by  high  resolution gas chromatography.
Environ.  Sci.  Technol. 25, 1311-1325.

Hinckley, D.A., Bidleman, T.F  and  Rice,  C.P  (1991)  Atmospheric  organochlorine pollutants
and  air-sea exchange of hexachlorocyclo-hexanes in the Bering  and Chukchi Seas.  J. Geophys.
Res. 96, 7201-7213.

Hitch, R.K.  and Day, HH.  (1992).  Unusual persistence of DDT in some western soils.  Bull.
Environ.  Contain. Toxicol. 48, 259-264.

Hoff, R.M. and Chan, K.W  (1987). Measurement of polycyclic aromatic hydrocarbons in the air
along the Niagara River. Environ. Sci. Technol. 21, 556-561.

                                          132

-------
 Hoff, R.M., Muir, D.C.G. and Grift, N.P (199la).  Annual cycle of polychlorinated biphenyls and
 organohalogen pesticides in air in southern Ontario.  1   Air concentration data.   Environ. Sci.
 Technol. 26, 266-275.

 Hoff, R.M., Muir, D.C.G. and Grift, N.P (1991b).  Annual cycle of polychJorinated biphenyls and
 organohalogen pesticides in  air in  southern Ontario.  2. Atmospheric  transport and sources.
 Environ. Sci. Technol. 26, 276-283.

 Holsen, T.M., Noll, K.E., Liu, S-P  and Lee, W-J (1991).  Dry  deposition  of polychJorinated
 biphenyls in urban areas.  Environ. Sci. Technol. 25, 1075-1081.

 Hopke, P.K. (1991).  An introduction to receptor modeling. In: Receptor Modeling for Air
 Quality Management. Hopke, P.K. (ed.), pp. 1-10. Elsevier, New York.

 HornbuckJe,  K.C., Achman,  D.A.  and Eisenreich,  S.J.   (1992).    Over-water and  over-land
 concentrations of PCBs in Green Bay, Lake Michigan. Environ.  Sci. Technol. 27, 87-98.

 Hov, 0.,  Penkett,  S.A., Isaksen, I.S.A.  and  Semb,  A.  (1984).  Organic  gases  in  the
 Norwegian Arctic. Geophys. Res. Lett. 11, 425-428.

 Husain, L., Webber, J.S., Canelli,  E., Dutkiewicz, V.A. and Halstead, J.A. (1984).  Mn/V
 ratio as a tracer of aerosol SO42'transport. Atmos. Environ.  18, 1059-1071.

 ICF (1992).  Atmospheric deposition of toxic chemicals  to surface waters: Identification  and
 summary of the recent literature. ICF Incorporated, Report, Fairfax, VA.

 ILZSG  (1984). World Directory: Lead and Zinc Mines and Primary  Metallurgical Works.
 International Lead and Zinc Study Group, London, U.K.

 ILZSG (1985). World Directory: Secondary Lead Plants. International  Lead and Zinc Study
 Group, London, U.K.

 Jenkins, J.J., Cooper, R.J. and Curtis, A.S. (1990). Comparison of pendimethalin airborne and
 dislodgeable residues following application to turfgrass. In: Long-Range Transport of Pesticides.
 Kurtz, D.A. (ed.),  pp. 29-46. Lewis Publishers, Chelsea, Michigan.

 Johnson, N.D., Scholtz, M.T., Cassaday, V. and Davidson, K. (1990). MOE Toxic Chemical
 Emission Inventory for  Ontario  and Eastern North America.  ORTECH International, Draft
 Report No.P.89 -  50 - 5429/OG, Mississauga, Canada.

 Johnson, N.D., Scholtz,  M.T., Cassaday, V.,  Davidson, K. and Ord, D.  (1992).  MOE  Toxic
 Chemical Emission Inventory for Ontario and Eastern  North America.  Final report to the Air
 Resources Branch, Ontario Ministry of the Environment,  by Ortech International  Report No
P92-T61 -5429/OG; 402 pp.

Jury,  W.A.,  Spencer, W.F. and Farmer, W.J.  (1983).  Behavior  assessment model for trace
 organics in soil. I. model description. J. Environ. Qital.  12, 558-564.

                                         133

-------
Jury, W.A.,  Farmer,  W J  and Spencer, W.F  (1984a).  Behavior assessment model for trace
organics in soil. II  chemical classification and parameter sensitivity,  J. Environ. Qua!  13, 567-
572.

Jury, W.A.,  Spencer,  WF  and Farmer, W.J. (1984b).  Behavior assessment model for trace
organics in soil. III. application of screening model. J. Environ. Qual.  13, 573-579.

Jury, W.A.,  Spencer,  W.F. and Farmer, W.J. (1984c).  Behavior assessment model for trace
organics in soil. IV review of experimental evidence, J. Environ. Qual.  13, 580-586.

Kamens, R., Bell, D.,  Dietrich, A., Perry, J., Goodman, R.,  Claxton, L. and Tejada, S.  (1985).
Mutagenic transformation  of dilute  wood  smoke systems  in the presence of 63  and NO2'
analysis of selected HPLC  fractions from wood smoke particle extracts. Environ. Sci. Technol.
19, 63-69.

Kamens,  R., Guo, Z., Fulcher,  J. and Bell, D.  (1988).   Influence  of humidity, sunlight, and
temperature on the daytime decay of PAHs on atmospheric soot particles. Environ. Sci. Technol.
22. 103-108.

Kamens, R.M., Karam, H., Guo, J.,  Perry,  J.M. and Stockburger, L. (1989).  The behavior of
oxygenated polycyclic aromatic hydrocarbons on atmospheric  soot particles.   Environ.  Sci.
Technol. 23,801-806.

Kamens, R.M., Guo, J., Guo, Z. and McDow, S.R. (1990). Atmos. Environ. 24A, 1161-1173.

Keeler, G.J., Stevens, R. and Olmez,  I (1992) Sources of toxic air pollutants measured the Lake
Michigan Urban Air Toxics Study (LMUATS). Presented at: The International Symposium on
Measurement of Toxic and Related Air Pollutants, Raleigh,  N.C.

Keeler, G.J., Japar,  S.M., Brachaczek, W.W.,  Gorse,  R.A., Jr. and Pierson, W.R. (1990)
The sources of aerosol elemental carbon at Allegheny Mountain. Atmos. Environ. 24A, 2795-
2805.

Keeler, G.J., Spengler, J.D., Koutrakis, P., Allen,  G.A., Raizenne, M. and Stern, B. (1990)
Transported acidic aerosols measured in southern Ontario. Amos. Environ. 24A, 2935-2950.

Keeler, G.J. and Samson, P.J. (1989)  On  the spatial representativeness of trace element
ratios.  Environ. Sci. Technol. 23, 1358-1364.

Keeler, G.J. (1987) A hybrid approach for source apportionment of atmospheric pollutants in the
Northeastern United States. Ph.  D. Thesis, The University of Michigan, Ann Arbor.

Kenski, D.M., Wadden, R.A., Scheff, P.A. and Lonneman, W.A. (1991).  Receptor modeling of
VOCs  in Chicago, Beaumont, and Detroit. Presented at the 84th  Annual Meeting of the  Air and
Waste  Management Assoc., Vancouver, B.C., June, paper no. 91-82.3.
                                          134

-------
 Khalil, M.A.K. and Rasmussen, R.A. (1984). Statistical analysis of trace gases in Arctic haze.
 Geophys. Res. Leu. 11, 437-40.

 Kjeller,  L-O., Kulp, S-E.,  Bergek,  S., Bostrom, M, Bergqvist, P-A., Rappe,  C, Jonsson,  B.,
 deWit, C., Jansson, B. and Olsson, M (1990).  Levels and possible sources of PCDDs/PCDFs in
 sediment and pike samples from Swedish lakes and rivers, part I. Chemosphere 20, 1489-1496

 Kleindienst,  T.E., Shepson, P.B., Edney, E.G., Claxton, L.D. and Cupitt, L.T. (1986).  Wood
 smoke:  measurement of the mutagenic activities of its gas- and particulate-phase photooxidation
 products. Environ. Sci. Technol. 20, 493-501.

 Knap, A.H.  and Binkley, K.S. (1991).  Chlorinated organic compounds in the troposphere over
 the western North Atlantic Ocean measured by aircraft. Atmos. Environ. 25 A,  1507-1516.

 Koester,  C.J. and Hites, R.A. (1992). Wet and dry deposition pf chlorinated dioxins and fiirans.
 Environ.  Sci. Technol. 26, 1375-1382.

 Komarov, B. (1980). The Destruction of Nature in the Soviet Union. Sharpe Inc., New York,
 NY.

 Krieger,  M.S.  and Hites, R.A.  (1992).   Diffusion denuder  for the collection of semivolatile
 organic compounds.  Environ. Sci. Technol. 26, 1551-1555.

 Kurtz, D.A. and Atlas, E.L. (1990).  Distribution of hexachlorocyclo-hexanes in the Pacific Ocean
 basin, air and water. In: Long-Range Transport of Pesticides. Kurtz,  D.A. (ed.), pp.  143-160.
 Lewis Publishers, Chelsea, Michigan.

 LaFlamme,  R.E.  and  Hites, R.A.  (1978).   The global distribution of polycyclic  aromatic
 hydrocarbons in recent sediments. Geochim. Cosmochim. Acta 42, 298-303.

 Lamb, B., Guenther, A., Gay, D. and Westberg, H.  (1987).   A national inventory of biogenic
 hydrocarbon emissions.  Atmos. Environ. 21, 1695-1705.

 Lane, D.A., Johnson, N.D., Barton,  S.C., Thomas, G.C., Schroeder, W.H. and Strachan, W.M.J.
 (1988).  Development and evaluation of a novel gas-and-particle sampler for semivolatile organic
 compounds in ambient air.  Environ.  Sci. Technol. 22, 941-947.

Lane, D.A., Johnson, N.D., Hanley, M-J.J., Schroeder, W.H.  and Ord,  D.T.  (1992a).  Gas and
 particle phase concentrations of alpha-hexachlorocyclohexane, gamma-hexachlorocyclohexane,
and hexachlorobenzene in Ontario air. Environ. Sci. Technol. 26, 126-133.

Lane, D.A., Schroeder, W.H. and Johnson,  N.D. (1992b). On  the spatial and temporal variations
in  atmospheric concentrations  of hexachlorobenzene and hexachlorocyclohexane isomers  at
 several locations in the province of Ontario, Canada. Atmos. Environ. 26A, 31-42.

Larsson, P and Okla, L. (1989).  Atmospheric transport of chlorinated hydrocarbons to  Sweden
 in 1985 compared to 1973. Atmos. Environ. 23, 1699-1711.

                                         135

-------
Larimer, J.S., Hoffman, E.J., Hoffman, G., Fasching, J.L.  and Quinn,  J.G. (1990)   sources of
petroleum hydrocarbons in urban runoff.  Water, Air, Soil Pollut. 52, 1-21.

Lewis, R.G., Martin, B.E., Sgontz, D.L.  and Howes, Jr., I.E. (1985). Measurement of fugitive
atmospheric emissions of polychJorinated biphenyls from hazardous waste landfills. Environ. Sci.
Technol.  19,986-991.

Lewis, R.G., Coutant,  R.W., Callahan,  P.J. and Chuang,  J.C. (1988)  Phase distribution of
polynuclear aromatic hydrocarbons in ambient air.  Presented at : 196th National Meeting of the
American Chemical Society, Los Angeles, CA.

Li, S.-M. and Winchester, J.W. (1990). Haze and  other aerosol components in  later winter
Arctic Alaska, 1986.  J. Geophys. Res. 95, D2, 1797-810.

Ligocki, M.P and Pankow, J.F. (1985). Trace organic compounds in rain III. particle scavenging
of neutral organic compounds. Atmos. Environ. 19, 1619-1626.

Lindstrom, G., Rappe, C. and Sjostrom, M. (1989). Multivariate data analysis applied in studying
the distribution of PCDDs and PCDFs in human milk.   Chemosphere 19, 745-750.

Lowenthal, D.H. and Rahn, K.A. (1985).  Regional sources of pollution aerosol at Barrow,
Alaska during winter 1979 -  80 as deduced from elemental tracers. Atmos. Environ.  19, 2011-
24.

Mackay,  D., Paterson, S. and Schroeder, W.H. (1986). Model describing the rates of transfer
processes or organic chemicals between atmosphere and water.  Environ. Sci.  Technol. 20, 810-
816.

Mackay,  D. (1992) Source Investigation for Lake Superior,  Report to the Virtual Elimination
Task Force, International Joint Commision, Windsor,  Ontario.

Majewski, M.S., Glotfelty, D.E. and Seiber, J.N. (1989). A comparison of the aerodynamic and
the theoretical shape profile methods for measuring pesticide evaporation from soil.  Atmos.
Environ.  23, 929-928.

Majewski, M.S., Glotfelty, D.E., Tha Paw U.K. and  Seiber, J.N. (1990).  A field comparison of
several methods  for measuring pesticide evaporation  rates from soil. Environ. Sci. Technol. 24,
1490-1497.

Manchester-Neesvig,  J.B.  and  Andren,  A.W. (1989).  Seasonal variation in the  atmospheric
concentration of polychlorinated biphenyl congeners.  Environ. Sci. Technol. 23, 1138-1148.

Masclet.  P., Movier, G. and Nikolaou, K.  (1986). Relative decay index and sources of polycyclic
aromatic hydrocarbons  Atmos. Environ. 20, 439-446.

MBB (1986). Non-ferrous Metal Works of the World. Metal Bulletin Books Ltd.,  4th Edition.
Park House, Surrey, U.K.

                                          136

-------
 McConnell, L.L. (1992)   Air-Water Exchange  of Chlorinated Organic Compounds.   Ph.D
 dissertation, Dept. of Chemistry and Biochemistry, University of South Carolina, Columbia, SC ,
 190pp.

 McConnell,  L.L.,   Gotham,  W.E.  and  Bidleman,  T.F   (1992).     Gas   exchange  of
 hexachlorocyclohexanes in the Great Lakes. Environ. Sci. Technol. (in press).

 McDow, S.R. and Huntzicker, J.J, (1990).  Vapor adsorption artifact in the sampling of organic
 aerosol:  face velocity effects. Atmos. Environ. 24A,  2563-2571.

 McVeety,  B.D.  and  Hites, R.A.  (1988).   Atmospheric  deposition of polycyclic aromatic
 hydrocarbons to water surfaces:  a mass balance approach. Atmos. Environ. 22, 511-536.

 Miller, D.F., Schorran,  D.E., Hoffer, I.E., Rogers, DP., White, W.H. and Macias, E.S. (1990)
 An analysis of regional haze using tracers of opportunity. J. Air Waste Mgmt. Assoc. 40, 757-
 761.

 Miller, M.S, Friedlander, S.K. and Hidy,  G.M (1972) A chemical element balance for Pasadena
 aerosol. J. Colloid Interface Sci.  39, 165-178.

 Mischke, T., Brunetti, K., Acosta, V., Weaver, D. and Brown, M.  (1985). Agricultural Sources of
 DDT Residues in California's Environment, Environmental Hazards Program, California Dept. of
 Food and Agriculture, Sacramento, California, 42 pp.

 Mobley,  J.D.,  (1992).  Procedures   for   verification   of  emission  inventories.   U.S.
 Environmental  Protection Agency, Final  Draft  Report,  Research  Triangle Park,  NC, 29
 September, 1992.

 Mueller, P.K.,  Hidy, G.M, Basket, R.L.,  Fung,  K.K., Henry,  R.C., Lavery, T.F., Warren,
 K.K. and Watson, J.G.  (1983). The Sulfate Regional Experiment: Report of Findings 3, Palo
 Alto, CA., Electric  Power Research Institute, EA-1901.

 Muller, R., Lach, G. and Parlar, H. (1988).  Vergleichende untersuchungen uber das vorkommen
 von toxaphenruckstanden. Chemosphere 17, 2289-2298.

 Muller, M.D., Schlabach, M. and Oehme, M. (1992). Fast and precise determination of alpha-
 HCH enantiomers in environmental samples using chiral high resolution  gas  chromatography.
Environ. Sci. Technol. 26, 566-569.

 Murphy, T.J. (1983).   In: Physical  Behavior of PCBs in the Great Lakes.   Mackay, D. and
 Paterson, S. (eds.). Ann Arbor Science, Ann Arbor,  Michigan.

Murphy, T.J., Formanski, L.J., Brownawell, B. and Meyer, J.A. (1985). Polychlorinated biphenyl
 emissions to the  atmosphere in  the Great  Lakes  region, municipal landfills and  incinerators.
Environ. Sci. Technol. 19, 942-946.

 NAPAP (1990). Acidic Deposition: State of Science and Technology, Report 8.

                                         137

-------
Nash, R.G. and Hill, B.D. (1990)  Modeling of pesticide volatilization and soil Celine under
controlled conditions. In: Long-Range Transport of Pesticides. Kurtz, D.A  (ed ).  pp.  17-28.
Lewis Publishers, Chelsea, Michigan.

Neme, C. (1990).  Electric utilities emissions of toxic air pollutants.  Center for Clean Air
Policy, Washington, D.C.

Ngabe, B. (1992).  Organic Contaminants in Air and Runoff Water. Ph.D. dissertation, Dept. of
Chemistry and Biochemistry, University of South Carolina, Columbia, SC, 157 pp.

Nishioka, M.G., Howard, C.C., Contos, D.A., Ball, L.M. and Lewtas, J. (1988).  Detection of
hydroxylated nitro aromatic and hydroxylated nitro polycyclic aromatic compounds in an ambient
air paniculate extract using bioassay directed fractionation. Environ. Sci. Technol. 22, 908-915.

Nishioka, M.G. and Lewtas, J. (1992).  Quantification of nitro- and hydroxylated nitro-aromatic
PAHs in selected ambient air daytime winter samples. Atmos. Environ. 26A, 2077-2087.

Nriagu, J.O. and Pacyna, J.M. (1988). Quantitative assessment of worldwide contamination of
air, water and soils by trace metals. Nature, 333, 134 - 139.

Pacyna, J.M. (1980). Coal-fired power plants as a source of environmental contamination by trace
metals and radionuclides. Habilitation thesis, Technical University of Wroclaw, Poland.

Pacyna, J.M., Ottar, B., Tomza, U. and Maenhaut. W. (1985). Long-range transport of trace
elements to Ny Alesund, Spitsbergen. Atmos.Environ. 19, 857 - 65.

Pacyna, J.M. and Oehme, M. (1988).  Long-range  transport of some organic compounds to the
Norwegian Arctic. Atmos. Environ. 22, 243 - 57.

Pacyna, J.M. and Munch, J. (1988).  Atmospheric Emissions of Arsenic, Cadmium, Mercury,
and  Zinc in Europe in  1982.  Norwegian Institute  for Air Research,  NILU  OR 17/88,
Lillestram, Norway.

Pacyna, J.M. and  Ottar,  B. (1989).   Origin  of  natural constituents in the Arctic aerosol.
Atmos. Environ.  23, 809-815.

Pacyna, J.M., Bartonova,  A.,  Cornille,  P. and Maenhaut,  W. (1989).  Modelling  of long-
range transport of trace elements. A case study. Atmos.Environ. 23, 107 -  114.

Pacyna, J.M. (1989). Technological parameters affecting atmospheric emissions of trace elements
from major anthropogenic  sources. In: Control  and Fate of  Atmospheric Trace Metals. J.M.
Pacyna and B. Ottar (eds,). Kluwer Academic Publishers, Dordrecht, the Netherlands.

Pacyna, J.M. and  Shaw, G.E. (1991). Arctic Haze and Air Pollution. Invited overview paper
at the Int.Conf. on the Role of the Polar  Regions in Global Change, Fairbanks, AL, 11-15
June, 1990.
                                          138

-------
 Pacyna, J.M. (1992). The Baltic sea Environmental Programme.  The Topical Area Study for
 Atmospheric Deposition of Pollutants.  Final Technical Report. Norwegian  Institute for  Air
 Research, NILU OR 46/92, Lillestrem, Norway.

 Pait, A.S., DeSouza, A.E. and Farrow, D.R.G. (1992).  Agricultural Pesticides in Coastal Areas:
 A National Summary,  Office of Ocean Resources  Conservation and  Assessment,  National
 Oceanic and Atmospheric Administration, Rockville, Maryland, 112 pp., (pre-publication copy).

 Pankow, J.F. (1987).  Review and comparative analysis of the theories on partitioning between
 the gas and aerosol paniculate phases in the atmosphere. Atmos. Environ. 21, 2275-2283.

 Pankow, J.F. and Bidleman, T.F.  (1992).  Interdependence of the slopes and intercepts from log-
 log correlations of measured gas-particle partitioning and vapor pressure.  I. Theory and analysis
 of available data. Atmos. Environ. 26 A, 1071-1080.

 Parmele, L.H.,  Lemon,  E.R.  and Taylor, A.W.  (1972).  Micrometeorological  measurement of
 pesticide vapor flux from bare soil and corn under field conditions.  Water, Air, Soil Pollut. 1,
 433-451.

 Paterson, S., Mackay, D., Tarn,  D. and  Shiu, W-Y. (1990).  Uptake of organic chemicals  by
 plants: a review of processes, correlations, and models.  Chemosphere 21, 297-331.

 Peters, J.A. (1981).  An environmental assessment of POM emissions from residential wood-fired
 stoves and fireplaces. In:  Proceedings of the Fifth International Symposium  on Polynuclear
 Aromatic  Hydrocarbons, Cooke, M.  and  Dennis,  A.  (eds.), pp.  571-582.  Battelle  Press,
 Columbus, Ohio.

 Piccot,  S.D., Watson,  J.J.  and  Jones, J.W. (1992).  A global inventory of  volatile organic
 compound emissions from anthropogenic sources. J. Geophys. Res. 97, 9897-9912.

 Pistikopoulos, P., Masclet, P  and Movier, G. (1990).   A receptor model adapted to reactive
 species:  PAH evaluation  of source  contributions  in  an  open urban site.   I.   paniculate
 compounds. Atmos. Environ. 24A, 1189-1197.

Pitea, D., Todeschini, R., Lasagni,  M., Moro, G., Bonati,  L.  and Chiesa,  G. (1989).   The
 combustion  of  minicipal solid wastes:   PCDD  and  PCDF  in MSW and  in emissions,  a
chemometric approach. Chemosphere 19, 751-757.

Podoll, R.T., Jaber, H.M. and Mill, T. (1986). Tetrachlorodibenzo-dioxin:  rates  of volatilization
and photolysis in the environment. Environ. Sci. Technol. 20, 490-492.

Prospero, J.M. and Seba, D.B. (1972).  Some additional measurements of pesticides  in the  lower
atmosphere of the northern equatorial Atlantic Ocean. Atmos. Environ. 6, 363-364.

Pryor, S.C. and Hoffer, T.E. (1992).  A case study of pollutant transport from Los Angeles to the
desert southwest. Atmos. Environ. 26A, 243-250.
                                         139

-------
Radian (1992).  Software Technical Specification for the Emission Factor Information System.
Draft Report. Radian Corporation, Research Triangle Park, NC.

Rahn,  K.A. and  Lowenthal, D.H. (1984). Elemental tracers  of  distant regional pollution
aerosols. Science.  23,  132-139.

Rahn,  K.A. and Lowenthal,  D.H. (1985).  Pollution aerosol in the Northeast: Northeastern-
Midwestern contributions. Science. 228, 275-284.

Ramdahl,  T., Alfheim, I.  and Bjorseth, A.  (1983), In: Mobile Source Emissions Including
Polvcvclic Organic Species. Rodia, D., Cooke, M.  and Haroz, R.K. (eds.),  pp. 277-298.  D.
Reidel Pub. Co., Dordrecht, The Netherlands.

Ramdahl,  T., Zielinska, B., Arey, J.,  Atkinson, R., Winer,  A.M. and  Pitts, Jr., J.N.  (1986).
Ubiquitous occurrence of 2-nitrofluor-anthene and 2-nitropyrene in air.  Nature 321, 425-427.

Rapaport, R.A.,  Urban, N.R., Capel.  P.D., Baker, I.E.,  Looney,  B.B., Eisenreich,  S.J. and
Gorham,  E.  (1985).   "New"  DDT  inputs  to North America - atmospheric  deposition.
Chemosphere 14, 1167-1173.

Rappe, C., Marklund, S., Kjeller, L-0. and Lindskog, A.  (1989). Long range transport of PCDDs
and PCDFs on airborne particles. Chemosphere 18, 1283-1290.

Rappe, C. (1992).   Sources of PCDDs and PCDFs --  introduction,  reactions, levels,  patterns,
profiles, and trends.  Chemosphere 25, 41-44.

Rice, C.P., Samson, P.J. and Noguchi, G.E. (1986).  Atmospheric transport of toxaphene to Lake
Michigan. Environ. Sci. Technol.  20, 1109-1116.

Risebrough, R.W., Huggett, R.J., Griffin, J.J.  and  Goldberg,  E.D. (1968).  Pesticides:  trans-
Atlantic movement in the northeast trades. Science 159,  1233-1236.

Rosen, H. and Hansen, A.D.A.  (1984). Role of combustion-generated carbon particles in the
absorption of solar radiation in the Arctic haze. Geophys. Res. Lett. 11, 461 - 4.

Ross, L.J., Nicosia, S., McChesney, M.M., Hefner, K.L., Gonzalez, D.A. and Seiber, J.N. (1990).
Volatilization, off-site deposition,  and dissipation of DCPA in the field.  Science 19, 715-722.

Saeger, M., et al.  (1989).  The 1985 NAPAP Emissions Inventory (Version 2): Development of
the Annual Data and Modelers' Tapes. U.S. Environmental Protection Agency,  Research Triangle
Park, NC, Report No. EPA-600/7-89-012a (NTIS PB88-225792).

Schantz, M.M., Benner, B.A., Chesler.  S.N., Koster. B.J., Hehn, K.E., Stone,  S.F., Kelly, W.R.,
Zeisler, R. and Wise,  S.A. (1990).  Preparation and analysis of a marine reference sediment
reference material for the determination of trace organic constituents.  Fres. Z.  Anal. Chem.  338,
501-514.
                                          140

-------
 Scheff, P.A. and Wadden, R.A. (1991)  Receptor modeling for volatile organic compounds. In:
 Receptor Modeling for Air Quality Management.  Hopke, P.K. (ed.), pp. 213-253.  Elsevier, New
 York.

 Scheff, P.A., Wadden,  R.A.,  Bates, B.A.  and Aronian, P.F  (1989). Source fingerprints  for
 receptor modeling of volatile organics. J. Air Pollut. Com. Assoc. 39, 469-478.

 Scholz, M.T. and Voldner, E.G. (1992).  Estimation of pesticide emissions to the air resulting
 from agricultural application. Presented at the Ninth World Clean Air Congress and Exhibition,
 Montreal, Aug. 30 - Sept. 4.

 Schonherr, J. and Reiderer, M. (1989). Foliar penetration and accumulation of organic chemicals
 in plant cuticles. Rev. Environ. Toxicol. Chem. 108, 1 -70.

 Schroeder, W.H. and Lane, D.A.  (1988).  The fate of toxic airborne pollutants.   Environ. Sci.
 Technol. 22, 240-246.

 Schroeder, W.H., Lindquist, O., Munthe, J.  and Xiao, Z. (1992) Volatilization of mercury from
 lake surfaces.  Sci. Total. Environ.  125,47-66.

 Schulz, D.E., Petrick, G. and Duinker, J.C. (1989). Complete characterization of polychlorinated
 biphenyl congeners  in commercial  Aroclor and  Clophen mixtures by  multidimensional gas
 chromatography - electron capture detection. Environ. Sci. Technol. 23,  852-859.

 Schwartz, T.R., Stalling, D.L. and Rice, C.L.  (1987).   Are polychlorinated biphenyl residues
 adequately  described by Aroclor mixture  equivalents?   Isomer-specific principal component
 analysis of such residues in fish and turtles. Environ. Sci.  Technol. 21, 72-76.

 Seiber,  J.N.,  Madden,  S.C., McChesney,  M.M. and  Winterlin,  W.L.  (1979).   Toxaphene
 dissipation from treated cotton field environments:  component residual behavior on leaves, and in
 air, soil, and sediments  determined by capillary gas chromatography. J.  Agric. Food Chem. 27,
 284-291.

 Semb, A. and Pacyna, J.M. (1988). Toxic Trace Elements  and  Chlorinated Hydrocarbons:
 Sources, Atmospheric Transport and Deposition. Nordic Council  of Ministers, NORD 1988:
 74, Copenhagen, Denmark.

 Sheridan, P.J. (1989).  Characterization of size segregated particles collected over Alaska and
 the Canadian high Arctic, AGASP-II  flights 204-206. Amos.Environ. 23, 237 - 86.

 Simoneit, B.R.T. (1989). Organic matter in the troposphere — V application of molecular marker
analysis to biogenic emissions into the troposphere  for source reconciliations.  J. Atmos. Chem. 8,
251-275.

 Slivon,  L.E.,  Gebhart I.E.,  Hayes,  T.L.,  Alford-Stevens, A.L.  and  Budde,  W.L.  (1985).
Automated  procedures  for  mass spectrometric determination  of polychlorinated  biphenyls as
isomer groups. Anal. Chem. 57, 2464-2469.

                                         141

-------
Smith, R.M.. O'Keefe. P W., Aldous. KM., Valente, H ,  Connor, S.P  and Donnelly, R.J. (1990).
Chlorinated dibenzofurans and dioxins in atmospheric samples from cities in New York.  Environ.
Sci. Technol. 24, 1502-1506.

Spencer, W.F. and Cliath, M.M. (1990).  Movement of pesticides from soil to atmosphere. In:
Lone-Ranee Transport  of Pesticides.  Kurtz, D.A.  (ed.),  pp.  1-16. Lewis Publishers, Chelsea,
Michigan.

Spies, R.B., Andresen, B.D. and Rice, Jr., D.W. (1987).  Benzthiazoles in estuarine sediments as
indicators of street runoff. Nature 327, 697-699.

Standley, L.J. and Simoneit, B.R.T. (1987).  Characterization of extractable plant wax, resin, and
thermally matured components in smoke particles from prescribed bums. Environ. Sci. Technol.
21, 163-169.

Stevens, R.K., Dzubay, T.G., Shaw, R.W., Jr., McClenny, W.A., Lewis, C.W. and Wilson,
W.E. (1980).  Characterization of the aerosol in the Great Smoky Mountains.  Environ. Sci.
Technol. 14,  1491-1498.

Stevens, R.K., Zweidinger, R.F., Lewis, C.W. and Dzubay, T.G. (1989). Volatile hydrocarbons as
mobile source tracer  species for receptor modeling. In: Man and his Ecosystem, Proceedings of
the Eighth World Clean Air Congress, Brasser,  L.J., Mulden, W.C.  (eds.), pp. 201-205. The
Hague, The Netherlands, Elsevier Science Publishers, Amsterdam.

Strachan, W.M.J. and Eisenreich, S.J. (1988). Mass Balancing of Toxic Chemicals in the Great
Lakes:  The Role of Atmospheric Deposition. Appendix  1 from the Proceedings of the Workshop
on the Estimation of Atmospheric Chemicals to the Great Lakes Basin,  Scarborough, Ontario,
Joint Water Quality Board/Science Advisory Board/International Air Quality Advisory Board of
the International Joint Commission, Windsor, Ontario, 113  pp.

Sturges, W.T. and Barrie. L.A. (1989). Stable lead isotopes in Arctic aerosols: Evidence for
the origin of arctic air pollution.  Atmos. Environ. 23, 2513 -  19.

Swackhamer, D.L., McVeety, B.D. and Kites, R.A. (1988).   Deposition and evaporation  of
polychlorobiphenyl congeners to and from Siskewit Lake, Isle  Royale, Lake Superior.  Environ.
Sci.  Technol. 22, 664-672.

Swackhamer, D.L. and Hites, R.A.  (1988).  Occurrence and bioaccumulation of organochlorine
compounds in fishes from Siskewit Lake, Isle Royale.  Environ. Sci.  Technol. 22, 543-548.

Swackhamer, D.L. and Eisenreich,  S.J. (1991).  Processing organic contaminants in lakes. In:
Organic Contaminants  in the Environment:  Sources.  Environmental Pathways, and Effects.
Jones, K.C. (ed.), pp. 33-86. Elsevier Science Publishers.

Lakes. Vol. Ill, Schmidtke. N.W (ed.), pp. 389-427. Lewis Publishers, Chelsea, Michigan.
                                          142

-------
 Swanson, S.E., Rappe, C., Malmstron, J. and Kringstad. K.P  (1988). Emissions of PCDDs and
 PCDFs from the pulp industry.  Chemosphere 17, 681 -691.

 Sweet, C.W. and Vermette, S.J. (1992).  Toxic volatile organic compounds in urban air in Illinois.
 Environ. Sci.  Technol. 26, 165-173.

 Takada, H., Onda, T. and Ngura, N. (1990). Determination of polycyclic aromatic hydrocarbons
 in urban street dusts and their source materials by capillary gas chromatography.  Environ. Sci.
 Technol. 24, 1179-1186.

 Takeoka, H., Ramesh, A., Iwata, H., Tanabe, S., Subramanian, A.N., Mohan, D., Magendran, A.
 and Tatsukawa, R. (1991).  Fate of the insecticide HCH in the tropical coast area of south India.
 Marine Pollut. Bull. 22, 290-297.

 Tan, Y.L. and Heit, M. (1981).  Biogenic and abiogenic polynuclear aromatic hydrocarbons in
 sediments from two Adirondack lakes. Geochim. Cosmochim. Ada 45, 2267-2279.

 Tan, Y.L., Quanci, J.F., Borys, R.D. and Quanci, M.J. (1992). Polycyclic aromatic hydrocarbons
 in smoke particles from wood and burning duff. Atmos. Environ. 26A, 1177-1181.

 Tiernan, T.O., Wagel,  D.J., Vanness,  G.F.,  Garrett, J.H., Soldi, J.G. and Harden, L.A. (1989).
 PCDD/PCDF in the ambient air of a metropolitan area in the U.S. Chemosphere 19, 541-546.

 Tsai, W., Cohen, Y., Sakugawa, H. and Kaplan, I.R.  (1991). Dynamic partitioning of semivolatile
 organics  in gas/particle/rain phases during rain scavenging.  Environ. Sci. Technol.  25, 2012-
 2023.

 Tuncel,  S.G.,  Olmez,  I.,  Parrington, J.R.,  Gordon,  G.E. and  Stevens,  R.K.  (1985).
 Composition  of the  fine  particle regional  sulfate  component  in  the  Shenandoah  Valley.
 Environ.  Sci.  Technol. 19, 529-537.

 Tuominen, J.,  Salomaa, S., Pyysalo, H., Skytta, E., Tikkanen,  L., Nurmela, T., Sorsa, J., Pohjola,
 V., Sauri, M. and Himberg,  K.  (1988). Polynuclear  aromatic compounds and genotoxicity in
 paniculate  and  vapor  phases of ambient air:  effect  of traffic,  season,  and meteorological
 conditions.  Environ. Sci. Technol. 22,  1228-1234.

 Tysklind, M. and Rappe, C.  (1991).  Photolytic transformation of polychlorinated dioxins and
 dibenzofarans in fly ash. Chemosphere 23, 1365-1375.

 Uliasz, M. and Pielke,  R.A. (1990). Application of the receptor oriented approach in mesoscale
 dispersion modeling. In: Proc.  of the 18th  International Technical Meeting  on Air Pollution
 Modelling (NATO/CCMS), Vancouver, Canada, May 1990.

 Van Vaeck,  L.  and  Van  Cauenberghe, K.  (1984).   Conversion  of polycyclic  aromatic
 hydrocarbons  on diesel paniculate matter upon  exposure to ppm levels  of ozone.   Atmos.
Environ.  18, 323-328.
                                         143

-------
Vial, J. Bruckman. L. and Buckert. J. (1992). Development of an emissions inventory of toxic
air  contaminants  for  the  Great  Lake  states. Presented at the  Inter. Conf.  .ind Course  on
Emission Inventory Issues, Durham, NC, October 19 - 22,  1992.

Villeneuve, J-P., Holm, E. and Cattini,  C. (1985).  Transfer of chlorinated hydrocarbons in the
food chain lichen-to-reindeer-to-man.  Chemosphere 14,  1651-1658.

Villeneuve, J-P., Fogelqvist, E. and Cattini, C. (1988).  Lichens as bioindicators for atmospheric
pollution by chlorinated hydrocarbons. Chemosphere 17, 399-403.

Voldner, E.G. and Smith L. (1989). Production, Usage,  and Emissions of Fourteen Priority Toxic
Chemicals.  Appendix 2 to the Proceedings of the Workshop on the Estimation of Atmospheric
Chemicals to the Great Lakes Basin,  Scarborough, Ontario, Joint Water Quality Board/Science
Advisory Board/International Air  Quality Advisory Board of the International Joint Commission,
Windsor, Ontario, ISBN 1-895085-20-9, 94 pp.

Voldner, E.C. and Schroeder, W.H. (1989). Modelling of atmospheric transport and deposition
of toxaphene into the Great Lakes ecosystem.  Atmos. Environ.  23, 1949-1961.

Welch, H.E., Muir, D.C.G.,  Billeck, B.N., Lockhart, W.L., Brunskill,  G.J., Kling, H.J., Olson,
M.P. and Lemoine, R.M. (1991). Brown snow:   a long range transport event  in the Canadian
Arctic. Environ. Sci. Technol. 25, 280-286.

Westerholm, R.N., Aimen, J., Ki,  H., Rannug, J.U., Egeback, C-E., Gragg, K. (1991).  Chemical
and biological characterization  of paniculate,  semivolatile, and gas-phase associated compounds
in diluted heavy-duty  diesel exhausts:  a comparison of three different semivolatile phase samplers.
Environ. Sci. Technol. 25, 332-338.

White, W.H., Macias, E.S., Miller, D.F., Schorran, D.E., Hoffer, T.E. and Rogers, D.P. (1990).
Regional transport of the urban workweek:  methylchloroform cycles  in the Nevada - Arizona
desert. Geophys. Res. Lett. 17,  1081-1084.

Wilber, G.G., Smith, L. and  Malanchuk, J.L.  (1992).  Emissions inventory of heavy metals and
hydrophobic orbanics in the Great Lakes basin. In: Fate of Pesticides and Chemicals  in the
Environment. Schnoor, J.L. (ed.),  pp. 27-50. J. Wiley and Sons, New York.

Willis, G.H., McDowell,  L.L.,  Southwick, L.M. and  Smith, S. (1985).  Toxaphene,  methyl
parathion, and fenvalerate disappearance from cotton foliage in the mid-south. J. Environ. Qual.
14,116-120.

Willis, G.H. and  McDowell, L.L. (1987).   Pesticide  persistence on  foliage.   Rev  Environ.
Toxicol. Chem.  100

Wind,  J.J.  and Burke, L. (1990).  Estimation  of Atmospheric Deposition of Toxics to the
Chesapeake   Bay     Watershed,    Prototype    Emissions    Inventory   and   Ptorotype
Concentration/Deposition  Model., Prepared  for the U.S.  Environmental Protection Agency.
Chesapeake Bay Liaison Office, by American Management Systems, Inc

                                           144

-------
 Wise,  S.A., Banner, B.A.,  Chesler,  S.N., Hilpert, L.R.,  Vogt,  C.R.  and May, W.E.  (1986)
 Characterization of the polycyclic aromatic hydrocarbons from two standard reference material air
 paniculate samples.  Anal. Chem. 58, 3067-3077.

 Wise, S.A.,  Hilpert, L.R., Rebbert, R.E.,  Sander, L.C., Schantz, M.M., Chesler, S.C. and May,
 W.E. (1988).   Standard reference  materials for the determination of polycyclic aromatic
 hydrocarbons. Fres. Z.  Anal. Chem. 332, 573-582.

 Wittlinger, R. and Ballschmiter, K. (1990).  C6-C14 organohalogens (alpha- and gamma-HCH,
 HCB, PCB,  4.4'-DDT, 4,4'-DDE, cis- and trans-chlordane,  trans-nonachlor, anisols) in the lower
 troposphere of the southern Indian Ocean.  Fres. Z. Anal. Chem. 336, 193-200.

 Woodrow, I.E., McChesney, M.M. and  Seiber,  J.N. (1990).   Modeling the volatilization  of
 pesticides and their distribution in the atmosphere. In: Long-Range  Transport of Pesticides.
 Kurtz, D.A.  (ed.), pp. 61-81. Lewis Publishers, Chelsea, Michigan.

 World Meteorological Organization (1989).  Atmospheric Inputs of Trace  Species to the World
 Oceans, Report  of the Group of Experts on Scientific Aspects of Marine Pollution (GESAMP),
 Report No. 38, IMO/FAO/UNESCOAVMOAVHO/IAEA/UN/UNEP.

 Yokley,  R.A.,  Garrison, A.A.,  Wehry,  E.L.  and  Mamantov, G.  (1986).   Photochemical
 transformation of pyrene and  benzo(a)pyrene vapor-deposited on eight coal stack ashes. Environ.
Sci. Technol. 20, 86-90.

Zhang, X. and McMurray, P.H (1991).  Theoretical aspects of evaporative losses of adsorbed or
absorbed species during  atmospheric aerosol sampling. Environ. Sci. Technol. 25, 456-459.

Zweidinger,  R.B.,  Stevens,  R.K.  and  Lewis,  C.W.  (1990).  Identification  of volatile
hydrocarbons as mobile source tracers for fine-particulate organics.   Environ.  Sci. Technol.
24, 538-542.
                                         145

-------