United States
             Environmental Protection
             Policy, Planning,
             And Evaluation
June 1989
The Potential Effects
Of Global Climate Change
On The United States
Appendix F
Air Quality
                                          Printed on Recycled Paper

     ^         ON THE UNITED STATES:

              APPENDIX F - AIR QUALITY
             Editors: Joel B. Smith and Dennis A. Tirpak
                  WASHINGTON, DC 20460

                       MAY 1989

                              TABLE OF CONTENTS


      J.E. Peimer, PJS. Cornell, D J. Wuebbles, and C.C. Covey

      Ralph E. Morris, Mike W. Gery, Mei-Kao Liu, Gary E. Moore, Christopher Daly,
      and Stanley M. Greenfield


The ecological and economic implications of the greenhouse effect have been the subject of discussion within
the scientific community for the past three decades.  In recent years, members of Congress have held hearings
on the greenhouse effect and  have begun to examine its implications for public policy. This interest was
accentuated during a series of hearings held in June 1986  by the Subcommittee on Pollution of the Senate
Environment and Public Works Committee.  Following the hearings, committee members sent a formal request
to the EPA Administrator, asking the Agency to undertake two studies on climate change due to the greenhouse

        One of the studies we are requesting should examine the potential health and environmental
        effects of climate change. This study should include, but not be limited to, the potential impacts
        on agriculture, forests, wetlands, human health, rivers, lakes, and estuaries, as well as other
        ecosystems and societal impacts.  This study should be designed to include original analyses, to
        identify and fill  in  where important research gaps exist,  and to solicit the opinions of
        knowledgeable people throughout  the country through a process of public hearings and
To meet this request, EPA produced the report entitled The Potential Effects of Global Climate Change on the
United States.  For that report, EPA commissioned fifty-five studies by academic and government scientists on
the potential effects of global climate change.  Each study was reviewed by at least two peer reviewers.  The
Effects Report summarizes the results of all of those studies.  The complete results of each study are contained
in Appendices A through J.

                               Appendix                       Subject

                                 A                           Water Resources
                                 B                           Sea Level Rise
                                 C                           Agriculture
                                 D                           Forests
                                 E                           Aquatic Resources
                                 F                           Air Quality
                                 G                           Health
                                 H                           Infrastructure
                                 I                            Variability
                                 J                            Policy

The goal of the Effects Report was to try to give a sense of the possible direction of changes from a global
warming as well as a sense of the magnitude. Specifically, we examined the following issues:

        o   sensitivities of systems to changes in climate (since we cannot predict regional climate change, we
            can only identify sensitivities to changes in climate factors)

        o   the range of effects under different warming scenarios

        o   regional differences among effects

        o   interactions among effects on a regional level


        o  national effects

        o  uncertainties

        o  policy implications

        o  research needs

The four regions chosen for the studies were California, the Great Lakes, the Southeast, and the Great Plains.
Many studies focused on impacts in a single region, while others examined potential impacts on a national scale.


The Effects Report studies used several scenarios to examine the sensitivities of various systems to changes in
climate. The scenarios-used are plausible sets of circumstances although none of them should be considered to
be predictions of regional climate change. The most common scenario used was the doubled CO2 scenario
(2XCO2X which examined the effects of climate under a doubling of atmospheric carbon dioxide concentrations.
This doubling is  estimated to raise average global temperatures by 1.5 to 45°C by the latter half of the 21st
century. Transient scenarios, which estimate how climate may change over time in response to a steady increase
in greenhouse gases, were also used. In addition, analog scenarios of past warm periods, such as the 1930s, were
used. •

The scenarios combined average monthly climate change estimates  for regional grid boxes  from General
Circulation Models (GCMs) with 1951-80 climate observations from sites in the respective grid boxes. GCMs
are dynamic models that simulate the physical processes of the atmosphere and oceans to estimate global climate
under different conditions, such as increasing concentrations of greenhouse gases (e.g., 2XCO2).

The scenarios and GCMs used in the studies have certain limitations.  The scenarios used for the studies assume
that temporal and spatial variability do not change from current conditions. The first of two major limitations
related to the GCMs is their low spatial resolution. GCMs use rather large grid boxes where climate is averaged
for the whole grid box, while in fact climate may be quite variable within a grid box. The second limitation is
the simplified way that GCMs treat physical factors such as clouds, oceans, albedo, and land surface hydrology.
Because of these limitations, GCMs often disagree with each other on estimates of regional climate change (as
well as the magnitude of global changes) and should not be considered to be predictions.

To obtain a range of scenarios, EPA asked the researchers to use output from the following GCMs:

        o  Goddard Institute for Space Studies (GISS)

        o  Geophysical Fluid Dynamics Laboratory (GFDL)

        o  Oregon State University (QSU)

Figure 1 shows the temperature change from current climate to a climate with a doubling of CO2 levels, as
modeled by the three  GCMs.  The figure includes the GCM estimates for the  four regions.  Precipitation
changes are shown in  Figure 2. Note the disagreement in the GCM estimates  concerning the direction of
change of regional and seasonal precipitation and the agreement concerning increasing temperatures.

Two transient scenarios from the GISS model were also used, and the average decadal temperature changes
are shown in Figure 3.

                       FIGURE 1. TEMPERATURE SCENARIOS
                   GCM Estimated Change in Temperature from 1xCO2 to 2xCO2
   Great Southeast Great  California United
   Lakes        Plains       States*
Great  Southeast Great California  United
Lakes         Plains        States*
Great  Southeast  Great California  United
Lakes         Plains        States*


                                                                               * Lower 48 States

               GCM Estimated Change in Precipitation from 1xC(>2 to 2xC(>2
> 0.8
3 0.6-
8 0.4-
t °-2'
s 0.0-
J -0.2
2 -0.4-

^J-| 1


I In
jjj ^^ *"""

i.u •

.0.6 .



1 1












Great  fiouthoast
Great California United
Plains       States*
Great  Southeast Great California United
Lakes         "lalns       States*
Great  Southeast Great California  United
Lakes         Plain*         States*


                                                                           * Lower 4» States


    O   3







EPA specified that researchers were to use three doubled CO, scenarios, two transient scenarios, and an analog
scenario in their studies. Many researchers, however, did not nave sufficient time or resources to use all of the
scenarios. EPA asked the researchers to run the scenarios in the following order, going as far through the list
as time and resources allowed:

       1. GISS doubled CO2

       2. GFDL doubled CO2

       3. GISS transient A

       4. OSU doubled CO2

       5. Analog (1930 to 1939)

       6. GISS transient B


The studies contained in these appendices appear in the form that the researchers submitted them to EPA.
These reports do not necessarily reflect the official position of the VS. Environmental Protection Agency.
Mention of trade names does not constitute an endorsement

                          J.E. Penner
                          PS. Cornell
                         D J. Wuebbles
                          C.C Covey
              Lawrence Lwermore National Laboratory
                     University of California
                      Livennore, CA 94550
            Interagency Agreement No. DW89932676-01-1


        The working tide of this report and the title that appears in the LLNL Scope of Work is Estimates of
 the Interactions Between Urban/Regional Air Quality and Changes in Tropospheric Chemistry and Climate.
 The current report title was chosen to emphasize the importance of interactions at all spatial scales between
 local/urban and global in understanding the effects of climate change from the perspective of atmospheric
 photochemical processes.
        'This work was performed under the auspices of the UJS. Department of Energy under Contract W-
7405-Eng-48, with the partial support of the U.S. Environmental Protection Agency under Interagency Agreement
DW89932676-01-0.  However, it does not necessarily reflect the Agency's views and no official endorsement
should be inferred from it.  We thank Stan Crotch for useful discussions concerning his research on GCM
intercomparison and allowing us to use some unpublished results. We would also like to thank John DiBari,
Matthew Hopper, and Steven Giles, who participated in the ISME-sponsored LLNL summer study program, for
their contributions.  Thanks  are also due to Peter Fmkelstein, Joseph Bufalini, Anne Thompson, and Ralph
Cicerone for their useful comments and careful reviews of the original manuscript


             CO,	   1-8
             CH:	   1-9
             N-O	  1-10
             Cf and fir-containing Industrial Compounds  	  1-10
             Other Trace Constituents of Importance With Possible Trends	  1-11
                   Nitrogen Oxides	  1-12
                   Carbon Monoxide	  1-12
                   Natural and Anthropogenic Higher Hydrocarbons  	  1-13
                   Global Background Stratospheric Aerosol  	  1-13
                   Global Background Tropospheric Aerosol	  1-13
       THE CLIMATE SYSTEM	  1-14
             Global Annual Average Radiative Forcing  	  1-14
             Additional Climate Variables	  1-16
             Direct Radiative Impacts	  1-19
             Climate Feedback Response to Direct Radiative Forcing Changes	  1-20
             Photochemical Forcings of Global Climate Change	  1-20
             Climate Responses to Composition Changes from the Historical Record	  1-22
             Global Climate Projections 	  1-23

             Formation of Oxidant 	  1-24
             Effect of Climate Change on Oxidant Formation 	  1-25
             Formation of Acidic Species	  1-35
             Effect of Climate Change on Acid Rain and Acid Deposition	  1-36

       TROPOSPHERIC OH 	  1-38
             Sources and Sinks of OH	  1-38
             Impact of Climate Change on Tropospheric OH	  1-39
             Climate Implications of OH Abundance Change	  1-41
       TROPOSPHERIC O,	  1-41
             Sources and Sinks of O3	  1-43
             Impacts of Climate Change on Tropospheric Ozone	  1-44
        COMPOSITION	  1-45

             Bridging Across Spatial Scales	  1-65





                                            CHAPTER 1


        Man's ability to impact the local environment has been evident for some time.  Since the pioneering
studies of Haagen-Smit in the 1950s to explain and understand the photochemical formation of smog, we have
been aware that fossil fuel emissions can impact local air quality (see Leighton, 1961, and references therein).
During the 1960s and 1970s, we became increasingly aware of the impacts of the long-range transport of fossil
fuel emissions.  These impacts manifest themselves on a more continental scale, the regional scale of acid rain
and acid deposition, for example (Albritton et aL, 1987).

        During the last two decades as welj, it has become increasingly clear that man's activities can also result
in changes to much larger areas, areas which are global in scope (see NRC, 1983, and NRC, 1984). On these
scales, not only may the chemistry of the atmosphere change, but tile climate of our planet may change as well.
The most obvious changes to climate are driven by increases in COg, a species whose lifetime is so long that
other than its impact on temperatures and therefore reaction rate constants, it has no obvious interaction with
chemistry.  However, more recently, trends in more minor species, for example, CH4, N2O, and halocarbons,
have been documented.  These species interact radiatively in the atmosphere to alter temperatures. They also
interact chemically in the atmosphere to  impact other  species whose lifetimes  are much shorter (e.g.,
tropospheric OH and stratospheric O3).

        With the increasing awareness of climate change and the possibility of global chemical changes to the
atmosphere, it becomes important to ask whether these global changes might impact regional chemistry in such
a way as to obviate or otherwise hinder some of the  programs already legislated or under contemplation and
designed to help reduce local and regional air pollution.  The issue is non-trivial because the effects of climate
change are manifold and wQl be felt differently in different  seasons and different areas around the world.
Furthermore, the atmospheric chemical system as a whole will impact regional chemistry by impacting boundary
and background concentrations leading to a need to understand not only regional chemical changes but global
ones as welL  Finally, global chemical changes have the potential to feed back and change climate so that the
effects cannot be fully known until the entire) climate/chemistry system is understood.

        This paper attempts to outline and estimate the interactions between climate change and atmospheric
chemistry that need investigation on both local and/or regional and global scales. The problem is not simply
one of estimating temperature change and running chemical models we already possess to get an answer.  The
ways that a changing climate influences atmospheric chemistry include not only temperature and precipitation
changes, but changes to atmospheric transport processes, changes to  the budgets of species with biological
sources (which respond to temperature and moisture changes), changes to vegetative cover which would alter
deposition rates, changes in the rate of export of pollutants from the urban/regional environment to the global
one, etc. Furthermore, changes hi atmospheric chemical composition will lead to climate change.  Figure 1.1
gives an overview of the climate/chemistry system and the interactions that it will be necessary to understand
in order to predict the future state of our planet.  The responses of these various subsystems may be highly
coupled to both each other and the climate change itself. For example, climate change has the potential to alter
water vapor concentrations. Changes in the concentration of water vapor would profoundly alter many species
concentrations by  impacting atmospheric OH concentrations, because OH acts as a scavenger for many
pollutants.  Changes to HJO will also impact tropospheric O3 concentrations.  Because O- absorbs infrared
radiation, these changes wffl feed back to alter the climate.  As another example, climate change also has the
potential to alter the biological sources and sinks of certain gases.  Because biological processes supply many of
the species of radiative importance in the troposphere, changes  in these sources wUl feed back to alter the
climate response.

                                         Trends In
                                        N,O, CFC'S
                                  Changes In climate:
                                      •   Temperature
                                      •   Precipitation
                                      •   Water vapor
                                      •   Winds
                                      •   Radiation fields
        Oxldant (O3)
        Acid deposition
Trends In:
   CH4, NMHC'S
            Figure 1.1.  Processes interconnecting climate and global and regional chemistry.


        Changes in global chemistry will be intimatetytied to changes in regional chemistry and changes on one
scale will affect changes on others. Just as the globally averaged temperature response to a given change in
composition will be manifested by different changes in various different regions, the "average" composition
change win be different in different regions. Determination of how this all ties together to feed back on the
climate  change is a formidable problem indeed  In this paper, we make  a first attempt at examining this
problem, by (1) a discussion of possible compositional changes on both regional and global scales from the
variety of changes  that may be introduced  by climate change, and potential climate changes resulting from
compositional change and (2) a discussion of needed research.

        In the following, we begin with a review of donate change and a discussion of the evidence that leads
us to believe that it is indeed an issue that win be facing man in the near future. Chapter 3 attempts to define
the ways in which climate change might impact local and regional oxidant formation and regional acid rain
formation. Chapter 4 outlines how climate change might impact global  chemistry, in so far as tropospheric OH
and tropospheric Oj are concerned. Chapter 5 discusses the variety of  models available today to address these
issues. A short review of the capabilities of general circulation models  to treat climate change is followed by a
discussion of our ability to treat "global" chemistry (Le., large-scale compositional changes generally treated by
a great  deal of spatial  averaging) and regional chemistry (Le., higher resolution  treatments with predictions
confined to specific locations).  A central problem that we face is that the grid resolution of general circulation
models is much larger than that typically needed in regional chemistry models.  We have not yet developed the
tools needed to translate the meteorological changes predicted by general circulation models to the changes that
wiU drive regional  chemistry changes. A related problem concerns the fact that no two general circulation
models agree as to what the regional meteorological changes related to  a change in climate might be. A similar
problem plagues our ability to link global chemistry models to regional chemistry models. In this case the
treatments available today not only differ in grid resolution but also in dimensionality—with "global" chemical
models being at most two dimensional, while the most  sophisticated and realistic regional models are three
dimensional.  Chapter 5 summarizes some of these issues and Chapter 6 gives a set of research areas that will
need substantial support and effort in order to better define how climate change might impact atmospheric
chemistry. These issues concern not only the issues of predicting changes across several spatial scales, but also
issues concerning our understanding of atmospheric budgets.  Because  of the spatial heterogeneity in the
troposphere, the budgets of many of the gases contributing to changes in the composition and climate are poorly
understood.  This is a second area where substantial effort is needed.


                                           CHAPTER 2


        This report attempts, in part, to provide a synopsis of our ability to define how certain kinds of changes
in the global climate may affect atmospheric chemistry in both a regional and global context. Thus, the global
climate changes considered here  include both secular trends  in average climate  statistics and changes in
moments of the distribution of climate variables related to the controlling variables of regional air  chemistry.
While understanding the basis of possible future change in quantities such as annual average global surface
temperature is a difficult problem receiving current attention, understanding the potential for changes in higher
moments of the  distributions of  climate variables is even more complex and much less  studied.  But
consideration of such changes remains important given the desire to characterize extreme events  as well as
average properties of regional air chemistry.

        The external driver for the  closely coupled terrestrial atmosphere/biosphere/ocean/land surface system
is the solar energy input and the details of its spectral distribution.  Intrinsic solar variability and the Earth's
orbital elements produce weather  and climate trends and variability on time scales ranging from seasonal to
cosmic, some systematic and some apparently random.  We are interested here in trends and variability in
addition to the celestial component, which can be ascribed, in many cases, to human population and industrial
and agricultural activities. The time scales of these changes are generational or political, ranging from decades
to centuries.  In this chapter, we will present an overview of relevant aspects of the climate system and the
reasons for expecting significant trends on these time scales.  An overview of the interactions of climate with
atmospheric trace constituents and chemistry are shown in Figure 2.1.


        From the standpoint of regional air chemistry, which involves questions of near surface abundances of
oxidants like ozone, wet and dry deposition of acidic species and transport and lifetimes of trace atmospheric
constituents, the climate variables of interest include moments of the distributions of temperature, precipitation,
clouds, and boundary layer meteorology. In the basic global sense, these variables are controlled by surface and
atmospheric temperature and water. The distributions of temperature and water are in turn controlled by solar
and long-wave radiation transfer involving the  surface and the  atmosphere.  The radiative properties of the
atmosphere, that is absorption and emission of energy in the range from the infrared to the ultraviolet by
atmospheric constituents, and the albedo of clouds and the Earth's surface are then the fundamental quantities
and processes of interest in understanding the climate. It happens that both the radiative properties of the
atmosphere, through the medium of the abundances of trace constituents, and the surface albedo are susceptible
to modification by modem man (Wang et al., 1986; Dickerson and Cicerone, 1986; Ramanathan et al., 1987).

        More or less well-mixed atmospheric constituents include those species listed in Table 2,1.  Of these
constituents, in order of relative importance according to current concentrations, HJO, COy O*, CH4, N2O and
potentially CFC-11 and CFC-12 are significant in atmospheric radiative transfer (Luther and Ellingson, 1985;
Ramanathan et aL,  1987).  Of these all but HjO and O3 have established abundance trends thought to be
directly traceable to emissions of human industrial and agricultural activity.  (Note: there are indications that the
distribution of O, is also changing-)  HJO and O,, in  addition to many other of the trace constituents listed
above, may also be increasing, but  trend detection is made difficult because of inadequate monitoring and/or
spatial and temporal heterogeneity of distributions.  Ozone and other photochemicaUy produced species are
subject to indirect modification of their abundances and distributions by trends in directly emitted source species.

        We summarize below and  in Table 22 current knowledge of global-scale observed abundance trends
and budgets of many Of these directly emitted atmospheric constituents. Atmospheric abundances of these trace
constituents are controlled by the rates and spatial distributions of emissions (production) and losses.

   401 MSTSW/TS-753
      (202) 260-3944
                                                                                Direct effects
                                                                 	^ Feedback loops
           Equilibrium Values
        Trends and Budgets
    Trace Gas Atmospheric
                                            Shortwave (VIS/UV)
  Longwave (IR) Absorption
       and Emission
Atmospheric Radiation
             Industrial Production
          Energy and Combustion
         Agriculture, Land use
        and Animal Husbandry
           and Geologic
          Trace Gas
           Wet and Dry
        Surface Deposition
     Stratospheric Oxidation
        and Photolysis
       Oxidation (OH)
   Trace Gas Loss
      Climate Feedbacks
     Climate System
                                   Ozone Column
                                     and Profile
                                 Stratospheric (NOy, CIOy,...)
    Tropospheric (HO ,...)
Photochemical Processes
    and Intermediates
                                  Figure 2.1. Climate/chemistry system.

                                           Table 2.1
       Trace Constituent                    	
                                                               ppm         ppb
Nitrogen (NO                               78
Oxygen (OO                                21
Water (H-D)                              0-2
Argon (Ar)                                  0.9
Carbon dioxide (COO                         0.034
Neon (Ne)                                                182
Helium (He)                                              52
Methane (CH4)                                           1.7
Krypton (Kr)                                              1.1
Hydrogen (HO                                            0.56
Nitrous oxide (N2O)                                                    300
Carbon monoxide (CO)                                                 90
Xenon (Xe)                                                           87
Ozone (O3) troposphere                                               20-400"
Ammonia (NHO                                                       <1
Ethane (CJL.)                                                         0.8
Methyl chloride (CH-Cl)                                                 0.6
Carbonyl sulfide (OCS)                                                  0.5
CFC-12 (CF2C10                                                       0.35
CFC-11 (CFCL)                                                        0.2
Formaldehyde (GHLO)                                                  0.2
Sulfur dioxide (SOO                                                     0.1
Nitrogen oxides (NO, NOO                                               0.05-100*
Other CFCs                                                           <0.15
Many other species                                                     <0.1
 Higher values representative of highly polluted conditions.

                      Table 2.2.  Established and Suspected Trace Species Trends
Trace Constituent
Annual Increase
        Importance for Climate



Tropospheric O,
Stratospheric O-




Tropospheric OH









Absorbs infrared radiation; affects
  stratospheric O3                         d, 1
Absorbs infrared radiation; affects
  tropospheric O. and OH; affects
  stratospheric H^O and O3                a, 1
Absorbs infrared radiation; affects
  stratospheric O3                         k, 1
Absorbs infrared radiation; affects
  stratospheric O3                         b, 1
Absorbs infrared radiation; affects
  stratospheric O3                         c, 1
Absorbs infrared radiation; affects
  stratospheric O3                           1
Absorbs infrared radiation; affects
  stratospheric O3                         e, 1
Absorbs infrared radiation; affects
  stratospheric O3                         f, 1
Absorbs uv and infrared radiation             i
Absorbs uv and infrared radiation            j
Involved in tropospheric O3 and OH        g, h
Involved in O3 and OH cycles
  and precursor of acidic nitrates
Involved in tropospheric O3 and OH
Produces CCN which can alter cloud
  albedo; forms SO2
Forms aerosol in stratosphere which
  alters albedo
May be major source of OCS in the
Major  precursor of acid rain
Major  natural source for SO2
Scavenger for many atmospheric pollutants,
  including CH4, CHgCClg, CHgFjCl
^ In lower troposphere.
  In upper stratosphere.

References:   a. Blake and Rowland (1988)
             b. Cunnold et al. (1983a)
             c. Cunnold et al. (1983b)
             d Keeling (1983)
             e. Khalil and Rasmussen (1981)
             f. Khalil and Rasmussen (1985a)
             g. Khalil and Rasmussen (1988)
             h. Risland and Levine (1985)
             L  Xiao et al.  (1986)
             j.  Watson (1988)
             k. Weiss (1981)
             1.  WMO(1985)


 Observed non-zero trends indicate an imbalance of production and losses related to changes in either or both
 processes. Changes in production or loss rates can be currently continuing or, particularly for long-lived species,
 the result of past events.  Uncertainties in the understanding of the budgets of most of the important source
 species makes the prediction of future abundance trends difficult, even for species with well-defined current
 abundance trends.  Although trends are  listed  for tropospheric  and stratospheric ozone based on recent
 evaluations of available data from ozonesondes, ground-based measurements, and satellite measurements, there
 are many uncertainties associated with the data that determine these trends, and therefore the trends for ozone
 should be regarded cautiously.


        Carbon dioxide is the single most important trace constituent from the standpoint of global climate
 change, as its  radiative  impact is  second only  to H^O and  its emission source has a  significant  direct
 anthropogenic fraction (Wang  et al., 1986; Ramanathan et al.,  1987).  It does not participate  directly in
 photochemistry and is well-mixed in the troposphere and stratosphere. Many recent primary sources  and
 reviews are available discussing the many complex elements of the global biogeochemical carbon cycle (DOE,
 198Sb; NRC, 1983). We  touch here only on the most salient features.

        Millenial scale long-term natural variability of the CO2 atmospheric abundance is closely related to
 climate variability. Records generated both by proxy techniques and recovery of ancient air trapped in glacial
 ice show the CO2 abundance cycling between 200 and 300 ppm as the climate cycled from glacial to interglacial
 conditions (Barnola et al., 1987; Genthon et al., 1987). Inferred rates of change at the end of the last glacial
 period are similar in magnitude to that currently observed.  However variability during the millennial prior to
 1800 (and the start of growth in large scale land use changes and industrial activity) appears smaller, in the
 range of  260 to  285  ppm.  In the period since 1800, both direct  and proxy methods, including archived
 spectroscopic plates, tree  rings, bubbles in ice and wet chemical measurements, show a secular trend of increase
 to the current level of 345 ppm (Gammon et al. in DOE, 1985b).

        A continuous record of CO2 abundance at Mauna Loa,  Hawaii, using a precise technique, is available
 starting in 1958 (Keeling, 1983). More recently, continuous surface abundance records from additional sites at
 a variety of latitudes  in both hemispheres provide  sufficient information to portray seasonal and latitudinal
 variations superimposed on the secular global abundance trend (WMO, 1985).  Currently, the secular trend in
 global surface abundance  averages about 1.4 ppm/year (3 x 1015 gC/year), or about 0.4%/year.  At Mauna Loa,
 the trend of increase has  been monotonic since the beginning of the record in 1958, with some indication that
 the rate of increase has slowed over the last decade (Keeling, 1983; WMO, 1985). A seasonal cycle in CO2
 abundance is observed at  northern middle and high latitudes peaking in the winter with a maximum amplitude
 of about 15 ppm  (3 x 1016 gC).  Interannual variability in the secular trend appears to be on the order of 1
 ppm, about the, same  magnitude as the current annual trend.

        Biogeochemical cycling of carbon among the atmosphere, biosphere, surface, oceans and sediments
 covers time scales from geologic to seasonal The largest  reservoir of carbon (> 99.999%) and the largest loop
in the cycle is the formation and weathering of carbonate rocks, involving tectonic processes and taking place on
 a geologic time scale.  Total burdens in the atmosphere, oceans,  biosphere and soils (exclusive of subfossil and
fossil deposits) are all of  the order 1 x 1018 gC, with perhaps a  500-year characteristic  time constant for CO,
loss from  this system. Annual respiration of the  oceans  and biosphere is of the order 1 x 1017  gC.  Annual
variability in the processes controlling biological productivity and  near surface ocean uptake, driven for example
by El Nino/Southern Oscillation (ENSO) events, can be of the order of 1 x 1015 gC.  The current anthropogenic
exogenous CO, production from land use changes and fossil fuel burning is about  6 x 1015 gC/year (WMO,

        It appears that the* CO^atmospheric abundance increase between 1800 and the early 20th century was
driven by changes in land use affecting the partitioning of carbon between biosphere and atmosphere. Direct
emission by fossil fuel combustion became the dominant term in CO2 increase  early in  this  century and



continuing to  the present.  A significant fraction,  perhaps  40%, of the anthropogenic  release of CO2 is
partitioned into the oceans by a net annual ocean uptake (WMO, 1985).

        Much effort has gone into a quantitative understanding of the historical CO, record and associated
human activities as a basis for  prediction of future trends. Quantitative models describing uptake and chemical
and biological processes in the ocean, the sensitivity to vegetation changes of carbon fluxes from terrestrial
ecosystems, the role of subfossil (e.g.,  peat) deposits, and fossil fuel consumption are required to derive the
small net terms as differences  of source and sink fluxes in the carbon budget. Significant uncertainties remain
in all of these areas.

        Predicting the future atmospheric abundance of CO,, and thus its radiative impact and contribution to
climate change, requires knowledge of the carbon cycle and how it partitions non-equilibrium CO2 emission
inputs.  The  ability to predict future  emissions is  based  on the  ability to predict future  trends in global
population and economic growth and energy consumption patterns as well as any possible  effects of future
responses to observed climate  change and alternative energy-producing technologies. Clearly uncertainties will
remain large.  A possible view of the future is an average annual increase in atmospheric abundance over the
next century of about 0.5%/year (NRC, 1983; Edmonds et al., 1984; Bolin et al., 1986; Mintzer, 1987). The CO2
abundance at the end of the next century would, in this case, be about twice the average abundance over the last
several thousand years.


        The modern abundance record for CH4 (Rinsland  et al., 1985b) starts in 1951 with information from
spectroscopic plates which can be precisely analyzed to give an abundance of 1.14 ppm in Europe at that time.
Sporadic  measurements  between  1965  and  1977  during  the  development  of  the high-precision  gas
chromatography/flame  ionization detection method  for CH4 show a clearly increasing abundance over this
period (Ehhalt, 1985).  Finally, continuous records with high  precision are available since  1979, with globally
distributed sampling yielding  information on the latitudinal  dependence of seasonal variability (Blake and
Rowland,  1986,1988; Khalil and Rasmussen, 1983a). The record has also been extended back over thousands
of years by analysis of air encapsulated in glacial bubbles.

        The glacial air bubble  record shows a constant CH4 abundance of about 0.7 ppm until several centuries
before present, followed by an observed increasing abundance (Craig and Chou, 1982; Khalil and Rasmussen,
1982; Rasmussen and Khalil, 1984; Stauffer et al., 1985, 1988).  Connecting the 1.14 ppm abundance in 1951
with a similarly derived spectroscopic value of 1.58 ppm in 1981 implies an average annual increase over 30 years
of 1.1% (Rinsland et al., 1988). The continuous record  between 1979 and  present shows an average annual
increase around 1% with significant interannual variability. A seasonal cycle is observed at most latitudes,
generally with a minimum  in  the summer and with  a poleward increase in amplitude.  The present average
abundance is about 1.7 ppm in the northern hemisphere and 1.6 ppm in the southern hemisphere.

        The atmospheric abundance of CH4 represents a balance primarily between biogenic sources and
photochemical sinks in the troposphere and stratosphere. The solubility of CH4 is sufficiently low that the ocean
is not an important reservoir. The current atmospheric burden is about 3.5 x 1015 gC (Wuebbles and Edmonds,
1988). The atmospheric lifetime resulting from reaction with  OH can be estimated by comparison to another
species, CRjCCL, also predominantly destroyed by OH and with a similar lifetime.  From knowledge of the
kinetic rate constants for reaction of OH with  CH4 and CHgCClg and the ALE observational data on the
lifetime of CRjCCLj (Prinn et aL, 1983), the CH. lifetime  must be about 11 years and the calculated annual
CH4 loss by reaction with OH is around 32 x 101* gC. Uptake by biological processes in dry soils contributes
perhaps 7-8 x 1012 gC/year to  the sink. Accumulation in the atmosphere is observed to be about 1% or 0.017
ppm/year, corresponding to 3.5 x 1013 gC. The global source term is thus constrained to be roughly 3.6 x 10U
gC/year (WMO, 1985; Wuebbles  and Edmonds, 1988).   The equilibrium atmospheric abundance for this
emission term and lifetime is around 2.0 ppm, so that the methane abundance should be expected to increase,
exclusive of compensating atmospheric changes or source reductions.


        Apportioning the total source term among the many known CH4 sources is a difficult problem (WMO,
1985; Wuebbles and Edmonds, 1988). Three major sources, enteric fermentation in ruminant animals, rice
paddies and natural wetland emissions, are thought to be of roughly equivalent magnitude (7-10 x 1013 gC/year).
Abiogenic emissions related to fossil fuels and biomass burning may be important (3-4 x 1013 gC/year).  The
sensitivity of emission to many variables in an ecosystem and to such things as specific strains of rice restricts
the potential to generalize individual field studies to the globe. It appears that perhaps 60% of the total source
term can be directly ascribed to human activities. Investigation of the 13C/12C ratio of atmospheric CH4 and
its trend, in comparison to the isotopic composition of known source emissions, may result in better definition
of relative source magnitudes in the future.

        The currently observed imbalance of the source and sink strengths in the atmospheric CH4 budget can
arise from either an increasing source strength or an increasing atmospheric lifetime (decreasing sink) or both.
Documentation of increases in the global cattle population and rice paddy cultivation area exist over the last
several decades.  The area of natural wetlands may be decreasing with land use development. Several recent
studies (Khalil and Rasmussen, 198Sb; Levine et al., 1985; Thompson and Kavanaugh,  1986) ascribe a fraction
(0.2-0.8%) of the CH4 increase to a decrease in global OH abundance.  Because of the many uncertainties in
source characterization, little confidence should be placed at present in predictions of  future CH4 abundance,
although it appears likely that levels above 2.0 ppm will be reached sometime in the next century.
        Nitrous oxide is a long-lived (lifetime approximately 150 years), well-mixed trace constituent with a
current global average abundance of about 310 ppb (WMO, 1985). Continuous records at several sites since
1978 show a 1 ppb deficit in the  southern hemisphere relative to the northern and a global trend of increase
of about 0.2%/year (Weiss, 1981; Khalil and Rasmussen, 1983b).  Trends measured at northern hemisphere
stations  are  larger than for the  southern  stations, suggesting  an increasing  interhemispheric disparity.
Measurements extending back to 1961 are consistent with the 0.2%/year trend (WMO, 1981).  Preliminary
results on bubbles of glacial air suggest an N2O abundance of 270-280 ppb in 1600 (Pearman et al., 1985).

        The sink for atmospheric N2O appears to be limited to photolysis in the stratosphere and reaction with
electronically excited atomic oxygen.  From the observed distribution and models of stratospheric photochemical
processes (WMO, 1985), the magnitude of the sink can be calculated to be 1 x 1013 gN/year.  The global
atmospheric burden is 1.5 x 1015 gN, implying a lifetime of 150 years. The observed global trend in abundance
corresponds to an annual increase of 2-3 x 10   gN, or as much as 25% of the implied source strength of about
1.4 x 1013 gN/year. Probably about half of this source is contributed by natural emissions from tropical and
subtropical forests, attributed to microbial nitrification processes in the soil.  Perhaps a third of the source can
be assigned to conversion of fuel nitrogen in combustion processes, dominated by coal combustion but with a
contribution from biomass burning.  The remaining sources are spread among ocean emissions, indicated by
observed N2O supersaturation of surface waters, and soil emissions from temperate and boreal forests, cultivated
soils and fertilized soils. The magnitude of the combustion source is similar to the  magnitude of the implied
source imbalance, leading  several authors  to suggest that the  current increase can be  attributed essentially
entirety to the anthropogenic combustion source (e.g., Weiss, 1981).

        The sink term is essentially fixed in the absence of large changes in stratospheric ozone abundance, so
that a continued positive trend is likely.  An equilibrium abundance around 400 ppb is implied by the current
source strength. However, continued increase in the fossil fuel combustion source term is also likely as use of
coal increases. Changes in tropical land uses may also have short- and long-term effects that may contribute
to or mitigate source strength increases.  These considerations  support projecting the observed trend of
0.2%/year into the future.

Cl- and Br-Containing Industrial  Compounds

        Sporadic surface and balloon measurements of the atmospheric abundance of CFC-11 and CFC-12
showed,  starting in the late 1960s,  that these industrially produced compounds were  accumulating in the



atmosphere and were essentially well-mixed in the troposphere. A program of precise, continuous, absolute
measurements at five sites sampling marine air at mid and low latitudes in both hemispheres commenced in
1978 (Cunnold et aL, 1983a,b; WMO, 1985). The average annual growth rates observed are around 5% for both
CFC-11 and CFC-12, with current levels of about 220 ppt for CFC-11 and 350 ppt for CFC-12. Most  emission
sources are in the northern hemisphere, producing an  interhemispheric gradient of about 10%, with higher
abundances in the northern hemisphere.  The current southern hemispheric trend is larger than that in the
northern hemisphere, however, tending to reduce the gradient.

        From the observed increase hi atmospheric burden and the history of emissions, atmospheric lifetimes
of around 75 years for CFC-11 and 111 years for CFC-12 can be derived (WMO, 1985). These are consistent
within error limits of model-derived values, considering loss only by stratospheric photolysis.  There is no
evidence that processes other than stratospheric photolysis are significant atmospheric sinks for CFC-11 and

        Although lifetimes for CFCs with respect to stratospheric photolysis will be  somewhat sensitive to
perturbations of the stratospheric ozone column, future CFC atmospheric abundances  can be calculated with
some confidence assuming knowledge of future global emission rates. The current burden of CFC-11 is about
5.5 x 1012 g, with current annual emissions of about 2.6 x 1011 g.  For CFC-12, the current burden is about 7.7
x 1012 g, global annual emissions  about 4.1 x 1011 g.  If the current levels of emission were to continue
indefinitely, CFC-11 and CFC-12 would eventually reach abundances of around 800 ppt and 2.6 ppb, respectively,
assuming no change hi lifetimes.

        Future emission rate projections for these compounds are, however, highly uncertain. Even neglecting
the possibility of future global emissions regulations connected to the interaction of CFCs with stratospheric
ozone, economic projections of future production and emission depend on estimates of uncertain growth in
quantities including population, gross national product per capita, and development of new technology, among
others (Hammitt et al., 1987).  Especially uncertain are projections in new markets and developing (Third
World) economies.  With the announcement of the 1987 Montreal Protocol  designed to limit future emissions
of CFC-11 and CFC-12 in particular, continuing compounded growth of emissions seems unlikely. But achieving
emission rates in equilibrium with current abundances would require substantial emission reductions, to 25% of
the current emission rate for CFC-11 and  only about l/8th  of the current rate for CFC-12.  Consequently,
without substantial cuts in emissions, some continued increase in atmospheric abundance is inevitable.  The
atmospheric concentrations of several brominated halocarbons, namely CF2ClBr and CF^Br, which are used in
fire extinguishing, are increasing rapidly (10-30%/year). Although then* concentrations are small (~ 1 pptv), they
could have potentially significant effects on atmospheric chemistry and climate in the next century (WMO, 1985;
Wuebbles and Edmonds, 1988).

Other Trace Constituents of Importance With Possible Trends

        There are many other surface-emitted  trace atmospheric constituents with direct or indirect potential
climatic impacts for which good global records  and firm trend indications are lacking at present. For many of
these constituents, the difficulties in determining global  average abundances and secular trends are related to
short atmospheric lifetimes and the implied spatial and temporal heterogeneity of their distributions. We will
briefly discuss nitrogen oxides (NO  = NO+NO,), carbon monoxide (CO), higher hydrocarbons and global
background tropospheric aerosol NO^ CO, and NMHC (non-methane hydrocarbons) affect climate processes
indirectly through their participation in tropospheric photochemistry affecting both OH and O- (see Chapter 4).
Tropospheric O- has important direct effects on radiation, and OH is predominantly responsible for controlling
the lifetimes  of oxidizable species, including  CH4 and hydrogen-containing halocarbons (e.g.,  CHXCLj).
Tropospheric aerosol can  scatter solar  radiation  and  absorb both solar  and longwave energy.  The  net
surface-troposphere effect of increased aerosol can be either warming or cooling, depending on the albedo of
the underlying surface and  other factors.


        Nitrogen Oxides

        The nitrogen oxides, NO and  NO2 as natural constituents play several roles in tropospheric and
stratospheric chemistry.  They are involved in tropospheric photochemistry producing both ozone and hydroxyl
radicals, which in turn act as major atmospheric oridants.  NO and NO2 are observed to be globally distributed
trace tropospheric constituents, with abundances ranging from 10 ppt in remote clean marine air to the ppb level
at clean continental sites.  There is no direct observational evidence for a secular trend in the global average
tropospheric abundance, but there are many reasons to infer an increased level relative to levels earlier in this

        The tropospheric odd-nitrogen sources total perhaps 5 x 1013 gN/year with a large uncertainty (Logan,
1983).  Fossil fuel combustion accounts for about 40% or 2 x 1013 gN/year, and biomass burning, lightning, and
soil emissions for perhaps 20% each. There is a small contribution from the stratosphere, where odd-nitrogen
is produced in the oxidation of N2O.  The loss processes include dry deposition and photochemical oxidation
to HNO, followed by rainout.  Both are highly variable in time and space.  The extent of correlation of the
source emission with the tropospheric sink is important hi determining the NO+NO2 lifetime, which can vary
from a day to a week. Specifying a global average lifetime and thus a global burden or average abundance is
therefore difficult.

        The inference of a positive global abundance  trend follows  from the importance  of fossil fuel
combustion as a source of nitrogen oxides and the evidence that the increase in fuel  nitrogen emissions is
responsible for the observed NgO increase.  Further, there is evidence for nitric acid increases in precipitation
from ice cores from 1895 to 1978 (Neftel et al., 1985).  Likewise, estimates of global NOX emissions from fossil
fuel usage indicate growth from 1966 to 1980 (Hameed and Dignon, 1988). However, even a rough estimate of
the future trend cannot be made based on current knowledge. There is accumulating evidence, which will be
discussed below, for the importance of PAN and other organic nitrates as longer-lived reservoirs for the nitrogen
oxides  with the ability to transport odd nitrogen on global scales in the mid-troposphere.

        Carbon Monoxide

        The reaction of carbon  monoxide (CO) with the hydroxyl radical (OH) serves as the dominant sink
process for CO as well as a major conversion pathway of OH to other forms of HO». Abundance records for
CO are currently being generated at several points around the  globe (WMO, 1985).  The current northern
hemispheric average surface abundance is  around 150 ppb, about twice the average southern hemispheric
abundance.  A seasonal cycle with a winter maximum is evident at or below the 25% level, with higher
amplitude at higher latitudes. Substantial variability, related to the spatial inhomogeneity of sources and the 1-
to 4-month atmospheric lifetime,  is also observed. This variability and the generally short duration of continuous
records makes the detection of a secular trend difficult.  Early reported results range from no trend different
from zero, discernible at high southern latitudes, to 5% per year observed at Cape Meares, Oregon, between
1979 and 1982 (WMO, 1985). Analysis  of spectroscopic plates from 1951 and  1981 shows an average annual
increase of about 2% over Europe between these times (Rinsland and Levine, 1985).  Recent measurements at
six sites indicate a global trend of about  1%/year (Khalil and Rasmussen, 1988).

        The global atmospheric burden of CO is about 2 x 1014 gC. The global sink can be estimated at about
1 x 1015 gC/year, fixing the global source at about the same value. About 85% of the sink term is the reaction
of CO with OH forming CO, and HOj, with the remainder attributable  to soil uptake. From studies of
combustion processes and  analysis of the global extent of fossil fuel combustion and biomass burning, direct
anthropogenic emissions of CO comprise about half of the total source term (Logan et aL, 1981; Kavanaugh,
1987; WMO, 1985).  The remaining sources are natural burning and the atmospheric oxidation of CH4, natural
tropical and temperate higher hydrocarbons, and anthropogenic hydrocarbons.

        The importance of anthropogenic combustion as a source of CO and the larger abundance observed
in the northern hemisphere, where the sources are concentrated, suggest strongly that  CO levels are higher
now than earlier in this century and before.  The trends derived from the recent data records indicate that the



CO abundance is continuing to increase. Predicting the future behavior of both the source and sink terms is,
however, difficult at present. In considering the sink term, the tropospheric OH abundance is affected by many
other species, including HgO, CO, O3, NOX, and CH4, in a complicated photochemical interdependence.  The
prediction of a future OH  trend is problematical. Because the short lifetime of CO keeps sources and sinks
in approximate balance, a continued increase in CO abundance will depend not just on continued emission by
combustion but on a continuing increase in the rate of combustion emissions of the sort that produce CO.

        Natural and Anthropogenic Higher Hydrocarbons

        A very large number of Cn (n=2-8+) hydrocarbons can be found throughout the global troposphere
in abundances ranging from 10's of ppt to the ppb level and higher (NRC, 1984). Values near source regions,
such as forests and urban  areas, are  at the high end of this range  and can be significantly higher locally.
Lifetimes range from hours to several days with respect to oxidation by OH.  The paucity of observations and
the large variability in abundances do not support global average characterizations at present.  The importance
of the anthropogenic component  of the source term varies with individual hydrocarbons but can be significant
for the lighter species. Because they are sources of CO  and are consumed reactants in the NOX/HOX/O3
photochemical system, future changes in higher hydrocarbon abundances  must be considered in predicting the
climate interaction with atmospheric composition and chemistry.

        Global Background Stratospheric Aerosol

        The chief aerosol found in the stratosphere is composed of condensed sulfuric acid vapor (HjSO^) and
water vapor.  The globally distributed unperturbed background aerosol found in the lower stratosphere, termed
the Junge layer, has a visible optical depth less than 0.01 (WMO, 1985).  Its source is thought to be oxidation
of surface-emitted carbonyi sulfide, OCS (NRC,  1984; WMO, 1985).  OCS is present at about 500 ppt in the
troposphere, has a long tropospheric lifetime, and is mixed efficiently into  the stratosphere.  OCS oxidation can
be initiated by photolysis  and by  OH, producing SOX species and eventually sulfuric acid.  The extremely low
vapor pressure of sulfuric acid, about 60 ppt at 215 K, allows condensation into aerosol at very low abundance
levels.  Other surface-emitted sulfur-containing species, for example SOg, DMS, and CSg, do not persist long
enough in the troposphere to be transported to the stratosphere.

        A state of unperturbed background stratospheric aerosol may be relatively rare, however, as frequent
volcanic eruptions inject significant quantities of SO, directly into the lower and mid stratosphere. Recent major
eruptions include Agung in 1963 and El Chichon in 1982. The subsequent sulfuric acid aerosol clouds can over
a period of months be distributed globally at optical densities that overwhelm the natural background aerosol.
The stratosphere's relaxation to background optical density has a characteristic time on the order of years, so
that the average state of stratospheric aerosol could be considered as partially perturbed by volcanic emissions.

        The  Antarctic stratospheric aerosol implicated in the  spring ozone hole phenomenon is  probably
different in composition from the global sulfuric acid aerosol  In addition to water, it probably contains a
significant proportion of nitric acid (HNO3) and may incorporate hydrochloric acid (HC1) as well  The chief
Antarctic aerosol precursors are stratospheric water vapor and gas phase NOX and Cl-containing species.  Its
formation is related to the very low stratospheric temperatures reached in the Antarctic winter.

        Given the importance of episodic volcanic injections to the global stratospheric aerosol burden, it is
unlikely that human activities could introduce an observable trend  into the stratospheric aerosol abundance.
In the case of Antarctic aerosol, Blake  and Rowland (1988) have made the suggestion that the secular trend in
methane may have increased formation through the increase in stratospheric water vapor formed by methane
oxidation in the stratosphere.

        Global Background Tropospheric Aerosol

        Background tropospheric aerosol is  variable in its composition as well as its spatial and temporal
distribution (Wang et aL, 1986). The major types of aerosol composition are soot, sulfate, maritime, crustal, and



        While quantitative predictions of trends are not currently available, development of techniques for such
prediction is being pursued   Aerosol physicochemical processes, important at the regional level, include
coagulation, growth/shrinkage by condensation/evaporation, nucleation of fresh aerosol at some threshold size,
thermodynamic equilibrium between aerosol and the gas phase, cloud processing of aerosol, and wet and dry
removal processes. Model representations of the microphysics of aerosol formation and growth and the aerosol
size distribution have been discussed by Suck and Brock (1979) and Seigneur et aL (1986).


Global Annual Average Radiative Forcing

        The most fundamental planetary climate variable could be considered to be the planetary annual average
surface temperature. Considering an atmosphere transparent to radiation and hi the absence of a significant
internal planetary heat  source, this temperature is controlled by the solar energy input and the surface albedo
of the planet. The surface temperature adjusts so that the planetary emission of heat to space balances the solar
energy input.  Annually and globally averaged, about 340 W m~2 of solar energy strikes the earth and 30% is
reflected back to space, leaving about 240 W m~2 to be absorbed.  Of the 100 W m'2 reflected back to space,
clouds account for about two-thirds of the loss and the surface for about an eighth, with atmospheric Rayleigh
scattering reflecting the rest. Were the atmosphere transparent to infrared radiation, the surface temperature
would simply adjust to balance the absorbed insolation and would be about 25S°K.

        Although the total absorbed and emitted energy of the surface-atmosphere system balance at 240 W
m  , the fluxes of processes transferring energy within the system, between the surface and atmosphere and
within the atmosphere itself, can exceed the net 240 W m'2 net solar input Actually, about 70 of the 240 W m"
2 in absorbed insolation are absorbed by trace constituents, aerosol, and clouds in the atmosphere (Ramanathan
et aL, 1987).  In addition, since the earth's observed average surface temperature is 288°K, about 390 W m*2 is
radiated upward from the surface, so that trace atmospheric constituents must be responsible for trapping the
excess surface-emitted  infrared radiation.  The earth-atmosphere system must be in balance with the solar
energy input, so that the effective radiating temperature of the earth and its atmosphere must be lower than the
surface temperature. This is accomplished by the climate system in two steps; first, the atmosphere is not
transparent  to outgoing infrared radiation by virtue  of HgO, CO^ clouds, and other trace constituents and
second, the temperature of the atmosphere decreases with altitude in the troposphere.

        The surface, then,  absorbs about 170 W m'2 of solar insolation and about 320 W m"2 of infrared energy
emitted by the atmosphere. To balance the 490 W m'2 total input, the earth's surface releases about  100 W
m  in the form of sensible and latent heat in addition to the 390 W m'2 emission of infrared radiation.  About
20 W m  of the surface radiation is emitted directly to space, primarily in the 7-13 p  wavelength "window"
between the HgO vapor and CO2 absorption bands.  The atmosphere's energy input totals 540 W m'2, 70 in
direct solar, 370 in surface infrared emissions, and  100 in sensible and latent heat  Of the 540 W m'2 in
atmospheric infrared emission, 320 is emitted downward to the earth's surface and 220 is emitted to space, with
about 90 emitted by clouds and the remaining 130 W m'2 by trace constituents.  Water vapor, clouds, and CO2
are responsible for about 90% of the atmospheric radiation trapping.  Methane, N,O, and O3 contribute the
remaining 10%. A schematic diagram of the global average components of the earth s energy balance is shown
is Figure 22.

                      340 Wm
                                                OUTGOING RADIATION

                                          Shortwave            Longwave
    Absorbed by
    trace gases
    and aerosols
                                                           Net Emission
                                                           Water Vapor,
                                                               by clouds
                                                             Water Vapor,
                                                     by Clouds
         Absorbed by
            by Surface
                                              LONGWAVE RADIATION
                                                                              Heat Flux
         Heat Flux
     320   25
  Figure 22. Schematic diagram of the global average components of the earth's energy balance.


        In the troposphere, dynamics driven by the direct solar surface warming and the hydrologic cycle
 actively mix the air on time scales shorter than the time  scales of energy transport by radiative processes.
 Consequently, the surface and the troposphere can be considered as a tightly coupled system from the radiative
 standpoint. In the consideration of climate change, the details of the vertical profiles of absorption and emission
 of energy within the troposphere are much less important than the  net radiative flux changes measured at the
 tropopause (Ramanathan et al., 1987; Porter and Cess, 1984). Because the stratosphere is, on the other hand,
 stable with respect to vertical mixing and can be considered to be  in radiative equilibrium, the nature of the
 vertical profiles of absorbing and emitting species like O3 are of greater importance to the stratospheric climate.

 Additional CKmate Variables

        The actual response of the  climate system to the observed trends in atmospheric constituents discussed
 above involves more than the direct changes in radiative forcing, with respect to predicting both annual average
 temperature changes  and the specific climate  variables that most directly affect regional  air  quality and
 chemistry. Strong feedbacks within the surface/atmosphere/ocean/cryosphere system modify the net global
 radiative response to changes in atmospheric constituent abundances. Couplings also exist between atmospheric
 dynamics and hydrology and the budgets and photochemistry of the trace constituents, so that potential changes
 in dynamics and clouds and precipitation will produce another level of feedback to the climate.

        Many potentially important climate feedback processes have been identified.  Figures 23 and 2.4 give
 a schematic portrayal  of important feedback processes. Because water vapor and clouds are the principal
 constituents responsible for reflecting, absorbing and emitting radiation in the atmosphere, while snow and ice
 cover on the surface  contribute significantly to surface albedo at higher latitudes, most feedbacks  involve
 processes affecting the hydrologic cycle. These include the  dependence of absolute humidity on temperature,
 the extent of cloud cover and cloud properties (optical depth and radiating temperature), the dependence of the
 tropospheric temperature profile on humidity, and the temperature dependence of the extent of surface coverage
 by snow and ice. The abundance of water vapor in the stratosphere appears to be controlled by the temperature
 of the tropical tropopause, through which air entering the stratosphere passes, and the photochemical oxidation
 of  CH4.   Increases  in the  temperature  of  the  low-latitude tropopause  may result from  an  altered
 surface-atmosphere radiative balance and may increase stratospheric water vapor, with subsequent feedback on
 surface temperatures  and stratospheric photochemistry. Some of these feedback processes remain poorly
 quantified. In the case of the various cloud feedbacks, even some of the signs are in doubt

       The close coupling of the surface and the troposphere, however, appears to enforce at least a degree
 of simplification on the consideration of climate change impacts on tropospheric regions.  In a global sense,
 the climate change remains a result of the  change in net fluxes at the  tropopause.  The exact  nature of the
 original source of radiative perturbation (for example, the particular trace gas emission scenario)  and the
 combination of resulting climate feedbacks  are of lesser importance in determining  climate change than the
global net radiative result.  The extent to which regional climate change is independent of the original forcing
is unclear. This general conclusion  probably fails, for example, at high latitudes where the importance of snow
and ice albedo impacts may be much greater with respect to other types of forcing than they are globally.

       Refined understanding of potential changes in the climate variables that directly affect local and regional
air  quality and chemistry, that  is regional temperature extremes and distributions, precipitation patterns and
distributions, wind patterns and near surface temperature profiles, requires an understanding of how net
tropopause radiative forcing changes affect the details of the surface-troposphere system.  Two lines  of
investigation are being followed in the climate research community. First, three-dimensional general circulation
model depictions of the physics of the atmosphere,  land surface, and upper oceans, including more  or less
detailed treatment of atmospheric radiative transport, can produce climate variable statistics on the spatial scale
of a hundred kilometers. The capabilities and limitations of such models will be discussed below. Second, the
record of CO2 abundance over geologic history shows a variation that correlates with the temperature record.

           H20, N2,02,C02,03,etc.
                            Air-Ice Coupling          Evaporation

                        ICE       t      Heat Exchange  ft wind Stress
      Changes of
Atmospheric Composition
                        Changes of Land Features,
                         Orography, Vegetation,
                              Albedo, etc


i j

Changes of Ocean Basin,
Shape, Salinity, etc
                                                    Atmosphere-Ocean Coupling
Figure 23.      The prinitipal interactions among the components of the atmosphere-ocean-ice-land surface
                climate system and some examples of external changes that may cause climatic variations.
                Source:  Gates (1979).

                                                  /LATENT HEAT
                                                  V   FLUX
            ^ f OUTGOING LONGWAVE
                                                                    f  SENSIBLE HEAT AND   A
                                                                    IPOTENTIAL ENERGY FLUX
                       ABSORBED SOLAR
                          RADIATION    J
                                               NET ENERGY BALANCE
                                                                         SUBSURFACE A
                                                                        HEAT STORAGES
                                                                OCEAN HEAT A
                                                                   FLUX    J
               PlarwUry albedo

                                                                                    Latitude &
                                                                  Temperature gradient

                     Snow area
                Geography  f
                                                 Atmospheric moisture
Horizontal wind

                             Atmospheric moisture
                       Relative humidity
 Figure 2.4.   Schematic illustration of climatic cause and effect (feedback) linkages and variables that are often
              included in numerical models of the climate system.  Source:  Robock (1985).


From the relative timing of fluctuations in the CO, and temperature records, there is some indication that the
CO, changes may be causal, in addition to orbitaTvariability and other factors. If other climate variables can
be deduced from proxy records with precision, the past climate information maybe useful for predicting future
response to trace constituent trends.


        The evidence for observed nonzero trends in many trace atmospheric constituents and the prospects
for continued changes have been briefly summarized above. The direct global annual average radiative effects
of constituent abundance changes can be calculated and compared in the absence of climate, biological, and
photochemical feedback processes. The climate feedback terms, chiefly involving water vapor and clouds, are
determined by the total radiative forcing change and do  not depend on the nature of the assumed trace
constituent change scenario.  This is not true, however, of the photochemistry and source and sink feedbacks,
which are unique to each constituent.  Quantitative estimates in the following section were taken largely from
Ramanathan et al. (1987) and Wang et al. (1986).

Direct Radiative Impacts

        The direct, feedback-free radiative effect of an abundance change in a trace atmospheric constituent can
be estimated using radiative-convective models in which humidity and cloudiness, for example, are fixed while
the constituent abundance is altered.   For the prototypical case of the doubling of CO^ the change in net
radiative flux at the tropopause is about 4 W m .  About two-thirds of the effect is tropospheric reduction of
outgoing infrared radiation, the remaining third an increase in downward emission from the stratosphere.
Because the atmosphere is already optically thick in the wavelength regions of CO, absorption, the radiative
effect of changing COu is a logarithmic function of the abundance. The extensive overlap of the CO2 absorption
bands with those of H^O must be carefully considered in radiative transfer calculations for CO2.

        The current abundances of CH4 and N2O together contribute a total radiative forcing of around 3.5 W
m"2, but as absorbers at wavelengths with larger atmospheric transmittance, forcing changes per molecule are
around 30 and 130 times that for COj, respectively.  Because they are present at much lower abundances than
COy the absolute impact of their increase is reduced with respect to CO2.  However, assuming the continuation
of current growth rates, the radiative impact of doubled CH4 (after 70 years) would be about 25% of the impact
of the 50% increase in CO2 that would be expected in 70 years. N2O and CH4 are considerably less important
than CO2 in affecting stratospheric temperatures.

        Many of the industrially emitted CFC compounds have their principal infrared absorption bands around
10 ft. These compounds are generally much more efficient infrared absorbers than CO2 on a per molecule basis
because of larger band strengths, high atmospheric optical transmittance at their absorption wavelengths, and
the peaking of the surface emission in the 10-p wavelength region. Much smaller abundance increases, of the
order of 1 ppb, can produce a radiative forcing change as large as 20% of the forcing change of a CO2 doubling.

        Ozone absorbs both incoming solar radiation in the ultraviolet  and visible regions and terrestrially
emitted infrared radiation in a band centered at 9.6 ft. Acting as a net source of heat in the stratosphere, ozone
absorbs about 12 W m"2 of solar  radiation and 8 W  m'2 of terrestrial infrared radiation. Stratospheric ozone
reradiates about 4 W m*2, 60% of that to space. Because the direct absorption of incoming solar radiation in
the stratosphere by ozone is the major stratospheric heating term, stratospheric ozone has both heating and
cooling influences on the surface-troposphere system. The net direct radiative effect of stratospheric ozone
change depends on the balance of increased  (decreased) tropospheric solar flux with decreased  (increased)
downward infrared emission from the cooler (wanner) stratosphere. For ozone in the troposphere, however,
both direct solar absorption and infrared trapping warm the surface-troposphere system.  Although the current
tropospheric ozone column is only about 10% of the stratospheric column, representative computations show
that a 25% decrease in tropospheric ozone results in a net flux change of  -0.5 W m*2 at the tropopause, while
a 25% decrease in stratospheric ozone (a larger decrease than is currently projected) produces a net flux change



 tropopause, coincidental^ a region where a complex combination of photochemical and transport phenomena
 control the ozone abundance.

        The net effects of aerosols on radiative fluxes are very sensitive to aerosol optical properties so that, in
 many instances, conclusions about the sign of the effect are difficult In the case of Arctic aerosol, which appears
 to be related to long-range transport of continental pollution and conceivably could have a secular trend, the
 effect is likely to be tropospheric wanning.

 Climate Feedback Response to Direct Radiative Forcing Changes

        On the principle that the surface-troposphere system responds to the total net radiative forcing of the
 several trace constituents that contribute, the climate feedback response should be relatively independent of the
 particular emission  scenario assumed.  The flux changes resulting from individual constituent changes also
 appear to be approximately additive. This results from the high optical transmissivity of the atmosphere near
 10 /i where most important trace constituent absorption features occur and from the lack of significant overlap
 of the various absorption features.

        The radiative forcing for a doubling of CO2 is around 4 W m'2, and the estimated surface temperature
 increase in the absence of any climate feedback processes is calculated to be in the range L2 to  L3°K at
 equilibrium.  From the results of radiative-convective and general circulation models that incorporate climate
 feedback responses, the surface temperature change associated with doubled CO2 is in the range LS to 4-5°K,
 with most GCM results falling hi the upper end of this range. The donate feedback factor then, denned as the
 ratio of the  computed surface temperature change  to the zero feedback change, falls in the range L2 to 3.75.
 The expected range of the global surface temperature change from the direct radiative effects of any combination
 of trace constituent abundance changes can then be scaled, given the results above, by comparison of the radiative
 model results for the net flux change for the doubled CO2 scenario with a trace gas scenario.

 Photochemical Forcings of Global Climate Change

        In addition to changes in direct radiative forcing by trends in trace species, changes in the atmospheric
 radiation balance can be forced through photochemical modification of atmospheric composition.  Many of the
 radiativeh/ important atmospheric trace constituents are also photochemicaDy important participants in processes
 that control ozone abundances and the lifetimes and budgets of trace species. We noted above that small
 changes hi net radiative  forcings are additive with respect to their impact on the surface-troposphere system.
 On the other  hand, photochemical processes affecting atmospheric composition are in general  coupled and
 nonlinear, and their effects on atmospheric composition are not additive. The estimation of the net radiative
 impact of changes mediated by photochemistry is usually specific to the details of a scenario of coupled trace
 gas abundance and emission trends.

        We will discuss the climate-photochemistry process interactions in two identified significant  cases,
 tropospheric OH and tropospheric and stratospheric ozone, in  some detail in Chapter  4.  Here, we briefly
 present a summary of results from  several  recent  studies  estimating the  indirect  climate  impacts  of
 radiative-photochemical interactions using various  types of radiative models.  Although the effects of the
 photochemical feedback processes can be described in isolation, the various feedbacks  are, in general, also
 closely coupled with each other, so that the net effect depends on the simultaneous operation of an the processes.
 The abundances and distributions of three families  of photochemicalh/ active species, O^ HO^ and NOy, are
 prominent hi many of the chemical feedbacks on climate.

        The global  average tropospheric  abundance of the short-lived, photochemicauy generated hydroxyl
radical, OH, is the crucial quantity in determining the lifetimes of many reduced atmospheric trace constituents
 and the oxidizing or scavenging ability of the atmosphere.  The  tropospheric OH abundance can be  affected
through perturbing HOX production processes,

                            O3 + hi/ (< 310 nm) -  O^D) + O2


                            O3 + hi/ (< 310 nm) = O(1D) + O2

                                  CK'D) + HjO = 2 OH

                                 OH + HC = ... = nHOx + other products       (2.1)

perturbing loss processes,

                                OH + NO2 + M = HNO3 + M

                                    HO2 + HO2 = H2O2 + O2

                                     OH + HO2 = H2O  +  O2
                                     OH + CH4 = H2O + CH3                  (22)

and perturbing the OH/HO2 ratio,

                               OH + CO (+O2) = HO2 + CO2

                                     HO2 + NO = OH + NO2                  (23)

        The tropospheric relative humidity in climate models is relatively insensitive to surface/troposphere
temperature change, so that surface/tropospheric warming produces an increase in absolute water vapor
abundance.  All else being equal,  an increase in tropospheric  OH follows.   Depending  on assumptions
concerning the  distribution of NO^ reductions in CH4 and O3 (through HOX  catalyzed  loss) can result
(Thompson and Cicerone, 1986a,b).  In one study (Hameed and Cess, 1983) this negative feedback on radiative
warming of the troposphere resulted in a 10% reduction on the surface temperature increase.

        On the  other hand, positive  trends in the CH4 and CO abundances can bring about reduction in OH,
with a subsequent positive feedback on the lifetime and abundance of CH4 (and other hydrocarbons, including
partially halogenated species). In the presence of a sufficient NOX concentration, tropospheric ozone can also
be increased by the mechanism

                              OH + CH4 (+O2) = H2O + CHgOg

                                         + NO = O^O + NO2
                                     NO2 + hi/ = NO + O
Model studies indicate that the radiative impacts of increased ozone produced by this mechanism can be of
similar magnitude to the direct CH4 radiative forcing change (Wang et al., 1986).

       The importance of the distribution and abundance of global tropospheric NOX is highlighted in the two
preceding paragraphs. As discussed above, there is no concrete evidence for a secular trend, but a strong
inference can be made that increasing anthropogenic combustion (including aircraft operating in the upper
troposphere) results in increasing abundances. As NOX is the catalyst for tropospheric ozone production,
increased tropospheric ozone can also be inferred.  Ramanathan et aL (1987) estimate that the resulting
surface-troposphere radiative heating by this process during the 70s could be a third of the CO2-induced
warming over the same period.


        A second category of climate-chemistry interactions involves stratospheric ozone and stratospheric
climate.  Stratospheric ozone is the primary absorber of solar radiation between 200 and 350 nm and the
primary determinant of the surface flux of UVB radiation.  Current indications are that positive trends in N2O
or CFCs, separately, will reduce stratospheric ozone, with the Cl-containing compounds of greater concern at
present. Together, the effect of increased N2O is to mitigate the CFC-related ozone decrease. Positive trends
in COy CH., and stratospheric water vapor act to increase stratospheric O^ by reducing the efficiency of the
NO and C1OX catalyzed odd oxygen loss processes. The greatest radiative impact on the surface-troposphere
system arises from changes in lower stratospheric ozone. Predicting the sign of these changes requires a full
treatment of the  complex ozone-controlling stratospheric  photochemistry, including seasonal and latitudinal
dependence.  Ramanathan et al. (1987) suggest that the importance of these considerations for surface warming
can be on the order of 25% of the direct CFC radiative effect, for constant current CFC emissions at steady

Climate Responses to Composition Changes from the Historical Record

        Assembling an estimate of the direct and indirect radiative effects of trends in atmospheric composition
as  discussed above  depends  on  the  completeness  and  veracity  of  a  variety  of radiative  and
radiative-photochemical  models.  However, evidence from the climate  record of responses to compositional
change would directly incorporate the effects  of feedback processes, and there exists the possibility that
conclusions on climate change could be made independent of models.  Indeed, there is strong evidence that
significant compositional  variability with respect to CO, extends  into the past, prior to any anthropogenic
perturbation of the atmosphere (Genthon et aL, 1987). The CO2 variability also seems to be tied intimately to
climate variability on at least two widely separated time scales and by two independent processes.

        On time scales exceeding 500,000 years, the abundance of atmospheric carbon dioxide  is maintained
by the balance of losses through weathering of calcium-silicate rocks by rain containing dissolved  CO, and
emission as a byproduct of tectonic processes (Kasting et al., 1988).  This system provides a negative feedback
between temperature and CO2 that is proposed to maintain habitable conditions on Earth over a range of solar
energy input

        On the shorter time scale of the glacial-intergladal cycle (160,000 years), recently reported  results of
records with a sampling resolution of 2,000-4,000 years of the climate-related 18O and D isotopic abundances
(Jouzel et al., 1987) and occluded CO2 from the Vostok Antarctic ice core (Barnola et aL, 1987) show a strong
correlation of the profiles extending back to the last interglacial (warm) period. There are many subtleties and
complexities in interpreting the ice core records, but hi a simple  linear multivariate analysis Genthon et al.
(1987) concluded that more than half of the 11°C glacial-interglacial Antarctic surface temperature variability
could be assigned to the CO2 abundance variations. The effect on surface temperature of the direct global
radiative forcing of the 200 to 300-ppm CO, change was calculated to be 0.6°C, compared to a direct orbital
forcing effect of 0.2°C.  It appears, subject to the difficulties mentioned below, that the long term  or glacial
condition climate feedback factor (5 to 14) may be larger than the climate feedbacks included in GCM models
of the present atmosphere  (12 to 3.75). Difficulties in interpretation include the uncertain global significance
of the Antarctic surface  temperature record and the extent to which the effects of the 200- to 300-ppm CO,
glacial-interglacial variability can be related to the current situation with CO2 at 345 ppm (and increasing) and
with contributing increases in other constituents. It is dear from the ice core record that the global climate and
atmospheric infrared radiative transfer are intimately connected.

       On the still shorter time scale  of decades to centuries before present,  construction of the global
temperature record and attribution of any trends to the effects of changes in trace constituent radiative forcing
have received much attention (DOE, 1985a). Difficulties arise in evaluating the accuracy and representativeness
of the temperature record  and in accounting for  the time lag in climate response to radiative forcing changes
that results from the time constant of the atmosphere-ocean interaction.  Accounting for  the observed NH
temperature increase of about 0.6°C since 1850 with current GCM models requires that  about half of the
calculated equilibrium climate response to the CO2 increase has not yet been realized because of slow transport


of heat in the ocean.  The observed warming is  thus not  inconsistent with the observed CO2  increase
(Ramanathan et al., 1987), but firm conclusions are hampered by natural climate variability.

Global riimate Projections

        There have been a number of recent studies of the future climate impacts of trends in atmospheric
composition,  including trends in CO^ and other trace constituents discussed above. The climate impact can be
quantified in terms of global equilibrium surface temperature change, with consideration of the time  response
of the climate system (particularly the ocean response) necessary to predicting the actual time evolution of the
global temperature change over the next century. The general conclusion from the body of recent studies is that
the trends in  trace constituents other than CO, can be expected to approximately double the surface warming
that is expected to occur as a result of the CO2 increase in isolation (see, for example, Ramanathan et al., 1987).

        Predicting the future climate response with the use of models of various sorts requires decisions in five
broad areas.  First, the basic scenario of future  trace constituent source emissions must be developed.  As
discussed above, emission scenarios can involve questions in demographics, energy use and policy, economics,
technology, agriculture and land use, atmospheric, photochemistry, biostience on a range of scales, marine and
geosciences, and  many interdisciplinary areas.  Scenarios have  often been  based  on extension of currently
observed atmospheric abundance trends, but it can be important to account for economic or physical couplings
of emissions strengths that make particular combinations of source emissions or abundance trends more or less

        The second and third steps are the calculation of the direct and indirect radiative effects of abundance
changes. These require the capability to treat radiative transfer in the atmosphere in both solar and longwave
regions, as well as photochemical processes in both the troposphere and stratosphere.  These  two steps are
coupled, especially in the stratosphere where radiatively driven temperature change  can have profound impact
on the photochemistry, so that these areas are best treated together.

        The fourth problem is the magnitude of the climate sensitivity or response to radiative forcing. Under
consideration here are the  climate feedback processes that amplify the response of the surface-troposphere
system to the net radiative forcing change imposed by the trace constituent changes in the assumed scenario.
With  the effects of these processes established, global equilibrium surface temperature perturbations can be
quantified.  Fifth and last, the time response of the surface-atmosphere-ocean system must be taken into account
so that time-dependent surface temperature change can be predicted.

        As a consequence of the many large uncertainties in at least the scenario and climate response aspects,
climate change  predictions are more appropriately made in terms of (broad) likely ranges rather than specific
values.  (Similarly, the impacts on the regional topics discussed in this report can be construed as sensitivities
over a range of possible input and boundary condition values.) As an example of studies of climate-chemical
interactions projecting climate change into the next century, Ramanathan et aL (1987) consider three scenarios
generally characterized by liberal, conservative, and intermediate views of the extension of currently observed
abundance trends. The range of global equilibrium temperature change for the 50-year period ending in 2030
is 0.8 to 4.1°K, incorporating uncertainties in both scenario and climate sensitivity. The range of realized
temperature change over this period is about 0.5 to L2°K, including uncertainty in  the rate of ocean thermal
diffusrvity. These area substantial fractions of the 11°K glacial-interglacial temperature contrast deduced from
the Vostok Antarctic ice core.




        The above discussion shows that a significant climate change can be expected to occur during the next
 century. The effect of CO2 doubling alone can be expected to lead to global average temperature increases of
 between 1.5 and 4.5°C.  Additional temperature increases are expected from the increased concentrations of
 CFCs, CH4, and N2O. Furthermore, more subtle changes that have not yet been fully established may also be
 occurring, such as an increase in tropospheric O3. These changes would also add to the predicted temperature

        The temperature increases  associated with the  trends hi trace species discussed  above are globally
 averaged temperature changes.  Not all regions, however, will experience the same temperature increase;
 furthermore, changes other than just a temperature increase can be expected.  These changes could include,
 for example, changes hi cloud types  and amounts, changes in meteorological conditions in a given region, and
 changes hi background free tropospheric concentrations of a variety of species.  This variety of changes will
 affect the chemical processes that take place on urban and regional scales and lead to oxidant formation and
 acid deposition. It is important to try to understand the effects of these changes on regional chemistry, because
 we have already implemented emissions control policies that may not be appropriate for a climatically different
 future. New policies that are implemented should account for the possibility that climate change will alter the
 formation of oxidant and acid deposition. Below we discuss some of the possible changes that need better


        Elevated urban and regional oxidant concentrations result from the interaction of  sunlight, NOg, and
 NMHCs.  The tendency to form oxidant (primarily ozone)  depends, therefore, on the concentrations of
 hydrocarbons and nitrogen oxides, on the availability of sunlight, and on the rate coefficients of certain key
 photochemical reactions  (which depend  on  temperature).  As will be shown, climate change can alter  this
 tendency dramatically, although few quantitative estimates are available. Below, we first discuss the mechanisms
 of oxidant formation, and then  describe  the ways hi which climate change might impact urban and regional
 oxidant formation.

 Formation of Oxidant

        In very clean areas, where  NOX concentrations are about 20-50 ppt, O3 is generally  destroyed by
 photochemical reaction sequences initiated by the reaction of CO with OH:

                                      CO + OH = CO2 + H

                                   H +  O2 + M = HO2 + M

                                      HO2  + O3 =  OH + 2 O2

                                 Net: CO + 03 = CO2 + O2                   (3.1)

As NOX is added to this system, the reaction of HO2 with O, hi (3.1) becomes slower than the corresponding
reaction of HO2 with NO, creating a reaction sequence which produces O3:

                                      CO + OH = CO2 + H

                                   H  + O2 + M = HO2 + M

                                     HO2 + NO = OH + NO2

                                      NO2 + hi/ = NO + O
                                Net: CO + 2 02 = CO2 + O3                  (3.2)

        If NMHCs are then added to the system, the reaction sequence leading to O3 production can be even
faster. This is because as NMHCs react, they tend to increase the pool of hydroxyl radical species (OH, HO,,
as well as higher organic forms of peroxy and peroxyl radicals), leading to more conversions of NO to NO2 in
reaction sequences similar to (32). The sequence of reactions leading to O3 production can be initiated by the
reaction of OH with NMHCs as well as by reaction of OH with CO. Thus the sequence becomes:

                            NMHC + OH + O2 = RO2

                               RO2 + NO + O2 = HO2 + NO2 + RCHO

                                     HO2 + NO = OH + NO2

                                    2(NO2 + hi/ - NO + O)

                                 2(0 +  O2 + M = O3 + M)

                            Net: NMHC + 4 O2 = 2 O3 + RCHO                (33)

where RCHO is a carbonyl compound and R- is a hydrocarbon radical.  These processes can be moderated
locally. In areas with high NO  and lower NMHC/NO ratios, the reaction NO2 + OH + M -»• HNO3 + M can
serve to decrease  hydroxyl, which is necessary to initiate the  O^-producing sequence.  Also, nitrogen oxides
which are primarily emitted as NO remove O3 directly via: NO + O3 -*• NO,  + O,.  This second reaction
leads to a net reduction in O3 until a steady-state ratio for NO/NO2 is established. Aner this, the reaction of
NO with O~ is balanced by photolysis of NO2 followed by O atom attachment to O2 to reform O3. However,
the potential for higher oxidant concentrations where emissions of NMHC and NOX are high is dearly possible
and experienced in many urban areas.  A greater degree of complexity in the photochemical mechanism also
follows with the production from radical precursors of peroxyacetyl nitrates (PAN compounds) and other organic
nitrates. Therefore,  any climate change that substantially alters these concentrations can alter O3 formation.
Such possibilities are discussed below.

Effect of Cliniate Change on Oxidant Formation

        Several scenarios of climate  or global chemical change have the potential for altering oxidant levels.
Table 3.1 summarizes the types of climate change parameters that are important for regional chemistry. Below,
the specific effects of these changes on oxidant formation are discussed.

        A change in temperature win alter oxidant formation rates directly by influencing the rates of chemical
reaction. Figure 3.1 shows the peak ozone concentration reached in a box model calculation of ozone formation
as a function of the assumed atmospheric temperature and of the initial HC to NOX ratio (see Atherton and
Penner, 1988, for a description of the box model and initial conditions). For urban conditions, with a HC/NOy
ratios near 7, a 5 degree temperature  increase (from 298 to 303°K) might be expected to raise O, levels by 0.05
ppm  or 20%.  Most of the O3 response to temperature is due to the strong temperature dependence of the


              Table 3.1.  Climate Change Parameters Important for Regional Chemistry

 1.    A change in the average maximum or minimum temperature and/or changes in their spatial distribution
      and duration leading to a change in reaction rate coefficients and the solubility of gases in cloud water

 2.    A change in stratospheric O3 leading to a change in photolysis rate coefficients.

 3.    A change in the frequency and daily pattern of cloud cover and types of cloud formed leading to a
      change in photolysis rate coefficients and heterogeneous rates of conversion of SO2.

 4.    A change in the frequency and intensity of stagnation episodes or a change in the depth of the planetary
      boundary layer and its diurnal  cycle leading to more or less mixing of polluted air with background air.
     A change  in background boundary layer and/or free tropospheric concentrations of water vapor,
     hydrocarbons, NO^and O3 (due to changes in mixing processes or sources such as lightning for NOX)
     leading to more or less dilution of polluted air in the boundary layer with background air and layering
     the chemical transformation rates hi both the boundary layer and the free troposphere.
 6.   A change in the vegetative and/or soil emissions of hydrocarbons and NOX which are sensitive to
     temperature and/or light levels leading to changes in their concentrations.

 7.   A change  in  deposition rates to vegetative surfaces  whose  stomatal resistance is  a function of
     temperature, light intensity and other factors leading to changes in concentrations.

 8.   A change in energy usage  or technology leading to a change  in energy-related emissions and their

 9.   A change in secondary aerosol formation leading to changes in photolysis rates, the planetary albedo,
     and heterogeneous reaction rates.

10.   A change in circulation and/or precipitation patterns leading to a change in the abundance of pollutants
     deposited locally versus exported off the continent

                                                  HC/NOK - 7. INCREASED BL
                                     20          26          30
                                    TEMPERATURE ( frgrw C )
Figure 3.1. The effect of temperature on the peak O« concentrations predicted in a box model calculation of
          urban O3 formation. Calculations are shown for three hydrocarbon to NO. ratios. The effect of
          increasing the boundary layer depth for the case with a hydrocarbon to NOX ratio of 7 is also


thermal decomposition of PAN:
                                          PAN = RCO3 + NO2

                                             k = 2xl016e12542/Ts-1
When PAN forms by combination of peroxy acetyl radicals (RCO3) with NO- it sequesters the NO and peroxy
radicals that contribute to ozone formation in a non-reactive form (although PAN itself is a harmful irritant).
When temperatures are raised, more NO and RCO3 radicals are available for O3 production, and, as shown
in Figure 3.1, over an 8-hour period, the effect on O3 production can be pronounced, especially at intermediate
HC/NO ratios. At very high or very low HC/NOX ratios, the effect is not as pronounced. AthighHC/NOx
ratios, tne reaction sequence which produces O3 is very fast so that  the peak O3 concentration is not greatly
changed by the sequestering of NOX by PAN. At low HC/NOX ratios, the mixture of hydrocarbons and NOX
are not reactive enough to significantly affect O3 buildup in an 8-hour period,  but one might expect a larger
change at later times or over a several day period. Apparently, an increase of temperatures can significantly
exacerbate oxidant problems.

       A change in stratospheric O, concentrations would alter oxidant formation by altering the surface UV
flux, thereby changing photolysis rates.  O3, and in particular, stratospheric Og, is responsible for absorbing
much of the  UV radiation below 300  nm. Changes to the column abundance of O3 would thus change the
amount of shortwave radiation available in the troposphere. In the reaction sequence outlined above, only the
photolysis of NO, is shown. The photolysis of NO, is not expected to be significantly altered by a change hi the
column concentration of O3, because most of its photolysis occurs by absorption of longer wavelength photons
(between 350 and 400 nm). However,  several species that are intimately involved in the production of O3 have
significant absorption cross-sections at shorter wavelengths. These include H2CO, CHgCHO, HjO- O3,  etc.
Table 3.2 gives an estimate of the photolysis cross-sections for these species calculated at ground levefTor zenith
angle of 45 degrees for three levels of column O3 abundance calculated using the LLNL one-dimensional model
of the stratosphere and troposphere (Wuebbles, 1981). The maximum to minimum change in O3 represents a
25% decrease in column O3.

        Table 3.2.  Photolysis Rate Constants at the Surface as a Function of Overhead O3 Column

 PRODUCTS: CH3 + CHO   H2+CO      H + HCO

 Os Column*

 8.0E18     4.4E-6

 7.2E18     4.9E-6

 6.0E18     5.7E-6
       Photodissociation rate constants (s  J)

2.50E-5    2.09E-5    7.90E-3    7.10E-6    4.80E-4    2.18E-5

2.55E-5    2.20E-5    7.92E-3    7.30E-6    4.84E-4    2.60E-5

2.63E-5    2.40E-5    7.95E-3    7.80E-6    4.91E-4    3.47E-5
* Note:  Units are molecules cm"2.


        Figure 32 gives the ozone production during an 8-hour simulation using these three sets of photolysis
rate constants for a simulation in the box model with a HC/NCL ratio of 7. Similar runs at HC/NOX ratios of
2 and 28 showed a much smaller O3 response. According to these calculations, if stratospheric O3 decreases
by 25%, urban oxidant formation could increase by about 10% or by 0.03 ppm.

        These results may be contrasted with those reported by Gery et al. (1987) and Liu and Trainer (1987).
Gery et al. report peak O3 changes ranging from -7% to +47% for an O, column decrease of 33%. Their study
only considered conditions in which an urban area had already attained the 0.12 ppm standard for O3-  The
range of results they report is apparently a function of the specific HC/NCX ratio used in each model run and
also of the latitude, base temperature, and meteorology used.  Liu  and Trainer (1987) also calculated the
response of tropospheric ozone to a column ozone change.  They considered a range of NOX conditions and
scaled the NMHC concentrations with NOX for NOX concentrations above 1 ppb. For areas impacted by urban
emissions NO^ ranging from 1 to 10 ppb), they calculated a 10-15% increase in surface ozone resulting from a
20% decrease in column ozone. In very dean areas, photochemistry acts to destroy ozone, and they calculated
a decrease of up to 35% for surface ozone resulting from a 20%  decrease in column ozone.

        Cloud cover changes can directly alter O, production by changing photolysis rates. For example, on
very cloudy days, no significant photochemical production occurs, because too little sunlight reaches the surface.
Urban and regional oxidant models have varying sophistication in their ability to account for cloud cover.  One
simple technique is to decrease all photolysis rates in proportion to the amount of partial cover.  Using this
technique, one can estimate the impact  on O3 production using the box model.  The impacts are illustrated in
Figure 33 for a HC/NOX ratio of 7.  Clearly, as illustrated there, the amount of cloud cover is a significant
climate element for oxidant formation.  Furthermore, cloud amounts reflect the degree of convective intensity.
Such activity acts to redistribute pollutants vertically with potentially important consequences for global as well
as regional chemistry.

        The  highest oxidant concentrations are developed during stagnation episodes.  These episodes are
characterized by surface winds that are generally small (0-3 m/s) leading to little mixing of pollutants in the
horizontal, high-pressure systems with higher than normal temperatures, and decreased ventilation associated
with decreased vertical mixing and boundary layer depth. If the  frequency of these episodes were to change,
this would clearly impact the number  of days in which high  oxidant concentrations were experienced. In
addition, of course, as discussed above, any change in the intensity of any one of these characteristics would alter
the intensity of the oxidant episode.

        The depth of  the boundary layer influences the level of oxidant formation by restricting or enhancing
mixing of pollutants into tropospheric background air. Over continental areas, the boundary layer is generally
low at night, grows to heights as high as 1,000-2,000 m, during daylight hours as the ground temperature heats,
and then reforms near the surface again at night If the maximum height of the boundary layer were to increase
as a result of a higher surface heat flux, concentrations would decrease within the boundary layer, allowing for
decreased oxidant formation. This is illustrated using the box model in Figure 3.1. Doubling the ventilation rate
decreases the amount  of ozone formed during an 8-hour period by about 40% for a HC/NOX ratio of 7.  The
results obtained here  with the simple box model are in qualitative agreement with the results of sensitivity
studies  from  trajectory  models  (e^, Derwent  and Hov, 1988)  and  from more complete  grid-based
photochemical models (see, for example, the studies quoted in the review by Seinfeld, 1988; as well as Penner
et aL, 1983; and Seigneur et aL, 1981).  The above studies  also explore the effects of altered meteorological
conditions. Of course, specific areas will have different sensitivities.

        Pollutants which have mixed throughout the boundary layer the previous day, will become part of the
"free" troposphere during this process.  If the winds above the boundary layer are strong, these pollutants will
be adverted and diffused away from the region of concern, but with light winds they may be incorporated into
the boundary layer during its growth on the following day. Pollutants which are part of the "free" troposphere
are  generally represented in a photochemical model through boundary conditions.   Thus, to the extent that
boundary  layer injection processes are altered by climate  change, we can also expect changes to the free
tropospheric concentrations of NOX and hydrocarbons.  Other climate-induced changes will, of course, add to


                      1       I       I       I       I       I       I       r     T
.10* REDUCTION    —
200           300
  TIME (irin)
 Figure 32. The effect of changes hi the column O, abundance on urban O- formation.  The ambient O3
           column was assumed to be 8.0 x 1018 molecules cm'2. The other two cases represent column O3
           decreases of 10 and 25%.

   -»    .4
                          .6              .7              .8              .9
Figure 33. The effect of cloud cover changes on the peak O, concentrations predicted in a box model
          calculation of urban O, formation. The presence of clouds was assumed to decrease all photolysis
          rates by the same ratio.


this dynamically caused alteration of free tropospheric concentrations.  These changes include changes in the
water vapor concentration as well as changes in the anthropogenic emission of NO  and NMHCs due to a
change in energy usage. Also, a change in the natural fluxes of hydrocarbons and mtrogen oxides due to the
climate change would lead  to  changes in free tropospheric concentrations.   Also,  because  lightning is a
significant source of free tropospheric NOX, a change in the frequency of lightning could alter background NOX
concentrations. Finally, changes to background free tropospheric HC and NOX concentrations will likely cause
changes to the free tropospheric ozone concentration. These background concentrations, as well, impact urban
and regional oxidant concentrations.  Sensitivity studies of the response of grid-based models to changes in
boundary conditions give some indication of the importance of this source of change (Seinfeld, 1988; Penner et
al., 1983; Seigneur et al., 1981).

        As stated above, changes to the natural emissions of hydrocarbons and NCL can alter  background
tropospheric concentrations.  Both soil NOX emissions and vegetative hydrocarbon emissions appear to increase
exponentially with temperature (Williams et al., 1987; Lamb et al.,  1985). Vegetative hydrocarbon emissions
are also dependent on light intensity. Changes in  emissions will also contribute directly to changes in urban
and regional concentrations of HC and NOX that contribute to O3  formation.  In the urban atmosphere, the
sensitivity may not be large because the proportion of ozone formation that can be attributed to vegetative
hydrocarbon emissions and soil NOX emissions are generally small  (see Penner, 1984; Lurmann et al., 1983).
Figure 3.4 demonstrates the changes in peak O3 concentration over an 8-hour simulation for a simulation in
which vegetative HC fluxes were increased exponentially in response to a temperature increase following the
relationship determined by Lamb et al. (1985). (See Penner, 1984, for a description of the treatment of natural
hydrocarbon emissions in the box model). The curve is quite similar to that obtained by a straightforward
change in the temperature. Of course, the response of regional ozone may be more dramatic.  In addition to
changes in natural hydrocarbons and NOX emission rates, deposition may change. Deposition of many pollutants
takes place on vegetative surfaces.  The stomatal resistance of leaves is a function of temperature as well as
light intensity and other parameters (Baldocchi et al., 1987). Deposition rates might therefore be expected to
change as a result of climate change, possibly affecting air chemistry.

        A change in energy usage and the emissions from energy usage as a result of climate  change will
obviously lead to HC and NGL emission changes as well.  If temperatures increase, we can expect summertime
usage of air conditioners to increase, perhaps with a proportional  increase in emissions from power plants.
Evaporative emissions from vehicles and refueling would also increase. A simple estimate of the episodic impact
of these emission increases on urban ozone can be obtained from studies of the effect of emissions reductions.
Increases in  NOX will, in many urban areas, lead to local decreases  in O3 (although downwind concentrations
could increase).  Increases in hydrocarbon emissions would increase local and downwind O3 concentrations.
The study of Penner and Connell (1987), for example, showed that a 30% increase of NOX emissions led to an
O3 decrease of 25 to 35% at a location near the urban center where the largest ozone concentrations were
experienced, but an expanded area of O3 increase above the federal standard downwind of the central urban
region. A 30% increase in hydrocarbon emissions increased peak O3 at all locations, with an increase of 25 to
35% at the location where  the highest ozone concentration was recorded.  These changes were calculated for
changes in areawide emissions from all sources. The quantitative model results will be different, of course, for
different urban areas or if only increases in power usage were considered (see also Seigneur et aL, 1981; Penner
et al., 1983; Derwent and Hov, 1988, as well as studies quoted  hi Seinfeld,  1988).

        Finally, changes in the amount  of secondary aerosol can affect oxidant formation through effects on
photodissodation processes.   The  amount of secondary aerosol would  likely change as  a result  of both
temperature and water vapor changes (Schere, 1988). The estimates  of potential urban and regional O3 change
from various climate-related perturbations are summarized in Table 33. Because some climate perturbations
lead to increases in O3 and some lead to decreases,  further work is needed to fully assess the expected change.
However, as shown in Table 33, the response of urban and regional ozone to climate change is potentially


                            Table 33 Estimated Change in Urban Ozone
            Climate-driven            Estimated A
            Perturbation              O3 Change                    Notes

T increase from                        +20%
Increase biogenic emissions of           3-10%                   Could be larger in
 NMHC consistent with a                                        regions where biogenic
 a 5°C temperature                                             emissions are more
 increase                                                      important

25% decrease in column                 +10%                   This work and Liu and
 O3 abundance                                                 Trainer (1987)

                                 -7% to +47%                 Gery et aL (1987)

Increase cloudiness leading              -10%
 to a 10% decrease in
 photolysis rates

Increase boundary layer                -50%
 depth by 2
  Estimated using a simple box model of urban photochemistry with initial HC/NOX
  See Atherton and Penner (1988) for details.

                    I     I     I     I     I     I     I     I    I    I    I    I    I    T
I     I     I     1    I    I    I    I     I     I     I     I     I	L
                   25                        30
                  TEMPERATURE ( dMMM C )
 Figure 3.4. The  effect of changes in temperature and biogenic hydrocarbon emission on the peak O3
            concentrations predicted in a box model calculation of urban O, formation.  The solid line shows
            the effect of temperature alone. The short-dashed curve shows the increase expected when the flux
            of biogenic hydrocarbons is increased along with the temperature. The long-dashed curve shows
            the change expected if the initial concentration of biogenic hydrocarbons is increased in proportion
            to the expected increase in the flux of biogenic emissions.



       Emissions of NOX and SO2 into the atmosphere lead to the formation of nitric and sulfuric acid via a
number of complex heterogeneous and homogeneous chemical pathways. In addition, organic acids may form
from non-methane hydrocarbons (NMHC), which add to the abundance of acidic species. Once conversion to
acid species has occurred, surface deposition may take place via either dry or wet processes. The effect of a
particular change in climate on the rates of transformation of NOX, SO~ and NMHC to then- acidic forms and
on their method of delivery to the  Earth's surface will depend on which pathway is dominant in any given
situation. Here we review the chemical pathways thought to be primarily responsible for the conversion of NOX,
SOj, and NMHC to NO^, SO4"2, and organic acids. We speculate about the impacts that any given climate
change might have on these conversion rates. No comprehensive statement about sensitivity is possible, because
no comprehensive model results that integrate over a number of specific situations are available. However, we
indicate situations wherein a change might occur.

Formation of Acidic Species

       Nitrogen oxides (NO and NO,) are emitted primarily as NO.  NO is rapidly converted to NO9 via
    •    •*» *^
reaction with O3:
                                     NO + O3 = NO2 + O2                   (3.5)
with smaller contributions from reactions with HO2 and ROu (see reactions 33).  NO and NO^are converted
to a variety of other species by a number of different gas phase reactions.  Conversion to HNO3 takes place
primarily in the gas phase during daylight hours via:

                               NO2 +  OH + M = HNO3 + M                  (3.6)

At night and in clouds, other reaction sequences, such as:

                                     NO2 + O3 = NO3 + O2

                                    N03 + N02 = N205
                               N2°s + ^2°^ • 2 HNOa or 2 N(V
                                 NO3 + H^aq) • HNO3 or

                                 NO3 + RCHO = NO3 +  RCO                (3.7)

may also be important In reaction (3.7) above, RCHO indicates an aldehyde, but similar gas phase reactions
of NO3 with aromatics and terpenes also occur (Morris and Niki, 1974; Atkinson et aL, 1984).

        Conversion of SO, to SO4'2 occurs via a number of pathways that involve both homogeneous and
heterogeneous (aqueous phase) mechanisms. Gas phase conversion is thought to occur via:

                                     SO2 + OH = HSO3

                                    HSO3 + O2 = HO2 + SO3

                                    SO3 +  H2O = H2S04                     (3.8)

while the aqueous phase conversion occurs primarily via:


                               SO2 + O3 + HgO = H2S04 + 02
At high concentrations, metal ions in solution can also catalyze the conversion of SO, to SO4  by C*2.  Sulfate
particles may be formed by attachment of the gas phase HgSO^ formed in reaction (3.8) to pre-existing particles
or by the evaporation of cloud drops whose residue consists of H^SOj formed in the aqueous phase.  Which
process is most important  has not been determined.  Further, the relative importance of the subsequent
incorporation of sulfate particles into precipitation versus the aqueous conversion of SO2 to SO4  during
precipitation events remains uncertain.

        Formic and acetic acid can account for a significant fraction of the acidity in precipitation (Galloway et
aL, 1982; Keene et aL, 1983; Keene and Galloway, 1984; Norton, 1985),  although their relative importance
globally to that of the acidity from nitric and sulfuric acid is currently unknown.  Their formation mechanism
involves both gas-phase and aqueous phase processes, but the  relative importance of these processes is also
unknown (see discussion in Albritton et al. (1987) and also Atkinson and Lloyd (1984), Calvert and Stockwell
(1984), Chapman and Sklarew (1986), Graedel and Goldberg (1983), and Chameides (1984)). Because so little
is known, the effect of climate change on the conversion of NMHC to organic acids is not considered below.

Effect of Climate Change on Acid Rain and Acid Deposition

        Given the mechanisms of acid formation discussed above, it is immediately clear that changes to climate
could impact the rates and mechanisms leading to acid deposition.  For  example, cloud cover may change,
leading  to a change in the  rates of heterogeneous conversion  of SO2 to  SO4  .  Alternatively, precipitation
patterns may change, leading to a change in the spatial areas affected by acid precipitation. Prediction of these
changes is currently beyond the capability of GCMs (see Chapter 5). Furthermore, except for models of
episodic events, we still  cannot properly delineate the  relative role of  conversion rates  in clouds versus
conversion rates in dry air  so that estimating the global  and seasonal effects of cloud cover changes is not
possible. However, we can illustrate the climate and chemistry changes that are expected to impact the rates of
SO2 conversion to sulfate. These types of climate change include the following:

1) A change in average tropospheric temperatures and/or their regional distribution.

2) A change in stratospheric O3 concentrations.

3) A change in the background concentrations of O3, HjO, NOX, and NMHCs

4) A change in circulation patterns.

5) A change in the frequency and type of clouds formed.

6) A change in precipitation patterns.

        An increase in temperature can impact conversion of SO2 to SO4~2 in several ways. Walcek and Chang
(1988) used results from the RADM model (Chang et al.,  1987) to estimate that gas phase conversion rates of
SO2 to SO4  would be increased as a result of temperature increase in both the boundary layer and free
troposphere. However, in their model aqueous phase conversion rates were a factor of 3 larger than the gas
phase conversion rates. Increases in temperature caused a decrease of the HjO, production rate, which may
be associated with decreases in aqueous conversion rates of SO2^ The solubility of SO2 also decreases at higher
temperature, although the aqueous oxidation rate constants increase.

        A decrease in stratospheric O3 concentrations can  alter the conversion rate of SO2 to SO4~2 by
increasing the UV flux in the troposphere, thereby increasing photolysis rates.  The most important species
impacted will be H^O?  Since  H^ acts to efficiently convert SO2 to SO4'2 in cloud water and because its


concentration can be limiting in this process, an increase in HgOj concentrations will quantitatively increase the
conversion of SO2 to SO4
or of NO  to HNO3
        A change in the background concentrations of the species involved in the conversion of SO2 to SO/2
         j, to HNO3 will  alter acid formation rates.  These species include O~, H^O (which determines the
concentrations of HjOg and OH), NOX, and NMHCs.  The background concentrations of these species can be
altered by climate change  through a number of different mechanisms, none of which is easily quantified given
the state of today's knowledge. For example, a general wanning of the climate is expected to lead to increases
in the concentration of water vapor, if it is assumed that  the relative humidity remains constant.  Also, as
discussed above, both NOX and NMHCs have natural sources that are sensitive to temperature and precipitation.
Thus, a change in climate that affects temperature and precipitation patterns will lead to a change in the natural
fluxes of NOX and NMHCs. Further, both these species have large fluxes associated with energy usage.  If this
is altered as a result of climate change, the flux of NOX and NMHCs from energy usage will change.  It is not
possible, at this time, to  estimate how background concentrations might  respond to these  flux changes;
three-dimensional models are needed that, account for all the sources of these species in a realistic manner.
However, it can be stated that changes are expected and that these changes will alter the rates of acid formation.
Sensitivity tests with the RADM model confirm this observation (Brest, 1988), but it is not clear,  at the present
time, how to relate these tests to climate change.  Given that HJO, NO^ and NMHC concentrations will change,
it can also be stated that the background concentration of O3 will change, since  its chemistry depends intimately
on the levels of these species. Walcek and Chang (1988), for example, estimated that a 10% increase in ozone
concentrations would increase the gas phase oxidation of SO2 in the boundary layer, while decreasing SO4'2
production in the free troposphere. A 10% increase in background water vapor was estimated to increase gas
phase production rates of SO/ and HjOj in both the boundary layer and free troposphere.  However, these
estimates were not based on fully coupled model calculations.

        Potentially the most significant climate-change effect could be a change in average  cloud liquid water
content.  Hales (1988) used a cloud/chemistry model of a cyclonic storm to  estimate that changes hi
temperature would lead to substantial changes in precipitation  amount at constant  relative humidity.  The
deposition rate of SO/2 is thereby strongly affected. By changing the pH of cloud water, these effects could
also alter the roles of HgOg and O3 hi the conversion of SO2 to SO4  .

        Finally, if circulation and precipitation patterns change, the amount of acid deposited locally can change.
Galloway et aL (1984) have estimated that the amount of sulfur exported off the east coast of North America
is 30 to 35% of that emitted over land areas. A slowing of winds or a change of circulation patterns, possibly
as an increased seasonal duration of summer-like conditions, for example, could change the ratio of deposition
to export in a manner that significantly alters regional deposition.  Sensitivity studies reported by Brost (1988)
confirm that changes in meteorology can be quite important Further, precipitation can occur in a variety of
storm types which have different impacts on the redistribution of pollutants and on then- mixing.  Therefore,
the venting of acid precursors to  the upper atmosphere and/or the reaction times available for formation of
sulfate could change if the types of clouds and storm systems change.


                                          CHAPTER 4


        It has been established through long-term measurement records that global changes in radiatively
important species (e.g., CO^ CH4, N-O, CFCs) are occurring at significant rates, and models have been used
to demonstrate that these will drive development of a significant change to climate. As the above discussion
shows, changes in climate and global chemistry have the capacity to influence both acid deposition and urban
and regional oxidant formation.  Other climate-related changes in global atmospheric composition that have
significant implications may also be inferred.  These include perturbation of tropospheric and stratospheric
ozone driven by emissions of hydrocarbons and NOX  (in the troposphere) and emissions of CFCs and other
halocarbons (in the stratosphere). In addition, composition changes induced by climate change can feed back
through various processes to produce further climate change. One example of a feedback loop involves change
in tropospheric OH concentration, which then alters the lifetimes and transport of radiatively active species.
Another involves climate-driven change in tropospheric O3, which in turn directly impacts climate in its role as
an absorber and emitter of solar and infrared radiation.  Furthermore, O3 is  intimately involved in production
of hydroxyL Below, we outline some of the feedback loops and estimate, where possible, the sensitivities and


        The hydroxyl radical, OH, is the primary tropospheric chemical scavenger of CH3CCLj, CH^Cl, CH3Br,
and other hydrogen-containing halocarbons, as well as of CH4, NMHCs, CO, DMS, H^S, NO? and1>O2. Some
of these species (several halocarbons and CH4) are directly responsible through their radiative properties for
part of the expected climate change.  A change in  global OH concentration will alter the lifetimes of some of
the species driving the climate change, and thereby modify their abundances and radiative effects. Further. OH
and other closely related  species in the HOX family play a central role in the conversion of SO2 to SO/  and
NO2 to  HNO3 in the homogeneous phase, oy direct reaction of SO- and NO-with OH.  In solution, FL/),,
formed homogeneously by HO, disproportionation, acts to convert SO2 to SO4 in clouds.  Likewise, the fiO
family plays a central role in O33 production by oxidizing NO to NOj, by removing active forms of NO , and
by initiating the oxidation of hydrocarbons in the atmosphere.  Below we describe first the sources ana sinks
of OH,  then we consider how climate  change could influence its concentration, and finally we speculate on
related climate-chemistry feedback processes.

Sources  and Sinks of OH

        The major  source for  OH and for HOL (HO,  HO2, HJQU HNO3, etc.) in the  background global
troposphere is the reaction of an excited state ofatomic oxygen, 6( T>X with HjO:

                                  H2O  +  O(1D) = 2 OH                       (4.2)

The O(1D) in reaction (4.1) is generated by photolysis of O3 near 300 nm:

                          O3  + v\a> (A< 310 nm) = O2 + O(1D)                 (4.2)
In regions of high NMHC concentration, the oxidation of these species can also add to the HOX family by
radical chain-branching reactions:

                                  NMHC + OH = ..JK)2 + RO2               (43)

which lead to OH production after reactions with NO or O3 that convert HO2 and RO2 to OH.


       In addition to these direct sources of OH (and HOXX the abundance of the hydroxyl radical is
controlled by fast reactions interconverting OH with other members of the HOX and ROX families (HO~ HjOg
HNO3, RO, RO~ RCO,, etc.).  The rates of these radical chain-propagating reactions depend on the local
concentrations ofNO, CO, O3, CH4, NMHCs, and on local photolysis rate constants, particularly that for HgOg.
       A variety of radical chain-terminating reactions remove OH and HOX from the troposphere including:

                                   OH + HjOj = HjO + HO2

                                     OH + HO2 - HjO +  O2

                               OH + NO2 + M - HNO3 + M                   (4.4)


                                    HO2 + HO2  = H^ + O2                  (4.5)
followed by the heterogeneous removal of HLO2 and HNO3 by precipitation.  A suite of other reactions also
remove HOX but are currently thought to be less important than those listed here.

Impact of Climate Change on Tropospheric OH

       Given the importance of HjO, CH4, CO, NO, O3, and tropospheric solar fluxes to tropospheric OH
abundance, it is clear that, to the extent that climate would impact the concentrations of these species and
photolysis rate constants, a change in climate would alter OH. As noted above, reaction with OH provides the
major chemical scavenging mechanism for a variety of species in the troposphere.  In particular, the primary
removal mechanisms for CO and CH4 are by reaction with OH.  These reactions are also the most important
direct conversion reactions for OH on a global basis.  They cycle OH to other forms in the HO  family,
modifying the  ratio of OH to HOX.   The importance of these interactions in HOX chemistry implies that
increases in CO and CH4 abundances can lead to decreases in OH.  Similarly, increases in tropospheric NO
and O3 can lead to increases in OH by enhancing cycling reactions that convert HOL to OH.  Whether these
effects on OH will be offsetting is a  problem requiring higher-dimensional tropospheric chemical models to
understand and integrate in detail.

       Because water vapor is a parent compound for OH and other HOX species, changes to its concentration
would alter the concentration of tropospheric OH. Tropospheric water vapor is in balance with an evaporation
and transpiration source from the oceans, soils, and plants and the precipitation sink. Global increases in
temperature,  driven by a  climate change, are  expected to lead to changes in  the tropospheric water
concentration.  If the sources of water vapor are not perturbed by changes in vegetative cover and if circulation
patterns do not lead to more frequent precipitation events, then the concentration of HgO might be expected
to increase.  We note that global average relative humidity tends to remain almost constant with warming in
climate model experiments (Ramanathan et aL,  1987).  Of course, changes in precipitation patterns and/or
frequency and changes  in vegetation  patterns would make this simple picture more complex.  As a simple
estimate, a 2 degree increase in temperature could be associated  with a 10-30% increase in tropospheric H-O
levels, implying a few percent increase in OH and other HOX family members (Thompson et aL, 1988), aD else
being equal

       Reaction  of O3 with HO,  cycles HO. back to OH.  Therefore,  increases in  the  tropospheric
concentration of O3 would lead to increases in OH.  As discussed below, tropospheric O3 concentrations may
be increasing as a result of direct emissions of NO^ NMHCs, and CH4 as well as through indirect changes in
the emissions of these species (and others) which are driven by a climate change.  Because of the role of NO
in partitioning HOX and because of the large variation in the concentration of NO between remote oceanic
areas and continental areas, an increase in O3 by a factor of 2 could increase OH by perhaps 10% over ocean
areas and by probably greater than 10% over the continents (Thompson et al, 1988).  These changes might,



 of course, feed back on the concentration perturbation of O3 and would be better estimated using a fully
 coupled global model.

        Also, increases in the concentrations of halocarbons and N2O are driving changes that are calculated
 to lead to a net decrease in stratospheric O-. Stratospheric cooling, as driven by increases in CO2 and other
 infrared emitters, acts to increase stratospheric O3 (Penner, 1980).  These direct effects of emissions and
 indirect effects  of climate  change are calculated to have caused only small changes in stratospheric O- at
 present.  However, larger changes could occur in the future if the existing balance of processes is perturbed.
 Because most of the ozone column resides in the stratosphere and because ozone is responsible for much of the
 atmospheric opacity below 300 run, these changes would alter tropospheric OH by changing photolysis rate
 constants for any species with significant absorption hi this range (see Table 3.1). In particular, absorption by
 ozone leading to the formation of O(1D) would change, creating a change in the direct source of HOX and OH.
 Further, an increase in the  photolysis rate for HJOu could affect HO  loss rates.  A decrease in column O3 of
 20% would lead to an OH increase of roughly 15% over continental areas (Thompson et al., 1988; Liu and
 Trainer, 1987).

       Another mechanism for climate impact on OH concerns the magnitude of the source of OH created by
 the oxidation of NMHC.  Trainer et al. (1987) have estimated that in  areas of  low NOX concentrations,  the
 concentration of OH might be decreased by as much as a factor of 2 hi the first  100 m above ground level as
 a result of biogenic emission of isoprene and terpenes. The emission of biogenic hydrocarbons is exponentially
 dependent on temperature  (Lamb et al., 1985).  An increase of 5°K could lead to an increase in the biogenic
 emission of hydrocarbons by more than a factor of 3. An estimate of the impact  on global OH concentrations
 is difficult because the importance of  the role of NMHC in the global budget  for OH is not well defined.
 However, it is clear that locally, at least, these changes would lead to decreases in OH, although the global
 consequence (for removal of CFCs, CH4, etc.) cannot be estimated at this time.

       Likewise,  temperature increases can impact the bacterial source of CH..  Reaction of CH4 with OH is
 a net sink for OH  hi a low NOy environment, although the net effect of CH4 oxidation on HO  depends on the
 oxidation pathway (Logan et al., 1981). Because the bacterial source of CH. is about 50% or the total source
 (Khalil and Rasmussen, 1983a) and because CH4 emissions are exponentially dependent on temperature, an
 increase of 5°K could lead to almost a 50% increase hi the total (bacterial plus anthropogenic) source strength
 of CH4 (Hameed and Cess, 1983).  The estimated increase in atmospheric abundance of CHjWould be larger
 than the increase in its source strength because the concentration of OH would decrease as CH4 is increased.

       As discussed above, reaction of NO with HO2 is important in partitioning Hg between HO, and OH,
 particularly over continental areas where the concentration of NO is elevated. Bacterial sources of NOX could
 also be increased by an increase in temperature.  Using the relationship of Williams et al.  (1987), a 5 degree
 increase in temperature, for example,  could lead to an increased bacterial emission of about a factor of 3.
 Because the sources of NO^ are so poorly defined (Logan, 1983), the  effect of this change in emissions on
 continental NOX concentrations is uncertain. However, rough estimates suggest  a local increase  hi OH over
 continental areas of approximately 5% for a 10% increase hi NOX (Thompson et aL, 1988).

       Bacterial processes in soils act  as a small sink for atmospheric CO, and emission of CO by plants acts
 as a source. Temperature changes could also impact these sources and produce a small change on tropospheric

       Finally, changes hi tropospheric temperatures themselves create important changes in reaction rates that
 are strongly temperature dependent  In particular, the reaction of CH4 with OH has an activation energy of
 1710°K. An increase of 5°K would increase this reaction rate by 10%. The reaction of OH with NO, is
sensitive to temperature, as are the reactions of OH and HO2 with O3.  Model estimates using the LLNL1-D
troposphere/stratosphere model (Wuebbles, 1981) imply that an increase in temperature by 5°K could decrease
OH by a few percent, if all other concentrations are held constant.


       Any prediction of future changes in global-average tropospheric OH abundances depends on accounting
for the simultaneous  action of the many  coupled HOx-controlling processes outlined above.  Significant
uncertainties area encountered even in investigation of implied trends of OH abundance in the recent past.  In
studies of the CH4 trend, it is considered plausible that a decline in OH abundance (and associated increase in
CH4 lifetime) accounts for 20-80% of the observed increase in atmospheric CH4 abundance over the previous
few centuries (Khalil and Rasmussen, 198Sb; Thompson and  Cicerone, 1986a,b; Thompson and Kavanaugh,
1986; Levine et al., 1985).  But unresolved uncertainties in CO, NO , and NMHC trends, as well as global
distribution of NOX sources, lifetimes, and abundances contribute to the wide range of possible recent change.
Considering the troposphere as a whole, it is thought that the current situation is NO-poor with respect to net
HO  production by CH4 and NMHC oxidation (Logan, 1987).  Projected increases in these compounds are thus
likely to lead to continued decrease in total tropospheric OH abundance, as is shown by Isaksen and Hov (1987)
in a 2-D model study of coupled perturbations to the troposphere.  This average conclusion includes, however,
significant regional diversity, in which some areas of the troposphere may be characterized by OH increase.
Isaksen and Hov (1987) also find that OH is increased slightly in their model when NOX, CO, CH4, and NMHC
are assumed to increase concurrently.  The  changes in OH from the various climate-related perturbations
discussed are summarized in Table 4.1.  Clearly, the changes are subtle and will require fully coupled models
that can address local as well as global changes.

Climate Implications of OH Abundance Change

       The connection between average tropospheric OH abundance, CO and CH4 lifetimes, and climate has
been mentioned above.  If,  as seems likely, CO and CH4 continue to increase,  the average OH abundance
should decrease, subsequently enhancing the tropospheric concentration of CH4 above the value resulting from
the direct effect of increases in the emissions of these species (Penner et al., 1977) and leading to a larger
climate impact.  Other species of radiative or stratospheric photochemical importance (e.g., hydrogen-containing
halocarbons) that are primarily destroyed by reaction with OH will be similarly affected in what appears to be
a positive feedback mechanism. This direct  effect of a decrease in OH, however, is moderated by the increases
in O3 that also occur as a result of CH4 and CO increases (in regions of sufficient NOX) (see, for example,
Isaksen and Hov, 1987, and Thompson and Cicerone, 1986a,b).  As discussed above, increases lead to OH
increases. Thus, to evaluate the effect of these changes it is important to account for all these species in a
coupled system that also accounts for the long-range transport of the longer-lived species from source to remote


       In Chapter 3, we discussed how changes in climate could alter O3 production on urban and regional
scales. One factor discussed there, which could contribute to these changes, was a change in the background
tropospheric ozone concentration.  A change  in background O3 (see also  Chapter 2) is also important for a
change in climate itself. O3 absorbs infrared radiation near 9.6 ft. At current concentrations it absorbs roughly
4% of the longwave energy emitted at the surface of the earth. An increase in tropospheric O3 of 15% could
lead to a temperature  increase  of about 0.1 degrees (Ramanathan et  al., 1987).  Troposoheric ozone
concentrations are the result of a balance between transport of roughly 5 x 1010  molecules cm   s"1 from the
stratosphere (Levy et al., 1985), surface deposition to land and vegetation estimated to be of similar magnitude,
and photochemical sources and sinks whose magnitudes are estimated to be comparable to or larger than the
stratospheric source (NRC, 1984). Because tropospheric ozone abundances can be controlled variously by global
scale transport, regional scale transport, and photochemistry, the distribution and lifetime of tropospheric ozone
are significantly heterogeneous in time and space.  Perturbations  to the photochemical  source  and sink for
tropospheric O3 could come about directly via modulation of the emissions of anthropogenic hydrocarbons and
NOX, and/or indirectly via changes to the natural sources of these species.  Below we discuss first the sources,
sinks, and chemistry of global tropospheric ozone and then a set of mechanisms that could connect changes in
climate to changes in tropospheric ozone.

                            Table 4.1. Estimated Change in Global OH
        Change in OH

10-30% increase in

2 x O3 in troposphere

Increase biogenic NMHC emissions

10% increase in NOX

Simultaneous increase in CO, CH.,
few percent

few percent

-10% over ocean

Current emissions decrease
 OH by a factor of 2 near
 surface; no estimate for
 effect of increased emissions


Decrease in OH
Simultaneous increase in CO, CH4,     Increase in OH
LLNL 1-D model

Thompson et al. (1988)

Thompson et al. (1988)

Trainer et al. (1987)

Thompson et al. (1988)

Isaksen and Hov (1987)

Isaksen and Hov (1987)

Sources and ginlcs of Q.

       The production and accumulation of ozone in the stratosphere produces a mole fraction gradient in the
region of the tropopause down which ozone is transported by eddy motions, mainly at micflatitudes.  This source
of tropospheric ozone is observed directly and can be inferred from observations of radioactive tracers of
stratospheric origin.  Agreement among observations and models of various dimension sets the global average
magnitude of this source in the range of 3 to 12 x 1010 molecules cm*2 s"1 (4-15 x 1014 g/year) (see Levy et at,
1985 and papers quoted therein). This ozone is introduced into a region in which ozone photochemical lifetimes
are long (order of months), so that further mixing by transport within the troposphere occurs over global scales.

       Ozone is known to be taken up on contact with leaf surfaces, soils, and to a lesser extent water and
snow surfaces. Global average estimates of the total surface deposition of ozone lie in the range 5-15 x 1014
g/year (NRC, 1984), essentially the range of the estimated stratospheric source. General circulation model
simulations including the stratospheric source and surface loss, but no tropospheric photochemistry, successfully
emulate in remote regions the large-scale observed features of the free tropospheric ozone distribution and its
seasonal dependence (Levy et al., 1985).

       However, estimates of tropospheric photochemical production and loss magnitudes are, in most cases,
larger than the stratospheric source and surface loss terms (see, for example, Logan et al., 1981). It is clear
that in some regions of the troposphere, with urban smog being the extreme example, photochemical processes
dominate the ozone budget In regions with little NOX, however, net O, loss can occur. The ozone produced
near urban areas can be transported to remote areas, as well, contributing to global O3 production (Vukovich
et aL, 1985).  Thus, the sign and magnitude of the  net  tropospheric  photochemical term for O3 remains

       Tropospheric photochemical ozone loss can occur through the reactions:

                           O3 + hi/ (A< 310 nm)  = O(1D) + O2

                                  O(1D) +  HO  = 2 OH                       (4.6)

                                       OH + O3 = HO2 + O2

                                      HO2 + O3 = OH + 2 O2                  (4.7)

The importance of the second loss mechanism depends on both the HOX and NOX abundance.  The primary
source for tropospheric HOy is reaction (4.6) above. In the presence or sufficient NO, HO9 reacts to form
                                     HO2 + NO = OH + NO2                   (4.8)

The ozone lost by reaction with OH in (4.7) can be regenerated, through reaction (4.8), after photolysis of NO2:

                         NO2 + hi/ (A< 410 nm) = NO + O

                                   O + O2  + M»O3 + M                      (4.9)

Measurements of ozone in the remote marine troposphere, where NO was observed at levels of only 10 ppt
(Davis et aL, 1987; Ridley et al., 1987), imply the importance of local photochemical ozone loss.

        If peroxy radicals are produced by reactions other than the reaction of OH and O3, then reactions (4.8)
and (4.9) provide  a photochemical  source of ozone.   Tropospheric photochemical ozone production is



 qualitatively similar to production in photochemical urban smog.  In general, peroxy radicals are generated in
 the OH-initiated oxidation of CO, CH,, and higher hydrocarbons (NMHC) which are capable of oxidizing NO
 to NO/M leading to ozone formation after NO, photolysis. For CO, one ozone molecule can be produced for
 each CO oxidized. In the complete oxidation 67 CH4 to CO2 and HgO where sufficient NO is present, between
 3 and 4 ozone molecules can be produced for each CH4 reactant molecule,  with HjCO as an important
 intermediate (Logan et  al., 1981).  Generally, higher  hydrocarbon complexity is associated with higher
 ozone-forming potential

        Observations at Niwot Ridge, Colorado, show the importance of photochemical formation of ozone in
 clean continental  air (Fehsenfeld et al., 1983). A clear diurnal variation in the measured ozone abundance is
 seen in the summer, which is related to oxidation of anthropogenic and natural hydrocarbons. Based on a
 theoretical interpretation of their observations, Liu  et aL (1987) propose that summertime photochemical
 production of ozone over the United States as a region may reach 1 x 1012 molecules cm"2 s , an order of
 magnitude larger  than the global average stratospheric source. They suggest also that the accumulation of
 anthropogenic NOX in the winter may contribute to  observed seasonal enhancement of springtime ozone at

        Given the wide range of NOX and O3 photochemical lifetimes encountered in the troposphere and the
 complexity of NOx-related photochemistry, understanding the current global  tropospheric ozone budget will
 depend on the availability of diagnostic models.  These models of photochemistry and transport must include
 the complexity of anthropogenic and  natural source distributions and strengths for NO,, and a variety of
 hydrocarbons as well as boundary  layer and free  tropospheric transport  at a spatial  resolution better than
 existing global circulation models.

 Impacts of Climate Change on Tropospheric Ozone

        In Chapter 3 above, we discussed the potential impact of  climate  change on ozone and oxidant
 formation in urban and regional settings that are often already strongly perturbed by anthropogenic emissions.
 In individual urban cases, die effects of climate change are commingled with the dominant effects of local trends
 in emission characteristics and fluxes. The global sum of all local anthropogenic emissions may also be affecting
 ozone in the previously unperturbed background troposphere  (see, for example, Vukovich et al., 1985). There
 is evidence of an increase in tropospheric ozone over the  last 10-20 years from several Northern Hemisphere
 observing stations with consistent long-term records, including some remote sites (Oltmans and Komhyr, 1986;
 Logan, 1985).  Liu et al. (1987) propose that the statistically significant winter and spring increasing trends at
 Mauna Loa and Hohenpeissenberg are the result of increasing anthropogenic emission fluxes of NOX and
 NMHC directly affecting atmospheric trace constituent composition.

        But as large uncertainties remain in the understanding of the current tropospheric ozone budget and
 controlling processes, predictions of the effects of climate change on tropospheric ozone must be considered
 speculative.  A number of processes related to climate and composition change, including stratospheric ozone
 perturbation and  trends in NO., CH4, CO, and NMHC emissions and budgets, are  potentially capable of
 affecting tropospheric ozone. These processes are important, as with tropospheric ozone itself, over a range
 of spatial and temporal scales. A comprehensive coupled analysis over all scales is a problem for the future, but
 speculations can be made about various aspects.

        Two possible areas in which  stratospheric  properties can  affect tropospheric chemistry are  the
 downward flux of ozone from the stratosphere and the downward flux of solar UV radiation impinging on the
 troposphere.  The magnitude of the source of ozone mixed downward across the tropopause depends on the
 lower stratospheric ozone abundance and the dynamics of the stratosphere-troposphere exchange process. Some
 current photochemical model calculations show that for some stratospheric perturbations of the sort that might
be expected to occur over the next several decades, upper stratospheric ozone depletion is associated with lower
stratospheric ozone increase,  often termed "self-healing" (WMO, 1985). Upper tropospheric ozone might then
be expected to increase, in the absence of dynamical changes  reducing the stratosphere-troposphere exchange


       Decreases in the overhead burden of ozone result in increases in the photochemically active solar flux
that drives ozone photochemical production. Model results simulating clean continental or urban air (Liu and
Trainer, 1987; Whitten and Gery, 1986) show that the increase in actinic flux produces an increase in ozone.
In the remote troposphere with very low NOX, this relationship does not hold, however (Liu and Trainer, 1987).

       The direct impact of composition trends on global tropospheric ozone operates through ozone forming
photochemical precursors. In photochemical model calculations of clean continental or urban air, increases in
NOX,  CH4, CO,  and NMHC are all related to  increased tropospheric ozone, via the reaction mechanisms
discussed above.  There are many subtleties, however, which complicate quantitative predictions.  Liu et al.
(1987) infer from data obtained at Niwot Ridge,  a clean continental site, that ozone production is a nonlinear
function of NOX and NMHC emission strength.  The impact on predictive modeling is that calculations of
tropospheric ozone distributions will be sensitive to details of the dispersion of NO and related reservoir species
such as PAN. Models of lower dimension, in which NOX emissions are effectively rapidly dispersed zonally or
over large regions, can overpredict photochemical ozone formation rates. Isaksen and Hov (1987) have applied
a 2-D tropospheric model to the question of the  interaction of trends in CO, CH4, and NOX and their effects
on ozone and OH over a period representing 1965 to 1995. Using simple assumptions of combined annual rates
of growth or decline, they find that, hi general, the sense and magnitude of ozone change is tied to the assumed
NOX trend.  When NO  is held fixed or increases,  increases  in CO or CH4 result in increasing O, (and
decreasing OH).  When NOX is assumed to decline, O3 decreases also, even though CO, CH4, and NMHC are
concurrently increasing.

       These results and those for tropospheric  OH clearly demonstrate the importance of understanding the
global distribution of NOX in three dimensions over the range of scales from urban to global.  Of particular
pertinence is the significance of long-range transport of NOX in the form of organic nitrates and peroxynitrates.
The summer lifetime for NO  is on the order of a day or less in the planetary boundary layer. The lifetime of
PAN with respect to thermal decomposition is highly dependent on temperature and can be on the order of
weeks or months in the middle troposphere. Singh et al. (1985) have speculated that PAN could serve to convey
NO  released in the boundary layer to the remote troposphere, where it plays  an important role in affecting
the background global abundances of OH and ozone.


       We  summarized  in  Chapter 2 the trends  in atmospheric  composition that are the drivers for
perturbation of atmospheric radiative transfer and climate change. Chapter 3 and Chapter 4 above discuss the
effects that the expected climate change may have  on regional and global tropospheric photochemistry.  The
temperature dependence of CH4 and NMHC emissions by plants and soils mentioned  above  demonstrates
another level of interaction. The changing emissions of trace species that are expected to drive climate change
are also functions of climate change variables, through both natural and anthropogenic (economic) mechanisms.

       A distinction can be made, in considering the interaction of climate change and anthropogenic source
gas emissions, between the effects of actual climate change and those of anticipated climate change. Anticipated
climate change can bring about reductions, which may or may not be mandated, in production and emissions
related to efficiency improvements in equipment and processes, exploitation of alternatives and straightforward
regulation of emissions. A current example is the Montreal Protocol on the production of several halocarbon
compounds thought to act to deplete stratospheric ozone abundance. The decreases (or slowing of the rates of
increase) in atmospheric abundances of several CFCs and the related stabilization of stratospheric ozone, which
are expected to result from application of this agreement, will affect the future global atmosphere hi the areas
of surface and tropospheric ultraviolet fluxes, stratospheric climate, and surface-tropospheric climate.  It is
possible that means may be found to reduce combustion-related emissions in anticipation of global warming.


        Analysis of the actual effects of climate change on anthropogenic emissions is far beyond the scope of
this report, but we offer a few general speculations. Domestic energy consumption is a function of ambient
temperature and should respond to changes in the temperature frequency distribution.  Changes in temperature,
frost dates, precipitation patterns, and water availability can certainly affect the choice of crops and could, for
example, affect methane emissions.

        Natural biospheric sources of CO^ CH4, CO, NMHC, and NO will also be affected by climate change.
The CO2 atmospheric abundance record shows both the seasonal variability of photosynthesis and respiration
patterns of plants and variability related to ENSO climate perturbation events (WMO, 1985).  It is not dear
whether long-term temperature and precipitation changes can  produce a secular trend in the distribution of
carbon between the atmosphere, the biosphere,  and the oceans, which would affect the CO2 atmospheric
abundance trend  For CH4, wetland emissions are thought to be die major natural source.  Clearly changes in
the area! extent of wetlands resulting from climatological changes in the hydrologic cycle, as well as  the
temperature dependence of biologic activity, will affect the CH. source flux. This effect could be exacerbated
by the enhanced warming in tundra and permafrost regions that is predicted by some general circulation models.

        Finally, other interdependences  of photochemistry and the climate system can be envisioned which
cannot be quantified at present.  For example, cloud cover plays a major role in atmospheric radiative transfer
in both the visible and infrared wavelength regions. Biological activity in the oceans produces the short-lived
sulfur species HJS and dimethyl sulfide.  They are oxidized in part to produce sulfate aerosol particles, which
act as cloud condensation nuclei. If the  fluxes of these species are affected by warming of the ocean surface
water, changes in CCN concentration and cloud  cover or properties could result The DMS/cloud/albedo
interconnection has been invoked recently to explain the Cretaceous-Tertiary extinction event hi the following
manner (Rampino and Volk, 1988).  If a large proportion of ocean plankton were suddenly killed by darkness
(following comet impact), the DMS emissions,  CCN, and global cloud cover and albedo would all decrease.
The Earth would then warm substantially, disrupting the saurian ecosystem.  This is, of course, conjecture, but
it illustrates the complexity of the climate system.




        So far we have discussed a number of interactions between climate change and air chemistry that may
affect the future state of the atmosphere.  In this section we focus our attention on the availability of models to
treat these scientific issues. Our purpose here is twofold.  We want to assess the capabilities of present day
models, and also to advocate construction of new models that are needed to improve the assessments of effects
discussed above.  To obtain better predictions of the effect of climatic changes upon global, regional, and local
air chemistry, we must combine a number  of different models that simulate different physical processes and
operate on varying scales-for example, global-scale climate models, regional-scale weather  models, and
chemistry models. The question of how to  couple models in a workable fashion is dealt with in the final part
of this section.

        Why use models?  Put bluntly, the answer is that we have no choice if we wish to predict the future, or
even to predict a range of possibilities for the future (the typical situation in environmental forecasting).
Measurements of relevant quantities in the laboratory and in the field are indispensable for understanding the
atmosphere.  Such measurements, of course, cannot by themselves tell us what to expect as Earth's climate
changes; to answer that question we must use some sort of theory or model to extrapolate from the present to
the future.

        Models in science can be extremely simple, but in atmospheric science they are often complex enough
to require substantial use of supercomputers. Such is often the case with the models discussed below. These
models are physically based; that is, they solve mathematical equations that express the laws of physics and the
experimentally determined behavior of atmospheric constituents (e.g., chemical reaction rates).  Accordingly,
one might hope for the construction of one "perfect" model encompassing all relevant atmospheric processes,
which would yield detailed and reliable forecasts of atmospheric conditions at  all  points on the globe.  In
practice, however, such  a  goal is not feasible.  A  model that kept track  of all chemical and atmospheric
properties of interest on local scales, and that extended over the entire globe, would overwhelm the capacity of
even the best present day or foreseeable supercomputer. And even if a "super-supercomputer" were to arise,
uncertainties in the input  to  such a  machine—such as chemical reaction rates—would  probably render the
"perfect" model's output questionable.

        Fortunately, there  is a realistic alternative.  A hierarchy of models have  already been developed and
applied to the various atmospheric processes (Schneider and Dickinson, 1974). For processes of interest in this
report, the models have served to conceptualize theoretical understanding and thereby provide a foundation for
the construction of coupled models. As we shall see, joining different models must be approached with caution,
but could result in a useful degree of predictive power for effects connecting urban, regional, and global scale
air chemistry and global scale climate.

        Prediction of air quality on regional and local scales from the global climate obviously requires models
which resolve the variability of wind, temperature, and other parameters of interest over the three dimensions
of latitude, longitude, and altitude. A detailed discussion of these three-dimensional models will be found below.
First, however, it is appropriate to mention models with more limited resolution; these have been and will
continue to be useful in identifying key factors affecting global climate and its simulation by three-dimensional
models. Table 5.1 depicts the hierarchy of one-, two-, and three-dimensional models. Most climate models fall
into one of three categories: energy balance models (EBMs), radiative-convective models (RCMs), and general
circulation models (GCMs).   EBMs, as their name suggests, compute the average surface temperature by
assuming a balance between solar energy absorbed by the Earth-atmosphere system and infrared radiation (IR)
radiated by Earth to space (North et aL, 1981); typically this calculation is performed for a number of latitude
bands from the Equator to the Poles, but the variation of temperature with altitude is not explicitly considered.
Thus, EBMs generally cannot handle the vertical exchanges of IR and solar radiation, or the vertical transport
of heat by means of turbulent convection, by any means other than crude empirical parameterizations. RCMs,



                                             Table 5.1

    TYPE               TYPICAL RESOLUTION             LIMITATIONS

                   Altitude     Latitude     Longitude
none 5°-10°
0.1 km none
1-5 km 2°-10°
1-5 km 2°-10°
overly empirical
no horizontal variations
no longitudinal variations
sub-gridscale processes
in contrast, are specialized to calculate just such radiative and convective transfers of energy (Ramanathan and
Coakley, 1978)--but at the price of averaging over all horizontal variations, i.e., lumping land and ocean areas
into one "average" surface type and ignoring the significant variations in solar energy amounts that are received
at different latitudes, seasons, and times of day.

        GCMs simulate not only radiative and convective energy transports, but also atmospheric circulation
(winds), together with temperature,  cloudiness, and precipitation over the globe; these are the most detailed
models available to simulate  climatic change (Saltzman, 1983).  Some GCMs are  two-dimensional, resolving
their parameters in latitude and altitude, but not longitude.  These models can either compute atmospheric
circulation from the input of solar energy, with assumed constraints on atmospheric temperature (Wuebbles et
al., 1987) or compute both atmospheric temperature and circulation self-consistently (Saltzman, 1978); in both
cases, however, longitude-varying features ("eddies") cannot be explicitly calculated  and must, therefore, either
be parameterized using approximate empirical relationships or ignored altogether. Two-dimensional circulation
models have been particularly useful in investigating connections among solar and infrared energy transfer,
atmospheric circulation, and chemistry (Wuebbles et al., 1987).  It has been difficult to address such problems
with three-dimensional GCMs without consuming prohibitive amounts of computer time.  However,  many
important atmospheric phenomena-for  example,  the  development and movement of the  "fronts" that
characterize midlatirude weather systems-cannot be represented in a model that ignores or parameterizes
eddies.  This limitation may apply to important physical processes that transport chemical species. Thus it is
likely that in the  future, efforts to  understand the interconnection between climate and air chemistry will
increasingly make use of three-dimensional general circulation models.


        Three-dimensional GCMs are essentially the same as the computer models used for  modern weather
forecasting (Haltiner and Williams, 1980). The output of GCMs, however, is used not to forecast the occurrence
of specific weather events, such as  a storm occurring at a particular time and  place, but rather  to  study
longer-term changes in climate that occur over broader scales, such as changes in globally averaged temperature,
or regional drought. A key assumption in GCM studies is the notion that while the forecast of specific weather
events beyond a few weeks into the future is impossible, the prediction of more general climatic trends is

        GCMs calculate in straightforward fashion the time evolution of atmospheric structure and circulation,
using well-known laws of classical physics such as conservation of mass, energy, and momentum. To solve the
equations on a computer, however, it is necessary to divide the atmosphere into a large number of boxes and
to solve  for only the average temperature, wind, etc, within each box. With today's computers each box can



be made no smaller than roughly a kilometer in height and a few degrees of latitude and longitude in its
horizontal dimension; otherwise there would be too many boxes for the computer to handle in a reasonable
amount of time. Therefore, present day GCMs must either ignore the "sub-gridscale" phenomena that take
place on horizontal scales smaller than roughly a hundred kilometers, or parameterize these phenomena in
terms of empirical relationships and grossly simplified physics,  rather than explicitly simulate them from
well-understood physical laws.  Such phenomena include many of concern to this report, including the formation
of clouds and the turbulent mixing of atmospheric pollutants near the planetary surface (Figure 5.1).

        What justification, then, do we have for believing the results of GCMs? As we shall see, there are many
details of GCM simulations of climate that are flawed, but there are good reasons for optimism about the ability
to predict climatic change with GCMs.  The fact that GCMs correctly simulate the cycle of the seasons, which
is in effect a climatic change  driven by known forcing (the variations in the amount of sunlight received at
different latitudes over the  year), is one confidence-builder (Schneider and Lender, 1984).  Another is the
recent series of successes in GCM simulations of past climates that are known through the geologic record
(COHMAP, 1988) and even, to a limited  extent, of the atmospheric states of other planets (Leovy, 1985).

        It is therefore not unrealistic to hope that GCMs can provide analysis of global and regional climatic
change of relevance to local, regional, and global air chemistry. GCMs are the most "complete" models available
for studying the atmosphere and climate in all its aspects—the transmission of solar and infrared radiation, the
generation of winds ("dynamics"), the hydrological cycle, and the interaction of the atmosphere with the oceans.
GCM simulations of some of these processes are available only in crude form and require further development,
but even in their present state GCMs are useful for studying interactions among the various components of the
climate system. The following discussion  of GCM flaws and limitations should be taken in a spirit of cautious

        The main impediment to using GCM simulations of climatic change on a regional scale is that present
day models  give inconsistent results on this scale.  As an example, we consider the increase in surface air
temperature (AT) caused by the greenhouse effect of doubling the atmospheric concentration of carbon dioxide.
The globally averaged temperature increase obtained by different GCMs spans the range from roughly 2°C to
5°C (NRC, 1982) with more recent models converging toward the higher end of this range (Schlesinger and
Mitchell, 1985; Manabe, 1986; Wilson  and Mitchell, 1987). As we consider increasingly detailed  aspects of
climate prediction, however, the agreement among the models diminishes and finally disappears. Grotch (1988)
has compared AT among the more recent models, using as a quantitative measure of consistency the statistical
correlations between results for different model pairs; 100% correlation would indicate perfect agreement, while
scores much less than 50% indicate essentially no agreement. Grotch finds that for the Northern Hemisphere
winter season, correlation scores for broad latitude bands are in the range of 80-95%, but when more limited
areas, 12" latitude by 15° longitude, are considered, the correlations drop to 50-80%; these fall further to 40-70%
when 40° by 5° areas are used for comparison between the models. In other words, although the models display
considerable agreement on global and broad continental scales, they show virtually no agreement when their
results are compared over smaller areas—for example, individual states in the UJS.~which are precisely the
scales we are interested in for  predictions of local and regional air quality.

        Of course, consistency among models is not by itself a sufficient condition for accuracy of simulation
(consistent models could all be equally wrong) but it is surety necessary to first understand why the models give
different results.  This problem is being directly addressed  in  a major initiative, sponsored by the  U.S.
Department of Energy, to intercompare the GCM-simulated climatic changes (Grotch, 1988).  Since the models
all solve the same equations  for large-scale atmospheric circulation  and temperature  structure, the  fault
presumably lies  in  the  parameterization of  small-scale  processes  discussed  above.   Sub-gridscale
parameterization has been identified as responsible for one inconsistency among GCMs, the presence or absence
of summertime drought in the  American Midwest as a consequence of CO, doubling (Manabe and Wetherald,
1986; MacCracken et aL, 1986). Mitchell and Warrilow (1987) and Meehl and Washington (1988) conclude that
the appearance of this drought in GCMs depends critically upon the assumptions made about the ability of the
soil to hold moisture.  In the words of Mitchell and Warrilow (1987), "the magnitude of them simulated summer
drying is dependent on the [assumed] physical attributes of the soU, and in certain regions can be reduced or




Con.'«rvation laws for mass,
   Momentum, energy and
   water vapor
                 Radiative transfer equations
      Exchanges of energy,
      momentum and H2O
                                                    Subgrid-scale mixing
Physical constants


Atmospheric composition
    Conservation laws for
       energy, water, snow
       and sea-ice
                                    Water run-off
                                    Sea-ice growth and
                                    Ocean mixing

                                    Deep sea heat storage
   Surface albedo

   Surface roughness
    Figure 5.1. Processes and interactions represented in a typical climate model (from Taylor and Crotch, 1987).


even reversed by an alternative, but equally plausible, treatment of run-off [of water into streams and rivers]..."
The lesson is that seemingly minor differences in a model's parameterization of small-scale processes can lead
to major errors in he model's simulation of regional and local details of climate.

        A number of processes come to mind as candidates for improved parameterization. Some are of direct
import to the transport and dispersal of atmospheric pollutants, such as turbulence in the boundary layer near
the  planetary surface (Deardorff, 1972;  Mellor  and Yamada, 1974) and convection and cloud formation
(Arakawa and Schubert, 1974; Kuo, 1974). Others do not directly influence local and regional air quality but are
significant factors in the determination of climate, for example, in addition to the question of soil moisture
discussed above, the behavior of sea ice (Semtner, 1984) and the upper mixed layer of the ocean (Niiler and
Kraus, 1977).  A variety of workable parameterizations for GCMs are given in the references cited above; what
is needed now is a systematic intercomparison of how the parameterizations affect GCM simulations in detail.
Of course, the GCM results must be compared not only with each other but also with the real world's climate.
Toward this end an enhanced global observing system, probably emphasizing remote sensing from space, is
needed (NASA, 1986).

        Finally, we must emphasize that realistic time-dependent analysis of climatic change is not generally
available from GCMs.  For example, the simulations of CO2 greenhouse warming discussed above all refer to
an equilibrium response; in the simulations, atmospheric CO, concentration is abruptly doubled and maintained
thereafter  at a constant level until the model achieves a steady-state climate, which is then analyzed as the
"response to CO, doubling."  The actual climatic change the world undergoes in response to steadily increasing
carbon  dioxide is a complicated affair,  with different  components of the climate system—air, land, and
ocean—approaching equilibrium at different rates; the resulting climate patterns do not necessarily resemble
the equilibrium response (Schneider and Thompson, 1981). Simulations of this time-dependent process are
difficult to perform because it is  necessary to consider not only the circulation  of the atmosphere but also the
circulation of the ocean  as it slowly responds to the new climate; what is required is an ocean GCM coupled to
an atmospheric GCM, an unwieldy  combination that consumes great amounts of supercomputer time.  The
small number of coupled time-dependent ocean/atmosphere simulations performed to date (e.g., Spelman and
Manabe, 1984; Nihoul,  1985;  Sperber et al., 1987) involve unsatisfactory simplifications such as idealized
geography (Thompson and Schneider,  1982) and limited run times.

        In summary, GCMs are essentially the same weather prediction models that have been extensively used
for decades-with steadily increasing accuracy (Haltiner and Williams, 1980)-by national weather services around
the world.  GCMs are used to forecast not specific weather events but rather more general climatic trends.
Their ability to do so is difficult to determine since the opportunities for comparison with the real world are far
more limited than is the case with numerical weather prediction, but indications are that current GCMs give
generally reliable climate simulations for the largest spatial scales, that is, the scales of large continents or of the
entire globe.  On the smaller scales of  interest in this report, simulations are not yet reliable, but correction of
the problem may be found by a thorough testing of sub-gridscale parameterizations, a straightforward task given
commitment of sufficient resources.


        Development of computational models over the last two decades to study atmospheric chemical
processes has revolutionized our understanding of global atmospheric chemistry. These models have allowed us
to analyze, as well as appreciate, the nonlinear interactions between atmospheric chemical processes. They give
us insight into the complexities of the interactions of these processes with radiative and dynamical processes in
determining the distributions of  atmospheric trace constituents.  Global atmospheric models have also been
extensively used in attempts to understand the impact that human activities are having on atmospheric chemistry,
in particular on ozone concentrations hi the troposphere and stratosphere.


        Until the last few years, the one-dimensional (1-D) horizontaOy averaged model has been the mainstay
for theoretical chemistry studies of the troposphere and stratosphere. Because of their computational efficiency,
such models can treat the full complexity of the chemistry schemes needed to simulate these regions. Typical
calculations can involve the simultaneous, coupled interactions of more than 30 species and over a hundred
chemical and photochemical reactions, with detailed calculations of the diurnal variations in the solar flux as a
function of altitude. Radiative processes can also be accounted for in these models, and it is now standard for
temperature feedbacks on atmospheric chemistry to be included.

        A severe limitation of the 1-D model lies in its crude representation of atmospheric transport processes.
These models parameterize all transport effects in terms of vertical diffusion.  This diffusive treatment is purely
empirical, with a vertical diffusion coefficient based on observations of the temporal and spatial distributions of
selected tracers.  The derived diffusion coefficient is then applied equally to all atmospheric constituents hi
current models,  although methods  have  recently been proposed (Holton,  1986)  for species-dependent
dynamically based transport parameterizations.

        A  further limitation  of 1-D models is  the  obviously implied use  of  average global rates  for
photochemical reaction rates.  Spatial correlations of short lifetime species cannot be directly simulated and this
complicates interpretation of 1-D results, leading to controversy over the meaning of calculated distributions of
various species.  Nonetheless, the 1-D model remains a useful theoretical tool because it works—simulating well
many of the important features in the vertical distributions of atmospheric constituents and comparing well with
the globally averaged results of two-dimensional models for perturbed atmospheres.

        Zonally averaged two-dimensional (2-D) models  are now becoming the workhorse for studies of the
global atmosphere.   While the 1-D model can only provide information on the horizontally and annually
averaged vertical  profile, the  2-D  model includes the important latitudinal  and seasonal dependences in
atmospheric processes. Two-dimensional models can, in principle, simulate the effects of zonally averaged
meridional and vertical mean and eddy transports, while also accurately accounting for the meridional variations
in photochemistry.

        Recent 2-D models can include essentially the same chemistry and radiative transfer treatments as have
been  traditionally associated  with 1-D  models.  As an  example, the  LLNL two-dimensional  chemical-
radiative-transport model calculates the concentrations  of 30 atmospheric trace constituents at about 300 grid
zones extending from pole to pole and from the ground to the stratopause.  A complete radiative transfer
submodel is used  to determine solar heating and infrared cooling rates that then drive the diabatic circulation
in the model  and  allow for trace species feedback effects on stratospheric temperatures.  This and other 2-D
models are being applied to extended time-dependent studies of anthropogenic effects  on global ozone hi
research sponsored by NASA.

        The ultimate fidelity of the two-dimensional model depends primarily on the accuracy with which  the
zonally averaged transport of trace constituents can be represented, and is limited by spatial and temporal
heterogeneity of  short-lived species.   Theoretical advances have  produced  major improvements in  the
understanding and formulation of 2-D models in recent years. The theoretical basis for 2-D models, along with
their associated limitations, is reasonably well understood Significant uncertainties remain, especially regarding
the treatment of eddy transport processes in these models, including the  treatment of interactions between eddy
and mean transport  processes.  Existing two-dimensional models do, however, provide a generally good
description of the spatial and temporal variations of trace constituents, within the limitations  of current
observational data bases.

        While the twordimensional model  has been brought to the level of being a sophisticated tool  for
diagnostic and  assessment studies, its fundamental limitations suggest  that we should  ultimately turn  to
three-dimensional (3-D) models. Besides the difficulty hi treating three-dimensional atmospheric dynamics hi
1-D and 2-D  models, these models are especially inadequate in the study of photochemical  processes in the
lower  troposphere. The short  chemical lifetimes of many lower atmospheric constituents when combined with
the significant variations in land-based versus ocean-based source strengths  require the development of 3-D



chemistry models. The development of these models has been slow, however, because of the computational
expense in treating realistic global atmospheric chemistry with a three-dimensional model.  As a result, while
there have been many studies of stratospheric chemical processes, theoretical modeling studies of chemistry
processes in the troposphere have been limited in scope. While many studies with 1-D models exist and a few
with 2-D models (e.g, Isaksen and Hov, 1987; Thompson and Cicerone, 1986; Thompson et al., 1988; Liu et al.,
1987), global tropospheric chemistry studies are necessarily limited by the unavailability of 3-D models.

        Although 3-D models with complete chemistry will not have meaningful results for several years, such
models are being developed and already are being applied, in a limited manner and with limited chemistry, to
scientific problems.  Walton et aL (1987), using the LLNL 3-D model GRANTOUR,  have investigated the
effects of anthropogenic NOX emissions on nitrate deposition on a global scale. Prather et al. (1987) have used
the 3-D tropospheric chemistry model being developed at NASA GISS to study the  relationship between
emissions of chlorocarbons and observations expected at various sites. Each of these models uses winds,
temperatures, and other input parameters based on calculations from GCMs.

        In the future, it will be possible to fully couple 2-D and 3-D models with GCMs and other climate
models.  This coupling win be important to determining the role of chemistry-climatic feedbacks on the
atmosphere, and to determining the effects that future climate change may have in modifying global chemistry,
particularly to ozone and OH concentrations in the troposphere and stratosphere.

        The above discussion has highlighted many of the current capabilities and limitations in modeling of
global atmospheric chemistry. In addition, there are many remaining uncertainties that also affect our ability to
model global chemistry. Uncertainties remain in the reaction rates and chemical schemes used in these models.
Feedbacks between chemical processes and other atmospheric processes represented in the models are still
poorly understood For meaningful evaluations of future climate and chemical interactions in the  atmosphere,
further research and model development are required to overcome current limitations and to reduce remaining


        Many different types of models have been developed to study atmospheric chemistry problems on urban
to regional scales.  These range from  simple box models, with no consideration of meteorology effects, to
complex three-dimensional models with internally derived mesoscale meteorology. In general, the complexity
of actual meteorological structure, the effects of topography, the variability in emission sources, and the
nonlinearity of relevant chemical mechanisms have indicated that, for many problems, simple box and trajectory
(or Lagrangian puff) models are insufficient tools and that higher dimensional models are required. For this
reason,  our discussion of urban and regional models will be confined to describing  the  capabilities  and
limitations  of existing multidimensional  Eulerian  models that  include detailed treatments  of emissions,
atmospheric chemistry, transport, and other relevant processes.

        Figures 5.2 and S3 give an indication of the complex interactions of atmospheric processes that are
treated  in urban/regional models.  Current models have achieved a high degree of sophistication in  their
treatments of these processes.  Some of the more important examples of urban and regional models are the
Regional Acid Deposition Model (RADM) developed for the UJS. Environmental Protection Agency at the
National Center for Atmospheric Research and at SUNY Albany (Chang et al., 1987), the Regional Oxidant
Model (ROM) developed by the Environmental Protection Agency (Lamb, 1982,1983,1985,1987), the Regional
Transport Model developed by Systems Applications Inc. (Liu et aL, 1984; Morris et aL, 1987), and the
Livermore Regional Air Quality model (LJRAQ) developed at Lawrence Livermore National Laboratory
(MacCracken et aL, 1978; Penner and ConnelL 1987). Tables 5.2 through 5.5 give a short description of each
of these models.  These tables provide  a guide to comparing the different approaches used by each group in
modeling of urban and regional air quality.

                                                            AND SURFACE
  Figure 5.2. Elements of a mathematical model for relating pollutant emissions to ambient air quality (from
            McRae et al., 1982).

& ^
boundary ~~

• 3-d arfvection
• vertical

Regional Acid
Deposition Model

ir«c« gn
1^ concentration
— ^ & deposit

Cloud effects) IDry deposition) IGas chemistry 1
• vertical
• luriwlent •
redistribution • sublayer
• wet removal
• aqueous

• surface
reshttncei •

36 species including
22 organic*
77 reactions
diurnal, seasonal.
latitudinal, height
varying photolysis
rates with
cloud effects

i ,
| Emissions)
• NOandNOa
•so, and see
• voC(IOclanes)
• NH.
• diurnal, season!.
• point and are*
Figure 53. Overview of a regional add deposition model - RADM (from Chang et al., 1987).


        The primary difference between urban models and regional models is the physical domain they attempt
 to simulate. An urban model, such as the one at LLNL mentioned above or the model at the California Institute
 of Technology (McRae et al., 1982), considers an area about 200 km by 200 km or smaller, a size appropriate
 to analyzing the production and transport of photochemical oxidant and other trace constituents important to
 air quality in an urban area on a single or several-day basis. Regional models, such as RADM or RTM, simulate
 atmospheric chemical and physical processes on much larger domains, on the order of 2,000 km by 2,000 km,
 about the size of the eastern third of the United States. These models are also used to study oxidant formation,
 with coarser resolution than urban models over these larger domains, and can, when appropriately designed, be
 used  to  study regional  acid  deposition ("acid  rain").   Each  of the  models described  here  is  spatially
 three-dimensional, containing multiple vertical layers to account  for the effects of variations in transport and
 mixing with altitude, and to include interactions with the free troposphere.

        Each of these models is designed to consider the effects of area and point source emissions of pertinent
 pollutants and pollutant precursors, chemical transformation processes, meteorological conditions influencing
 horizontal and vertical transport and dispersal, natural background trace gas concentrations, and dry deposition.
 In addition,  models designed to study  regional acid  deposition, such as RADM, also include  detailed
 representations of effects associated with clouds such as aqueous chemistry, scavenging, and sub-gridscale vertical
 transport of trace species. In general, the models designed primarily to study oxidant formation are not designed
 to study the formation and deposition of acidity; however, simulating either oxidant formation or acid deposition
 necessitates that similar treatments of many of the chemical and physical processes be included in both types of

        The primary drivers in the simulation of air  quality in these  models are the inventories of emitted
 species and of meteorological conditions used as input parameters to the model derivations. Uncertainties in
 either the emissions or in the wind speed and direction can have a significant influence on the derived trace
 constituent concentrations.  For example, Morris et al. (1987) suggest that their  model's tendency in some
 calculations to underpredict ozone peak levels is likely related to deficiencies  in the complex-terrain wind fields
 and in the inability of the model  to  resolve  local NO  emissions. Many  of the models have  based their
 representations of meteorology on interpolations and analyses from available observations. However, because
 spatial and temporal resolution of these measurements is often not sufficient, newer models such as RADM
 are using primitive equation meteorological models to generate the required input information. These mesoscale
 models allow for better internal consistency of dynamical, thermodynamic, and hydrologic processes. However,
 uncertainties still exist with this approach, as predicted meteorological fields tend to diverge from observed values
 after a few days unless observed data assimilation approaches are used (NAPAP, 1987).

       A major problem with urban/regional  models, particular in the assessment of impacts  from climatic
 change, is their limitation to simulating episodes of only a few days. Computational expense and the lack of
 sufficient high-quality input data tend to limit the model calculations to being run episodically.  The difficulty
 of representing seasonal to annual changes in climatology with such episodic-based models has not yet been
 resolved. Likewise, aggregating a subset of events to determine longer-term  effects has not yet been shown to
 be a viable approach.

       Overall, the current urban/regional models have demonstrated the capability to simulate many chemical
 and physical processes important to the transport and transformation of trace  constituents determining air
 quality. However, many uncertainties still exist The detailed chemical mechanisms used in these models have
uncertainties associated with the kinetics of NMHC species, particularly aromatics,  and with heterogeneous
chemical processes. Parameterizations of gas and aqueous phase chemical interactions, the effects of clouds, and
dry deposition need to be verified by observations.  As discussed earlier, uncertainties in externally based input
parameters, such as emission rates, initial and boundary conditions, temperature, winds, and other meteorological
conditions,  can produce errors in the model-derived species distributions.


             Table 5.2.  The Regional Oridant Model - ROM (Shere, 1988)
Physical Domain
     60 grid cells E-W
     42 grid cells N-S
     3 prognostic, 1 diagnostic vertical layers of spatially and temporally variable depth
     (<3000 m agl)
     Typical domain size 1100 x 800 km.

Spatial Variables
     A/:  horizontal curvilinear coordinates from grid origin in degrees.
     z :  distance above ground level in meters.

Grid Resolution
     Horizontal:  1/6 degree latitude by 1/4 degree longitude
     Vertical: Generally layers 1 and 2 represent the mixed layer during the day. Layer
     1 is also used to represent any internal boundary layers present. Layer 3 is the cloud
     layer, extending from the cloud base to the tops of any convective-type clouds present.
     The depth of the diagnostic layer 0 is typically on the order of 30-50 m.

Chemical Scheme
     Carbon bond 4 (CB-4.0 with 28 chemical species.
     Hydrocarbon classes:   ETH (ethylene), OLE (olefins), PAR (paraffins), FORM
     (formaldehyde), ALD2 (higher aldehydes, TOL (toluene), XYL (xylene and other
     aromatics),  ISOP (isoprene), and  NONR  (non-reactive hydrocarbons  except

Homogeneous Chemistry
     70 reactions.

Heterogeneous Chemistry
     Sulfur compounds and primary and secondary aerosols not currently included.

Atmospheric Structure
     Determined from observed data. Twice daily radiosonde reports and hourly surface
     meteorological data determine the basic meteorological variables. Temporal and
     spatial interpolation of  observed variables: temperature, dew point, pressure,
     horizontal winds, and cloud cover and types.

Photolysis Treatment
     Photolysis rate constant for NGL, O~ (O(1D)path), FORM (radical products), FORM
     (stable products), and ALD2 (radical products) are determined individually during
     execution as functions of solar zenith angle and grid cell elevation. Table lookup of
     dear sky values (91 zenith angles and 10 elevations).  Solar flux values from Dave
     radiative transfer model. The final values are scaled by a cloud transmissivity factor.
     Other values are linearly related to NO2 photolysis constant.

Wind Field Generation
     Hourly gridded horizontal wind fields are determined from interpolations of observed
     winds, based on mass conservation and minimum kinetic energy constraints. The only
     exception is for layer 1 during nighttime conditions when a prognostic shallow-water
     fluid model is used to determine the wind flows in the stable inversion layer closest
     to the ground



               Table 5.2. The Regional Oridant Model - ROM (Schere, 1988) (continued)

       Transport Treatment
            Horizontal transport and diffusion are solved numerically in ROM with a biquintic
            polynomial technique that is independent of the Courant condition. Vertical exchange
            between layers is accomplished by mass conservation and the  specification of
            turbulent vertical velocities at the layer interfaces. A unique convective cloud flux
            is also able  to transport material from near the ground directly into the base of
            growing cumulus clouds in the third model layer, with  little entrainment  into
            intervening layers.

       Chemistry Solution
            Predictor-corrector version of the QSSA (quasi-steady state algorithm) that  also
            computes a variable time step based on the photochemical lifetimes of key species.
            No explicit steady states are used and all species are transported.

       Time-dependent Integration
            The modeled concentrations are given by:
            where F is the advective transport solution, 7 is the normalized chemistry solution,
            and ^ is the normalized vertical flux solution.  Each component is solved by forward
            time differencing with algorithms and time steps unique to each.  Integration of the
            vertical flux and chemistry solutions  is performed at 5-minute  intervals, and
            integration of all three components is performed at 30-minute intervals.

       Boundary Conditions
            Lateral boundary concentrations for ozone are determined from selected monitoring
            stations located near the upwind edges of the domain and are updated every 12
            hours.  Values for other species are assumed to be near tropospheric background
            values, subject to the constraint that they be brought into chemical equilibrium with
            each other and with the specified ozone value. Boundary conditions at the top of the
            domain are assumed to be spatially and temporally invariant, and are the chemically
            equilibrated tropospheric background values.

       Heterogeneous Deposition
            No heterogeneous deposition.

       Surface Deposition
            Resistance theory model with deposition velocities taken from the literature.

       Diurnal Effects
            Incorporated  in vertical layer structure.   Layer  1,  during the  day,  models
            approximately the lower 100 m of the well-mixed layer; at night, the layer expands to
            represent the radiative inversion layer, capped by a nocturnal jet  Layer 2, during the
            day, represents the bulk of the mixed layer; at night, it is a transport layer for the
            photochemical products. Layer 3, during the day, is the cloud layer, including direct
            vertical entrainment of near-surface air; at night, the layer is a transport layer for this
            material as well as photochemical products.


       Table 5.2. The Regional Oridant Model - ROM (Schere, 1988) (continued)

Snb-gridscale Processes
    Layer  0 attempts to model  the expected  sub-gridscale  variations of modeled
    concentrations based on the non-homogeneity of emissions sources within a grid cell
    No transport is considered in this layer and NO scavenging of ozone within source
    plumes is  considered to  be  instantaneous.   Concentrations  and concentration
    variances are estimated in this diagnostic approach.


                       Table S3.  The RTM-H1 Model (Gery and Morris, 1988)
       Physical Domain
            User specified. Examples include:
            Eastern UJS. (2080 x 1840 km)
            Western Europe 34° x US' (2232 x 1388 km)
            Midwest/Southeast U.S. 42° x 35° (3820 x 3885 km)
            Central California Phase I (210 x 340 km)
            Central California Phase U (520 x 320 km).

       Spatial Variables
            3-D fields of constituents, T, HjO (g), layer depths, u, v, w
            2-D fields of surface wind speed,  precipitation, cloud cover, emissions, surface

       Grid Resolution
            User specified, to date 10-55 km.

       Chemical Scheme
            Carbon Bond 4 (CB-4.0) with 28 chemical species
            Hydrocarbon classes:  ETH (ethylene), OLE (olefins), PAR (paraffins), FORM
            (formaldehyde), ALD2 (higher aldehydes), TOL (toluene), XYL (xylene and other
            aromatics),  ISOP  (isoprene),  and  NONR (non-reactive  hydrocarbons  except
            Expansion to 33 species underway.

       Homogeneous Chemistry
            64 reactions, expansion to 81 reactions underway.

       Heterogeneous Chemistry
            Not treated

       Atmospheric Structure
            Homogeneous tropospheric model

       Photolysis Treatment
            Light and  spectral distribution vary duimally (zenith angle).  Aerosol loading,
            molecular scattering, ozone column density and albedo also considered.  Photolysis
            varies with light intensity and the spectral distribution for each species.

       Wind Field Generation
            Application dependent
            Objective analysis (Eastern UJS.)
            Dynamic meteorology models (Western Europe, Midwest/Southeast UJS.)
            Hybrid diagnostic/dynamic models (Central California).

       Transport Treatment
            Sharp and Smooth Transport Algorithm (SHASTA).

       Chemistry Solution
            Quasi-Steady-State relationship with Crank-Nicholson integration scheme.


          Table 53. The RTM-ffl Model (Geiy and Morris, 1988) (continued)

Time-dependent Integration
    Time steps range from 3 to 30 minutes depending on stability and convergence

Boundary Conditions
    User specified, may be spatially and temporally varying.

Heterogeneous Deposition
    Gaseous wet deposition uses solubility approach
    Particle wet deposition uses algorithms of Scott (1978).

Surface Deposition
    Resistance theory model, diagnostic surface layer.

Diurnal Effects
    Implemented for photolysis of individual species, emissions, and deposition.


             Table 5.4. The Regional Add Deposition Model - RADM (Chang et al., 1987)

       Physical Domain
            Horizontal 2000 x 2000 km
            Troposphere depth.

       Spatial Variables
            3-D fields of constituents, T, u, v, ^
            2-D fields of precipitation, deposition, P^-

       Grid Resolution
            80 km horizontal, 80-8000 m vertical

       Chemical Scheme
            36 species, 22 organics, 11 radicals.

       Homogeneous Chemistry
            77 reactions.

       Heterogeneous Chemistry
            8 reactions including aqueous SO2 oxidation, nitric acid formation from aerosol/NO3.

       Atmospheric Structure
            Predicted from meteorology model.
       Photolysis Treatment
            Full 5-Eddington radiative transfer model of troposphere with cloud effects.

       Wind Field Generation
            Meteorology model with hourly resolution.

       Transport Treatment
            Smolarltiewicz (corrected upstream).

       Chemistry Solution
            Exponential-assisted predictor-corrector.

       Tune-dependent Integration
            Time steps range from <10 to 150 seconds.

       Boundary Conditions
            User specified, varying.

       Heterogeneous Deposition
            1-D cloud chemistry/scavenging model

       Surface  Deposition
            Resistance theory model

       Diurnal Effects
            Fully implemented.


      Table 5JS. The LLNL Regional Air Quality Model (Penner and Council, 1987)

Physical Domain
    Flexible, 100-200 km in horizontal
    Currently 2 layers in vertical to approximately 2000 m.

Spatial Variables
    Fixed horizontal spatial grid, time varying vertical grid
    3-D fields of trace constituents, u, v, w
    2-D fields of deposition.

Grid Resolution
    1-5 km horizontal, 50-2000 m vertical

Chemical Scheme
    22 species, 5 lumped hydrocarbon classes, 3 organic radicals.

Homogeneous Chemistry
    60 reactions.

Heterogeneous Chemistry
    Not treated.

Atmospheric Structure
    Daily sounding and surface measurements of wind speed, cloud cover.  Model is
    currently isothermal

Photolysis Treatment
    Table  lookup  based on  zenith angle  from  values generated  by LLNL 1-D
    troposphere/stratosphere model, including effects of multiple-scattering and cloud

Wind Field Generation
    From analysis of data with mass-consistent interpolation model, hourly resolution.

Transport Treatment
    Smolarkiewicz (corrected upstream).

Chemistry Solution
    Predictor-corrector with steady-state assumptions for some species.

Tune-dependent Integration
    Operator-splitting with base 10-minute advection  time step.  Variable time step
    chemistry solution.

Boundary Conditions
    User specified  constituent fields typically representing rural or ocean abundance
    levels, varying.

Heterogeneous Deposition
    Not treated.


        Table 5.5.  The LLNL Regional Air Quality Model (Penner and Cornell, 1987) (continued)

       Surface Deposition
            Resistance theoiy model.

       Diurnal Effects
            Zenith angle dependence of photolysis constants
            Diurnal variation of mixed layer depth.


Bridging Across Spatial Scales

        This chapter has discussed the tools available for evaluation of the interactions between climate change
and atmospheric chemistry.  Climate change is essentially the product of general circulation models.  These
predict precipitation patterns, wind patterns, global cloudiness, and temperature change with a resolution no
better than several hundred kilometers.  Because many important processes take place on smaller scales, these
must be parameterized in the models, leading to inconsistent results on  the scales important to regional
chemistry.  Figure 5.4 shows characteristics of typical current regional, global chemistry, and global climate
models. Note that there are no arrows showing connections between existing models.

        Regional and  urban chemistry and deposition models need meteorological  and climate change
information on much smaller scales. Typically, an urban oxidant model would occupy much less than one grid
square of a general circulation model  Regional  models require a detailed description of the meteorology
appropriate to, at most, several GCM grid points. A challenge for the future is to design a means for bridging
across these spatial scales (assuming the GCM predictions for climate change on a regional basis become more
robust). Some initial work in this direction has begun, but these initial efforts will require significant ongoing
support to understand the level of confidence 'that might be placed in their predictions. A further challenge is
to try to bridge the evident differences in time scales. GCM  calculations attempt to predict statistics of the
predicted climate averaged over a reasonably long period—Le., one week or more. Regional and urban models
have only been applied to particular episodes—with at most a several-day time period. A method for interpreting
the results of the episodic predictions, within the context  of the climate change statistics, must be devised.

        The connections between global  chemistry models and regional and  urban models are even more
tenuous.  Three-dimensional chemistry models capable of  treating  the  horizontally inhomogeneous gas
concentrations evident in the global troposphere  are only now beginning  to  be developed.  Sub-grid-scale
parameterizations that currently plague the GCM simulations will likely have impacts for the chemistry models
as weH Further, sub-grid-scale parameterizations of chemistry are likely to be important in these models as
well, because chemical processes that alter concentration distributions on urban and regional scales would need
to be treated in some realistic fashion. It will likely be important to develop schemes to make the chemistry on
these larger scales consistent with the chemical transformations that we know take place on  smaller scales.

        Unfortunately, little is known about the effect of chemical transformations across such spatial scales.
Much research needs to be done, and, as appropriate, new modeling tools  need to be developed to account for
the necessary bridging across spatial scales. Modeling capabilities are needed not only to provide more accurate
representation of urban processes in regional models or regional processes in global models, but also to provide
meaningful input data from global models to regional to urban models.

        Finally, regardless of our ability (or lack thereof) to physically link these modeling systems, our basic
knowledge of the global tropospheric  chemical system  needs refinement  and  tuning.  Our  knowledge of
biogeochemkal cycles is crude at best  Therefore, our ability to confidently predict trace gas trends is crude,
and certainly our ability to predict the impact of climate change on biologically emitted source gases is nearly
nonexistent Hence substantial effort is called for to truly understand the interactions of our planet's chemistry
and rfimatp systems.

Global chemistry models
       • Explicit chemistry
       • 1-D or 2-Dlmenslonal
       • Resolution:  Entire hemisphere
         or 100's of kilometers of latitude
       • Seasonal statistics
Climate models
       •  No chemistry
       •  3-Dlmenslonal
       •  Resolution ~ 100's kilometers
       •  Seasonal statistics
                          Regional chemistry models
                                 •  Parameterized HC chemistry
                                 •  3-Dimenslonal
                                 •  Resolution -5-10  kilometers
                                 •  Episodic event simulations
                         Figure SA. Characteristics of current models.


                                            CHAPTER 6

                             CONCLUSIONS AND RESEARCH NEEDS

        The primary purpose of this report has been to examine the possible interactions that may occur
between a changing climate and atmospheric chemical processes.  In particular, the focus  has been on
tropospheric chemistry and the interactions with climate change that could occur on spatial scales extending
from the urban to regional to global A number of conclusions can be drawn from the analyses presented, and
many of these win be  described below. While the potential for climate  change in coming decades is also
discussed in this report, it is not the major emphasis, and the discussion below will only attempt to cover a few

        The single most significant finding in this study is that  very little is known about the effects of such
interactions, either on climate or on air quality.  Few modeling studies or pertinent observational data exist that
allow quantification of the effects that climate change could have on urban or regional or global chemistry, or
vice versa Likewise, little is known about the importance of chemical transformations that may occur in
transition from the modeling of one spatial scale to another.

        Some of the major findings of the analyses reviewed or  developed in this report are:

1.  There is sufficient evidence to indicate that atmospheric emissions and concentrations of radiatively and
    chemically important trace gases, such as COy CH4, NgO, CF^CU, CFCLj, and CO, are increasing, and
    have been increasing for a long time. These increases derive  largely from human-related activities. Current
    analyses suggest that it is improbable that present trends toward increasing concentrations will be arrested
    or reversed in the near future.  Predictions of future scenarios for their growth are limited by uncertainties
    in trace gas budgets and in forecasting economic growth, energy use, and other factors.

2.  The direct effects from the radiative forcing from  these gases on climate are not in question.  However,
    ihere are many uncertainties associated with the climate feedback processes that will determine the eventual
    change  in temperature and other climatic variables.  Climate  models indicate that global surface
    temperatures for a doubling of COg, the radiative equivalence of which could occur by mid-21st century, is
    in the range of 1.5 to 4-5°K, with general circulation models giving results in the upper end of this range.

3.  General circulation models are in general agreement regarding the effects  of climate change on a global
    basis, but are in substantial disagreement on the effects over specific.regions. Representations of clouds,
    the planetary boundary layer, and surface processes all contribute to the uncertainties in determining the
    derived climate change.  Currently, the results from GCMs should not be regarded as reliable indicators of
    regional effects from climate change.

4.  The effects of climate change on urban and regional scale  chemistry could be  quite significant, but very
    little information currently exists on the sensitivity of air quality models to  climatic  parameters. The few
    studies available suggest that oxidant formation may be sensitive to changes in temperature, in stratospheric
    ozone, in cloud cover, in boundary layer depth, in background concentrations,  and in induced emissions
    responses. No information exists  on whether the  frequency of stagnation episodes would be affected by
    climate change.

5.  Likewise, little is known about how add deposition would be affected by climate change. Types of climate
    change that could  affect the conversion rate of SO2 to sulfate and the add deposition indude changes in
    temperature, stratospheric ozone, background concentrations, circulation patterns, frequency and types of
    douds formed, and precipitation patterns.

6.  Interactions  between  dimate  change  and  global tropospheric chemistry primarily center  around
    perturbations to the distributions of ozone and the hydroxyi radical. Changes in ozone, driven by increasing



     concentrations of CO and NOX for example, can have a direct impact on climate. At the same time, climate
     change can influence ozone concentrations.  Hydroxyl radical is the primary chemical  scavenger of such
     radiatively and  chemically important  gases as  CH*,  NMHC,  and CO; therefore,  changes in OH
     concentrations, whether due to direct emissions  of CH. or CO, or due to climate-induced changes in
     temperature or H2O, can have a significant impact on the lifetimes and transport of radiatively active gases.

7.   Studies of global tropospheric chemistry are currently limited due to unavailability of  three-dimensional
     global scale models that can adequately account for the spatial variations in trace gas emissions and

8.   Urban  and regional emissions, particularly of NOX and NMHC, may affect global scale tropospheric
     chemistry and thus affect climate. However, the chemical forms of these species by the time they reach the
     global scale are not well known nor represented adequately in current global models.  Likewise, the extent
     to which O, formed from these precursors within the urban environment is exported to affect hemispheric
     backgrounds is unknown.

9.   In general, the current modeling tools available, both in terms of climate  models and air quality models,
     are insufficient to delineate meaningful diagnostic or prognostic analyses of all of the changes in climate
     parameters of interest.

     The findings from  this study lead naturally  to the development of a list of  important research needs.
Research needed to provide a useful evaluation of the interactions between climate change and atmospheric
chemistry, and to determine the effects of climate change on air quality, includes the following:

1.   Biogeochemical cycles and the role of human-related emissions and removal processes  need to be better
     understood through monitoring, specific  process-oriented measurements, and theoretical model-based
     budget analyses.  Global scale emissions inventories are needed to understand the causes of the increasing
     concentrations of  important greenhouse  and  chemically active trace gases, and to  make  defensible
     projections of future emissions. The proper coupling of the effects of atmospheric chemistry and climate
     on trace gas concentrations with the dependence of future emissions on natural  and human-related factors,
     including economic, energy use, agricultural and technological developments,  needs  to be considered in
     developing such scenarios.

2.   Climate models, in particular general  circulation models, require  continued efforts  to  improve their
     simulation of climatic behavior, from global down to regional variations of climatic  parameters. Better
     validation against observations is needed, again down to the regional  scale. More accurate and physically
     complete treatments of important climatic feedback  processes need to  be investigated, including the
     representations of clouds, oceans, sea ice,  and snow cover.

3.   Global atmospheric models capable of treating the chemical, radiative, dynamical, and climatic interactions
     of the many important trace gases must be developed and tested.  This is an active research area, but one
     still in its infancy, and efforts should be expanded. Two- and three-dimensional chemistry-transport models
     coupled to climate  models are being developed

4.  Analogue techniques based on historical records need to be investigated and, perhaps through accounting
     for statistical and atmospheric chemical/physical  theory, developed  to provide an alternative means  to
    understanding potential climate change effects down to regional scales. A necessary aspect of such analyses
    will be the determination of past causes of climate change.

5.  The global distributions of essentially  all relevant tropospheric trace gases  and aerosols need to be
    determined Laboratory and field measurements  also need to assess the  relevant physical properties  of
    gas-phase and  aerosol constituents.  Of highest priority are measurements of O3, OH, NO^ CO, NMHCs,
    and aerosols.  The processes involved in the production and removal of tropospheric ozone need to be



6.  Theories and specialized models will need to be developed to aid in the interpretation of experimental
    data,  particularly for the synthesis and interpretation  of global distributions and  trend  data, and for
    understanding of various processes, including biological and surface exchange, formation and growth of
    aerosols, and cloud  and precipitation chemistry.  Likewise, chemical theory should be tested through
    appropriate field and laboratory studies.

7.  A high priority is the development of appropriate techniques  to bridge spatial scales to provide meaningful
    input data  for air quality models, and to provide more  accurate representation of local processes in the
    regional  models and regional processes in the global  models.  Realistic treatments  of  sub-grid-scale
    processes, such as chemical transformations or chemical-transport interactions, are needed to improve
    modeling capabilities. Analyses of the effects of climate change require higher resolution data than that
    directly available from climate models (even if their results  were reliable).  Localized mesoscale climate
    models, using input data from GCMs, may be necessary for determining regional effects of global climate

8.  Methods need to be developed for determining the effects on air quality from  climatic change. Sensitivity
    and uncertainty studies are needed to determine the key climatic variables requiring further analysis, but
    care must be taken  that model parameterizations and  simplifications of atmospheric processes do not
    unduly influence any sensitivity studies of urban and regional  effects.  These studies also need to determine
    the spatial and time scales for which climatic variables should be provided.

9.  Schemes need to be developed to use urban and regional models beyond their normal episodic time range.
    Longer term analyses are necessary for determining the  overall effects of climate change.

10. Development of existing air quality models  should  continue,  particularly in  terms  of improving model
    representations of cloud processes (e.g., vertical mixing, aqueous chemistry), photolysis,  and  boundary
    layer/canopy interactions. The potential importance of secondary aerosol formation needs to be evaluated.

11. Additional  atmospheric data bases are needed for  assessing, evaluating, and validating urban/regional
    models.  Critical treatments of average conditions  in  addition to pollution  episodes  are of particular
    importance from the standpoint of climate change and its effects on regional air quality.

12. Chemical kinetics data bases need further improvements for air quality studies, particularly for the rates
    of oxidation of aromatics and biqgenic hydrocarbons.

13. The response of biogenic and anthropogenic emissions to climate change needs to be examined.  Also, the
    latitudinal and seasonal dependence of the tropospheric water vapor/temperature feedback relationship
    needs to be studied



Albritton, D., F. Fehsenfeld, D., B. Hicks, J. Miller, S. Liu, J. Hales, J. Shannon, J. Durham, and A. Patrinos,
NAPAP, Interim Assessment /The Causes and Effects of Acidic Deposition, Vol. m, 1987.

Arakawa, A., and W.H. Schubert, Interaction of Cumulus Cloud Ensembles with the Large-scale Environment,
I, J. Atmos. ScL, 31, 674-701,1974.

Atherton, CS., and J.E. Penner, The Transformation of Nitrogen Oxides in the Polluted Troposphere, to be
published, Tellus, 1988; also Lawrence Livermore National Laboratory UCRL-96343, Rev. 1,1988.

Atkinson, R., and A.C.  Lloyd, Evaluation of Kinetic and Mechanistic Data for Modeling of Photochemical
Smog, J. Phys. Chem. Ref. Data, 13, 315-444,1984.

Atkinson, R., C.N. Plum, W.P.L. Carter, A.M. Winer, and J.N. Pitts, Jr., Rate Constants for the Gas-phase
Reactions of Nitrate Radicals with a Series of Organics in Air at 298 K, J. Phys. Chem., 88, 1210-1215,1984.

Baldocchi, D.D., B.B.  Hicks, and P. Camara, A Canopy Stomatal Resistance Model for Gaseous Deposition to
Vegetated Surfaces, Atm. Env., 21, 91-101,1987.

Baraola, J.M., D. Raynaud, Y.S. Korotkevich, and C. Lorius, Vostok Ice Core Provides 160,000-year Record of
Atmospheric COg, Nature, 329, 408-414,1987.

Blake, D.R. and FS. Rowland, World-Wide Increases in Tropospheric Methane, 1978-1983, J. Atmos. Chem.,
4, 43-62,1986.

Blake, D.R. and F.S. Rowland, Continuing Worldwide Increases in Tropospheric Methane, 1978-1983, Science,
239 1123-1131,1988.

Bolin, B., B.R. Doos, J. Jager, and RA. Worrick,  editors, The  Greenhouse Effect, Climatic Change, and
Ecosystems, J. Wiley and Sons, New York, 1986.

Brost, RA., The Sensitivity to Input Parameters of Atmospheric Concentrations Simulated  by a  Regional
Chemical Model, J. Geophs. Res., 93, 2371-2387,1988.

Calvert,  J.G., and W.R. Stockwell, The Mechanisms and Rates of the Gas Phase Oxidations of Sulfur Dioxide
and the Nitrogen  Oxides in the Atmosphere, in SOy NO, and NO, Oxidation Mechanism: Atmospheric
Considerations, edited by J.G. Calvert, pp. 1-62, Butterworth, Boston, MA, 1984.

Chameides, W.L., The Photochemistry of a Remote Marine Stratiform Cloud, J. Geophys. Res., 89,4739-4755,

Chang, IS,  RA,  Brost, ISA. Isaksen, S. Madronich,  P. Middleton, W.R. Stockwell, and CJ. Walcek, A
Three-Dimensional Eulerian Acid Deposition Model: Physical Concepts and Formulation, J. Geophys. Res.,

Chapman, E.G., and D.S. Sklarew, Organic-inorganic Ion Relationships in North American Precipitation, Paper
presented at the 191st American Ch'emical Society Meeting, New York, April, 1986.

COHMAP, Major Climatic Changes of the last 18,000 Years: Observations and Model Simulations, Cooperative
Holooene Mapping Project, 1988.


Craig, H. and C.C. Chou, Methane:  The Record in Polar Ice Cores, Geophys. Res. Lett., 9,1221-1224,1982.

Cunnold, D.M., R.G. Prinn, RA. Rasmussen, P.G. Simmonds, F.N. Ah/ea, CA. Cardelino, AJ. Crawford, PJ.
Fraser, and R.D. Rosen, The Atmospheric Lifetime Experiment, 3, Lifetime Methodology and Application to
3 Years of CFCLj Data, J. Geophys.  Res., 88, 8379-8400,1983a

Cunnold, D.M., R.G. Prinn, RA. Rasmussen, P.G. Simmonds, F.N. Alyea, CA. Cardelino, and AJ. Crawford,
The Atmospheric Lifetime Experiment, 4, Results for CF9CL Based on 3 Years of Data, J. Geophys. Res., 88,

Davis, DJD., J.D. Bradshaw, M.O. Rodgers, S.T. Sandholm, and S. KeSheng, Free Tropospheric and Boundary
Layer Measurements of NO Over the Central and Eastern North Pacific Ocean, J. Geophs. Res., 92,2049-2070,

Deardorff, J.W.,  Parameterization of the Planetary Boundary Layer for Use in General Circulation Models,
Mon. Weather Rev., 100, 93-106,1972.

Derwent, R. and O. Hov, Application of Sensitivity and Uncertainty Analysis Techniques to a Photochemical
Ozone Model, J. Geophys. Res., 93, 5185-5199,1988.

Dickinson, R.E. and R J. Cicerone, Future  Global Warming from  Atmospheric Trace  Gases, Nature, 319,

DOE, Detecting the Climatic Effects of Increasing Carbon Dioxide, edited by M.C. MacCracken and F.M.
Luther, 13S. Department of Energy Report ER-0235,1985a.

DOE, Atmospheric Carbon Dioxide  and the Global Carbon Cycle, edited by J.R. Trabalka, UJS. Department
of Energy Report ER-0239,1985b.

Edmonds, JA., J. Reilly, J.R. Trabalka, and D£. Reichle, An Analysis of Possible Future Atmospheric
Retention of Fossil Fuel COy US. Department of Energy, report DOE/OR/21400-1,1984.

Ehhalt, D.H.,  Methane in the Global Atmosphere, Environment, 27, (10): 6-12, 30-33,1985.

Fehsenfeld, F.C., MJ.  Bellinger, S.C. Liu, D.D. Parrish, M. McFarland, M. Trainer, D. Kley, P.C. Murphy,
D.L. Albritton, A Study of Ozone in the Colorado Mountains, J. Atmos. Chem., 1 87-105,1983.

Galloway, J.N., G.E. Likens, W.C. Keene, and M.M. Miller, The Composition of Precipitation in Remote Areas
of the World,  J. Geophys. Res., 87, 8771-8786,1982.

Galloway, J.N., DJVI. Whelpdale, and G.T. Wolff, The Flux of S and N Eastward from North America, Atmos.
Environ., 18, 2595-2607,1984.

Gates, W.L., The Physical Basis of Climate, in Proceedings of the World Climate Conference (WMO No. 537),
World Meteorological Organization,  Geneva, Switzerland, 112-131,1979.

Genthon, G, J.M. Barnola, D. Raynaud, C. Lorius, J. Jouzel, NJ. Barkov, YJS. Korotkevich, and V.M. Kotlyakov,
Vostok Ice Core: Climatic Response to CO, and Orbital Forcing Changes Over the Last Climatic Cycle, Nature,
329, 414-418,1987.

Gery, M.W., R.D. Edmond, and G.Z. Whitten, Tropospheric Ultraviolet Radiation Assessment of Existing Data
and Effect on Ozone Formation, SYSAPP-87/191,1987.


Graedel, T.E., and K.I. Goldberg, Kinetic Studies of Raindrop Chemistry, 1. Inorganic and Organic Processes,
J. Geophys. Res., 88,10865-10882,1983.

Grotch, S., Regionally Intercomparing GCM Predictions with Historical Climate Data, in preparation, 1988.

Hales, J., paper presented at LLNL-EPA Workshop on Impacts of Climate Change, Chapel Hill, North Carolina,
February, 1988 and Lawrence Livermore National Laboratory UCRL-99436,1988.

Haltiner, GJ., and R.T. Williams, Numerical Prediction and Dynamic Meteorology, 2nd ed., John Wiley and
Sons, New York, 1980.

Hameed, S., and R.D. Cess, Impact of a Global Wanning on Biospheric Sources of Methane and its Climatic
Consequences, Tellus, 35B, 1-7,1983.

Hameed, S., and J. Dignon, Changes in the Geographical Distributions of Global Emissions of NOX and SOX
from Fossil-fuel Combustion Between 1966 and 1980, Atmos. Env., 22, 441-449,1988.

Hammitt, J.K.,  F. Camm, P.S. Connell,  W.E. Mooz,  KA.  Wolf, DJ. Wuebbles, and A. Bamezai, Future
Emission Scenarios for Chemicals That May Deplete Stratospheric Ozone, Nature, 330, 711-716,1987.

Holton, J.R., A dynamically Based Transport Parameterization for One Dimensional Photochemical Models of
the Stratosphere, J. Geophys. Res., 91, 2681-2686,1986.

Hov, O., Application of Sensitivity and Uncertainty Analysis Techniques to a Photochemical Ozone Model, J.
Geophys. Res., 93, 5185-5199,1988.

Isaksen, IS A., and O. Hov, Calculation of Trends in the Tropospheric Concentration of Oa, OH, CO, CH.  and
NOX, Tellus, 39B, 271-285,1987.

Jouzel, J., C. Lorius, J.R. Petit, C. Genthon, N.I. Barkov, V.M. Kotlyakov, and V.M. Petrov, Vostok Ice Core:
A Continuous Isotope Temperature Record Over the Last Climatic Cycle (160,000 years), Nature, 329,403-408,

Kasting, J J7., O.B. Toon, and J.B.  Pollack, How Climate Evolved on the Terrestrial Planets, Scientific Am., 258,
February 90-97,1988.

Kavanaugh, M.,  Estimates of Future CO, N2O, and NOy Emissions from Energy Combustion, Atmos. Env.,

Keene, W.C, J.N. Galloway, and J.D. Holden, Jr., Measurements of Weak Organic Acidity in Precipitation
From Remote Areas of the World, J. Geophys. Res., 88,5122-5130,1983.

Keene, W.C., J.N. Galloway, Organic Acidity in Precipitation of North America, Atmos. Environ., 18,2491-2497,

Keeling, C.D., The Global Carbon Cycle:  What we know and could know from atmospheric, biospheric,  and
oceanic observations, in Proceedings of the CO, Research Conference: Carbon Dioxide, Science and Consensus,
DOE CONF-820970, pp. H3-H.62, ILS. Dept of Energy, Washington, DC, 1983.

Khalil, MAJL, and RA. Rasmussen, increase of CHC1F, in the Earth's Atmosphere, Nature, 292, 823-824,

KhaliV.MAJe,  and RA.  Rasmussen, Secular Trends of Atmospheric Methane (CHA Chemosphere,  11,
877-883; 1982.



Khalil, MA.K-, and RA. Rasmussen, Sources, Sinks, and Seasonal Cycles of Atmospheric Methane, J. Geophys.
Res., 88, 5131-5144,1983a.

Khalil, MAIL, and RA. Rasmussen, Increase and Seasonal Cycles of Nitrous Oxide in the Earth's Atmosphere,
Tellus, 35B, 161-169,1983b.

Khalil, MAJC, and RA. Rasmussen, Tricholorotrifluoroethane (F-113) Trends at Pt. Barrow, Alaska, in
Geophysical Monitoring for Climate Change, No. 13, Summary Report 1984, ILS. Department of Commerce,
ERL/NOAA, Boulder, CO, 1985a.

Khalil, MA.IC, and  RA. Rasmussen, Causes of Increasing Atmospheric Methane:  Depletion of Hydroxyl
Radicals and the Rise of Emissions, Atmos.  Environ., 19, 397-407,1985b.

Khalil, MA.K., and RA. Rasmussen, Carbon Monoxide in the Earth's Atmosphere:  Indications of a Global
Increase, Nature, 332, 242-245,1988.

Kuo, H.-L., Further Studies on the Influence of Cumulus Convection on Large-scale Flow, J. Atmos. Sci., 31,

Lamb, B.,  H. Westbert, G. Allwine, and T.  Quarles, Biogenic Hydrocarbon Emissions from Deciduous and
Coniferous Trees in the United States, J. Geophys. Res., 90, 2380-2390,1985.

Lamb, R.G.,  A Regional Scale (1000 km) Model of Photochemical Air Pollution, 1, Theoretical Formulation,
Office of Research and Development Report, Environmental Science Research Laboratory, UJS. Environmental
Protection Agency, 1982.

Lamb, R.G.,  A Regional Scale (1000 km) Model of Photochemical Air Pollution, 2, Input Processor Network
Design,  Office of Research and  Development Report, Environmental Science Research Laboratory, UJS.
Environmental Protection Agency, 1983.

Lamb, R.G., A Regional Scale (1000 km)  Model of Photochemical Air Pollution, 3, Tests of Numerical
Algorithms, Office of Research and Development Report, Environmental Science Research Laboratory, US.
Environmental Protection Agency, 1985.

Lamb, R.G., Diagnostic Studies of Ozone in the Northeastern United States Based on Applications of the
Regional Oxidant Model (ROM),  APCA Conference on the Scientific and Technical Issues Facing Post-1987
Ozone Control Strategies, Hartford, Connecticut, November 16-19,1987.

Leighton, PA., Photochemistry of Air Pollution, Academic Press, New York, v, 300 pp., 1961.

Leoyy, CJB., The General Circulation of Mars: Models and Observations, Adv. Geophys., 28, 327-346,1985.

Levine, JS., C.P. Rinsland, and G.M. Tennille, The Photochemistry of Methane and  Carbon Monoxide in the
Troposphere in 1950 and 1985, Nature, 318, 254-257,1985.

Levy, H., n, J.D. Mahlman, WJ. Moxim, and S.C. Liu, Tropospheric Ozone: The Role of Transport, J. Geophys.
Res., 90,3753-3772,1985.

Liu, MJC, R.E. Morris, and J.P. Killus,  Development of a Regional Oxidant Model and Application to the
Northeastern United States, Atmos. Environ., 18,1145-1161,1984.

Liu, S.C, M. Trainer, F.C. Fehsenfeld, D.D. Parish, E J. Williams, D.W. Fahey, G. Hubler, and P.C. Murphy,
Ozone Production in the Rural Troposphere and the Implications for Regional and Global Ozone Distributions,
J. Geophys. Res., 92,4191-4207,1987.



Liu, S.C., and M.Trainer, Response of the Tropospheric Ozone and Odd Hydrogen Radicals to Column Ozone
Change, J. Atmos. Chem. 6, 221-233,1988.

Logan, JA, MJ. Prather, S.C. Wofsy, and M.B. McElroy, Tropospheric Chemistry: A Global Perspective, J.
Geophys. Res, 86, 7210-7254,1981.

Logan, JA,  Nitrogen Oxides in the Troposphere: Global and Regional Budgets, J. Geophys.  Res., 88,

Logan, JA., Tropospheric Ozone: Seasonal Behavior, Trends, and Anthropogenic Influence, J. Geophys. Res.,

Logan, JA., private communication, 1987.

Lurmann, F.W., A.C. Lloyd and B. Nitta, Modeling Potential Ozone Impacts from Natural Hydrocarbons--!!.
Hypothetical Biogenic HC Emission Scenario Modeling, Atmos. Environ., 17,1951-1963,1983.

Luther, F.M. and R.G. Ellingson, Carbon Dioxide and the Radiation Budget, in Projecting the Climate Effects
of Increasing Carbon Dioxide, M.C. MacCracken and F.M. Luther, editors, US. Department of Energy, report

MacCracken, M.C., D J. Wuebbles, J J. Walton, W.H. Duewer, and F.E. Grant, The Livermore Regional Air
Quality Model: I. Concept and Development, J. Applied Met., 17, 254-272,1978.

MacCracken, M.C., M.E. Schlesinger, and M.R. Riches, Atmospheric Carbon Dioxide and Summer Soil Wetness,
Science, 234, 659-660,1986.

Manabe, S., and R.T. Wetherald, Reduction in Summer Soil Wetness Induced by an Increase in Atmospheric
Carbon Dioxide, Science, 232, 626-628,1986.

Manabe, S. and R.T. Wetherald, Large-scale Changes in Soil Wetness Induced by an Increase in Atmospheric
Carbon Dioxide, J. Atmos. ScL, 44,1211-1235,1987.

McRae, GJ., W.R. Goodin, and J.H. Seinfeld, Mathematical  Modeling of  Photochemical Air Pollution,
California Institute of Technology, Environmental Quality Laboratory Report No. 18,1982.

Meehl, G A., and W.M. Washington, A Comparison of Soil Moisture Sensitivity in Two Global Climate Models,
J. Atmos. ScL, (in press), 1988.

Mellor, GX., and T. Yamada, A Hierarchy of Turbulence Closure Models for Planetary Boundary Layers, J.
Atmos. ScL, 31,1791-1806,1974.

Mintzer, I.M., A Matter of Degrees:  The Potential for Controlling the Greenhouse Effect, World Resources
Institute Research Report #5,1987.

Mitchell, J.F.B., and DA. Wanilow, Summer Dryness in Northern Midlatitudes Due to Increased CO~ Nature,
330, 238-240,1987.

Morris, E.D., Jr., and H. Niki, Reaction of the Nitrate Radical with Acetaldehyde and Propylene, J. Phys. Chem.,

Morris, R.E., SD. Reynolds, and M.K. Liu, The Systems Applications, Inc. Regional Transport Model - Current
status,and future needs, APCA Conference on the Scientific and Technical Issues Facing Post-1987 Ozone
Control Strategies, Hartford, Connecticut, November 16-19,1987.



NASA, Earth System Science: Overview, National Aeronautics and Space Administration, Washington, DC,

National Acid Precipitation Assessment Program (NAPAP), Report on the peer review of the Regional Acid
Deposition Model (RADM), Washington, DC, 1987.

Neftel, A., J. Beer, H. Oeschger, F. Zurcher and R.C. Finkel, Sulfate and Nitrate Concentrations in Snow from
South Greenland 1895-1978, Nature, 314, 611-613,1985.

Nihoul, J. (ed.), Coupled Ocean-Atmosphere Models, Elsevier Oceanography Series, VoL 4, 767 pp., Elsevier,
Amsterdam, 1985.

Niiler, P.P., and E.B. Kraus, One Dimensional Models of the Upper Ocean, in Modeling and Prediction of the
Upper Layers of the Ocean, edited by E.B. Kraus, pp. 143-172, Pergamon Press, 1977.

North, G.R., RP. Cahalan, and J A. Coakley, Jr., Energy Balance Climate Models, Rev. Geophys. Space Phys.,
19, 91-122,1981.

Norton, R.B., Measurements of Formate and Acetate in Precipitation at Ntwot Ridge and Boulder, Colorado,
Geophys. Res. Lett., 12, 769-772,1985.

NRC, Carbon Dioxide and Climate: A Second Assessment, U.S. National Research Council, National Academy
Press, Washington, DC, 1982.

NRC, Changing Climate:  Report of the Carbon Dioxide  Assessment  Committee, VS. National Research
Council, National Academy Press, Washington, DC, 1983.

NRC, Global Tropospheric Chemistry: A Plan for Action, UJS. National Research Council, National Academy
Press, Washington, DC, 1984.

Oltmans, SJ., and WJ>. Kohmyr, Surface Ozone Distributions and Variations From 1973-1984, Measurements
at the NOAA Geophysical Monitoring for Climatic Change  baseline  observatories,  J. Geophys. Res., 91,

Pearman, G J J»., D. Etheridge, F. deSilva, and P J. Fraser, Evidence of Changing Concentrations of Atmospheric
COp N2O and CH4 from Air Bubbles in Antarctica, Nature, 320, 248-250,1985.

Penner, J.E., M.B. McElroy, and S.C. Wofsy, Sources and Sinks for Atmospheric H,: A Current Analysis with
Projections for the Influence of Anthropogenic Activity, Planet. Space ScL, 25,521-540,1977.

Penner, J£., Increases in CO,  and  Chlorofluoromethanes:   Coupled Effects  on Stratospheric  Ozone,
Proceedings of the Quadrennial International: Ozone Symposium, Aug. 4-9,1980, Boulder, CO, 1980.

Penner, JJ2, Modeling the Effect  of Biogenic Hydrocarbon Emissions in the San Francisco Bay Area, APCA
Specialty Meeting Proceedings on the Environmental Impact of Natural Emissions, Research Triangle Park,
North Carolina, March, 1984; also Lawrence Livermore National Laboratory UCRL-89170,1984.

Penner, JJL, and P.S. Connell, Pollutant Transport Study.  Bay Area to  North  Central Coast Air Basin,
Lawrence Livermore National Laboratory UCID-21287,1987.

Penner, J.E., J J. Walton, and T. Umeda, Air Quality Model Validation: Application to the San Francisco Bay
Area  and St Louis, published in the proceedings of the 76th Annual Meeting of  the Air Pollution Control
Association, Atlanta Georgia, June 19-24,1983.


Potter, G.L., and R.D. Cess, Background Tropospheric Aerosols: Incorporation Within a Statistical-Dynamical
Climate Model, J. Geophys. Res., 89,9521-9526,1984.

Prather,  M.,  M.  McElroy, S.  Wofsy,  G. Russell, and D. Rind,  Chemistry of the Global Troposphere:
Fluorocarbons as Tracers of Air Motion, J. Geophys. Res., 92, 6579-6613,1987.

Prinn, R.G., R A. Rasmussen, P.G. Simmonds, F.N. Atyea, D.M. Cunnold, B.C. Lane, CA. Cardelino, and A J.
Crawford, The Atmospheric Lifetime Experiments, 5, Results for CH-CCL based on three years of data, J.
Geophys. Res., 88, 8415-8426,1983.

Ramanathan, V., and J A. Coakley, Jr., Climate Modeling Through Radiative-Convective Models, Rev. Geophys.
Space Phys., 16,465-489,1978.

Ramanathan, V., L. Callis, R. Cess, J. Hansen, I. Isaksen, W. Kuhn, A. Lads, F. Luther, J. Mahlman, R. Reck,
and M. Schlesinger, Climate-Chemical Interactions and Effects of Changing Atmospheric Trace Gases, Rev.
Geophys., 25,1441-1482,1987.

Rampino, M.R., and T. Volk, Mass Extinctions, Atmospheric Sulfur and Climatic Warming at the K/T
Boundary, Nature, 332, 63-65,1988.

Rasmussen,  RA. and  MA.K. Khali], Atmospheric Methane  hi  the Recent  and Ancient Atmospheres:
Concentrations, Trends, and Interhemispheric Gradient, J. Geophys. Res., 89,11599-11605,1984.

Ridley, BA., MA. Carroll, G.L. Gregory, Measurements  of Nitric Oxide in  the Boundary Layer and Free
Troposphere Over the Pacific Ocean, J. Geophys. Res., 92, 2025-2047,1987.

Rinsland, C.P. and JS. Levine, Free Tropospheric Carbon Monoxide Concentrations in 1950 and 1951 Deduced
from Infrared Total Column Amount Measurements, Nature, 318, 250-254,1985a.

Rinsland, C.P., JJS. Levine, and T. Miles, Concentration of Methane in the Troposphere Deduced from 1951
Infrared Solar Spectra, Nature, 318, 245-249,1985b.

Robock, A., An Updated Climate Feedback Diagram, Bui. Amer. Meteor. Soc., 66 786-787,1985.

Saltzman, B., A Survey of Statistical-Dynamical Models of the Terrestrial Climate, Adv. Geophys., 20,183-304,

Saltzman, B., (ed.), Theory of Climate,  Part I: History and Application of General Circulation Models, Adv.
Geophys., 25,1983.

Schere, 1C, paper presented at LLNL-EPA Workshop on Impacts  of Climate Change, Chapel Hill, North
Carolina, February, 1988 and Lawrence  Livermore National Laboratory UCRL-99436,1988.

Schlesinger, MJE., and J.F.B. Mitchell, Model Projections of the Equilibrium Climatic Response to Increased
Carbon Dioxide, in The Potential Climatic Effects of Increasing Carbon Dioxide,  edited by M.C. MacCracken
and F.M. Luther, VS. Department of Energy, 1985.

Schneider, S.H., and R.E. Dickinson, Climate Modeling, Rev. Geophys. Space Phys., 12,447-493,1974.

Schneider, SJL, and SI/. Thompson, Atmospheric CO9 and Climate:  Importance of the Transient Response, J.
Geophys. Res, 86,3135-3147,1981.

Schneider, S.H., and R. Lender, The Coevolution of Climate and Life, Sierra Club Books, San Francisco, 1984.


Seinfeld, J.H., Ozone Air Quality Models, A Critical Review, JAPCA, 38, 616-645,1988.

Seigneur, C, T.W. Tesche, P.M. Roth, and L.E. Reid, Sensitivity of a Complex Urban Air Quality Model to
Input Data, J. Appl. Meteoro., 20,1020-1040,1981.

Seigneur, C, A.B. Hudischewskyj, J.H. Seinfeld, K.T. Whitby, E.R. Whitby, J.R. Brock, and H.M.  Barnes,
Simulation of Aerosol Dynamics: A Comparative Review of Mathematical Models, Aerosol Sci. and Tech., 5,

Semtner, AJ., Jr., On Modeling the Seasonal Thermodynamic Cycle of Sea Ice in Studies of Climatic Change,
Climatic Change, 6, 27-37,1984.

Singh, H.B., LJ. Salas, BA. Ridley, J.D. Shelter, N.M. Donahue, F.C. Fehsenfeld, D.W. Fahey, D.D.  Fairish,
EJ. Williams, S.C. Liu, G. Hubler, and P.C. Murphy, Relationship Between Peroxyacetyl Nitrate and Nitrogen
Oxides in the Clean Troposphere, Nature, 318, 347-349,1985.

Spelman, M J., and S. Manabe, Influence of Oceanic Heat Transport Upon the Sensitivity of a Model Climate,
J. Geophys. Res., 89, 571-586,1984.

Sperber, K.R., S. Hameed, W.L. Gates, and G.L. Potter, Southern Oscillation Simulated in a Global  Climate
Model, Nature, 329,140-142,1987.

Stauffer, B., G. Fisher, A. Neftel, and H. Oeschger, Increase of Atmospheric Methane Recorded in Antarctic
Ice Core, Science, 229,1386-1388,1985.

Stauffer, B., E. Lochbronner, H. Oeschger, and J. Schwander, Methane Concentration in the Glacial Atmosphere
Was Only Half that of the Preindustrial Holocene, Nature, 332, 812-814,1988.

Suck, SJi.  and J.R. Brock, Evaluation of Atmospheric Aerosol  Particle Size Distribution  Via Brownian
Coagulation: Numerical Simulations, J. Aerosol ScL, 10, 581-590,1979.

Taylor, ICE., and S.L. Crotch, Observational and Theoretical Studies of Greenhouse Climatic Effects, Lawrence
Uvermore National Laboratory UCRL-98087,1987.

Thompson, AJML, and RJ. Cicerone, Possible Perturbations to Atmospheric CO, CH4, and OH, J. Geophys.
Res., 89,10,853-10,864,1986a.

Thompson, A.M., and RJ. Cicerone, Atmospheric CH4, CO and OH from 1860 to 1985, Nature, 321,148-150,

Thompson, A.M., and M. Kavanaugh, Tropospheric CH ,/CO/NO^ The Next Fifty Years, in Effects of Changes
in Stratospheric Ozone and Global Climate, VoL 2, United Nations Environmental Program Report, 1986.

Thompson, A.M., R.W. Stewart, MA. Owens, and J A. Herwehe, Sensitivity of Tropospheric Oxidants to Global
Chemical and Climate Change, submitted to Atmos. Env., 1988.

Thompson, S.L., and S.H. Schneider, Carbon Dioxide and Climate: The Importance of Realistic Geography in
Estimating the Transient Temperature Response, Science, 217,1031-1033,1982.

Xiao, G.G, G.C. Reinsel, J Ji. Pedrick, G.M. Allenby, C.L. Mateer, AJ. Miller, and J J. DeLuisi, A Statistical
Trend Analysis of Ozonesonde Data, J. Geophys. Res., 91,13121-13136,1986.


Trainer, M., E.Y. Hsie, SA. McKeen, R. Tallamraju, DJ>. Parish, F.C. Fehsenfeld, and S.C. Liu, Impact of
Natural Hydrocarbons on Hydroxyl and Peroxy Radicals at a Remote Site, J. Geophys. Re&, 92,11879-11894,

Vukovich, F.M., J. Fishman and E.V. Browell, The Reservior of Ozone in the Boundary Layer of the Eastern
United States and Its Potential Impact on the Global Tropospheric Ozone Budget, J. Geophs. Res.,  90,

Walcek, C., and J. Chang, paper presented at LLNL-EPA Workshop on Impacts of Climate Change, Chapel
Hill, North Carolina, February, 1988 and Lawrence Livermore National Laboratory UCRL-99436,1988.

Walton, J J., D J. Wuebbles, and S. Hameed, Simulation of Global Wet and Dry Deposition of Nitric Acid in a
Three-Dimensional Chemistry-Transport Model, published in Acid Rain:  Scientific and Technical Advances, R.
Perry, R.M. Harrison, J.N.B. Bell, and J.N. Leater, editors, Selper Ltd^ London, 1987.

Wang, W.-C, D J. Wuebbles, W.M. Washington, R.G. Isaacs, and G. Molnar, Trace gases and Other Potential
Perturbations to Global Climate, Rev. Geophys., 24,110-140,1986.

Watson, R., Executive Summary of the Ozone Trends Panel, NASA, 1988.

Weiss, R.W., The Temporal and Spatial Distribution of Tropospheric Nitrous Oxide, J. Geophys. Res.,  86,

Whitten,  G.Z.,  and M.W. Gery, The  Interaction  of Photochemical  Processes in the Stratosphere and
Troposphere, in  Effects of Changes in Stratospheric Ozone and Global Climate, VoL  2, United Nations
Environmental Program Report,  1986.

Williams,  EJ., D.D. Parish,  and F.C. Fehsenfeld, Determination of Nitrogen Oxide Emissions from Soils:
Results from a Grassland site in Colorado, United States, J. Geophys. Res^ 92, 2173-2179,1987.

Wilson, C A., J.F.B. Mitchell, A Doubled CO, Climate Sensitivity Experiment with A Global Climate Model
Including a Simple Ocean, J. Geophys. Res., 92,13315-13344,1987.

WMO, The Stratosphere 1981: Theory and Measurements, World Meteorological Organization, Global Ozone
Research and Monitoring Project Report No. 11,1981.

WMO, Atmospheric Ozone: 1985, Vols.  1 and 3, World Meteorological Organization, Global Ozone Research
and Monitoring Project Report No. 16,1985.

Wuebbles, D J., A Summary of the LLNL One-Dimensional Model of the Troposphere and Stratosphere: 1981,
Lawrence Livermore National Laboratory UCED-19185,1981.

Wuebbles, D.,  P.  Connell,  K.  Grant,  R.  Tarp,  and  K.  Taylor,  Initial Results of  the  LLNL  2D
Chemical-radiative-transport Model, Lawrence Livermore National Laboratory UCDD-21178,1987.

Wuebbles, D J., and J.  Edmonds, A Primer on Greenhouse Gases, DOE/NBB0083 Report TR040,1988.

                    CLIMATE PARAMETERS
                         Ralph E. Morris
                         Mike W. Gery
                          Mei-Kao Liu
                         Gary E. Moore
                         Christopher Daly
                       Stanley M. Greenfield
                     Systems Applications, Inc.
                      101 Lucas Valley Road
                       San Rafael, CA 94903
            Contract Nos. CX-0001-4-0055 and PX-001-8-0436


      BACKGROUND	  2-1

      IN CLIMATE	  2-3
            VARIABLES 	  2-5
            Boundary Layer Heights 	  2-6
            Cloudiness  	   2-6
            Water Vapor Concentrations	   2-6
            Wind Fields	   2-7
            Stability	   2-8
            Emissions	   2-8
      UNCERTAINTIES	   2-9
            Definitions of the Climate Change Scenarios  	   2-9
            Model Inputs	  2-10
            Model Limitations	  2-10
            Central California	  2-11
            Midwestern/Southeastern Region	  2-13

      VARIABILITY	  2-85
      METHODOLOGY	  2-85
            Selection of Meteorological Variables	  2-86
            Classification of Meteorological Variables	  2-87
      THE DATA BASE	  2-87
            RTM-ffl Simulations	  2-87
            RTM-ffl Input Variables	  2-87
            Spatial Grouping of Variables 	  2-92
            Temporal Grouping of Variables	  2-93
            Ranges of the Meteorological Variables within Episodes	  2-95
            A General Conceptual Model 	2-101
            Correlational Analysis	2-101
            Cluster Analysis	2-108




                                            CHAPTER 1



        Global atmospheric changes are expected to occur within the next several decades because of increases
in levels of pollutants in the atmosphere. These changes are associated with (1) warming of the atmosphere due
to the greenhouse effect of trace gases, (2) depletion of the stratospheric ozone layer, and (3) modification of
tropospheric chemistry.   The emitted materials responsible  for those changes are carbon dioxide, carbon
monoxide, methane, chlorofluoro-hydrocarbons, nitrous oxide, and other trace gases.

        As atmospheric concentrations of the so-called greenhouse gases continue to increase, the potential
climate change and consequent environmental impacts have become issues of great concern worldwide.  Such
climate changes could significantly affect the chemistry and dynamics of the troposphere and ultimately endanger
human health and sensitive ecological systems.  Of particular concern to many parties, including the Environ-
mental Protection Agency (EPA) and National Park service (NPS), is the possibility that increased temperature
and depletion of stratospheric ozone may result in increases in tropospheric concentrations of ozone. The
resulting  increased reactivity of the troposphere would also  result in increased  concentrations of  other
photochemically active species,  such as hydrogen peroxide and peroxyacetyl nitrate. Hydrogen peroxide is
believed to be one of the principal chemicals in the formation of sulfate and consequently acid deposition, by
acting as an oxidizer of sulfur dioxide, while increased concentrations of ozone and peroxyacetyl nitrate may
result in significant damage to forest ecosystems.

        To provide an initial estimate of the possible effects of future climate changes on tropospheric oxidant
concentrations, the EPA, through an interagency agreement with the NPS, funded a preliminary study that
examined the effects of future climate changes on urban  ah* quality at several U.S. cities (Gery et  al., 1987).
Specifically, this study used a computer simulation model,  OZIPM3, a simple photochemical trajectory box
model, to study the effects of increased temperature and decreased stratospheric ozone on ozone formation. The
results indicated that, if anticipated climate changes do occur, most of the cities studied will not be able to meet
the current National Ambient Air Quality Standard (NAAQS) for ozone (0.12 ppm) without more emission
controls than are currently envisioned.

        This preliminary study, however, did not estimate the effects of future climate changes  on ozone
concentrations in the rural atmosphere. The simulation model used in the study, OZIPM3, does not completely
simulate the complex interaction between the processes of transport, diffusion, chemistry, and deposition nor does
it include  any feedback mechanisms between these processes.  In addition, because the model does not divide
the atmosphere into more than one vertical layer, it does not properly account for emission, meteorological, and
chemistry variations with height, which may be important in both urban and rural environments. Therefore, the
purpose of the study reported here was to examine the ability of another model to assess the effects of future
climate change on tropospheric air quality. This model, the RTM-ffl, is an Eulerian three-dimensional regional
oxidant model that has been developed over several years (Liu et al., 1984; Morris et  al., 1987).
        'Although the information in this report has been funded wholly or partly by the UJS. Environmental
   Dtection Agency under Contract Nos. CX-0001-4-0055 and PX-0001-8-0436, it does not necessarily reflect the
Agency's views, and no official endorsement should be inferred from it.




        This study involved two tasks. The first task was to examine past RTM-m calculations of regional ozone
concentrations in order to estimate the sensitivity of the model to changes in climate.  Ozone concentrations
predicted by the model were related to meteorological input parameters in order to gain insight into how
potential future perturbations of these meteorological parameters will affect ozone concentrations. The second
task was to determine the sensitivity of the RTM-m to changes in climate by simulating a base case of current
climate conditions and potential future climate sensitivity scenarios. (The results of the second task are reported
in Chapter 2, and those of the first task in Chapter 3.)

        To estimate the sensitivity of RTM-m calculations of ozone to climatic change, past model simulations
were re-analyzed and the meteorological conditions used as input were classified into sets of variables.  The
predicted ozone concentrations associated with these sets of meteorological conditions were then examined to
determine their sensitivity to variations in climatic conditions.

        We then applied an updated version of the RTM-m to two modeling domains: one  covering central
California and the other covering the midwestern and southeastern United States. For each modeling domain
the model was exercised for a base case of current temperature and ultraviolet light conditions, and for a future
climate scenario reflecting the effects of global wanning. The model's ozone concentration predictions were then
analyzed to determine the sensitivity of the model's predictions of tropospheric air quality to climate changes.

        This study provides a preliminary estimate of the sensitivity of calculations of air quality to climate
changes, and should help identify those climate parameters to which the calculations are most sensitive. These
results will help focus future research on the effects of possible changes in climate on air quality.


        The use of air quality simulation models (AQSM) to estimate changes in air quality due to changes in
climate  conditions is beset with limitations because of the quantitative  uncertainty associated with assumptions
about global climate change. Current climate models can only provide general indications of the direction of
future climate changes; AQSMs, however, require specific meteorological inputs. Thus, for this preliminary
estimate of the sensitivity of AQSM calculations of ozone concentrations to potential climate modifications, we
used an AQSM (the RTM-m) to simulate current climate meteorological conditions and a simple modification
of these conditions that reflects the general direction of potential climate changes. In no way do we claim that
the future climate scenarios described here represent actual future climate conditions, only that they include some
of the basic features predicted by climate models, namely, increased temperature and UV radiation.



                                FUTURE CHANGES IN CLIMATE

        The continued release of emissions of carbon dioxide (CO2) and other trace gases has in recent years
led to the concern that these trace gases will result in a global warming of the atmosphere by blocking the escape
of thermal infrared radiation. This phenomenon is commonly referred to as the greenhouse effect (Dickinson
and Cicerone, 1986; Ramanathan et al., 1985; Wang et aL, 1986).  To quantify the amount  of global warming
expected in the future, general circulation models (GCMs) and other climate models have been exercised with
various estimates of future loadings of trace gases in the upper atmosphere.  We analyzed the predictions of
climate change from four GCMs  to obtain two representative future climate scenarios: a 4°C temperature
increase and a combination of a 4°C temperature increase with a 10% reduction in stratospheric ozone.

        In this section we report the results of simulations of current climate conditions and two future climate
scenarios using a regional air quality model, the RTM-HI. The maximum daily ozone concentrations calculated
by the model are presented along with an interpretation of the impact of possible climate changes on air quality

        Our future climate scenarios were based on results from the following four GCMs:

        National Center for Atmospheric Research (NCAR) Community Climate Model (CCM) model

        National Aeronautics and Space Administration (NASA) Goddard Institute of Space Studies (GISS)

        National Oceanic and Atmospheric Administration (NOAA) General Fluid Dynamics  Laboratory
        (GFDL) model

        Oregon State University (OSU) model

        Current estimates of the emission and retention of manmade CO2 in the atmosphere indicate a distinct
possibility that atmospheric concentrations of CO2 will double within the next century. Under these conditions,
several GCMs predict an increase in the global average surface temperature at sea level of from 1 to 5°C The
above four GCMs models predict  that the doubling of CO2 concentrations would increase the global average
temperature in the range of 2 to 5°C (Wuebbles and Penner, 1988).  The presence of other trace  gases, in
addition to CO^ would increase the global warming further.  Although the four GCMs generally agree on the
level of increase of global average temperature, they do not agree in their predictions of temperature increases
in specific regions, such as California  Studies conducted by Lawrence Livermore National Laboratory (LLNL)
and others indicate that current GCMs cannot, as yet, provide meaningful results for specific regions of interest

        Because of these limitations, translating the output of GCMs to hourly average temperature increases
for a specific region of interest, as required by a regional air quality model such as the RTM-m, is problematic.
The temporal and spatial scales of a GCM (decades and thousands of kilometers) are not compatible with those
of a model like RTM-m (hours and 10 to 50 km).  Because of difficulties in adapting the  GCM output to a
regional model and the discrepancies in the predictions of  the GCMs for many climate  variables, we have
assumed simply that the temperature increases by 4°C throughout the modeling region. A temperature increase
of 4°C is in the higher range of increases in global average temperature predicted by the four GCMs (2°C to
5°C) yet is well within the range of'those predictions. In fact, temperature increases in central California and
the midwestern/southeastern U.S. could be substantially larger than 4°C since the GCM simulations we analyzed
considered the effects of doubling CO, concentrations on temperature increases. The effects of increases of
other trace gases, or a more than doubling of the CO2 concentrations, would result in a higher temperature
increase. Of course it is also possible that the GCMs may be overpredicting increases in temperature due to



uncertainties in the GCM calculations.  Nevertheless, the fact remains that atmospheric concentrations of CO,
are rising and all of the results of climate models we studied agree that this will result in an increase in global
average temperature.


        The 4°C temperature increase predicted by the climate models is a global, sea-level, annual average
value.  In general, the increases predicted by the GCMs are relatively higher in polar regions and lower at the
equator. The modeling regions in the study reported here lie in the mid latitudes, thus the temperature increases
should be between those at the equator and at the polar regions.  Although the four  GCMs are in general
agreement when predicting the global average temperature effects of a doubling of CO2 concentrations, they are
less in agreement when examining changes in temperature for specific regions.  For example, the four models
predict  that, under a doubling of the CO2 concentrations, the summer average temperature for the eight
westernmost states would increase in the range of 2^ to 5°C, while in the eastern United States the temperature
increase would range from 2^ to 5-5°C. These results support the assumption of a 4°C temperature increase
for the central California and midwestem/southeastern regions.

        A consideration in defining this increased  temperature scenario is how to  relate this 4°C average
increase to the temperature fields used as input to the RTM-HI, which vary diurnally and by vertical layer. For
the present study it was decided to add 4°C to each grid cell in each layer for each hour of the episode. While
this approach may be viewed as an oversimplification of the actual phenomena, it is not unrealistic, as is discussed

        None of the  output from the GCMs is reported to have sufficient temporal and spatial resolution to
resolve the temperature increases as hourly variations for a specific region. There is some speculation that the
increase in global average temperature would have a greater impact on the nightime low temperature than on
the daytime maximum.  However, in the absence of better information, it will be assumed that the 4°C increase
in surface temperature occurs uniformly throughout  the diurnal cycle.

        The next consideration in the definition of the increased temperature climate scenario is how to relate
the temperature increase to the vertical layers of the RTM-m. The RTM-m contains three vertical layers. The
first layer corresponds to the mixed layer. For daytime hours the top of this first layer is defined by an inversion
below which rapid connective mixing occurs.  It would appear that, because of rapid vertical mixing within this
layer, an increase in surface temperature would result in a uniform temperature increase throughout the layer.
For nighttime hours this first layer in RTM-m  is defined as a shallow layer, usually approximately 50 to 150
meters thick.  Since the top of this layer is usually defined by the magnitude of the wind speed and represents
a mechanically well-mixed layer from the friction velocity and Coriolis parameter (Yu, 1978),  any increase in the
surface temperature would occur throughout the mixed-layer of the RTM-m.

        The second layer in the RTM-m is a very thin layer  (200 to 650 meters thick) that represents  a
transition between the mixed layer and an inversion layer  (the third layer aloft).  This transition layer has a
special function in the application of the RTM-m to central California.  During the day the wind fields in the
San Joaquin Valley are dominated by upslope flows near the surface, which are compensated by return flows
aloft Between the upslope winds at the surface and downslope return flow aloft is a shallow transition layer in
which the slope flows counteract each other, resulting in a dead air space. Because of such convergence zones,
there is significant transport between this transition layer and the mixed layer. At night a similiar situation exists,
except that the direction of the flow regimes is reversed, with the mixed layer containing the downslope (drain-
age) flows. Again there is some interaction between the mixed layer and the transition layer in the convergence
zones, especially in the valley, where the pooling of the drainage flows causes an uplifting of  the air. It appears
that die second layer  is tied to the mixed layer quite strongly; this is borne out by the RTM-m simulations, in
which the ozone predictions in the second layer are closer to those in the mixed layer than the  third layer (Morris
et aL, 1987). Thus one could logically assume that any temperature increase in the mixed layer would also be
reflected by an increase in the second layer.



        Other climate effects researchers are assuming that the sea-level average temperature increase occurs
throughout the troposphere (Wuebbles, 1988). Thus, to be consistent with other research and in the absence
of any better information, we are assuming that the 4°C temperature increase also occurs in the third layer.


        The atmosphere is a complicated system in which many variables, such as pressure, temperature, and
winds, are in dynamic balance with each other.  This linkage is described mathematically by the primitive
equations (Pielke,  1984).  Thus any modification of the temperature may also affect other meteorological
variables. These other variables may include, but are not limited to, the following:

        Boundary (mixing) layer heights
        Frequency and patterns of cloud cover and cloudiness
        Water vapor concentrations
        Wind  fields, e.g., thermal winds
        Emissions  rates, both anthropogenic and biogenic
        Frequency and intensity of climate variations

Boundary Layer Heights

        One would expect that an increase in surface temperature would result in an increase in convective
activity and a resultant increase in the convective mixing height. For the central California modeling domain the
mixing heights were defined as the base of an inversion from temperature soundings.  Since we have assumed
that the 4°C temperature increase occurs uniformly throughout the vertical temperature sounding, the height of
the  inversion base and  mixing  height  would  remain  unchanged.    The mixing   heights for  the
midwestern/southeastern modeling domain were created by the Meteorological Processor for Diffusion Analysis
mixing height algorithm (Paumier et al., 1986), which estimates the convective mixing height as the base of the
inversion but also contains some algorithms that may be sensitive to changes in temperature.  However, changes
in the mixing heights input into the RTM-m would also involve adjusting the layer-average wind, temperature,
and water vapor fields.  Quite a bit of uncertainty is associated with the location of the mixing height because
of the spatial and temporal interpolation of the twice-daily mixing heights at radiosonde sites located 200 to 300
km. Therefore, the mixing heights for the climate change scenarios are the same ones used in the base  case.
As a result, the effects of temperature increases on ozone formation should be overstated by the model since the
emissions are trapped in a smaller volume of air.


        There were clear skies throughout the period of the California episode studied here (5-10 August 1981).
Thus it appears  that we should assume continued clear  skies for the climate  change scenarios, since the
appearance of douds would be unjustified  and would unreasonably confuse and complicate  the analysis.

        The episode selected for the midwestern/southeastern modeling domain (14 to 21 July 1980) includes
periods of clouds and some precipitation. Increased cloudiness would increase precipitation, decrease ultraviolet
light, and possibly increase vertical transport of pollutants, while a decrease hi cloudiness would have the opposite
effect Thus increased cloudiness would most likely result in lower ozone concentrations due to reductions in
UV radiation and possible  increases in the occurrence of aqueous-phase chemistry. Since the specification of
the increase, or decrease, of cloudiness associated with a 4°C temperature increase would be arbitrary, we have
assumed that cloudiness remains unchanged.  However, cloudiness should be considered as a climate change
variable in future studies.


Water Vapor Concentrations

        The relationship between temperature and water vapor concentrations can be determined via the
Clausius-Clapeyron equation. The change in water vapor concentration in the atmosphere, as the temperature
increases, depends on the supply of water at the surface.  If the surface is dry, the water vapor concentration will
not change. If there is a plentiful supply of water, the surface water flux would increase but the relative humidity
would stay nearly constant.  GCM results suggest that relative humidity would stay constant with increased
temperature (Ramanathan, 1988).  Thus, when atmospheric temperature T increases, the ratio of the vapor
pressure to the saturation vapor pressure (i.e., relative humidity) would remain constant, i.e.:

       e(TD )      e(TD)
    e(TD) = e(TD)
                es (T + AT)

                   e.CT + AT)
where TD is the dew point and e(TD) is the vapor pressure when the temperature has been increased by AT.
From the above equations water vapor concentrations for increased temperature conditions can be calculated
from the original water vapor concentrations by multiplying by the factor e$(T + AT)/es(T), which can be
estimated from the Clausius-Clapeyron equation as follows:

          exp [21348 - 5388/CT + At)]
          exp [21348 - 5388/T]
Assuming a temperature increase of 4°C and ambient temperatures of 250, 275, and 300°K, we get values for
A of 1.40,132, and 1.26, respectively. Thus a 4°C temperature increase will be expected to increase the water
vapor concentrations by 26 to 40%.

        Increases in water vapor would result in a higher rate of production of the hydroxy radical and thus
higher ozone concentrations.  In general, however, maximum ozone concentrations are not that sensitive to
changes in water vapor concentrations at normal levels of relative humidity.

Wind Fields

        It is  difficult to quantify the effects of increased temperature  on the wind fields.  For the central
California episode it appears the increased temperature could result in significant alterations of the flow fields.
Because of stagnant synoptic conditions during the California episode, the flow fields in the San Joaquin Valley
are dominated by thermal upslope and downslope flows. Increased surface temperature could result in stronger
slope flows. In addition, the increased temperature could also alter the temperature differential between the San
Joaquin Valley and air over the Pacific Ocean, resulting in modifications of the strength of the land-sea breezes.

        The wind fields for the central California application were generated by the Diagnostic Wind Model,
which relies on observations and parameterizatibns of processes to obtain its wind fields (Kessler and Douglas,
1988; Morris et  aL, 1987).  The Diagnostic Wind Model requires hourly values  of a domain-mean surface
temperature and lapse rate.  The surface temperature  and lapse rate  are used  hi the denominator of the
expression for calculating the Froude number, which describes the blocking and deflection of the synoptic wind



flows by complex terrain.  The lapse rate is also used to determine the strength of the thermally generated
slope flows. The assumption of a uniform increase in temperature for the future climate scenario would not
affect the lapse rate but may affect the Froude number calculations.  An increase of 4°C in the temperature
would change the Froude number by approximately 1%.  This change is very small and would not significantly
alter the flow fields, especially since the wind fields in the California episode (5-10 August 1981) are dominated
by stagnant conditions without large amounts of blocking and  deflection effects. Thus we are assuming no
changes in the wind fields for the climate change scenarios for the California modeling domain.

        The wind fields for the July 1980 midwestera/southeastern episode were generated by the Limited Fine
Mesh (LFM)  prognostic meteorological  model.  In principle it could be possible  to alter the  radiation
parameterization and initial temperature fields of a prognostic model, such as the LFM, to create new wind fields
under conditions of increased temperature. However, these fields were obtained from archived LFM output and
rerunning the model to obtain new wind fields is a difficult task; the effort involved would have exceeded the
resources and time constraints for this study.  Thus the same wind fields were used for both the base case and
climate change scenarios.

        It appears reasonable that a change in the thermal structure of the atmosphere due to a temperature
increase would modify wind patterns.  However, it is hard to quantify at this time how those patterns would
change,  and what effect they would have on the model calculations of changes in air quality for the  climate
change scenarios.

        It appears reasonable that increased surface temperature would result in an increase in convective activity
and a corresponding increase in the frequency of unstable atmospheric conditions. In most air quality simulation
models, such as the RTM-m, changes in stability affect plume rise and dispersion processes.  In the RTM-HI
stabilty is calculated using a method proposed by Turner (1970) that relates stability class to surface wind speed
and solar insolation.  Since this stabilty classification scheme does not explicitly use temperature, we have no
choice  but to assume that stability does not change in the climate change scenarios.

        Higher plume rise for point sources may result in local increases  in ozone (due to the lack of NO
titration) and possibly lower ground-level ozone values downwind. The occurrence of unstable conditions earlier
in the day would bring emissions of hydrocarbons and nitrogen oxides (NOX) together earlier in the day, resulting
in a longer period of photochemical activity.  Thus the lack of stability changes in the climate change scenarios
may understate changes in ozone concentrations.


        It is well known that increased temperature  results in increased  evaporation and hence  increased
emission of volatile organic compounds (VOC). Temperature is also a key factor in emissions from automobiles.
Since automobiles are the main class of reactive hydrocarbon emitters, and a major source for NOX emissions,
these emission changes could be significant.  Automobiles tend to emit more reactive hydrocarbons and less
NO  and CO at higher ambient temperatures. Higher temperatures during the summer months, as is assumed
in the California and midwestem/southeastern applications of the RTM-m,  would also result in increased use
of air-conditioning systems and corresponding increases in emissions from power plants supplying the electricity.

        Biogenic emissions are also known to increase with temperature as plant respiration increases.  How-
ever, at extreme temperatures some plants will close their stomata openings to reduce moisture loss, resulting
in decreased respiration and a lower rate of biogenic emissions.  The net effect of increased temperature on
anthropogenic and biogenic emissions is a definite increase in reactive hydrocarbons and possibly a decrease in
low-level (area source) NOX and increase  in elevated  (point  source) NOX, resulting in a more reactive


        It is difficult to estimate emission rates and distribution SO to 100 years in the future.  Reactive
hydrocarbon and NOX emissions are currently controlled to meet air quality standards. In addition, automobiles
that far in the future would be significantly more efficient than the current automobile fleet, and the widespread
use of electric cars is a distinct possibility. On the other hand, population growth will increase emissions as more
people use electricity, industrial products, and automobiles.

        Given the uncertainities associated with estimating emissions so far in the future, we elected to use the
base case 1980 NAPAP anthropogenic and biogenic emissions for the climate change scenarios. Future research
should include efforts to estimate future emissions under increased temperature conditions.


        Several climatic and atmospheric modeling studies have indicated that the stratospheric ozone layer is
decreasing due to chemical reactions with  halocarbons (Cicerone et al., 1983; Prather et al., 1984), methane
(Craig and Chou, 1982), and nitrous oxide (Weiss, 1981).  Since stratospheric ozone is the principal attenuator
of ultraviolet radiation,  the reduction in the stratospheric ozone layer would result in increased penetration of
UV-B to the troposphere. This phenomenon would not only likely result in an increase in skin cancer, but also
in increased photochemical reaction rates for specific atmospheric pollutants.  These increased photochemical
rates may, in turn, result in increased production of photochemical oxidants in the troposphere.

        In fact, according to a recent study, the depletion of stratospheric ozone may be a more important factor
than temperature increases in the formation of ozone in the urban environment (Gery et al., 1987). That study
assumed a reduction in the stratospheric ozone column of from 16.5 to 33% and calculated the effects of the
stratospheric ozone reduction on air quality in several urban areas.  However, since the study an international
agreement to control emissions of chlorocarbons has been signed. Based on the implementation of emission
reductions specified in the so-called Montreal protocol, new estimates of the stratospheric ozone reduction range
from 0 to 20%, with the latest predictions  being a few percent. However, it remains to  be seen whether the
proposed controls will be effective in the future.

        Thus for the climate change scenario that includes depletion of stratospheric ozone concentrations we
are assuming a  10% decrease.  For the base case simulations we assume an ozone column of 0300 on-atm.
This represents approximately the monthly average ozone column at North American latitudes during the months
of July and August Thus a 10% reduction in stratospheric ozone would result in an ozone column of 0270 cm-
atm. Photolysis rates were calculated by the method discussed by Gery et al. (1987).


        Three modeling scenarios were defined to estimate the effect of future climate changes on tropospheric
ozone concentrations. These scenarios consist of a base case and two future climate cases:

        Base case—current meteorological  and ozone column conditions

        Scenario #1~4°C temperature increase  and attendant increase in water vapor concentrations.

        Scenario #2~4°C temperature and water vapor increases and a 10% reduction in stratospheric ozone

        The RTM-m was exercised for ozone episodes of approximately one-week duration for the central
California and the midwestern/southeastern modeling domains for the base case and scenario #1.  Due to
limitations on time  and resources for  this work, scenario #2 was  modeled only for  four days from  the
midwestern/southeastern modeling episode.



        Before presenting the modeling results it is useful to review the uncertainties associated with assumptions
in the modeling.  These uncertaintes can be roughly divided into  three categories:   (1) uncertainty in the
assumptions used to define the climate change scenarios,  (2) uncertainties in the model inputs, and (3) model

Definitions of the Climate Change Scenarios

        Foremost among the uncertainties in the climate change scenarios is that the 4°C temperature increase
is assumed to occur uniformly throughout the  modeling  domain and over the diurnal cycle.   If most of the
increase in temperature occurs at night, rather than in the day when photochemical activity occurs, then the
assumption of uniform temperature change across the diurnal cycle may overstate the ozone concentrations,
since ozone formation occurs during the day. If the temperature increase above the ground is less than 4°C,
the assumption of a 4°C increase aloft may also overstate the ozone  concentrations.

        The assumption that mixing heights will not change with the increase in temperature will overstate ozone
concentrations. By not including changes in cloudiness in the climate change scenarios, the effects of increased
temperature will be either over- or underestimated, depending on  whether cloudiness actually increases or
decreases.  The  fact that stability is also not adjusted for the climate change scenarios may also result in an
understatement of ozone concentrations.

        The uncertainties associated with the water vapor concentrations stem from the assumption that the
temperature increase is uniform and that relative humidity will remain constant as the temperature increases.
As for the temperature, if daytime temperature  and temperatures aloft do not increase 4°C, then the assumed
water vapor concentrations will  be too high and the resulting effects of increased  temperature on ozone
concentrations will be overstated.

        Finally, the effects of holding emissions constant for the climate change scenarios will, most likely, result
in an understatement of ozone concentrations. A significant increase in hydrocarbon emissions can be expected
with a 4°C temperature increase and the population growth expected in the next 50 to 100 years. However, no
one can predict the emission reductions that will occur in the future, or the possibility of more efficient emission
controls. However, higher temperatures do result in an increase in emission rates, so that neglecting this change
in emissions would result in an understatement  of the changes in ozone concentrations.

Model Inputs

        Quite a bit of uncertainty is  associated with  the meteorological and emissions inputs used  in the
applications of the RTM-ffl, as is true for all air quality models. Of particular note are the wind fields  over
complex terrain in the central California application. The ramifications of these uncertainties should, in general,
be minimal since we are examining the sensitivity of a regional oxidant model to postulated changes in future
climate conditions, and any possible  deviations between model input  and actual conditions would be the same
for both the base case and climate  change cases.  However, since there is quite a bit of uncertainty in the
temperature and water vapor concentration fields used as input, this uncertainty does transfer over  to the
calculations of ozone concentrations under increased temperature conditions.  Depending  on the base  case
temperature,  an increase of 4°C will produce different effects in  the  photochemistry rates and  resultant
production of ozone.

Model Limitations

        In any modeling study we cannot know with full certainty whether a model is capturing all of the  effects
of changes in meterological and emission input data  In this study we decided to change a minimal number of
variables—temperature and water vapor—for the first climate change scenario.  Thus the primary uncertainty will
be in the chemical mechanism used.  We have attempted to limit this uncertainty by incorporating  into  RTM-
m the latest version of the Carbon Bond IV Mechanism,  the CB-IV (Gery et aL, 1988).




        The effects of increased temperature on tropospheric ozone concentations were studied in two separate
simulations using RTM-II:  a six-day ozone episode (5-10 August 1981) in California and an eight-day episode
(14-21 July 1980) in the midwestern/southeastern United States. The RTM-m was run using current climate
conditions (the base case) and an increased temperature scenario, which was characterized by a 4°C uniform
temperature increase and attendant increase in water vapor concentration assuming constant specific humidity.

        An Eulerian regional oxidant model, such as the RTM-m, will report multi-level gridded fields of hourly
average concentration for each species (about 30) in the  chemical mechanism. Clearly, not all of this data can
or needs to be presented here. Of particular importance, in terms of air quality regulations and health effects,
are the effects of climate changes on the maximum daily ozone concentrations. Thus in the following paragraphs
we discuss the calculated changes in maximum daily ozone concentrations due to an increase in temperature for
each day of the two episode simulations.

        Maximum daily ozone concentrations were calculated at each grid point in the central California
modeling domain for each day of the six-day episode. The results for the base case are displayed in isopleths
for each day of the episode in Figure 2-1.2 The results for the climate change scenario are shown in Figure 2-
2. The differences between the daily maximum ozone concentrations for the climate change scenario and the
base case are shown in isopleths in Figure 2-3. The highest maximum daily ozone concentrations predicted by
the RTM-m for each day are given in Table 2-1.

        August 5. 1981

        On this day the main area of elevated ozone concentration is downwind of the San Francisco Bay Area
for both the base case and climate change  scenario (compare Figures  2-la and 2-2a).  The increase in
temperature moves the location of the highest ozone concentration closer  to  the urban area.  In fact, the
increased temperature results in ozone concentrations that exceed the primary National Ambient Air Quality
Standard (NAAQS) for ozone (12 pphm) whereas no exceedance is calculated in the base case. As predicted
by the RTM-m, an increase in temperature of 4°C results in an area approximately 300 km exceeding the
primary ozone standard on this day.

        In areas away from the San Francisco urban area the effects of the temperature increase are small (see
Figure 2-3a). In many areas there is no change, but in some areas, such as downwind of Bakersfield, there is
a slight  increase in ozone.  However, in Bakersfield itself there is a slight decrease of over 0.5 pphm.  This
decrease appears to be the result of increased mixed layer NO concentrations from a point source whose plume
rise was reduced because of bouyancy flux due to the increase in the ambient temperature. The increase of NO
emissions in the mixed layer results in the local suppression of ozone due to titration.

        August 6. 1981

        The highest ozone concentrations for the entire episode were calculated for both the base case and
increased temperature case on this day (see Figures 2-lb, 2-2b, and 2-3b). The temperature increase results in
increases in ozone throughout the modeling domain; the highest maximum daily ozone concentration increases
from 15.0 pphm in the base case to 18.0 in the increased temperature scenario, a 20% increase (Table 2-1). The
        2A11 figures for this chapter are collected at the end of the chapter.


Table 2-1.    Highest Daily Ozone Concentrations Predicted by the RTM-m For Each Day of the Central
             California Episode For the Base Case and the Case of Increased Temperature
Maximum Daily

5 August 1981
6 August 1981
7 August 1981
8 August 1981
9 August 1981
10 August 1981




ozone peak is slightly farther downwind in the increased temperature case.  The region of predicted ozone
concentration hi excess of the NAAQS almost doubles from 3,700 km2 (in the base case) to 6,600 km2. Thus
under the increased temperature scenario the model predicts exceedances of the NAAQS in the city of San

        In two of the main cities in the San Joaquin Valley, Fresno and Bakersfield, the temperature increase
results in an increase of approximately 0.5 pphm (8%) in the maximum daily ozone  concentration. In the rest
of the valley ozone increases are on the order of 0.1 pphm.  At regions distant from emission sources, such as
the Sierra Nevada, there is little change in the ozone concentrations.

        August 7.1981

        On August 7 the highest ozone concentrations for the base case and increased temperature scenario
occur at the same site, approximately 80 km downwind of San Francisco (Figures 2-lc and 2-2c). However,
ozone levels increase from 11.7 in the base case  to 13.1 pphm under increased temperature.  Again, under
increased temperature the NAAQS is exceeded, where no exceedance occurred under current climate conditions.
In the heart of the Oakland urban area ozone decreases by approximately 1 pphm (Figure 2-3c).  This decrease
may be due to the faster photochemical reactions caused by increased temperature, such that the more reactive
hydrocarbons close to the urban areas are burned out before the time of peak photochemical production in the
middle of the day. This is partially confirmed by the fact that the highest ozone concentration occurs earlier
in the increased temperature scenario than in the base case.

        In areas outside the San Francisco Bay Area urban plume the temperature increase has very little effect
on ozone concentrations; the only significant effect is a slight increase (0.5 pphm) in the vicinity of Bakersfield.

        August 8.1981

        On August 8 the highest ozone level in the increased temperature scenario (13.7 pphm) is only slightly
higher than the highest value in the base case (13.5 pphm) (see Figures 2-Id and 2-2d).  There is almost no
difference between the two cases in the area! extent of the region in excess of the NAAQS. Again, the increased
temperature scenario results in a slight (0.1 pphm) decrease in ozone at the San Francisco - Oakland urban
center.  In the remainder of the modeling domain there are slight  increases  in ozone, except for a small area
around western Yosemite National Park, where there is a slight decrease in ozone (Figure 2-3d).

        August 9.1981

        Ozone concentrations on August 9  are lower than the previous days in both the base case and the
increased  temperature scenario.  Because  of a shift in wind to the northwest, the  highest daily ozone
concentration is located southeast of the San Francisco Bay Area near Stockton. Over the northwestern por-
tion of the modeling domain the temperature increase results in an increase in ozone (from 0.1 pphm in the base
case to 2.0 pphm). Again the temperature increase has little effect on ozone concentrations in  the remainder
of the modeling domain, away from the influences of the San Francisco Bay Area urban plume  (Figure 2-3e).

        August 10.1981

        On August 10 an increase in wind speeds from the northwest  results hi a "tongue" of ozone spreading
out from the San Francisco Bay Area down the San Joaquin Valley (Figures 2-lf and 2-2f). Although the highest
ozone concentration on 10 August is the lowest for the six-day episode, for both the base case (9.1 pphm) and
increased temperature scenario (9.8 pphm), ozone concentrations within the San Joaquin Valley are both higher
and more widespread than on previous days.  Throughout the entire modeling domain the ozone concentrations
in the increased temperature scenario are higher than in the base case.


Midwestern/Southeastern Region

        Maximum daily ozone concentrations were calculated at each grid point in the midwestern/southeastern
U.S. for an eight-day ozone episode (July 14-21,1980). The results for the base case are shown in isopleths in
Figure 2-4. The results for the first climate change scenario are shown in Rgure 2-5. The differences between
the climate change scenario and the base case are shown in isopleths in Figure 2-6. The highest maximum daily
ozone concentrations predicted by RTM-ffl for each day are given in Table 2-2.

        July 14.1980

        On the first  day of the  episode the increased temperature has almost no effect on  the  ozone
concentrations compared to the base case (Figures 2-4a and 2-5a).  In fact, the plot showing differences between
the two cases (Figure 2-6a) reveals that there are about equal areas with increases and decreases in ozone and
these differences rarely exceed 0.1 pphm in magnitude. The exception is in the vicinity of Pittsburg, where the
increased temperature scenario results in a decrease in ozone of approximately 1 pphm. For a region of this size
it would take over one day to minimize the effects of the initial conditions; thus, we cannot tell if this decrease
is significant or just an artifact of the initial conditions. The highest ozone concentrations calculated for the base
case and increased temperature case are the same (113 pphm).

        July IS. 1980

        On July 15  the  highest ozone concentrations for the base case (11.5  pphm)  and the increased
temperature scenario (11.9 pphm) both occur in the urban plume downwind of Chicago.  Except for an area
stretching from Arkansas to Alabama, ozone concentrations are higher everywhere under the increased
temperature scenario. The highest increase (approximately 1 pphm) occurs on the border between Minnesota
and Iowa (Figure 2-6b).

        July 16.1980

        The highest ozone concentration for the base case on July 16  (125 pphm) occurs in the urban plume
downwind of Detroit.  Under the increased temperature scenario the highest concentration (13.0 pphm) is still
downwind of Detroit but the predicted area of exceedances of the ozone NAAQS increases by almost a factor
of three from approximately 9,800 km2 in the base case to 27,000 km2. In the increased temperature  scenario
increases in ozone are mainly in the upper midwestern, southwestern, and southeastern portions of the modeling
domain (Figure 2-6c). In three areas with  elevated ozone (>10 pphm) hi the base case—the Ohio River valley,
New Orleans, and lowa-the increases in ozone in the increased temperature scenario are less (<0.2 pphm) than
is seen downwind of Detroit.

        July 17.1980

        On July 17 ozone levels in the increased temperature scenario are higher than in the base case over
almost the entire modeling domain (Figures 2-4d, 2-5d, and 2-6d).  Significant increases, greater than 0.5 pphm,
occur over an area connecting the cities of Chicago and Detroit. The highest calculated ozone concentration in
the region is 12.0 pphm, compared to 11.7  pphm for  the base case. There are slight increases in ozone (0.1 to
0.5 pphm) in other areas of base case predicted elevated ozone concentrations (> 10 pphm), Le., Nashville, New
Orleans, and Iowa. However, in the increased temperature scenario ozone levels over eastern Kansas are lower
than in the base case (10.7 pphm compared to 11.7 pphm).

       Julv 18.1980

        On Jury 18 ozone concentrations throughout the entire modeling domain increase significantly under the
increased temperature scenario.  Increases in excess  of 1 pphm occur over Iowa, Wisconsin, and Pennsylvania;
in most of the modeling domain increases are in excess of 0.5  pphm.  The highest ozone concentration under
the increased temperature scenario is 12.1 pphm, an exceedance of the NAAQS.



Table 2-2.   Highest Maximum Daily Ozone concentrations Predicted by the RTM-ffl for Each Day of the
           Midwestern/Southeastern Episode For the Base Case and the Case of Increased Temperature
                                 Maximum Daily Ozone
                                 Base    Increased     Percent
                    Date         Case   Temperature    Increase

                 14 July  1980   11.3       11.3            0.0
                 15 July  1980   11.5       11.9            3.5
                 16 July  1980   12.5       13.0            4.0
                 17 July  1980   11.7       12.0            2.6
                 18 July  1980   11.2       12.1            8.0
                 19 July  1980   13.8       14.8            7.2
                 20 July  1980   11.1       11.2            0.9
                 21 July  1980   12.6       12.3           -2.4


        July 19.1980

        Urban plumes from St. Louis, Chicago, Detroit, and emissions from the Ohio River valley create a
region of elevated ozone concentration covering Indiana, Illinois, Ohio, and Michigan. The highest calculated
ozone concentrations of the episode occur on this day downwind of Detroit over Toronto (Figures 2-4f and 2-
5Q: 13.8 pphm in the base case and 14.8 pphm in the increased temperature case. In the latter case ozone levels
are higher than in the base case over most of the region; the most significant increases (0.5 to 1 pphm) occur
in Minnesota, Iowa, Wisconsin, and Michigan.

        July 20.1980

        On July 20 the only significant increases in ozone under the increased temperature scenario occur in the
upper northwestern portion of the  region (Figures 2-4g, 2-5g,  and 2-6g). The highest ozone concentration
calculated for the base case on this day  (11.1 pphm) occurs over Pittsburgh; for the increased temperature case
it is 112 pphm, also over Pittsburgh. A more dramatic increase in ozone due to the increase in temperature
occurs downwind of Chicago, from 10.7 pphm in the base case to 11.2 pphm.

        July 21.1980

        On the last day of the episode  the increased temperature results in significant ozone increases over a
large portion of the region compared to the base case (Figure 2-6h). The main exception is near the center of
the western boundary, where ozone under the increased temperature scenario is less than in the base case (123
pphm compared to 12.6 pphm). However, significant increases occur around Chicago, where the maximum daily
ozone concentration in the Chicago plume increases from 11.1 in the  base case to 123 pphm in the increased
temperature scenario.  Despite the fact that the highest ozone concentration on this day is lower in the increased
temperature scenario than in the base case, the area of ozone concentrations in excess of the NAAQS increases
from approximately 9,800 km2 in the base case to 16,900 km2 in the increased temperature scenario.


        Because  of limitations on the study, simulations of climate change scenario #2 were limited to four days
of the midwestern/southeastern episode and the California episode was not simulated Here we briefly examine
the results of the simulation of the four days (July 14-17,1980) in the midwestern/southeastern UJS. episode.
In this second climate change scenario  the temperature is assumed to increase by 4°C, as in scenario #1, and
photolytic rates are assumed to increase due to a 10% reduction in the ozone column. Predicted maximum daily
ozone concentrations for the second climate change scenario are  shown in isopleths in Figure 2-7, and isopleths
showing the differences between this case and the base case are shown in Figure 2-8.

        As noted by Gery et al. (1987), the effects of increased ultraviolet irradiance or temperature on
tropospheric ozone concentrations depends on the oxidant-forming potential  of the system.  This in turn is
generally a function of meteorological conditions and the efficiency of the atmospheric system in converting oxi-
dant precursors to oxidants. For some atmospheric systems the increased energy results in increased reactivity
in the morning hours, depleting enough oxidant precursors from the system to limit afternoon ozone production
to levels lower than the base case.

        For example, on July  14 of the modeled episode the patterns of increases and decreases in ozone for
the two climate change scenarios are similar.  This seems to indicate that the ozone formation potential for this
day is precursor limited. Over most of the region the maximum daily ozone concentrations for the two scenarios
differ by only 0.1  pphm; the biggest difference occurs in the northeastern portion of the modeling domain. The
highest ozone concentration for all three scenarios on 14 July occurs over the Ohio River valley, where the value
under increased UV and temperature conditions is 11.6 pphm compared to 113 pphm for both the base case
and first climate  change scenario. Thus for the Ohio River Valley, the ozone formation potential may not be
precursor limited. Differences in the two climate change scenarios occur on the Pennsylvania-Ohio border; under



increased temperature alone (scenario #1) ozone is approximately 1 pphm lower than in the base case, whereas
under scenario #2 (increased temperature and UV) ozone is only 0.1 pphm lower than in the base case.

        On the other three days of the episode under the second climate change scenario, ozone  levels are
slightly lower (0 to 03 pphm) than under scenario #1. Both climate change scenarios produce a similiar pattern
of increases and decreases in concentrations when compared to the base case for this episode. This is illustrated
by the highest dairy maximum ozone concentrations for Jury 15-17: the values under the second climate change
scenario lie between the base case and first climate change case (see Table 2-2).

        These results are consistent with those found by Gery et aL, who showed that under conditions of
increased temperature and UV radiation the highest ozone concentrations were frequently lower in cities with
less oxidant precursors. This is because the increased energy, due to an increase in temperature and/or UV
radiation, bums out the oxidant precursors earlier in the day, resulting in less oxidant precursors in the afternoon,
the period of maximum ozone formation  potential   The  rather  coarse grid spacing used  in  the
midwestern/southeastern  modeling domain (approximately 50 km on a side) reduces the peak  precursor
concentrations from the urban areas because of dilution in the large grid cells.


        Although this preliminary model sensitivity analysis may be useful in anticipating the kinds of air quality
controls that may be needed in responding to potential global climate changes, the uncertainties associated with
predicting just how the climate may be modified preclude any definitive discussion here of regulatory controls.
These preliminary modeling results can only indicate possible general trends hi  exceedances of the ozone
standard and increases in the number of people exposed to unhealthy levels of ozone as a result of global climate
        Our study indicates that under most conditions (Le., immediately downwind of urban areas) increased
temperature tends to (1) increase ozone concentrations, (2) move the location of the peak ozone concentration
closer to urban areas, and (3) expand the area in which ozone concentrations exceed the primary ozone standard
of 12 pphm. Thus the modeling study indicates that global warming will not only lead to more exceedances of
the primary ozone standard over a larger area, but also to an increase in the number of people exposed to these
elevated ozone concentrations. Maps of the ozone concentrations in the vicinity of the San Francisco Bay Area
for the base case and first climate change scenario dearly show the increase in areas of exceedances of the
NAAQS due to increases in temperature (Figure 2-9).  Similar maps for the midwestern/southeastern modeling
domain are given in Figure 2-10. A crucial question is whether future increases hi global temperature will
jeopardize the goal of attainment of the ozone standard in regions in the United States and increase the number
of people exposed to unhealthy levels of ozone concentrations.

        Census  data from 1980 were mapped onto  the  two  modeling  regions (central  California  and
midwestem/southeastera U.S.) to determine the number of people that maybe exposed to higher levels of ozone
concentrations as a result of increased ambient temperature. Table 2-3 presents estimates of the number of
people-hours exposed to ozone concentrations in excess  of 8, 12, and 16 pphm for the base case and the
increased temperature case for the two modeling regions. These estimates were obtained by multiplying the
population in a grid cell by the total number of hours that the predicted hourly ozone concentration in that grid
cell exceeded 8,12, or 16 pphm. Thus the figures in Table 2-3 are overstatements of the actual population
exposure since people tend to spend 80 to 90% of their time indoors, and indoor ozone concentration levels
are typically 10 to 70% of the outdoor level

        Table 2-3 indicates that approximately three times as many people in the central California modeling
domain and 60% more people in the midwestern/southeastern modeling domain will be exposed to hourly ozone
concentrations in excess of the NAAQS as a result of a 4°C  temperature increase.  In addition, the modeling


Table 2-3.    Number of People-Hours of Exposure to Ozone Concentrations in Excess of 8,12, and 16 pphm
           For the Base Case and the Increased Temperature Case (Climate Change Scenario
                            Exposure  to      Exposure to       Exposure  to
                            0^ > 8 pphm      0^ > 12 pphm     0^ >  16 pphm

         Central California

         Base Case           70,509,216           660,876             0

         Increased          102,012,064         2,052,143        92,220
         Midwestern/Southeastern U.S.

         Base Case        1,722,590,208       29,805,348             0

         Increased        1,956,205,568       47,528,944             0

results suggest that, with the increase in temperature, people in central California will be exposed to ozone con-
centrations in excess of 16  pphm whereas  under current temperature conditions  the  modeled  ozone
concentrations do not exceed 16 pphm.

        In air quality regulation, geographic areas are divided into air basins for purposes of classifying air
quality,  although air basins are sometimes  further divided into local regions.  A region is considered
"nonattainment" of an air quality standard (NAAQS) if its estimated exceedance rate of the NAAQS is more than
once per year over three years. Exceedances are counted in terms of days. The number of monitors recording
an exceedance or the number of hours is not a factor in establishing an exceedance. Thus whether two monitors
in a region measure an exceedance of the NAAQS or one monitor measures several hours of exceedance on the
same day, that day counts as one exceedance of the NAAQS for that region.

        In the RTM-m simulation of the six-day central California episode the number of exceedances of the
NAAQS for the entire modeling domain doubles from two to four days due to increases in temperature (Table
2-1).  For the eight-day episode in the midwestern/southeastem ILS the number of days in which predicted
ozone concentrations exceed the NAAQS increased  from three for the base case to five for the increased
temperature case. These exceedances occur in several different areas: Chicago, New Orleans, Iowa, Illinois, and
downwind of Detroit

        To estimate the increases in exceedances and attainment status due to future increases in temperature
we estimated the number of exceedances in each grid cell of the central California and midwestern/southeastem
modeling domains.  As discussed above, attainment or nonattainment is usually defined for entire air basins,
although the definition of air  basins is somewhat  arbitrary and in some cases, at the discretion of the EPA,
subregions within air basins are classified separately.

        Table 2-4 gives the population exposed to exceedances of the NAAQS for ozone of 12 pphm. According
the the RTM-HI calculations for the California episode, a 4°C temperature increase will increase the number
of exceedances of the NAAQS from 2 to 3 and increase the number of people exposed to an exceedance.

        Out of a total of 5.5 million people living in the central California modeling domain in 1980, the
percentage of people exposed to one exceedance of the NAAQS increases from 3% (175,080 people) for the base
case to 8% (456,782 people) for the temperature increase case.  Approximately 70,000 and 26,000  additional
people would be exposed to, respectively, 2 and 3 exceedances of the ozone standard.

        Increased temperature also expands the area of exceedances. Regions experiencing one, two. and three
exceedances of the ozone NAAQS increase, respectively, from 5,200 to 7,100 km  , 3,000 to 4,400 km , and 0 to

        Similiar results are seen for the  midwestern/southeastem modeling domain.  With an increase in
temperature of 4°C, approximately 360,000 people would be exposed to two exceedances of the ozone standard.
Out of approximately 35 million people in the midwestern/southeastem modeling domain, the percentage of
people exposed to one exceedance of the NAAQS would increase from 25% (in the base case) to almost 30%.

        These results must be viewed with caution.  As discussed earlier, numerous simplifying assumptions were
made in modeling the  impacts of climate change  on ozone, and these assumptions add  significantly to the
quantitative uncertainty normally inherent in air quality modeling.  Some of these assumptions, e.g., that the
increase in temperature occurs everywhere, will tend to overstate the effects of increased temperature, while
others, e.g, that hydrocarbon emissions do  not increase under increased temperature conditions, tend to
understate the impacts. The  climate change  scenarios presented here are simplistic and most likely do not
completely describe the changes in climate associated with global warming.


        The model's calculations of ozone under the simplified climate change scenarios discussed here should
thus be viewed as possible trends rather than as conclusive impacts. The basic results of this modeling exercise
are that increases in temperature will likely result in increases in maximum dairy ozone concentrations, increases
in the areas impacted by high ozone concentrations, and increases in the number of people exposed to unhealthy
levels of ozone.  Under these circumstances, emission control requirements currently planned to achieve
attainment of the ozone standard may not be sufficient.


Table 2-4.    Population Exposed to Exceedances of the National Ambient Air Quality Standard for Ozone (12
            pphm) for the Base Case and Increased Temperature Case (Climate Change Scenario #1).
            Population Figures Represent the Total Population in all Model Grid Cells With a Specified
            Number of Exceedances
                                     Pop. in Grid  Pop.  in Grid  Pop.  in Grid
                                      Cells with    Cells  with    Cells with
                                      1  Exceed.      2 Exceed.     3  Exceed.

          Central California
Base Case
          Midwestern/Southeastern U.S.

          Base Case                 8,741,510              0              0

          Increased                10,199,734       359,753              0

!  ! j  I  j  !  ! !
   i-	i	i
                                   15       20

                                  X coordinates

Figure 2-1 a.   Predicted maximum daily mixed-layer ozone concentrations (pphm)

                  for the base case scenario: August 5, 1981.

Figure 2-1 b.
                     15       20       25       30
                    X coordinates
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the base case scenario: August 6, 1981.

                                    15       20
                                  X coordinates
Figure 2-1 c.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
                  for the base case scenario: August 7,  1981.

 Figure 2-1 d.
                      15       20
                    X coordinates
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the base case scenario: August 8, 1981.


                                    15       20
                                  X coordinates
Figure 2—1e.   Predicted maximum daily mixed—layer ozone concentrations (pphm)
                  for the base case scenario: August 9, 1981.

                                    15       20
                                  X coordinates
Figure 2-1 f.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
                 for the base case scenario: August 10, 1981.

I-	i
    !	!-•
 --T-4	- -•
                                    15       20
                                   X coordinates
Figure  2-2a.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
  'for climate sensitivity scenario #1 (temp, and water increase): August 5, 1981.


                                    15       20
                                   X coordinates
Figure 2-2b.  Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity  scenario #1  (temp, and water increase): August 6, 1981.


                                    15       20
                                  X coordinates
Figure 2-2c.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario #1  (temp, and water increase): August 7,  1981.

                                    15       20
                                  -X coordinates
Figure 2-Zd.  Predicted  maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario #1  (temp, and water increase): August 8, 1981.


l_4_4-*_4_ -4-4-4	
i  j  I i  j  i  i  j  I
                                    15       20
                                   X coordinates

figure 2-Ze.  Predicted maximum daily mixed—layer ozone concentrations (pphm)

  for climate sensitivity  scenario #1 (temp, and water increase): August 9, 1981.


                                    15       20
                                  X coordinates
Figure 2-2f.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
 for climate sensitivity scenario #1 (temp, and water increase): August 10, 1981.


I  !  I I  I  I  I  I   I  I  !  I  | I  !  !  !
                                15       20
                               X coordinates
Figure  2-3a.   Differences in maximum daily ozone concentrations (pphm)
       (climate sensitivity senario#1 - base case): August 5,1981.

                                                CKIiiaTaKe 'NAS
                                                        f t !  !
                                 15      20
                               X coordinates

Figure 2-3b.   Differences in maximum daily ozone  concentrations (pphm)
       (climate sensitivity senario#1 - base case): August 6,1981.


                    i  i  i  i  I  i i  i  i
    _  .L....1
JElaso ..Mob MS
                                 15       20
                                X coordinates
Figure 2—3c.   Differences in maximum dally ozone concentrations (pphm)
        (climate sensitivity senariorfH  - base case): August 7,1981.


                                                   !  !  I  I  |  I I  I  I
                                      i  «  j  i  i  i  I i»-**"»jMt. '
                          •  .Lenhoore-NAS .  .  .  .
             JPaso. Robins
                                                   	__: .  :	™i.......i	
                                                     1   •    i

                                                     •I	!•—•	i-

                                        15       20       25       30
                                      X coordinates
       Figure 2—3d.   Differences In  maximum daily ozone concentrations  (pphm)
              (climate sensitivity  senario#1  - base case): August 8,1981.


              r-£W'S landing^o- '  '  '

                                 15       20
                               X coordinates
Figure  2—3e.   Differences in maximum daily  ozone  concentrations (pphm)
       (climate sensitivity senario#1 — base case): August 9,1981.


                                15       20
                               X coordinates
Figure 2-3f.   Differences in maximum daily ozone concentrations (pphm)
      (climate sensitivity senario#1 —  base case): August 10,1981.

Figure 2-4a.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the  base case scenario: July 14, 1980.

Figure 2-4b.  Predicted  maximum daily mixed-layer ozone concentrations (pphm)
                   for the base case scenario: July 15, 1980.

Figure 2-4c.
Predicted maximum daily mixed—layer ozone concentrations (pphm)
    for the base case scenario: July 16,  1980.

Figure 2-4d.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the base case scenario: July 17,  1980.

Figure 2-4e.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the base case scenario: July 18, 1980.

Figure 2-41.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
     for the base case scenario: July 19, 1980.

Figure 2-4g.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the  base case scenario: July 20, 1980.

Figure 2-4h.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
    for the base case scenario: July 21, 1980.

Figure 2-5a.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
                for climate sensitivity scenario#1: July 14. 1980'.

Figure 2-5b.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario#1: July 15,  1980.

Figure 2-5c.
Predicted maximum daily mixed—layer ozone  concentrations (pphm)
  for climate sensitivity scenario^!: July 16. 1980.

Figure 2-5d.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario#1: July 17, 1980.

Figure 2-5e.  • Predicted maximum daily mixed-layer ozone concentrations (pphm)
                for climate sensitivity scenario#1: July 18.  1980.

Figure 2-51.   Predicted maximum daily mixed-layer ozone concentrations (pphm)
                for climate sensitivity scenario^!: July 19,  1980.

Figure 2-5g.
Predicted maximum dally mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario^!: July 20, 1980.

Figure 2-5h.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario#1:  July 21. 1980.

Figure 2-6a.   Differences in maximum daily ozone concentrations (pphm)
       (climate sensitivity scenario#1  - base case): July 14, 1980.

   Figure 2-6b.   Differences in  maximum daily ozone concentrations (pphm)
          (climate sensitivity scenario^! - base case): July 15, 1980.

Figure  2-6c.   Differences in maximum daily ozone concentrations (pphm)
       (climate sensitivity scenarlo#1 - base case): July 16, 1980.

  Figure 2—6d.    Differences in  maximum daily ozone concentrations (pphm)
          (climate sensitivity scenario^! - base case): July 17, 1980.

Figure 2-6e.   Differences in maximum daily  ozone concentrations (pphm)
       (climate sensitivity scenario#1 - base  case): July 18,  1980.

  Figure 2-6f.   Differences in maximum daily ozone concentrations (pphm)
         (climate sensitivity scenario^!  - base case):  July 19, 1980.

 Figure 2—6g.   Differences in maximum daily ozone concentrations (pphm)
        (climate sensitivity scenario#1 - base case): July 20. 1980.

   Figure 2-6h.   Differences in maximum daily ozone concentrations (pphm)
          (climate sensitivity scenario^! - base case): July 21. 1980.

Figure 2-7a.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario#2: July 14, 1980.

Figure  2-7b.
Predicted maximum daily mixed-layer ozone concentrations (pphm)
  for climate sensitivity scenario#2: July 15. 1980.

Figure 2-7c.
Predicted maximum daily mixed—layer ozone concentrations (pphm)
  for climate sensitivity scenario#2: July 16, 1980.

Figure 2-7d.   Predicted maximum daily  mixed-layer ozone concentrations (pphm)
                 for climate sensitivity scenarlo#2: July  17, 1980.

Figure 2-8a.   Differences in maximum daily ozone concentrations (pphm)
       (climate sensitivity scenario#2 - base case): July 14, 1980.

  Figure 2—8b.   Differences in maximum daily ozone concentrations (pphm)
         (climate sensitivity scenario#2 - base case): July 15. 1980.

Figure 2-8c,   Differences fn  maximum daily ozone concentrations (pphm)
       (climate sensitivity scenario#2 - base case): July 16, 1980.

 Figure 2-8d.   Differences  in maximum daily ozone concentrations (pphm)
         (climate sensitivity scenario#2 - base case): July  17, 1980.

                                                                                         Exceeds Standard
               ':::'::::!:::>::: I;:!:::1::::!:: I::: :!::::'::::': :»:!::!::'::: :!:::'::::!
                     iCr ow' s i Landing
                         Base Case
                               Crow's Landing \\\\\
                                                                                      i Castle! HAFB!!
                     Climate Sensitivity Scenario
     FIGURE 2-9a. Comparison of predicted maximum daily ozone concentrations (pphm) for the
                 base case and climate sensitivity scenario #1 (temperature  and water increase)
                                                 for August 5,  1981

                                                                       Exceeds Standard
                                                V     -i      -1      A     .4      -i      ^

                                                                     Exceeds Standard
                     r  i  [
San Jose
         Crow's Landing
                  Castle AFB
                                                       San Jose
                                                                      Q Modesto

                                                                Crow's Landing
.  a
                                                                         Castle AFB
            Base Case
                                                       Climate  Sensitivity Scenario #1

FIGURE 2-9c.  Comparison of predicted maximum daily ozone concentrations (pphm) for the
           base case and climate sensitivity scenario #1 (temperature and water increase)

                                        for August 7,  1981

                                                                              Exceeds Standard
         San. Jose
Crow's Landing
          Castle AFB
                     Base  Case
                                                                                     \  i
                                         San Jose
                                                                  Crow's Landing
                                      .  Q
                                                                            Castie AFB
                                         Climate Sensitivity Scenario
g FIGURE 2-9d.  Comparison of predicted maximum daily ozone concentrations (pphm) for the

S            base case and climate sensitivity scenario #1 (temperature and water increase)

                                          for August 8, 1981

                                                                                Exceeds Standard
                                                          V-     A      -i      4
                                                                              Exceeds Standard
     -  e	
                     Base  Case
                   I i
                                                                                Exceeds Standard
                     Base Case
Climate Sensitivity Scenario #1
FIGURE 2-rlOa, Comparison of predicted maximum daily ozone concentrations (pphm) for the
            base case and climate sensitivity scenario #1 (temperature and water increase)
                                           for July 14,  1980

                                                                                Exceeds Standard
                     Base Case
Climate Sensitivity Scenario #1
FIGURE  2-lob. Comparison of predicted maximum daily ozone concentrations (pphm)  for the
            base case and climate sensitivity scenario #1 (temperature and water increase)
                                           for July 15, 1980

                                                                               Exceeds Standard
                     Base Case                           .      Climate Sensitivity Scenario #1

FIGURE  2-10d. Comparison of predicted maximum daily ozone concentrations (pphm) for the
            base case and climate sensitivity scenario #1 (temperature and water increase)
                                           for July 17, 1980

                                                                               Exceeds Standard
                     Base Case                                 Climate Sensitivity Scenario #1

FIGURE 2-10e. Comparison of predicted maximum daily ozone concentrations (pphm) for the
            base case and climate sensitivity scenario #1 (temperature and water increase)
                                           for  July 18, 1980

                                                                                 Exceeds Standard
                                                         f      -\     -A      -i      -1      -i      -i
                                                                *V     '
                     Base Case
Climate Sensitivity Scenario #1
FIGURE 2-10f. Comparison of predicted maximum daily ozone concentrations (pphm) for the
            base case  and climate sensitivity scenario #1 (temperature and water increase)
                                            for July 19,  1980

                                                                                     Exceeds Standard
                          Base Case
Climate Sensitivity Scenario #1
    FIGURE 2-10g.  Comparison of predicted maximum daily ozone concentrations (pphm) for the
                 base case and climate  sensitivity scenario #1 (temperature and water increase)
                                                for  July 20, 1980

                                                                                 Exceeds Standard
                      Base Case                                 Climate Sensitivity Scenario #1

FIGURE 2-10h.  Comparison of predicted maximum daily ozone concentrations (pphm) for the
             base case and climate sensitivity scenario #1  (temperature and water increase)
                                            for July 21. 1980



        Changes in climate  that are forecast to occur  in the next several decades will probably manifest
themselves as changes in the frequency distribution of climatic variables. For example, a global wanning might
(1) shift the frequency distribution of temperatures toward wanner values, decreasing the chances of cold
temperatures occurring, or (2) change  the shape of thedistribution to include a higher frequency of high
temperatures, so that the chances of cold temperatures  occurring remain the same.  In other words, future
frequency distributions of climatic variables will largely be made up of events that have occurred historically, but
then- chances of occurring will change. Therefore, we are justified in using historical data to examine the impact
of future meteorological conditions on ozone concentration. In so doing, we can gain insight into the sensitivity
of ozone to potential changes in specific' meteorological variables.  We can also gain insight into which
meteorological variables are associated with particular weather regimes, which is useful for introducing realistic
climatic perturbations into  an air quality model.

We performed an analysis to:

        (1)    Assess  the sensitivity of ozone concentrations to meteorological variables.

        (2)    Evaluate methods of grouping the meteorological variables into weather types that are associated
              with ozone concentrations.

        If such groups prove to be useful in characterizing various levels of ozone, additional analyses can be
performed in the future to (1) produce realistic estimates of the type and degree of climatic perturbation that
might occur according to predictions of global climate models, and (2) determine the sensitivities of ozone
concentrations to climatic perturbations.

        Because data inputs  to air quality models are a convenient  source of relatively complete information
on meteorological  variables  that influence ozone concentrations, we examined  the data inputs and ozone
predictions for three applications of the PHOXA version of the RTM-m. The main difference between this
version and the RTM-ffl used in the study discussed in Section 2 is that the PHOXA version includes an early
version of the CBM-TV chemical mechanism (Whitten and Gery, 1985).


The analysis consisted of several steps:

        (1)    Meterological variables that potentially influence ozone concentrations were selected.

        (2)    Measures of these variables were developed, through spatial and temporal averaging, that would
              best relate them to ozone concentrations.

        (3)    The measures were grouped into climatic types that are associated with ozone concentrations.

        (4)    The feasibility of using such grouping schemes to predict ozone concentrations was assessed


 Selection of Meteorological Variables

        The RTM-m model inputs include all climatic information needed to model ozone formation and
 transport. These variables, which together constitute the "world" according to the RTM-m model, are the

        (1)    At each of the RTM-m layers, gridded and time-dependent arrays of:
                 wind speed and direction
                 water vapor mixing ratio

        (2)    Gridded and time-dependent arrays of:
                 mixing height
                 emission rates of ozone precursors
                 exposure class
                 precipitation rates
                 cloud cover

        (3)    Time-dependent boundary conditions of chemical species at the sides and top of each model

        In addition, the RTM-m derives the radiative flux for photochemical reactions (based on the date and
 time of day, latitude, and cloud cover).

        To make the selection of meteorological variables as straightforward aspossible,  it is necessary to develop
 simple techniques for reducing the thousands of data points available for each day to a dozen or so which have
 the greatest predictive power for maximum daily ozone concentrations. The following reduction techniques were

        (1)    Variables were combined into composite indexes based on a knowledge  of chemistry, dispersion,
               and emissions.  For example, wind speed and the mixing height can be multiplied into a product
               which is sometimes known as the 'ventilation'.

        (2)    Redundant data were eliminated by combining or removing intercorrelated variables and
               deemphasizing areas where there is no  need for representation by a classification variable.
               For example, water vapor mixing ratios and temperatures are of interest only in the mixed layer,
               where the bulk of ozone formation takes place.

        (3)    Data were systematically averaged in time and space.

        (4)    Variables were extracted from the model For example, the algorithm for producing estimates
               of radiative flux for photochemistry was taken out of RTM and used in conjunction with other
               input variables to make estimates of radiative flux.

        One complicating factor is that the maximum daily ozone is usually affected by the meteorology of the
preceding  hours.    In  addition,  there is occasionally a substantial buildup  and  carryover of ozone
concentrationsfrom one day to the next Since we have classified each day separately, we  have not treated
multi-day serial correlation of ozone levels.

        Once a dozen or so candidate variables have been selected, an objective method can be applied to
simplify the selection.  The creation of a correlation matrix for all of the discriminatory variables will reveal
redundancy in the form of correlation between variables.  Those variables can then be combined into more
powerful variables, or simply eliminated if their ability to predict ozone  is poor.


Classification of Meteorological Variables

        The goal of the analysis is to provide a scheme for classifying each day of RTM-m input data and
evaluate the ability of such a scheme to predict ozone concentrations.  There are two separate ways to proceed
The first is to divide the days on the basis of ozone levels. For example, one can define three ranges of ozone
concentration and then lump the days, first by ozone, and then by meteorology. This method insures a significant
discrimination of ozone levels.

        The second approach is to classify days only on the basis of meteorological variables, and to treat ozone
as a dependent variable.  While this approach results in unique meteorological classes, each class can include
such a wide range of ozone concentrations that the meteorological variables would not be useful in predicting
levels of ozone.

        For our purposes it was more efficient to use the former method, i.e., define ranges of ozone, then
classify the days falling under each ozone class by meteorology.  In this way the usefulness of the meteorological
variables in characterizing ozone levels (which is the information we seek) can be evaluated in a straightforward
manner.  For  example,  if one meteorological  class is associated with more than one  ozone  class, the
meteorological variables included in that  class are probably not good indexes of ozone concentration.


RTM-in Simulations

        The data base for this  analysis was developed from three RTM-m simulations (Table 3-1).

        Eastern UJS. Simulation:  This simulation was performed over a domain that encompasses nearly all of
the eastern United States and a large portion of the Midwest (Figure 3-1). The modeling period, 15 August to
15 September 1978, included several ozone episodes in various parts of the domain. The highest maximum daily
ozone predictions generally occurred in areas in or downwind of major emission areas, such as the mid-Atlantic
seaboard and the Ohio River valley. The cell size in this simulation, 80 x 80 km, was the largest of the  three

        Midwestern/Southeastern US Simulation:  The midwestera domain extends from the Great Plains to
just west of the mid-Atlantic seaboard, and from the  northern Great Lakes area to central Florida  (Figure 3-
2).  This domain overlaps about two-thirds of the modeling domain for the eastern United States simulation.
The simulation includes three separate ozone episodes. The first episode (April 1980) features elevated ozone
concentrations in eastern Texas and Louisiana.  The second episode (July 1980) features  elevated ozone
concentrations primarily in  the Ohio-Indiana region.  In the third  episode (August 1980), both of  the
aforementioned regions experience high ozone concentrations.

        Central California Simulation: The central California domain is the smallest of the three, covering an
area roughly bounded by the Pacific coast and Sierra  Nevada, and the San Francisco Bay Area and Tehachapi
Mountains (Figure 3-3).  The episode is  brief and features elevated ozone concentrations throughout the San
Joaquin Valley and San Francisco Bay Area. Features unique to this simulation are the small grid cell size (10
x 10 km),  an air flow regime dominated by thermodynamic processes acting on complex terrain, and a lack of
cloud cover during the episode.  Unlike the other lengthier simulations, this brief episode does not include
periods  of relatively low ozone  concentrations.

RTM-in Input Variables

        Meteorological variables influencing the formation and dispersion of ozoneare  contained in RTM-m
gridded input fields. Of these,  five are thought to have a significant influence on ozone concentrations and are

TABLE 3-1.  RTM-III simulations used in the historical data analysis.
             Number of    Subregion
Cell Size    Subregions      Days
Northeastern    2080 x 1840 km   80 x 80 km
1855 x 1942 km   45 x 55 km
320 x 520 km     10 x 10 km

                                               93       15 August - 15 September 1978
 32       17-25 April 1980
 15       7-21  July 1980
 12       18-29 August 1980

 12       5-10  August 1981

FIGURE 3-1.  Northeastern United States modeling  region with subregions.

3-2.  THe midwestern United States modeling region, with subregions.

                              15        20
                                "X. coor
FIGURE 3-3.  The central  California modeling  region,  with  subregions,
Hatched portions of the domain were not included  in  the  analysis.


 relevant to climate change studies. They are: wind speed, mixing height, temperature, water vapor mixing ratio,
 and solar radiation. Wind speed and mixing height might also be effectively combined into one variable, termed
 ventilation (wind speed x mixing height), that is a measure of the general capability of the atmosphere to
 transport and disperse pollutants. Solar radiation is calculated by RTM-m from cloud cover, latitude, time of
 year, and time of day.

        A model is, by definition, reality simplified.  The way in which  the input variables influence the
 calculations of ozone will not be the exact way in which they influence ozone in reality. For our purposes, it is
 important to understand the basic differences between the "RTM-m world" and reality, as they apply to our
 analysis. These differences deal primarily with approximations of vertical structure and processes:

        Vertical Stratification: RTM-m employs a four-layer vertical structure: three prognostic layers (mixed,
 inversion, and above inversion) and one diagnostic surface layer within the mixed layer.  Because the mixed layer
 is not stratified into more finery divided layers, phenomena that vary with height in the mixed layer in reality are
 represented by a layer average in RTM-m. Such phenomena include temperature lapse rate, wind speed
 increase with height, and directional wind shears.  When RTM-m is used to study the effects of climatic change,
 the earth's skin temperature is approximated by the mixed-layer temperature.

        Vertical Motion: RTM-m does not address the concept of inversion strength. Gases do not diffuse
 through the inversion base (defined by the mixing height), but material may pass  through this barrier by
 horizontal transport from areas of higher mixing height to areas of lower mixing height, or through positive
 vertical velocities produced by convergence in the wind field.

        In reality, the inversion strength varies as a function of the amount of subsidence and the temperature
 lapse rate.  Material may pass through the inversion base if it possesses sufficient energy to do so. For example,
 plumes of warm air or vertical flow produced by cloud pumping can produce updraft venting of material through
 the inversion base.   As previously mentioned, RTM-m has  no provision for  inhomogeneities in  the
 temperature/energy structure of the mixed layer that might lead to such events.  For our purposes, this means
 that the RTM-m formulation  tends to isolate  the  surface layer from other vertical layers, minimising the
 influence of meteorological variables above the mixing height and maximizing the influence of those in the mixed
 layer. This property of RTM led us to concentrate only on the mixed-layer values of the meteorological variables
 in this analysis.


        In an RTM-m simulation, meteorological variables are  defined at each grid point every hour or three
 hours; ozone concentrations are predicted each hour at each grid point To develop indexes that relate the
 meteorological variables to ozone concentrations, spatial and temporal averaging scales must be determined.

 Spatial Grouping of Variables

        Since the goal of this study is to investigate relationships between meteorological variables and regional
 ozone concentrations, the obvious choice is to  average variables over each modeling domain.  In practice,
 however, averaging over such large areas smooths over much of the interesting ozone events occurring within
 the domain; that is, variability in ozone concentrations and the meterological variables, which is necessary for
 the development of robust statistical relationships, is reduced.

        The reason for this problem is that large regional modeling domains encompass several major emission
source areas and are often larger than the typical synoptic scale of a particular weather regime (e^, storm, high-
pressure ridge). For example, in the northeastern U.S. simulation, southern New England experienced an ozone
episode on 16 August in which predicted ozone concentrations exceeded 10 pphm; meanwhile, the Chicago area
experienced much different conditions, with predicted ozone concentrations near background levels (<6 pphm).


        To avoid averaging over highly variable ozone patterns and to maximize the strength and coherence in
the relationships among variables, each  modeling domain was divided into  subregtons.   Each  subregion
encompasses an area in which the patterns of ozone concentrations vary in a coherent fashion.  The divided
modeling domains are shown in Figures 3-1 to 3-3.  The northeastern U.S. was divided into three subregions:
northwestern, covering the Ohio Valley and much of the Great Lakes; northeastern, covering the northeast
corridor; and southern, covering the southern one-third of the domain.

        The  domain for the midwestern  U.S.  was divided into four equal subregions: northwestern, which
includes the Chicago, St. Louis, and Kansas City urban areas; northeastern, which includes the Ohio River valley
and Great Lakes region; southeastern, covering the southeastern states; and southwestern, which includes the
Dallas, Houston, and New Orleans urban  areas.

        The  central California domain,  although relatively small, contains large gradients in many of the
meteorological variables because of the presence of large elevational gradients.  We found that those areas of
large gradients (i.e., the western slope of  the Sierra Nevada and the coastal mountains) would have required
excessive subdividing to produce subregions that exhibited an appreciable amount of homogeneity. Therefore,
the Sierra Nevada and part of the Coast Range were omitted from the analysis. That part of the domain showing
relatively little elevational gradient was divided into two subregions, northern and southern.  The northern
subregion includes the San Francisco Bay Area, Stockton, and Sacramento; the southern subregion covers the
southern San Joaquin Valley, including Fresno and Bakersfield.

Temporal Grouping of Variables

        The effects of climate change on exceedances of the primary 1-hour ozone standard is the chief interest
in this study;  therefore, we have chosen the maximum daily one-hour average concentration as the measure of
ozone. The maximum daily ozone concentration is influenced by meteorological processes occurring throughout
the day, so it is necessary to choose one or more  measurement periods for each variable that potentially best
relate it to the maximum hourly ozone concentration. The meteorological variables of concern are wind speed,
mixing height, temperature, watervapor mixing ratio, solar radiation, and  rainfall (Table 3-2).

        Wind speed is  a measure of the rate of advection and transport, and is often a good indicator of the
weather regime  affecting an area.  Low windspeeds are typically associated with slow-moving high-pressure
systems and stagnant air masses, which are often characterized by above  average ozone concentrations.  High
wind speeds are often associated with weather disturbances or the advection of new ah* masses into an area; both
are characterized by relatively low ozone concentrations. Wind speed may affect both the nocturnal buildup and
afternoon dispersion of ozone.  Therefore, the selected data periods for wind speed are (1) average windspeed
over all hours of the day preceding the ozone maximum, and (2) average wind speed during the afternoon, before
the ozone peak occurs.  The timing of the maximum daily ozone varies depending on location, weather regime,
and time of year; however, for the sake of consistency, 1800  hours was chosen as the time of ozone maximum.
Maximum ozone values typically occur before this time, so the selected data periods for windspeed win typically
include more hours than actually occur before the ozone maximum.

        Mixing height  is a measure of the depth of the atmosphere over which most emission, transport,
transformation, and dispersion processes occur.  Generally, small mixing depths are conducive to the buildup of
pollutants, resulting in higher ozone concentrations. Elevated mixing heights allow a relatively large  amount of
vertical dispersion, resulting in lower ozone concentrations.  As in the case of wind speed, the mixing height
may affect both the nocturnal buildup and afternoon dispersion of ozone.  Therefore, the selected data periods
for mixing height are the same as for wind speed.

        Temperature influences photolysis reaction rates, and so is of greatest importance during daylight hours.
Maximum temperatures and peak ozone  concentrations often occur at the same time, and temperature can
directly influence ozone production and destruction during the buildup of ozone. Therefore, the selected data
periods for temperature are (1) average  mixed-layer temperature over all daylight  hours  up to  the ozone
maximum (1800  hours), and (2) the maximum daily mixed-layer temperature.


TABLE 3-2.  RTM-III input variables expected to be important in the formation and buildup of ozone
Wind speed
Mixing height
Solar radiation
3 hours*
3 hours*
3 hours*
3 hours*
3 hours**
3 hours*
1 hour**
Mixed layer
Mixed layer
Mixed layer
Mixed layer
Mixed layer
Data Selected for Analysis
0-1800 average; afternoon (1200-1800) average
0-1800 average; afternoon (1200-1800) average
Daylight (0700-1800) average; maximum 3-hour value
Daylight (0700-1800) average; maximum 3-hour value
Daily (0-2400) total; morning (0-1200) total
0-1800 average; afternoon (1200-1800) average
Daily total (0-2400)
 * One hour in the central California simulation.
** Not available in the central California simulation.


        Humidity influences the rates and dispositions of many chemical reactionsin the atmosphere. It can
affect the rate of ozone destruction during the night, but its primary effect is on ozone production during the day.
Maximum water vapor mixing ratios often occur near the time of the peak daily ozone concentration, and can
affect the rates of ozone formation and destruction at this time.  Therefore, the selected data periods for
humidity are (1) the average water vapor mixing ratio in the mixed layer during the daylight hours leading up
to the ozone maximum (1800 hours), and (2) the daily maximum mixed layer water vapor mixing ratio.

        Solar radiation directly dictates photolysis rates and thus is importantduring all daylight hours, especially
the morning hours.  A value for solar radiation is calculated by RTM-ffl by considering cloud cover and sun
angle. The selected data periods are: (1) total solar radiation received at the surface during the day, and (2) total
solar radiation received at the surface during the morning hours.

        Rainfall scavenges ozone and its precursors from the atmosphere  and therefore has  a negative
relationship  with maximum  daily ozone concentration.  Also, rainfall is usually accompanied by increased
windspeeds,  mixing heights, and cloudy sides, which also tend to depress ozone concentrations. The accumulated
rainfall in an area serves as a measure of the ozone-producing potential of the atmosphere, so the selected data
period is total daily rainfall.

        Ventilation, defined as the product of the  mixed layer wind speed and depth of the mixed layer, is a
measure of  the overall ability of the atmosphere to transport and disperse  pollutants, both vertically and
horizontally. It is not a primary meteorological variable, but has a strong influence on the buildup of pollutants
during the night; during the day, atmospheric ventilation directly influences the maximum concentrations of ozone
in the mixed layer. The selected data periods for ventilation are the same as for wind speed and mixing height.

Ranges  of the Meteorological Variables within Episodes

        If robust relationships  are to be  developed between the meteorological  variables and ozone
concentrations, there must be sufficient variability in both the independent and dependent variables. Table 3-
3 shows the  mean and range of values for predicted ozone and the meteorological variables  for each modeling
domain and episode.  In the northeastern simulation (Table 3-3a) all meteorological variables exhibit a relatively
large  range of values.  Average rainfall is very low, primarily due to the averaging of spotty rainfall events over
relatively large subregjons. The temperature range is substantial (about 12 K) and the water  vapor mixing ratio
varies over a factor of two.  Maximum ventilation is over 10 times the minimum, and maximum solar radiation
is about three times the minimum. Ranges of mixing height and wind speed reveal that most of the range in
ventilation (mixing height x wind speed) is accounted  for by variation in the wind speed, which ranges over a
factor of 10; mixing height ranges over a factor of two to three. Maximum ozone varies least, ranging over a
factor of less than two. However, the minimum of 0.047 ppm is essentially a background ozone concentration,
while the maximum  of 0.082 ppm indicates that high values are widespread over a subregion.  (Predicted
maximum ozone concentrations  rarely exceed 0.12 ppm for  any grid cell in any simulation analyzed.)

        Episode 1 in the midwestern region, in April, is characterized by relatively low temperatures and water
vapor mixing ratios; both  have somewhat larger  ranges  than in the  northeastern episode (Table  3-3b).
Temperature ranges over about 15 K, and water vapor mixing ratios vary over a factor of three. Mixing heights
are relatively high, averaging 30-40% higher than in the northeastern episode;  in contrast, rainfall is low. The
ranges and values of the other variables are roughly similar.

        Episode 2 in the midwestern region occurred in July; this is reflected  inthe highest temperatures and
water vapor mixing ratios of any of the simulations (Table 3-3c). In general, the ranges of the descriptive
measures are somewhat less than those in the northeastern episode; the range of ozone concentrations is almost
identical to the two previously discussed episodes, but the average is higher.

        Episode 3 in the midwestern region, in August, exhibits means, maxima, and minima similar to episode
2, except that solar radiation and mixing heights (and  hence, ventilation) are slightly lower (Table 3-3d).

Table 3-3a.   Ranges of Calculated Measures and Daily Ozone Concentrations for the Northeastern United
             States Modeling Region (AH Subregions Included), IS August -14 September 1978
Rainfall (inches)
Daylight temperature (°K)
Maximum temperature (°K)
Average water vapor (ppm)
Maximum water vapor (ppm)
Average ventilation (km /m)
Afternoon ventilation (km2/m)
Daily radiation (w/m )
Morning radiation (w/m2)
Average mixing height (m)
Afternoon mixing height (m)
Average wind speed (m/s)
Afternoon wind speed (m/s)
Maximum ozone (ppm)

Table 3-3b.   Ranges of the Calculated Measures and Daily Ozone Concentrations for the Midwestern United
             States Modeling Region, Episode 1 (All Subregions Included), 17 - 25 April 1980
Rainfall (inches)
Daylight temperature (°K)
Maximum temperature (°K)
Average water vapor (ppm)
Maximum water vapor (ppm)
Average ventilation (km2/m)
Afternoon ventilation (km2/m)
Daily radiation (w/m )
Morning radiation (w/m2)
Average mixing height (m)
Afternoon mixing height (m)
Average wind speed (m/s)
Afternoon wind speed (m/s)
Maximum ozone (ppm)


Table 3-3c   Ranges of the Calculated Measures and Daily Ozone Concentrations for the Midwestern United
             States Modeling Region, Episode 2 (All Subregions Included), 7 - 21 July 1980
Rainfall (inches)
Daylight temperature (°K)
Maximum temperature (°K)
Average water vapor (ppm)
Maximum water vapor (ppm)
Average ventilation (km /m)
Afternoon ventilation (km2/m)
Daily radiation (w/m2)
Morning radiation (w/m2)
Average mixing height (m)
Afternoon mixing height (m)
Average wind speed (m/s)
Afternoon wind speed (m/s)
Maximum ozone (ppm)

Table 3-3d
Ranges 01 ine caicuiaieo Measures ana uauy uzone vxrocenirauuns IOT me miuwesiem u
States Modeling Region, Episode 3 (AH Subregions IndudedX 18 - 29 August 1980
Rainfall (inches)
Daylight temperature (°K)
Maximum temperature (°K)
Average water vapor (ppm)
Maximum water vapor (ppm)
Average ventilation (km /m)
Afternoon ventilation (km2/m)
Daily radiation (w/m2)
Morning radiation (w/m2)
Average mixing height (m)
Afternoon mixing height (m)
Average wind speed (m/s)
Afternoon wind speed (m/s)
Maximum ozone (ppm)

Table 3-3e.   Ranges of the Calculated Measures and Daily Ozone Concentrations for the Central California
           Modeling Region, 5 - 10 August 1981
                Variable              N    Minimum   Maximum    Mean

     Rainfalla  (inches)                —   —
     Daylight temperature (°K)        12   298.6      303.5     301.3
     Maximum temperature (°K)         12   300.9      307.3     304.5
     Average water  vapor (ppm)        12   4814       8350      6433
     Maximum water  vapor (ppm)        12   6887       8959      7690
     Average ventilation (km2/m)      12   1.68       5.24      3.25
     Afternoon  ventilation (km2/m)    12   3.51       13.64     7.59
     Daily radiation (w/m2)b
     Morning radiation (w/nr)"
     Average mixing height (m)c
     Afternoon  mixing height (m)c
     Average wind speed  (m/s)°
     Afternoon  wind speed (m/s)c
     Maximum ozone  (ppm)	12   0.074      0.091     0.081

     a No rainfall  in region during episode.
       Skies were cloud-free during episode.
     c Not used in  analysis; see text.


        The central California episode has fewer meteorological variables (Table 3-3e). There was no rainfall
or cloudiness during the episode, which is typical for this region. Mixing height and wind speed are not broken
outinto separate variables (they are shown combined in the ventilation variable). A visual examination of the
variable fields revealed that the mixing height was relatively constant during the episode; therefore, ventilation
was a good proxy for wind speed. The ranges  of the variables are much smaller than in the other episodes.
Temperature ranges over  less than 7°K, and water vapor mixing ratios vary over a factor of less than 1.5.
Ventilation is generally lower than in the other simulations, and the range of ozone concentrations is much more
restricted.  The minimum ozone concentration of 0.074  ppm is by far the highest of any simulation.


A General Conceptual Model

        Based on experience and theory, we developed a conceptual model of the expected relationships between
maximum ozone concentration and the relevant meteorological variables. This model, illustrated in Figure 3-
4, is used as a reference point in comparing the results of the analyses. The model states that maximum daily
ozone  concentration is positively correlated with temperature, humidity (though only weakly so), and solar
radiation, and negatively correlated with mixing height, wind speed, ventilation (by definition), and rainfall.
Temperature, water vapor mixing ratios, and solar radiation all affect the rates of chemical reactions that produce
ozone-increases in the values of these variables increase the rates of these reactions. Mixing height, a measure
of volume of the mixed layer, often increases ozone concentrations when low; low wind speeds indicate stable
conditions, which are associated with high ozone concentrations; ozone and  its precursors are scavenged by
rainfall events, and such events are usually accompanied by cloudy skies.

Correlational Analysis

        A correlational analysis was performed to investigate relationships between (1) ozone concentration and
the meteorological variables and (2) the meteorological variables themselves.  The strength of relationships of
the former type is a measure of the ability of a variable  to predict ozone concentration, while that of the latter
type is a measure of the degree of redundancy within a set of variables. Relationships significant at the 95%
confidence level are shown in Table 3-4.


        In the northeastern region the maximum daily ozone concentration is significantly correlated (95%
confidence level) to rainfall, maximum  temperature, morning and total daily solar radiation, daily average and
afternoon ventilation, and  dairy average and afternoon  wind speed  (Table 3-4a).  Most highly correlated are
morning and daily total solar radiation (r = 0.56 and 053,  respectively) and  rainfall (r = -051).  This is
consistent with our conceptual model: higher radiation increases the rates of photolysis reactions that produce
ozone, and rainfall scavenges ozone and its precursors from the atmosphere. The cross-correlation coefficients
shows that rainfall is highly negatively correlated with radiation (rainfall is accompanied by cloudy skies), and
positively correlated with wind speed (storms often produce strong winds).  The solar  radiation measures are
positively correlated with temperature (high temperatures usually  occur under sunny skies), and negatively
correlated with wind speed (sunny skies often occur under stagnant high-pressure systems).

        In the midwestern episode 1 (April 17-25,1980) the maximum daily ozone is significantly correlated
(95% confidence level) with daily average and maximum temperature, daily average and maximum water vapor
mixing ratio, and daily total and morning solar radiation (Table 3-4b). Most highly correlated are daily average
temperature (r  * 0.67) and daily total solar radiation  (r - 054).  This follows our conceptual model-high
temperatures and sunny skies are conducive to ozone formation. As mentioned earlier, temperatures during this
episode were much lower than those of other episodes; it appears that this episode was caused by a brief period
of summer-like conditions (warm and sunny) in mid-spring. Cross-correlation coefficients show that temperature


                    Temperature Radiation
                    Wind    Ventilation  Rainfall
          FIGURE 3-4.  Conceptual model of the relationships between  maximum
          daily ozone concentration and relevant meteorological  variables.

TABLE 3-4a.  Correlations (significant at the 95 percent confidence level) between maximum daily ozone
concentration and the calculated measures—Northeastern U.S., all subregions (N = 93).
Temp. (Day)
Temp. (Max.)
H20 (Avg.)
H20 (Max.)
Rad (A.M.)
Rad (Total)
Vent. (Avg.)
Vent. (Aft.)
WS (Avg.)
WS (Aft.)
MH (Avg.)
MH (Aft.)
03 (Max.)




Temp. Temp.
(Day) (Max.)



H20 H20 Rad.
(Avg.) (Max.) (A.M.)













(Aft.) (Avg). (Aft.)







TABLE 3-Mb.  Correlations (significant at the 95 percent  confidence  level) between maximum daily ozone
concentration and the calculated measures—Midwestern U.S.,  episode  1, all subregions  (N = 36).
Rad. Vent.
(Total) (Avg.)
Temp. (Day)              ~      0.98    0.68    0.70           0.55
Temp. (Max.)             0.98    —      0.59    0.61           0.62
H20 (Avg.)               0.68    0.59    —      0.98
H20 (Max.)               0.70    0.61    0.98
Rad (A.M.)                                               —      0.63
Rad (Total)              0.55    0.62                    0.63
Vent. (Avg.)
Vent. (Aft.)                         '            0.34
WS (Avg.)
WS (Aft.)                                        0.35
MH (Avg.)
MH (Aft.)
03 (Max.)                0.67    0.64    0.42    0.44    0.37    0.54

     TABLE 3-4c.  Correlations (significant at the 95 percent confidence level) between maximum daily ozone concentration and
     the calculated measures—Midwestern U.S., episode 2, all subregions (N = 60).

Temp. (Day)
Temp. (Max.)
H20 (Avg.)
H20 (Max.)
Rad (A.M.)
Rad (Total)
Vent. (Avg.)
Vent. (Aft.)
WS (Avg..)
WS (Max.)
MH (Avg.)
MH (Aft.)
03 (Max.)
Temp. Temp. H20 H20 Rad. Rad.
Rain (Day) (Max.) (Avg.) (Max.) (A.M.) (Total)

Vent. Vent. WS WS MH MH
(Avg.) (Aft.) (Avg.) (Aft.) (Avg.) (Aft.)


0.88 0.35
0.63 — 0.86
-0.30 -0.29

TABLE 3-4d.  Correlations (significant at the 95 percent confidence level)  between maximum daily ozone concentration

and the calculated measures—Midwestern U.S., episode 3, all subregions (N  = 48).

Temp. (Day)
Temp. (Max.)
H20 (Avg.)
H20 (Max.)
Had (A.M.)
Had (Total)
Vent. (Avg.)
Vent. (Aft.)
WS (Avg.)
WS (Aft.)
MH (Avg.)
MH (Aft.)
03 (Max.)
Temp. Temp. H20
Rain (Day) (Max.) (Avg.)
0.33 0.97

-0.30 -0.34
H20 Rad. Rad. Vent.
(Max.) (A.M.) (Total) (Avg.)



(Avg.) (Aft.) (Avg.) (Aft.)

0.72 «


is positively correlated to solar radiation and water vapor mixing ratio (absolute humidity tends to increase as
the water-holding capacity of the atmosphere increases at higher temperatures).  As expected, solar radiation
is positively correlated with temperature.

        In the midwestern episode 2 (July 7-21, 1980) the relationships between the meteorological variables
and maximum daily ozone concentrations are relatively weak (Table 3-4c).  Those  significant at the 95%
confidence level are daily total solar radiation (r = 037), afternoon ventilation (r = -030) and afternoon mixing
height (r = -0.29). Again, results are in keeping with our conceptual model; ozone concentrations are highest
under conditions of sunny skies, low ventilation, and low mixing heights. Cross-correlation coefficients show
typical relationships with solar radiation-negative with rainfall and positive with temperature.  Ventilation shows
a negative relationship with water vapor mixing ratio; it appears that periods of relatively stagnant weather were
characterized by humid conditions in this episode.

        In the midwestern episode 3 (August 8-18,1980) the maximum daily ozone concentration is significantly
correlated  (at the 95% confidence level-)  with rainfall, daily average and maximum water vapor mixing ratio,
daily average and maximum ventilation, and daily average mixing height (Table 3-4d). All correlation coefficients
were negative and relatively low; none exceeded 034 (absolute value).  These results  are consistent with  our
conceptual model, except that water vapor mixing ratio is negatively correlated with ozone. Humidity appears
to have complex relationships with  other variables that  can alter its relationship with ozone.  For example,
increasing  humidity increases photochemical reaction rates, thus increasing the rate of ozone production;
however, humidity may be high during cloudy, rainy weather, which is not typically associated with high ozone
concentrations. Cross-correlation coefficients show humidity to be positively correlated with temperature (higher
temperature produces higher water-holding capacity) and ventilation to be negatively correlated with morning
radiation (sunny days often occur under stagnant, high-pressure conditions).

        In the central California episode no meteorological variables are correlated with maximum daily ozone
at the 95% confidence level  This maybe due to the small sample size (N = 12).  The strongest relationship
was that of daily average ventilation and ozone (r = -0.590). This is typical for the region, where complex terrain
can restrict air flow and produce an ozone episode if wind speeds and mixing heights are  low.


        Overall, solar radiation appears to have the strongest and most consistent relationship with ozone. One
or both measures of solar radiation (morning or total daily) were among  the  top three variables with  the
strongest relationship with ozone in three out of the five episodes analyzed. However, typically clear skies during
the ozone season in California make solar radiation a poor descriptor of ozone. Here  the dominant influence
on air quality is the interaction of mesoscale air flows with complex terrain; thus, wind speed and ventilation are
highly related to ozone in California.

        Other variables are occasionally highly related to ozone concentrations; their importance seems to be
tied to the magnitude of the variables involved. For example, in the midwestern episode  1 (April  1980) there
is a relatively strong positive relationship between temperature and ozone concentration. Temperatures during
this episode were much lower than during the other two midwestern episodes, which occurred in summer (see
Table 3-3). The  northeastern episode had the second lowest average temperature, and was the only other
episode in which temperature was significantly correlated with ozone. These results suggest that the lower the
temperature, the greater its influence on ozone concentrations.

        Rainfall is another example of a variable whose magnitude influences its relationship with ozone.
Average and maximum rainfall were highest in the northeastern episode and second highest in the midwestern
episode 3 (see Table  3-3); these two episodes also exhibit the strongest and second strongest relationships
between rainfall and ozone concentration, respectively.  From these results, it appears that if rainfall occurs
infrequently and in small amounts, it does little to influence ozone concentrations; conversely, when relatively
large amounts of rainfall occur, it has a-strong (negative) influence on ozone concentrations.

        We performed a duster analysis to investigate the possibility of devising a classification scheme for
 RTM-m meteorological input variables that has the ability to discriminate among low, medium, and high classes
 of maximum daily ozone concentration.  As discussed at the beginning of this section, such an analysis can
 proceed in two ways: (1) do a cluster analysis based on meteorology only, or (2) stratify the data by ozone class
 first, then cluster by meteorology.  The second method was chosen as the most efficient method of attaining the
 goal of thetanah/sis.  Thereare many ways to duster meteorological variables, some of which may or may not
 describe ozone concentrations well Stratifying by ozone before dustering isolates those dusters that have some
 discriminatory power.  The strength of the dusters serves as a measure of the usefulness of the dusters to
 discriminate among ozone levels.

        A duster analysis was performed for the northeastern episode  because it was the longest episode
 modeled (31 days in summer 1980). The list of meteorological variables was reduced, based on results from
 the correlational analysis and preliminary duster analyses.  Those variables that exhibited consistently low
 correlations with ozone and did not duster well were removed.  A list of the reduced variable set is shown in
 Table 3-5.

        The reduced data set was stratified by maximum daily ozone concentration. Three groups were formed
 by ranking the 93 ozone values (31 days for each of three subregions) and selecting the lower third (tow ozone),
 the middle third (medium ozone), and the upper third (high ozone). The data were then standardized to produce
 a mean of zero and standard deviation of one for all variables.


        Descriptive statistics for  the meteorological variables in each of the three ozone dasses for the
 northeastern episode are shown hi Table 3-5.  Maximum daily ozone concentrations  range from 0.056 ppm in
 the low ozone class to 0.074 in the high ozone dass.  While this appears to be a relatively small range, it must
 be recognized that the  concentrations are averages over subregions that are hundreds of square kilometers in
 size; hence, the standard deviation is very small (usually about 0.003 ppm).  Rainfall is quite variable within each
 ozone dass, but exhibits a consistent downward trend from low to high classes. Average temperature increases
 slightly from the low to high ozone classes, while daily solar radiation increases markedly. Water vapor mixing
 ratio and maximum windspeed decrease slightly from low to high  ozone dasses. Average ventilation and average
 mixing height reach maxima in the medium ozone dass, and therefore do not show consistent trends across the
 ozone dasses.

        The duster analysis produced several meteorological dusters and outlying cases in each ozone class.
 Two dusters in each dass accounted for most of the days hi that  dass.  In the low ozone class, 24 out of 31 days
 were grouped into two dusters; hi the medium ozone dass, 16  out of 31 were grouped into two dusters; two
 dusters accounted for 22 of 31 days in the high ozone dass.

        The major meteorological dusters are illustrated hi Figure 3-5.  The 025 standard deviation threshold
 was applied to the values of the standardized measures hi each duster to determine if it was significantly above
 or below the mean. In the top tier of dusters only rainfall and solar radiation show significant, consistent trends
 from the low ozone dass tothe high ozone dass. For example, rainfall is above average during low ozone days,
 about average for medium ozone  days, and below average for  high ozone days.  Solar radiation  exhibits the
 opposite trend. Ventilation and wind speed fall significantly below average during high ozone days, and are about
 average for the other two dasses.  All other measures remain about average  across all three dasses.
 Meteorological conditions characteristic of each dass are:  dondy and rainy hi the  low ozone dass, average
conditions in the medium class, and dear with low rainfall hi the high ozone dass. This follows our conceptual
model of ozone behavior, but the influences of certain variables, such as temperature, are notably absent. The
analysis suggests that warm temperatures may not necessarily be a factor in an ozone episode.


TABLE 3-5.  Mean (x) and standard deviation (a) for the meteorological variables
in each of three ozone classes.

                                  Low Ozone         Medium Ozone        High Ozone
  Meteorological Variable         x        o         x        °         X        °

Maximum ozone (ppm)            0.056     0.0037   0.066     0.0029   0.074     0.0032
Rainfall (inches)              0.053     0.046    0.028     0.026    0.017     0.022
Average temperature (°K)       292.7     3.12     293.1     2.29     293.2     2.29
Average water vapor (ppm)      16,243.0  3729.0   15,455.9  2851.0   15,063.0  2005.0
Average ventilation (km2/h)    14.4      7.87     15.0      7.68     10.7      5.37
Daily radiation (w/m2)         1726.0    383-4    1968.0    303.0    2089.8    238.2
Average mixing height (m)      657.0     123.0    676.3     93.2     643.4     85.4
Maximum wind speed (m/s)       5.58      2.25     5.50      2.13     4.38      2.10


[Cloudy; high rainfall]
       Warm            Cool
       Partly cloudy     Cloudy
       Low  ventilation   Low mixing height
       Moderate rainfall High  rainfall



                                                    [Average  conditions]
Warm            Cool
Mostly clear     Partly cloudy
Low ventilation   Low  mixing height
Low rainfall     Low  rainfall
                                            [Clear; low  rainfall]


Very  clear
Very low
Very low  rainfall
Low mixing height
Low rainfall
FIGURE  3-5.   Meteorological  clusters  for  three  classes  of maximum daily ozone concentration  for  the northeastern   episode.
Symbols in boxes are as  follows:  R = rainfall;  T =  temperature; H = humidity; V = ventilation; M = mixing  height;
W = wind  speed;  S =  solar radiation (see Table  3-5 for description  of variable used).    In the upper tier of boxes the  variables
are  ranked relative  to  the  mean  for the unstratified data.  In the lower tier  the ranking  is  relative to  the  mean of  the  data for
the one  ozone  class only.   An  up-arrow (t)  indicates a variable  is at  least  0.25 standard  deviation below average; a down-arrow
(|) indicates a variable is  at least 0.25  standard  deviation below  average; a  "0" indicates  a  variable is  within  0.25 standard
deviation of  the mean.


        When the second tier of clusters is compared to the first tier, we find that each duster has at least one
unique attribute (Figure 3-5).  Interestingly, only the clusters hi the high ozone class show significant differences
in ozone concentration; the "warm" cluster exhibits higher ozone than the "mild" cluster.  The warm duster,
characterized by very clear skies,  very low ventilation, and very little rainfall, is the dassic ozone-producing
situation.  We now see that although temperature is not a factor in the three-class ozone stratification, warm
temperatures are required for a severe ozone episode to occur.


        In the northeastern episode only daily solar radiation and daily rainfall exhibit systematic trends across
the three ozone classes. Ozone concentrations increase with increasing solar radiation and decreasing rainfall
These two variables are generally associated; for example, cloudy weather Cow solar radiation) is required for
rainfall, which  reduces ozone further through washout processes.  This finding is  in agreement with the
correlational analysis concerning solar radiation; however, the correlational analysis shows rainfall to be important
only in the northeastern episode and unimportant in the midwestern and California episodes.

        The clusters of meteorological variables within each ozone class do little to discriminate between ozone
levels within individual classes, except within the high ozone class.  This would suggest that clustering by
meteorology within each ozone class is not a useful excercise.   However, it is the high ozone cases we are
interested in, and for these we are able to produce clusters that have discriminatory power.


        From this exploratory analysis, we found that solar radiation is the best overall descriptive measure for
maximum daily ozone concentrations. Other measures, such as temperature and rainfall, may be important under
certain conditions. Meteorological variables associated with ozone levels hi central California are much different
than those in the Midwest or Northeast. Since cloudless and rainless sides  are the rule during the ozone season
in California, measures such as solar radiation and rainfall are not relevant to that area; measures dealing with
the dispersive character of the atmosphere, such as ventilation, are attractive descriptors for ozone in California.

        The formulation of solar radiation in RTM-ffl is not a fully accurate reflection of what occurs hi reality.
RTM-m only uses time of day, latitude, and doud cover to calculate solar radiation. Attenuation by ambient
aerosols and gases and fugitive dust are not treated Therefore, in a given region for a given day, only changes
in cloudiness would affect radiation levels in RTM-HI. It follows that RTM-ffl wfll be very sensitive to future
climatic scenarios that involve changes in cloudiness.

        The high incidence of cross-correlation among the meteorological variables illustrates the complexity
involved in grouping meteorological variablesby ozone level In the RTM-m application to central California
described in Chapter 2, temperature was increased to 4°C.  To be consistent, the absolute humidity also had to
be increased, since the twoare highly correlated. Likewise, if RTM-m were to be applied to the northeastern
episode with increased solar radiation, associated changes would also be necessary m wind speed, ventilation,
temperature, and rainfall to accurately depict the change (see Table 3-4a).

        The analysis of RTM-m  simulations to estimate  the sensitivity of calculations of ozone to dimatic
variability has been instructive and encouraging.  It has provided: (1) an indication of the sensitivity of ozone
concentrations to changes in meteorology; and (2) a better understanding (for the northeastern United States,
at least) of those variables that are likely to vary together as a group. We now know what degree of change in
meteorological variables is necessary to change, for example, a medium ozone day to a high ozone day. In
addition, we also know what degree of change in meteorological variables is necessary to change a high ozone
day to a very high ozone day.  Recommendations for further work are provided in Chapter 4.




        This study explored the sensitivity of a regional oxidant model, RTM-m, to variations in atmospheric
parameters in an effort to establish the usefulness of using a photochemical model to analyze the impact on air
quality of global climate changes.  The results of the study indicate that the ozone concentrations predicted by
a complex model using current atmospheric chemistry are sensitive to the climate change scenarios studied.
Within the uncertainties present, the modeling results suggest there could be potentially significant increases in
photochemical pollutants due to future climate changes.

        Given the preliminary nature of this study, and its limitations, it is recommended that in future studies
the potential air quality impacts of global climate change be  examined in more detail. While such details are
impossible to completely define because of the exploratory nature of such a study, it is possible to provide an
outline of possible recommended approaches.

        1.      Include more complev climate change scenarios.  In the preliminary study reported here only
               two meteorological parameters were changed: UV intensity at the surface and atmospheric
               temperature.   A third parameter, atmospheric water vapor,  was calculated  as function of
               temperature,  assuming that the specific humidity was held constant. Future  studies should
               examine an expanded set of linked meteorological parameters  (wind, relative humidity, cloud
               type and cover, precipitation, etc.) in addition to those used in this preliminary examination.

        2.      Consider climatic feedback and consistency in meteorological parameters.  If linked parameters
               are to be considered, it is imperative that  climatic feedback  be  treated in the simulations.
               Feedback is included, to a limited degree, in the scenarios produced by the General Circulation
               Models (GCMs). However, the GCMs do not at present report results at sufficient spatial and
               temporal resolution to be of use in applying regional air quality models.  Present-generation
               GCMs do not calculate  regional-scale cloud  distributions, winds, stabilities, etc  However,
               several mesoscale and one-dimensional boundary layer structure models are available that may
               be able to link the large-scale GCM output to the smaller scales required by regional models.
               These models maybe able to use the GCM outputs to generate sets of temporally and spatially
               varying meteorological parameters that are consistent with each other.

        3.      Broaden the range of climate perturbations. While we attempted to use the climate scenarios
               predicted by the GCMs in the calculations of regional air quality, it was clearly not possible at
               this time to include all variations in climate conditions contained in the GCM scenarios.  Only
               the effects of varying temperature and surface UV intensity were examined. Moreover, it was
               not possible to consider more than a few values of these two parameters. UV intensity at the
               surface was calculated assuming either 300 Dobson  or 270 Dobson; atmospheric temperature
               was assumed to be either baseline or 4°C above baseline. Given the uncertainties inherent in
               the construction of future climate scenarios, and the fact that GCM-predicted climate variations
               represent gross spatial and temporal averages, we  recommend that future studies include a
               broader range of meteorological parameters that are varied.

        4.      Analyse the impact of global change on the effectiveness of currepf air quality strategies.  The
               study reported here examined future air quality conditions assuming no change in precursor
               emissions from today.  Current regulatory efforts to reduce ozone concentrations involve
               reducing emissions of VOCs and NCL, which could  dramatically alter the chemical mix of the
               atmosphere and possibly the chemical response to potential changes in climate.  Future studies
               should examine the impacts of global climate change on an environment that more closely
               resembles the one that is likely to result from the implementation of possible control strategies.



        Such an examination will not only provide a more realistic picture of the impacts that could
        occur in the near future, but will also provide insight into the possible success of alternative
        emission control strategies.  This preliminary study has suggested that the greatest impact of
        climate change on photochemical pollutants is likely to be  in the  urban environment or
        immediately downwind.

5.      Increase the number of regions and meteorological conditions.  This preliminary study was
        severely restricted, by  time and resources, to an examination of the climatic sensitivity of a
        model for two specific ozone episodes hi two regions of the VS. Future studies should increase
        the number of regions and meteorological scenarios analyzed, thereby permitting somewhat
        more general conclusions about the impact of global climate change on air quality.

With respect to the analysis described in Chapter 3, we recommend the following:

1.      Extend the analysis of input data. The analysis of the input data for past RIM-HI applications
        described in Chapter 3 was both preliminary and incomplete. Although past applications of
        RTM-m to four regions of the United States were examined to relate climate variation to
        elevated ozone concentrations,  only the application to  the Northeast was subjected to duster
        analysis. The input data from the other applications (Midwest, Southeast, and West) should also
        be analyzed using cluster analysis.

        The analyses should be extended in several ways.   The formation of groups of days by
        meteorology and ozone conditions can be simplified and made more robust For  example, a
        principal-components analysis could determine the one or two components that describe most
        of the variability in the input data and predicted ozone concentrations.  (The analysis of the
        northeastern data reported here indicates  that as  much as 90% of the variability can be
        described using the first two components alone). The duster or group analyses can then be
        peformed on the principal components rather than on several original variables with substantial

2.      Extend the analyses to observed data  The analyses  reported here was based on RTM-IQ
        predictions rather than observations of ozone. Model predictions are calculated from input data
        sets that represent a limited and imperfect description of the meteorology during days when
        high regionwide ozone concentrations occur.  A better approach would be to apply the same
        statistical analysis described in this report to observed ozone and meteorological data. Such an
        analysis would have the following advantages:

               Meteorological parameters that are not induded in the input data for RTM-m but
               which are important to  ozone formation—such as inversion strength,  pressure and
               temperature gradients, and boundary-layer variables (convective and friction velodties)-
               -could be estimated from meteorological observations, possibly providing more robust
               meteorological variables to relate to elevated ozone concentrations.

               Observations provide a longer and more complete data set, giving an opportunity to
               analyze high as well as low ozone conditions and their frequency of occurrence.

3.      Changes in the frequency of ozone exceedances. The data analysis reported in Chapter 3 does
        not address how a change inclimate may result in an increase in the number of exceedances of
        the ozone  air quality standard.  Obviously this is an important component of analyzing the
        potential effects of climate change on air quality. A methodology should be developed that will
        relate changes in meteorological parameters to changes in the observed frequency of ozone
        exceedances with some estimate of the uncertainty induded.



 Bufltjes, PJ.H. 1987. "Calculation and Evaluation of the Photochemical Episode of July 22-26,1980." TNO,
 Apeldoorn, The Netherlands (87-278).

 BuOtjes, P JJL, and E. Luken. 1987. "Sensitivity Calculations of the Photochemical Episode of May 29 - June
 3,1982." TNO, Apeldoorn, The Netherlands (87-277).

 Cicerone, R J., S. Walters, and S.C. Ilu. 1983.  Nonlinear response of stratospheric ozone column to chlorine
 injections.  J Geophvs. Res- 88:3647-3661.

 Craig, H., and C.C Chou. 1982. Methane-the record hi polar ice cores. Geophvs. Res. Lett.. 9:1221-1224.

 Dickinson, RJL, and RJ. Cicerone. 1986. Future global warming from atmospheric trace gases.  Nature.

 Fawcett, EJB. 1977. Current capabilities in prediction at the National Weather Service's National Meteorological
 Center.  BuH Am. Meteorol Sotx. 58:143-149.

 Gery, M.W., R J). Edmond, G.Z. Whitten. 1987. Tropospheric Ultraviolet Radiation: Assessment of Existing
 Data and Effect on Ozone Formation.*  Systems Applications, Inc, San Rafael, California (SYSAPP-87/191).

 Gery, M.W., G.Z. Whitten, and J J». Kfflus.  1988.  "Development and Testing of the CBM-IV for Urban and
 Regional Modeling."  Systems Applications, Inc. San Rafael, California (SYSAPP-88/002).

 Kessler, R.C, and S.G.  Douglas.   1988.  "Analysis of Wind Fields  for the SCCCAMP 1985 Intensive
 Measurement Periods." Systems Applications, Inc, San Rafael, California (SYSAPP-88/055).

 Liu, MJL, RJL Morris, and JJ*. KOras.  1984. Development of a regional oxidant model and application to the
 northeastern United States.  Atmos. Environ- 18(6*1145-1161.

 Moore, G JL, RJL Morris, and CDaty.  1987.  "Application of a Regional Air Quality Model to the California
 San Joaquin Valley."  Systems Applications, Ino, San Rafael, California (SYSAPP-87/155).

 Moore, G E,  RJL Morris, and S.G. Douglas.  1988. "Application of a Regional Oxidant Model to  Central
 California. Volume L" Systems Applications, Inc, San Rafael, California (SYSAPP-88/045a).

 Morris, RJL 1987. •Preparation of Meteorological and Air Quality Inputs for the RTM-HJ for the TSDF and
 Rural Ozone Projects." Systems Applications, Inc, San Rafael, California (SYSAPP-87/077).

 Morris, RJL, R.C. Kessler, S.G. Douglas, and KJL Styles.  1987. Rockv Mountain Acid Deposition Model
 Assessment Project:  Evaluation of  Mesoscale Acid Deposition Models for Use in Complex Terrain. UJS.
 Environmental Protection Agency (EPA-600/3-88/008; NITS BP88-167481-AS).

 Morris, RJL, S J). Reynolds, and MJC Lin. 1987.  The Systems Applications, Inc. Regional Transport Model-
 -Current Status and Future Needs."  APCA Conference on the Scientific and Technical Issues Facing the Post
 1987 Ozone Control Strategies, Hartford, Connecticut, November 16-19,1987.

 Morris, RJL, and DA. Stewart 1987. "Refinement of a Regional Oxidant Model and Application to National
 Parks in the Eastern United States. Part I: Incorporation of a Nested Lagrangian Reactive Plume Model and
Application to the Eastern UJS." Systems Applications, Ino, San Rafael, California (SYSAPP-87/142).


Paumier, J., D. Stinson, T. Kelly, and C. Bellinger. 1986. MPDA-1:  A Meteorological Processor for Diffusion
Analysis. User's Guide.  U.S. Environmental Protection Agency (EPA-600/8-86-011).

Pielke, RA.  1974.  A three-dimensional numerical model of the sea breezes over southern Florida. Mon.
Weather Rev.. 102:115-139.

Pielke, RA. 1984. Mesoscale Meteorological Modeling.  Academic Press.

Prather, M J., M.B. McElroy, and S.C. Wofsy. 1984. Reductions in ozone at high concentrations of stratospheric
halogens. Nature. 312:227-231.

Ramanathan, V.  1988. The greenhouse theory of climate  change: A test by an inadvertent global experiment.
Science. 240:293ff.

Shuman, F.G.  1978.  Numerical weather prediction. Bull. Am. Meteorol. Soc.. 59:5-17.

Stewart, DA., R.E. Morris, and S.D. Reynolds.  1987.  "Application ofthe Regional Transport Model (RTM-
ffl) to Western Europe." Systems Applications, Inc., San  Rafael, California (SYSAPP-87/030).

Turner, D.B. 1970. "Workbook of Atmospheric Dispersion Estimates."  U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.

Weiss, R.F. 1981. Temporal  and spatial distribution of tropospheric nitrous oxide. J. Geophvs. Res.. 86:7185-

Wuebbles, D J., and J.E. Penner. 1988. "Sensitivity of Urban/Regional Chemistry to Climate Change: Report
of the Workshop." Lawrence Livermore National Laboratory, Livermore, California (UCRL 99436).

Yu, T. 1978.  Determining the height of the nocturnal boundary layer.  JAool. Met.. 17:28-33.