vvEPA
          United States
          Environmental Protection
          Agency
                Office of Air and Radiation
                Washington D.C. 20460
EPA 400/1-87/001G
December 1987
Assessing the Risks of
Trace Gases That Can
Modify the Stratosphere
         Volume VII:
         Technical Support Documentation
         Atmospheric Science Papers

-------
     Assessing The Risks of Trace Gases
      That Can Modify The Stratosphere
Volume VII: Technical Support Documentation
            Atmospheric Science Papers
       Senior Editor and Author: John S. Hoffman
              Office of Air and Radiation
         U.S. Environmental Protection Agency
               Washington, D.C. 20460
                  December 1987
                              U.S. Environmental Protection Agency
                              !Rp?i«n 5, Library (5PL-1S)
                              ,v,'',n S. Dearborn St -set, Room 1670
                              Chicago, iL   60.604

-------
                                  APPENDIX D

                               TABLE OF CONTENTS
                                                                         TAB
"Sensitivity of an Atmospheric Photochemistry Model to Chlorine
Perturbations Including Consideration of Uncertainty Propagation,"
by R.S. Stolarski (NASA) and A.R. Douglass (Applied Research
Corporation) (1986),  Journal of Geophysical Research.

Ozone Perturbations in the LLNL One-Dimensional Model - Calculated
Effects of Projected Trends in CFC's. CH4. C02. N20 and Halons
over 90 Years.  by Peter S. Connell and Donald J. Wuebbles,
Lawrence Livermore National Laboratory (1986)

Monte Carlo Uncertainty Analysis of Stratospheric Ozone in Ambient
and Perturbed Atmospheres, by Keith E. Grant, Peter S. Connell, and
Donald J. Wuebbles, Lawrence Livermore National Laboratory (1986)

A Parameterized Numerical Fit to Total Column Ozone Changes
Calculated by the LLNL 1-D Model of the Troposphere and
Stratosphere, by Peter S. Connell, Lawrence Livermore
National Laboratory (1986)

Global Modeling of the Ultraviolet Solar Flux Incident on the
Biosphere.  by George N. Serafino (Applied Research Corporation)
and John E. Frederick (University of Chicago) (Undated)

-------
                JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. D7, PAGES 7853-7864, JUNE 20, 1986



                     Sensitivity of an Atmospheric Photochemistry Model

                       to  Chlorine Perturbations  Including Consideration

                                       of Uncertainty Propagation

                                                   R. S. STOLARSKI

                 Atmospheric Chemistry and Dynamics Branch, NASA Goddard Space Flight Center, Greenbelt, Maryland

                                                   A. R. DOUGLASS

                                      Applied Research Corporation, Landover, Maryland

                  Models of stratospheric photochemistry are generally tested by comparing their predictions for the
                composition of the present atmosphere with measurements of species concentrations. These models are
                then used to make predictions of the atmospheric sensitivity to perturbations. Here the problem of the
                sensitivity of such a  model  to  chlorine perturbations ranging  from the  present influx of chlorine-
                containing compounds to several times that influx is addressed. The effects  of uncertainties in  input
                parameters, including reaction rate coefficients, cross sections, solar fluxes, and  boundary conditions, are
                evaluated using a Monte Carlo method in which the values of the input parameters are randomly
                selected. Results are probability distributions for present atmospheric concentrations and for calculated
                perturbations due to chlorine from fluorocarbons. For i 'ore than 300 Monte Carlo runs the calculated
                ozone perturbation for continued emission of fluorocarbons at today's rates had a mean value of —6.2%,
                with a 1-sigma width of 5.5%. Using the same runs but only allowing the cases in which the calculated
                present  atmosphere values of NO, NO2, and CIO at 25 km altitude fell within the range of measurements
                yielded  a mean ozone depletion of -3%, with a  1-sigma deviation of 2.2%. The model showed a
                nonlinear behavior as  a function of added fluorocarbons. The mean of the  Monte Carlo runs was less
                nonlinear than the model run using mean values of the input parameters.
                     1.  INTRODUCTION
   Stratospheric photochemical models have been used to at-
tempt  to predict the  effects of changes in the rates at which
various chemicals are added to the atmosphere. One problem
that has received much attention is the effect on ozone of the
addition of fluorocarbons 11 and  12 (CFC13  and CF2C12).
Since first proposed by Molina and Rowland [1974], this effect
has  been  evaluated  by  examining  the response  of one-
dimensional  and two-dimensional  models  to arbitrary  in-
creases in the input of fluorocarbons. Over about the last 7
years these  evaluations have been made most frequently by
determining the difference between the ozone calculated for an
atmosphere without fluorocarbons and the  ozone calculated
in the  steady state which would eventually be reached if fluo-
rocarbon release were continued at the present rate. This pres-
ent rate has remained  relatively constant  since about 1977
[Alexander Grant and Company, 1985]; however, some recent
evaluations  have suggested that future increases are probable
[e.g., Quinn et al., 1985].
   The change in the  column amount of ozone predicted by
models has  varied over the years from a decrease of nearly
20% to a decrease of only about 3% [see, for example, World
Meteorological  Oraani:ation (WMO), 1982,  National Acade-
my of Science,  1984], The calculations  have varied mainly
because of new  information on  reaction  rate coefficients and
photodissociation cross sections which are input parameters
to such models  and,  to a lesser extent, because of improve-
ments in the physical  and chemical details of the models. The
current calculations,  using the  chemical reaction  rate coef-

   Copynght by the American Geophysical Union.
Paper number 6D0166.
0148-0227/86/006D-0166505.00
ficients given by DeMore et al. [1985], yield results at the low
end of this  range; for example, the present model gives 5%
depletion. Such  relatively small calculated decreases  are ob-
tained not because the chlorine catalytic cycle has been found
to be slow but because interference reactions of the nitrogen
and hydrogen cycles with the chlorine cycle have been found
to have greater significance than previously thought.
   Recently,  Prather  et al. [1984] have  pointed out that the
capacity  of these  models to absorb chlorine  without  large
changes  in  ozone is  limited.  When  the  total  chlorine con-
centration becomes approximately equal to the total  odd ni-
trogen concentration,  the interference between these species
saturates, and catalytic removal of ozone by chlorine again
becomes  the dominant effect.  In fact, the added chlorine ti-
trates nitric oxide via the reaction  of NO with CIO, signifi-
cantly increasing the efficiency of chlorine catalysis.  Models
thus show significant nonlinear  response to added chlorine. In
the present model a 5% decrease in ozone column is obtained
for steady state emission of fluorocarbons at present rates, and
nearly 50% reduction is obtained for 4 times the present emis-
sion rates. This type of effect is implied by the results of Ci-
cerone et al. [1983],  but their calculations were never carried
beyond the present emission rates.
  This paper considers the problem of the injection of large
amounts  of chlorine  into  a  stratospheric  photochemistry
model. This problem is considered independent of consider-
ation  of whether such chlorine concentrations are likely to be
reached in the future because the calculations reveal interest-
ing information concerning the structure  and  limitations of
the conceptual model of the stratospheric  system. Specifically,
the robustness of conclusions  concerning the effects of large
amounts  of chlorine are tested by examining their sensitivity
to variations in input data. The sensitivity is examined in two
                                                          7853

-------
7854
                            STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                       RANGE OF
                                       MEASUREMENTS
                                  MODEL^
                                                     CIO -
                      0.001
                             0.01     0.1      1      10
                               MIXING RATIO (ppbv)
                                                100
                                                    70

                                                    60

                                                    50

                                                    40


                                                    20

                                                    10

                                                     0
                                                     0
                                                                RANGE OF MEASUREMENTS
                                                                             1MODEL
                                                                                            NO
                                                             1      10    100    103
                                                               MIXING RATIO (ppbv)

60

50
40

30
20
10

0.

-
RANGE OF
~ MEASUREMENTS
\/\
ULI
Jr sL
~ >/f /MODEL
r ^
-
\ iii mil i i i ii ml i i i mill
D1 0.1 1 10
	 I 	
-
-
-
-
-
-
-
HCI "
i i mill i i i i ii ii
100 1(


^
i
UJ
O
3
1-
H
_j
<


)3
                                                                     SUNSET
                                                                     MEASUREMENTS
                                                                   DAY
                                                                   MODEL
                                                              0.1
                                                                                                10*
                                MIXING RATIO (ppbv)
                                                             1      10     100     103
                                                               MIXING RATION (ppbv)
  Fig. 1.  Comparison of steady state model calculations of the present atmosphere for four species with the range of
atmospheric measurements. Total chlorine for the model runs is 2.5 ppbv Total odd nitrogen peaked at 19 ppbv. The
shaded areas indicate the range of measurements taken from  WMO [1982] (Figures 1-96 for CIO, Figures 1-100 for HCI,
Figures 1-62 for NO, and Figures 1-71 for NO2).
ways. The first is to consider extreme values of some of the
key  rate coefficients which  determine the  NOXC1O.,. interac-
tion. The second  is to employ  the Monte Carlo  method to
propagate uncertainties  in  all of the rate  coefficients, cross
sections, solar flux values, and boundary conditions through
the model calculation.
   One-dimensional models have been notably successful in
providing a first-order description of the processes governing
stratospheric  composition.  Altitude  profiles  of  the  con-
centrations  of  species such  as  O3,  O, Cl, CIO,  NO, NO2,
HNO3, HCI, and OH have been measured  and found to be in
general agreement with  calculations  from  one-dimensional
models. When examined closely, however,  a number of areas
of disagreement are found.  For instance, ozone in the upper
stratosphere is consistently lower in the models (both one and
two  dimensional)  than in the actual  atmosphere.  The upper
stratosphere is a  region in  which these models should  be  a
reasonable representation of stratospheric photochemistry and
composition. This paper will reexamine  some of these dis-
agreements  of models with the atmosphere, including the con-
sideration of photochemical uncertainties. Model comparisons
to data in the lower stratosphere will  also be examined, along
with the implications of these results  on ozone perturbation
calculations. This  is a region in which  there are known physi-
cal limitations  to the  one-dimensional model description of
the atmosphere. These limitations do not, however, negate the
basic photochemical conclusions of the model. It is not practi-
cal at present to do a full uncertainty propagation  calculation
in a multidimensional  model. The present  calculations will
provide a useful guide in the interpretation  of multidimension-
al model comparisons to data.

        2.   BASE CASE CALCULATION USING CURRENT
                        CHEMISTRY
   For the  following  calculations a one-dimensional  photo-
chemical steady state model is used. The domain runs from 0
                                                    to 60 km in 1.25-km  intervals. Vertical  transport is repre-
                                                    sented using a  standard diffusive  formulation; the diffusion
                                                    coefficient has a minimum value of 5.4 x 103 cm2 s~J at 19
                                                    km and then increases with altitude  to 2.4 x 104 cm2 s"1 at
                                                    30 km and 3.3 x 104 cm2 s'1 at 50 km. More than 30 chemi-
                                                    cal species are considered in the oxygen, nitrogen, hydrogen,
                                                    and chlorine families as well as the methane oxidation chain.
                                                    The model transports 15 species or families of species, with the
                                                    rest assumed to be  in photochemical steady state with the
                                                    transported  quantities.  Photochemical steady state is achieved
                                                    in the model by solving the continuity equation for each
                                                    species or family in sequence, assuming previously calculated
                                                    values for each of the  other species.  Iteration is then carried
                                                    out until  convergence  is  achieved such that the change in
                                                    species between successive iterations  is less than one part per
                                                    thousand. The original model was described  by Rundel et al.
                                                    [1978] and updated by Stolarski and Douglass [1985].  Diurnal
                                                    effects are considered  by  calculating daytime and nighttime
                                                    values for diurnally varying species. Relatively simple assump-
                                                    tions are used, such as  the instantaneous disappearance of O,
                                                    O('D), NO, OH,  HO2, and Cl at night. Special  attention is
                                                    paid to the conversion  of CIO to C1ONO2 and NO2 to N2O5
                                                    and vice versa through  the diurnal cycle.
                                                      The rest of this section shows a few results from this basic
                                                    model in order  to demonstrate its  behavior. First, the present
                                                    atmosphere  calculations will be examined, followed by con-
                                                    sideration of a  fluorocarbon perturbation that is constant at
                                                    today's fluxes. Then the effect of larger amounts of chlorine is
                                                    considered,  and the nitric  oxide  titration effect is  demon-
                                                    strated. Finally, the effects of  increasing nitrous oxide and
                                                    methane are considered. The impact  of uncertainties on these
                                                    calculations is considered in the following sections.
                                                      The model present atmosphere is obtained using boundary
                                                    conditions designed to  give an odd chlorine (Cl,) mixing ratio
                                                    of 2.5 parts  per billion by volume  (ppbv) in the upper strato-
                                                    sphere. The  only chlorine-containing source molecules injected

-------
                                    STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                                                                                        7855
                         10"
                                                10
                                                NO CONCENTRATION (CM'3)

                       Fig. 2.  Calculated NO concentration versus altitude for three values of total chlorine.
into the model are CH3C1, CC14, and the fluorocarbons 11
and 12  (CFC13 and CF2C12). The actual present atmosphere
(1985) may contain as much as 3 ppbv of Cl*. Many of the
measurements with which the model will be compared  were
taken several years ago, however, and correspond to less than
2.5 ppbv of C\x. Figure 1 shows a comparison of calculated
concentrations of four of the more important species, with the
range of data given by WMO [1982]. Note that for the four
species  shown the  model gives approximately the  same con-
centrations as measured. A  more  detailed  examination  indi-
cates a  number of discrepancies. For instance, the model CIO
profile is on the high side of the measurements up to about 35
km and then turns over to become smaller than measured
values by about 40 km.
       -  -10 -
       z
       5
       _i
       O
       (j
      O -20 -
      O
      CJ
         -30 -
  Fig. 3.
         -40-
               10          20          30
               CIX AT 60KM (ppbv)
Calculated ozone change versus total chlorine for arbi-
                                                     The HCl calculation lies within the range of measurements.
                                                   The difference in shape between the calculation and the en-
                                                   velope of measurements should not be considered significant,
                                                   as the measurements comprising the range show a number of
                                                   different shapes.
                                                     The calculated  NO profile is in the middle of the data en-
                                                   velope, except above the stratopause and near the tropopause.
                                                   Both disagreements are traceable to model  assumptions but
                                                   should  not significantly affect  the perturbation calculations
                                                   shown below. In the 50- to 60-km region, virtually all of the
                                                   odd nitrogen is present as NO. Its concentration is controlled
                                                   by the balance of transport from the source region below, with
                                                   local loss following the photolysis of NO and upward trans-
                                                   port to the sink region above 60 km. In this model the upward
                                                   transport process is represented by an upward flux at the
                                                   upper boundary  that  is related to the  loss expected to  take
                                                   place in the mesosphere  Because the odd  nitrogen contri-
                                                   bution to odd oxygen loss above 50 km is small, differences in
                                                   the NO  concentration here should not significantly affect the
                                                   ozone balance.  Near the  tropopause, NO  is  substantially
                                                   below measurements, while  NO2 is not. Again, this is a region
                                                   where odd nitrogen  processes make insignificant  contributions
                                                   to odd oxygen  loss. In the middle and  upper  stratosphere,
                                                   particularly in the altitude  region where  NO., loss processes
                                                   dominate ozone destruction, NO2 daytime and nighttime cal-
                                                   culations are near the middle of the range of measurements.
                                                     Many other examples could be considered to demonstrate
                                                   TABLE 1.  Asymptotic Value of Total Chlorine, Clx, as a Function
                                                   of the Release Rate of Fluorocarbons 11 and 12 in Factors Times the
                                                                          Present Rate
trary changes in key interference reactions. JPL 85-37 rates are from
DeMoreetal. [1985].
Factor Times
Present
Fluorocarbon
Flux
0
1.0
1.5
2.0
2.5
3.0
3.5

Asymptotic
a,,
ppbv
1.1
7.5
10.4
13 1
15.5
17.4
20.5

-------
7856
                                    STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                     DETERMINE SET OF
                    RATE COEFFICIENTS,
                  BOUNDARY CONDITIONS,
                           ETC
                        CALCULATE
                         AMBIENT
                       ATMOSPHERE
                        CALCULATE
                        PERTURBED
                      ATMOSPHERE (S)
     Fig. 4.  Logical flow diagram for Monte Carlo calculation.

how this particular one-dimensional model compares with at-
mospheric data.  The overall picture  is similar  to  that pro-
duced by other one-dimensional models (see, for example, Pra-
ther et  at.  [1984];  Wuebbles [1983a]), that is, general agree-
ment with measurements but many detailed differences. Com-
parison of model results with measurements will be considered
in more detail below when the Monte  Carlo uncertainty prop-
agation results are discussed.
   The model  has been run for the fluorocarbon perturbation
case, that is, comparison of the predicted  steady-state atmo-
sphere with today's emissions of fluorocarbons 11 and  12 to
the  predicted atmosphere  with no fluorocarbons.  This  has
been done for the idealized case in which  the fluxes of other
species, such as CH4 and N2O, are held constant. Because of
the nonlinear response  of the model for higher chlorine con-
centrations, similar to  that shown by Prather et al. [1984],
these calculations have been extended  to consider the effects of
constant fluorocarbon fluxes up to 3.5 times the present-day
release  rate. Even  if growth to these levels proves unlikely,
these results are  instructive concerning the expected behavior
of a chlorine-dominated atmosphere and the behavior during
the transition to that atmosphere.
   As in the work by Prather et al. [1984], a nonlinearity in
ozone change as a function of chlorine added to the model
was found. The calculated ozone depletion in the upper strato-
sphere  was found to   be  relatively linear, while the lower
stratosphere showed extreme nonlinear behavior. The lower
stratospheric change is positive for small perturbations due to
a near  cancellation  between  the direct  chlorine-catalyzed
ozone loss and the combination of self-healing (more UV pen-
etration causing increased production) and interference of Clx
with the NO.,, catalytic cycle. As Cl, increases, the interference
with NO,,, increases to  the point  where the titration of NO
takes place. This is illustrated  in  Figure 2, which shows  the
calculated NO concentration versus altitude for successively
larger C\x values. At 120 ppbv of Clx, which is obtained at 3.5
times the present fluorocarbon flux, a significant  bite has been
taken out of the NO  profile between 20  and  25  km. This
occurs  because the reaction  CIO +  NO—> NO2 + Cl  takes
over the conversion of NO  to NO2 and drives the NO to NO2
ratio to small values. In this region the same reaction is  the
dominant path for conversion of CIO back  to Cl. The catalytic
path CIO + O—> Cl + O2 is only a few percent of the conver-
sion of CIO to Cl. When the titration of NO has occurred, the
conversion of CIO to Cl slows down significantly because of
the absence of NO. Thus less chlorine is available as Cl atoms
for conversion to HC1. However, the rate of removal from the
HC1 reservoir remains high and is even slightly enhanced by
an increase in OH. Thus  the HO reservoir begins to drain,
and  the chlorine chemistry  of  this part of the  stratosphere
becomes dominated by  CIO and C1ONO2. The  resulting in-
crease in the  ratio of CIO to Clx leads to efficient ozone de-
struction.  When the  Clx concentration approximately equals
the  odd nitrogen concentration (which peaks at  18.6 ppbv in
this  model),  the  ratio C1O/C1X increases dramatically  from
about 0.01 to 0.07, with a change of only a few ppbv in Clx. To
demonstrate  this dependence  on the  odd nitrogen level,  a
second case  was run in which the odd nitrogen level  was
arbitrarily reduced to about 2 ppbv  by reducing the N2O flux
into the atmosphere by a factor of 10. In  this case,  C1O/C1X
increased  rapidly  at  a  much  lower Clx concentration (~2
ppbv).
  The  above explanations suggest  that  the  calculated large
ozone  changes due  to  adding  15-20 ppbv  of  Clx  may be
robust to changes in the chemical rate coefficient data. To test
this  assertion, runs of the model were made with arbitrary,
large changes in  the key  NO^-CIO,  interference reactions.
Figure 3 shows the  calculated  column ozone change versus
Cl, for the base case  and for cases in which the  reaction rate
coefficients  for   CIO + NO->  Cl + NO2  and  CIO + NO2
 + M —> C1ONO2 + M  were separately increased by a factor
of 10 and decreased  by a  factor of  100. Note that decreasing
either rate (by an amount that  is tantamount to setting it to
zero) causes the added chlorine  to destroy ozone  efficiently, so
that a large, linear response is obtained. Increasing either rate
causes NO^-Cl, coupling  to be so efficient that  chlorine has
little effect on ozone until  Clx is in the  15-20 ppbv range, and
then the interference rapidly saturates and  ozone responds
strongly to only small  amounts of additional chlorine. This
result suggests that the conclusion  that a large  ozone deple-
tion will be observed  at 15 20 ppbv Clv is robust but is based
on varying only one rate at a time. Results shown in section 3,
from a Monte  Carlo uncertainty propagation  in which all
rates are varied  at the same  time, demonstrate the limitations
of this conclusion.
  Indicated  in Table 1  are the fluorocarbon  fluxes necessary
to obtain  the given stratospheric Cl,. amounts for the present
chemistry case.  Note that as  the ozone decreases,  each in-
crement of fluorocarbon flux is less efficient at producing Clx
in the  stratosphere.  This  is due to  the  increased ultraviolet
solar radiation penetration,  which leads to decreased fluoro-
carbon lifetimes and smaller atmospheric concentrations. Thus
the effects of fluorocarbon release on stratospheric ozone are,
in a sense, self-limiting. Unfortunately,  to achieve  this self-
limitation, the fluorocarbons must reduce the ozone content of
the atmosphere by a significant  amount.
  All of the  above model runs have assumed changes in the
fluorocarbon fluxes into the atmosphere with  all  other bound-
ary fluxes held constant. This is certainly not the case in the
real atmosphere  where  a  number of other species, including
CH4 [Blake  et al., 1982;  Rasmussen and Khalil, 1984], CO2
[Keeling et al., 1984], and N2O [Weiss, 1981], are observed to
be  increasing. Furthermore, several others, including tropo-
spheric O3 [Loaan, 1985], CO  [Khalil and Rasmussen,  1984]
and NOX  [Crutzen and  Gidel, 1983], are strongly suspected to
be increasing. Two of these, CH4 and N2O, will be considered
to illustrate their possible interaction with the calculated chlo-
rine effects.  Increases in methane will tend to slow down the
chlorine catalytic effects through conversion of active ClO^ to

-------
                                    STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                                                                                                 7857
   1 -
    70
         75   8.0   8.5  90   95  100  10.5  110  115 120x10

            RATE COEFFICIENTS3 MOLECULE"1 SEC~1)
                                                      -12
  Fig. 5  Histogram of individual data points for the rate coefficient
       NO, -> O2 + NO. The distribution contains 62 data points
of
with a mean of 9.45 x
deviation of 0.74 x  10~12.
above parameters.
                                        s"  and a standard
                      The smooth curve is a Gaussian with the
the reservoir HC1. Nitrous oxide increases provide more NOX
to interfere with C1OX catalysis through  the CIO + NO and
CIO + NO2 + M reactions.
  Because constant fluxes were assumed, as the stratospheric
column of ozone begins to be depleted, more ultraviolet radi-
ation penetrates to  any given  altitude  and the loss rates for
CH4, N2O, and other species are increased. This shortens their
atmospheric residence time, resulting in a smaller abundance
in steady state for the assumed (constant) flux. For a pertur-
bation calculation  with  3.5 times the present fluorocarbon
flux, the steady state, ground-level mixing  ratio of CH4  is
reduced from 1.6 to 0.82 ppmv and that of N2O from 300 to
240 ppbv. To test the importance of changes in  N2O and CH4,
a few sensitivity tests were run with differing assumptions con-
cerning changes in the fluxes of CH4 and N2O.  At a fluorocar-
bon flux of 3.5 times  the present level the  CH4 flux was in-
creased by successively  larger  factors.  At 1.5  times the CH4
flux the ground-level concentration returned to a value slight-
ly above the ambient  1.6 ppmv, and the ozone depletion was
reduced from 40%  to 30%.  At double the CH4 flux its con-
centration in steady state was 3.7 ppmv, and the ozone  deple-
tion was further reduced to 23%. If enough CH4 were added,
the recovery in  the ozone perturbation  would end because
ozone  depletion  due to enhanced HOX would  begin to  domi-
nate. Increasing  the N2O flux  shows a more dramatic  effect.
At  1.5  times the present flux  the perturbation due to 3.5 times
the present  flux of fluorocarbons  is reduced  to  20%.  At  2
times the N2O flux  the fluorocarbon perturbation has become
15%, but at higher N2O fluxes, larger perturbations result due
to NOX effects on ozone.

             3   MONTE CARLO CALCULATIONS
  The Monte  Carlo method of uncertainty  propagation has
been used to examine the questions of comparison of  calcu-
lated atmospheric concentrations with measurements and to
determine their  relationship to  chlorine perturbations. The
basic method was described by Stolarski et al. [1978].  That
study is updated here  by using the  improved model described
in the  previous section, by using updated reaction rate and
cross-section data, and by consideration of uncertainties both
in the  preexponential factor of rate coefficients  and in the
temperature dependence. It is  shown that the  uncertainty in
the calculated  ozone depletions may be reduced significantly
by  requiring that calculated values of key constituents fall
within  a range of values specified by measurements.
  Figure 4 reviews  schematically the Monte Carlo technique,
as applied to this photochemical model. For calculation with
the base case chemistry, reaction rate coefficients, absorption
cross sections, solar fluxes, and boundary conditions are input
to the model. These are used with the difference formulation of
the constituent  continuity equations to  calculate an ambient
atmosphere  which approximates conditions previous to  any
fluorocarbon release. A  series  of  perturbed  atmospheres is
then calculated for the particular set of changes being tested,
in this case, increased levels of fluorocarbon fluxes. The results
of each  perturbed run are compared to the ambient case.  For
a Monte Carlo study one  additional step is  taken. A set of
random numbers is generated, one for each input  parameter
to be varied. The estimated probability distributions are then
used with each of these random numbers to pick a new value
for each input  parameters. These  are used in  the  model to
calculate an  ambient atmosphere  and a  series of perturbed
atmospheres. Then a new set of random numbers is generated,
and a completely new set of input  parameters is used in the
model. In this way, distributions of calculated concentrations
as well as perturbed concentrations and changes are accumu-
lated.
  A major  advantage of the Monte Carlo method  described
above over more conventional one-dimensional models is il-
lustrated in Figure 5. Shown is a histogram of all of the indi-
vidual measurements of the room temperature rate coefficient
of the  reaction  O + NO2^O2 + NO  by  three different
groups {Bemand et al., 1974; Slanger et al., 1973, Davis et al.,
1973]. These 62 data points form a distribution that is fit  well
by either a Gaussian or a log-Gaussian distribution with pa-
rameters given by DeMore et al. [1985]. These are a central
value of 9.4 x 10~12 and a one-sigma uncertainty  of ±10%.
Also shown in Figure 5 is a vertical line representing the mean
value which is normally used in a photochemical model. This
is not a complete representation of the known information on
                                                                    -10
                                                                    -20
                                                                 O
                                                                 u
                                                                 1-30
                                                                 O
                                                                 O
                                                                 3
                                                                   -40
                                                                   -50
                                                                                                 + 1o
                                       BASE
                                       CASE
                                                                                                      \MEAN
                                                                                FACTOR TIMES PRESENT
                                                                                 FLUOROCARBON FLUX
                                                              Fig. 6.  Monte Carlo results for ozone change  versus the factor
                                                            times the  present fluorocarbon fluxes Solid  line  is the  base case,
                                                            dashed line is the mean of the 329 cases. Vertical  bars are the one-
                                                            sigma uncertainty limits

-------
to
UJ
to

o
u.
O
oc
UJ
CO
   56
   40
30
   20
   10
DC
LU
m
5
1 I 1
1.0x PRESENT
FLUOROCARBON FLUX
J
r
: /
, nllnnrJ 9
i
J
	
a
-
1
   25
   20
   15
10
    -80    -60      -40     -20      0

             A03 COLUMN/03 COLUMN (%)
                                           +20
            2.5 x PRESENT

        FLUOROCARBON FLUX
           n iffln
    -80    -60      -40     -20      0

            AO3 COLUMN/Oa COLUMN (%)
                                           +20
3U


40
to
3 30
o
U_
O
cc
UJ
1 2°
Z
10
Q
1 	 1 	 r 	 1 	
1. 5 x PRESENT
FLUOROCARBON FLUX
-

_


J
Jll
If
IT
-80 -60 -40 -20
b


-

.

L



-
1
ZHl


40
CO
HI
< 30
CJ
u.
O
CC
UJ
m 20
Z
10
n
i i
2.0 x PRESENT
FLUOROCARBON FLUX
~

_


- . J
n LA™
,
n
1







I
I

c






J
11
/
~

-

"-
\
if
.
-
n
0 +20 -80 -60 -40 -20 0 +2
                                                          A03 COLUMN/OS COLUMN <%)
                                                                                                           A03 COLUMN/OS COLUMN(%)
                                                 in
                                                 UI
                                                 CO
                                              O

                                              IT
                                              UJ
                                              00
                                                    25
                                                    20
                                                    15
                                                 10
                                                         3.0 x PRESENT
                                                      FLUOROCARBON FLUX
                                                                                                     25
                                                                                                     20 -
CO

<  15

u.
O
cc
ul
CO
                                                                                                  10 -
            3.5 x PRESENT
        FLUOROCARBON FLUX
                                                            D


                                                            I
                                                                                                                                                           y,
                                                                                                                                                           o
                                                 _80     -60    -40     -20      0

                                                          A03 COLUMN/03 COLUMN (%}
                                                                                            +20
   -80    -60     -40    -20      0

             A03 COLUMN/03 COLUMN (%)
                                                                                                                                              +20
              Fig 1.  Frequency distributions for the calculated change in the column content of ozone for various values of the fluorocarbon flux. The

                    shaded area shows the cases for which the NO, NO2, and CIO concentrations all fell within the range of measurements at 25 km.

-------
                                    STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                                                                                                7859
       25
       20
   to
   LU
   CO
   UL
   O
   DC
       10
    3.5 x PRESENT
FLUOROCARBON FLUX
 PLUS 1.2x N2OFLUX
 PLUS 2.0 x CH4 FLUX
       -80     -60     -40      -20       0
                 A03 COLUMN/OS COLUMN (%)
                                      +20
  Fig. 8.  Frequency distribution for the calculated change  in the
column content of ozone for 3.5 times the present fluorocarbon flux,
1.2 x N2O flux, and 2.0 x CH4 flux. The shaded area shows cases in
which NO, NO2, and CIO concentrations all fall within the range of
measurements at 25 km.
this  reaction. Using the  Monte Carlo method,  an approxi-
mation to the complete distribution is considered rather than
a single value, and results should better  reflect the status of
knowledge concerning this rate coefficient. Of course, a well-
measured reaction was chosen for this illustration. Other reac-
tions have only been measured a few times (or only the mean
of the actual  runs were reported), and the measurements may
disagree with one another by substantial amounts. For these
cases it is not obvious how much improvement is obtained by
attempting to represent an entire distribution.
  The Monte Carlo technique was  applied to the  model de-
scribed earlier to generate a total of 329 cases  of varied inputs
for calculation of the atmosphere without fluorocarbons, with
enough fluorocarbons to  give a reasonable steady state repre-
sentation of the present atmosphere, and with the sequence of
increasing fluorocarbon perturbations to investigate the model
response to large chlorine amounts.  Figure 6 shows the calcu-
lated change in the column content of ozone compared to the
no-fluorocarbons  case as  a function of the injected fluorocar-
bon  flux shown in units of the present flux of fluorocarbons 11
and  12 (PFF). The solid curve, labeled the base case, gives the
result using the mean values for each input  parameter. The
dashed curve is the mean ozone depletion obtained from the
329 cases. The mean curve is significantly more linear than the
base case curve.  The mean perturbation calculated for the
PFF is -6.2%, as opposed to -5.0% for the base case. At 3.5
times PFF the calculated mean perturbation is -31%,  as
opposed  to —40%  for  the base case. The  vertical  ranges
shown  are the  one-sigma variance of the calculated distri-
bution of ozone change, but  these are only a general guide to
the width of the  distributions because they are  asymmetric.
The  distributions of calculated ozone changes are given in
Figure  7; the shaded areas  indicate the  distributions which
remain when  the calculated concentrations in NO, NO2, and
CIO  are required  to fall within a range of values specified by
measurements. This attempt to limit  the uncertainty  in the
calculated ozone  depletions is  discussed at the end  of this
section. For the case of 1.0 times the PFF the cases  ranged
from an increase  of slightly more than  2% to a decrease of
35%. The distribution is skewed, with the most probable value
between —2 and —3%. Figures 7 a-7/show the progression of
distributions obtained as the fluorocarbon flux was increased.
The mean increases as the flux increases,  but the positive cases
do not entirely disappear,  even for  the perturbation of 3.5
times the PFF. At 3.5 times PFF the  distribution has a bimo-
dal character with one group of cases having ozone changes
between about —35  to  —60%  and with the other group of
cases ranging from 0% to about —30%.
  As in the previous section, these fluorocarbon flux changes
have been made with all other boundary fluxes held constant.
To illustrate  the  effects of uncertainties on increased inter-
ference with the chlorine cycle caused by increased  CH4 and
N2O,  and the distribution of ozone column depletions at 3.5
PFF with  1.2 times the N2O flux and  2.0 times the CH4 flux is
given  in Figure 8. The mean ozone  depletion in this  case is
smaller than  for the 3.5 PFF case in which  CH4 and N2O
fluxes  are held fixed, and the bimodal character apparent in
Figure 7/is not observed.
  In an effort to understand these distributions and to reduce
the uncertainty in the calculated ozone depletions, compari-
sons have been made of measurements  with  calculated  con-
centration distributions for the  current atmosphere. The first
comparison considers the well-known problem that the calcu-
lated ozone concentrations near the stratopause are  tens of
percent less  than measurements (see,  for  example,  Butler
[1978] or Solomon et al. [1983]). In  Figure 9 the probability
distributions  of calculated ozone concentrations at  50, 40, 30,
and 20 km are compared to the 1-sigma variability ranges of
data given by Krueger and Minzner [1976]; the  Krueger-
Minzner (KM) data are in  close agreement with more recent
satellite data  [McPeters et al., 1984]. Although there are cases
which  fall within  the 1-sigma range of variability of the KM
data at each altitude, there  are no cases  which fall within this
range at all  four altitudes. This suggests that the problem of
disagreement between measured and calculated ozone is not
likely  to be solved within  the ranges of current rates, cross
sections, and  solar fluxes. This is consistent with the results of
Froideveaux et al. [1985]. There are only three cases which fall
within  2-sigma variability of the KM  data at all altitudes, and
these cases were examined individually to search for common
features. No  obvious single variation in production, loss, or
species concentration was  found. A common characteristic
was that all three of these cases  resulted  in small ozone deple-
tions, less  than 2% for 1 times PFF and  less than 14% for 3.5
times PFF. A large number of cases were found to  fall within
2.5-sigma  variability, but these cases correspond to the full
range of possible  ozone reductions. It is concluded  that con-
formity to the measured KM  ozone profile is  not a good
criterion for  reduction of the  uncertainty in  the  calculated
range of ozone depletions. This comparison does indicate that
the disagreement  between the measurements and  calculated
ozone concentrations at high altitude  is with the overall shape
of the  ozone  profile. This problem  cannot be  resolved  by
changes that will produce agreement at a single altitude.
  Next, consider  the comparison of the concentrations  of
some of the  other  minor species with measurements. Because
local concentrations  are more directly comparable  to model
output than are column abundances, comparisons are made
with measurements at single altitudes. Although  the physical

-------
7860
                                    STOLARSKI AND DOUGLASS  CHLORINE PERTURBATIONS
               2x1010 4x1010   6x1010  8x10'°
                 OZONE CONCENTRATIONS"3)
                                              1 x 1012
                                                                         OZONE CONCENTRATION (CIvT
       25
       20
     o
     cc
       10
                                             30km
2x1012   4x1012  6x1012   8x1012  1x1012
 OZONE CONCENTRATIONICM-3)
                                                                 25
                                                                 20
  1/3
  LLJ
  < 15
  O
  LI.
  O
  cc
  LLI
  2 10
                                                                                                     20 KM
                                                                            _d
            2x1012  4x1012   6x1012   8x1012
              OZONE CONCENTRATIONICM-3)
                                                                                                             1013
           Fig.  9.  Frequency distributions for the calculated ozone concentrations at 4 altitudes (a) 50 km, (b) 40 km, (c) 30 km,
         and (d) 20 km The shaded areas indicates the one-sigma variability range, as given in the Krueger-Minzner mid-latitude
         model which is based on rocket measurements.
representation of the dynamical processes near the ozone con-
centration peak is  poor in one-dimensional models, the calcu-
lated ozone column changes are highly correlated with the
local ozone changes at these altitudes.  This  is illustrated by
Figure 10,  a scatter plot of the local ozone change  at 25 km
versus the  ozone column change for the perturbation caused
by the PFF. The high  degree of correlation indicates that the
chemistry leading  to local  changes at 25 km should show a
high  correspondence  with  the  calculated  change  in  the
column.
  Figure 11 is a scatter plot  of the diurnal average NO con-
centration  versus  the  diurnal  average  NO2 concentration,
both at 25 km. Also shown are several lines of constant ratio
of NO2 to NO. There is a tendency for the cases  to cluster
about a ratio of approximately 4. This is  an indication  that
many of the input parameters that are  varied in the calcula-
tion  tend  to preserve  the NO2-NO ratio.  There  is also a
spread about this ratio ranging from about 2 to 10, indicating
that other input parameters tend to drive NO2/NO from the
mean ratio when they are varied. Also shown in Figure 11 are
pairs of horizontal and vertical solid lines. These indicate the
range of measurements as  taken from WMO [1982]. These
measurements have  been  adjusted to  correspond approxi-
mately to diurnal average values for comparison to  model
results. The majority of the Monte Carlo cases fall within the
box formed by  the intersection of these  lines.  A number  of
cases  fall outside the  range of NO2 measurements, some fall
outside the range of NO measurements, and a few fall outside
the range of both measurements.
  The behavior  of the ambient atmosphere concentrations re-
sulting from the random input parameter variations is illus-
trated  further in Figure 12, which is a  scatter  plot  of the
diurnal average  NO concentration versus  the diurnal average
CIO concentration, both at 25 km. Again the points tend to lie

-------
                                     STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                                                                                                   7861
        +20
        +10
  s
  $
   ro
  Q
   m
  o
 -10
        -20
        -30
        -40
          -40   -30   -20    -10
                                              +10   +20
  Fig.
             A03 COLUMN/OS COLUMN (%)
10.  Scatter plot  of  the change in the  local  ozone  con-
centration at 25 km versus the change in the ozone column content;
both changes are for a calculated perturbation  due  to addition of
fluorocarbons at the present rate
along  a  line, this  time the line of constant  product NO
 x CIO = 1  x 10'6  cm"6.  The solid lines are  the  range  of
measurements from WMO  [1982]. The range for CIO is taken
from measurements by  Anderson  and coworkers [Weinstock
et al., 1981], but the two largest measurements are eliminated
from the range as they  are many  sigma from the mean. The
range of measurements falls on  the  low side of the  Monte
Carlo cases. This was illustrated  earlier  in  Figure  1, which
compared the same measurements to  the base case. Figure  13
shows  the same type of scatter plot for  HNO3 versus NO.
                                                    10"
                       NOAT25KM(CM-J)
  Fig  11.  Scatter plot of the calculated ambient NO2 concentration
versus the calculated  NO concentration, both diurnally averaged at
25 km. The dashed lines indicate the range of measurements for both
species, as given by WMO  [1982J The solid lines indicate constant
values for the  ratio NO2/NO. Base case concentrations are NO2 =
2.0 x 10" cm"3 and NO = 4.5 x  108 cm"3
                                                                 I
                                                                 y
                                                           <
                                                           O
                                                                     10'
                                                                                   CIO AT25KM (CM
  Fig. 12.  Same as Figure 11 for NO versus CIO The solid lines
indicate constant  values for the product NO times CIO. Base case
concentrations are NO = 4.5 x 108 cm~3 and CIO = 2.5 x 107 cm" 3.
                                                       Again, the data encompass the majority of the points as for
                                                       NO versus NO2. Figure 13 combined with Figure 11 demon-
                                                       strates that in spite of the poor representation of the actual
                                                       physical processes that control total odd nitrogen  in the lower
                                                       stratosphere,  the  individual  odd  nitrogen  species  con-
                                                       centrations  are  generally  in  good agreement with measure-
                                                       ments. Even though  the total odd nitrogen at 25 km agrees
                                                       with measurements for the wrong reason (that is, vertical dif-
                                                       fusion is not  really  the dominant  transport mechanism), the
                                                       above results demonstrate that the chemistry done with the
                                                       odd nitrogen  that is there conforms to atmospheric measure-
                                                       ments within reasonable uncertainty.
                                                         The calculated ozone depletion due to increased emissions
                                                       of fluorocarbons depends  upon the  calculated ambient con-
                                                       centrations  of some species. Figure 14 is a scatter plot of the
                                                       calculated ozone depletion for  the present fluorocarbon flux
                                                       case versus the calculated diurnal  average  NO concentration
                                                       at 25 km for the present atmosphere. The crescent  shape of the
                                                                 10
                                                                              NOAT25KMICM-
                                                       Fig  13   Same as Figure 11 for HNO3 versus NO. Base case con-
                                                       centrations are HNO3 = 4.4 x 10" cm"3 and NO = 4.5 x 108 cm"3

-------
7862
    STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
        +10
                                                                    +10
       -40
                 2x109
           0        5x108      1x109      1.5x109
                        NOAT25KMICM-3)
  Fig. 14.   Scatter plot of the calculated change in the ozone column
for an injection of fluorocarbon at the present rate versus the calcu-
lated diurnal average concentration of NO at 25 km in the ambient
atmosphere. The dashed lines indicate the range of measurements for
NO,  as given by WMO [1982], Base case concentration  for NO is
45 x 108cm-3.

distribution of points is typical of all the scatter plots of ozone
change versus concentration of species in the nitrogen family.
At low NO values a wide range of ozone changes is calculated,
while at high NO values small ozone changes are always ob-
tained. The solid lines, the range of measurements as before,
indicate that  both the lowest and highest NO concentrations
are unacceptable. This removes  a number of cases with small
ozone depletion and most of the largest ozone depletion cases.
Figure 15  shows the same  type of plot  for CIO. The  corre-
lation is  quite  high, with a correlation  coefficient  of —0.87.
Because the model predictions of CIO tend to be on the high
side of the measurements, the cases selected by agreement with
CIO observations strongly favor the smaller ozone depletions.
Because of the relationship between CIO and NO, about one-
                     4x107
8x107
                                            1.2x108
                                        ~3
                       CIO AT25KM (CM
Fig.  15.   Same as Figure 14 for CIO. Base case concentration for
                    CIO is 2 5 x  107cirr3.
                                                                 0-10
                                                                 S
                                                                 Ill
                                                                 13 -20
                                                                 o
                                                                   -30
                                                                    -40
                                                                              SxlO6    1X106  1.5X106
                                                                                   OH AT25KM (CM-3)
                                                                     2X106   2.5X106
                                                             Fig 16.   Same as Figure 14 for OH. Base case concentration for OH
                                                                                  is 7.6 x  105cm~3.
third of the cases in which CIO is larger than the range speci-
fied by measurements correspond to cases which NO is lower
than its range of measurements. Some caution should be exer-
cised in the interpretation of these screened results because the
additional  uncertainty due to the vertical transport coefficient
was not considered. Its effect should be to broaden most of the
probability distributions somewhat. The interrelationships be-
tween NOj species, CL, species, and the calculated ozone de-
pletions will  be largely unaffected, and strong  distortions  of
the mean values are not expected.
  Another interesting correlation is that of ozone column
change with the calculated OH concentration. Because of the
critical role of OH in removal of chlorine from the reservoir
HC1, its concentration  is expected to  play  a crucial role  in
determining the catalytic efficiency of chlorine towards ozone.
This has been emphasized by previous calculations \_WMO,
1982; Wuebbles, 1983a, b"].  However,  the change  in  ozone
column shows no significant correlation with  the  OH con-
centration  at  25 km (correlation coefficient  —0.11), as illus-
trated by Figure 16.  The explanation  of this result requires
careful consideration of how the Monte  Carlo calculations
differ from the  previous calculations.  In  earlier calculations,
usually  only  a  few rate coefficients were changed, and the
changes have often been dominated by a single rate coefficient
in the HOX  family. Thus the response to  a  change  which
caused increased HOX concentrations would  be  an increase in
ozone destruction by chlorine and  a  decrease  in  ozone de-
struction by  NOX. The  present calculations change  all of the
rate  coefficients and  other input parameters for  every case.
The dominant changes are not necessarily directly in the HOX
family and affect other species as well as OH. For instance, a
change that results in increased O('£>) will yield a higher OH
concentration,  which  tends to give higher sensitivity toward
chlorine. That same change will give high NO., concentrations
through the  reaction  of Of1/)) with N2O. Higher  NO,, con-
centrations tend  to interfere  with chlorine catalysis of ozone
destruction, counteracting the direct effect of  HO.,. The overall
result for the different combinations which produce variations
in OH is the poor correlation of ozone depletion with ambient
OH. Because of the poor correlation,  no attempt  has been

-------
                                     SlOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
                                                                                                                   7863
 made to screen out ozone depletion cases by comparisons of
 measured with calculated OH.
   It  was  found that of the 329 Monte Carlo  cases,  125 had
 ambient concentrations of CIO, NO, and NO2  that fall within
 their measurement ranges. These 125 cases have smaller mean
 ozone depletions for all levels of chlorine perturbation than do
 the full set. The distributions of ozone depletions are given by
 the shaded areas  in Figure 7. At  1 times the PFF the distri-
 bution has a mean of —3.0%  and a one-sigma standard  devi-
 ation  of 2.2%.  None of the cases remaining have an ozone
 depletion greater  than 10%. At 3.5 times PFF the calculated
 ozone change is -23.9 + 13.6%. In this case the reduction in
 the mean resulted from  the removal  of more cases from the
 large-depletion part of the bimodal distribution than from the
 lower peak. Not all of the large-depletion cases were removed,
 however.  The  shaded area of Figure 8 shows the screened
 result for the 3.5 times PFF, 1.2 times the nitrous oxide flux,
 and  2.0  times  the methane  flux case.  The  result  is  now
 — 9.1 + 7.0%,  and  all cases above about  a 30% depletion
 have been removed.

              4.  SUMMARY AND CONCLUSIONS
   Using  a  one-dimensional   stratospheric   photochemical
 model, sensitivity studies in which key rate coefficients are
 changed by substantial amounts indicated that the calculated
 ozone depletions  exhibit nonlinear behavior as the flux of
 chlorine-bearing compounds  such as fluorocarbons is in-
 creased. This behavior is observed because as  chlorine is in-
 creased the nitric oxide is titrated from the lower stratosphere,
 leaving chlorine to act  upon ozone  with little or no  inter-
 ference. Without the interference of odd nitrogen species, addi-
 tion of chlorine results in large calculated ozone depletions as
 compared  to the  present atmosphere.  Monte Carlo analysis,
 designed to explore the  total extent of rate coefficient space,
 finds a much wider spread of potential conclusions concerning
 the impact of large chlorine  amounts. The Monte Carlo calcu-
 lation employs probability distributions for the values of pa-
 rameters,  such  as reaction rate  coefficients, cross sections,
 solar flux, and boundary conditions, and produces probability
 distributions for calculated concentrations and ozone changes.
 The ozone depletion due to fluorocarbon injection yields a
 skewed distribution for injection of the present fluorocarbon
 flux,  with a mean calculated depletion of —6.2% ± 5.5% in
 the ozone column.  This  is compared to the  calculation of
 — 5.0% for the  base case, using the central values of all the
 input  parameters. The  distribution  moves  toward higher
 ozone depletions as the fluorocarbon flux increases until at 3.5
 PFF a double-peaked distribution  is obtained, with peaks cen-
 tered around —15 and —45% change in the ozone column.
 The mean is  —31%, with a one-sigma standard deviation of
 + 17%. This compares to the base case calculation of -40%.
  Atmospheric measurements  have been used in an effort to
 reduce the uncertainty in the calculated ozone depletions by
 considering only those Monte  Carlo cases which fall within a
 range  specified by the measurements. One such study  con-
 cerned the ozone concentration profile itself. The upper strato-
 spheric problem in which  the  model obtains  lower ozone
 values than measurements was  confirmed; however, a signifi-
 cant number of the  Monte  Carlo  cases did  overlap with the
data.  At 30 km the model tended to predict  too much ozone,
and while  a significant number of cases again overlapped  with
data, no cases passed within  the one-sigma variability limits of
the  data at both 30 and 50 km. When the variability limits
 were extended to two-sigma, only three cases passed this test.
 At 2.5 sigma, more than 40 cases passed, and these showed the
 entire range of possible ozone  depletions, indicating that no
 reduction in the overall uncertainty could be obtained by re-
 quiring  agreement of  the  calculated  ozone  profile with
 measurements.
   A second data comparison was made for some of the minor
 constituents. Comparisons were made at 25  km  because here
 the calculated local ozone depletion is best correlated with the
 calculated column  ozone depletion.  When the Monte Carlo
 cases were screened, such that only cases for which the calcu-
 lated concentrations of NO, NO2,  and  CIO at 25 km  fell
 within the range of measurements,  more than one-third of the
 cases remained. All cases  remaining had a  calculated  ozone
 depletion of less than  10% for the  PFF. The distribution was
 symmetric, with a mean of -3% and a standard deviation of
 ±2.2%.  Thus  these measurements tend to select against  the
 large ozone depletion  cases. As  the fluorocarbon flux was in-
 creased, there was also a tendency to select against high ozone
 depletion cases, but by 3.5 PFF a significant number of cases
 of large ozone depletion were obtained.
   Although this study is only a beginning  of what is possible
 in considering Monte  Carlo results in the context of model-
 measurement intercompansons, several tentative conclusions
 can be drawn:
   1.  It  is  possible to reduce  the uncertainty in  calculated
 ozone depletions by requiring  calculated  concentrations  for
 key minor constituents to fall  within the  range specified by
 measurements.
   2.  Relatively low  ozone depletions  are  associated with
 cases  in  which  calculated  concentrations in NO,  NO2, and
 CIO fall within the range specified by measurements.
   3.  For  larger chlorine injection rates at  least part  of the
 bimodal  nature of  the ozone depletion prediction  passes the
 minor constituent measurement  test, thus retaining the  possi-
 bility  of large ozone depletions.
  4.  The  problem of fitting  ozone concentrations  in the
 upper stratosphere  is actually one of profile shape and simul-
 taneously fitting in both the upper and middle stratosphere.
  5.  Conformity of the calculated ozone profile to measure-
 ments does not appear to be a good criterion for reduction of
 the uncertainty in the calculated  range of ozone depletions.

  Acknowledgments   We wish to thank JoAnn Wadkins, a partici-
 pant in the GISS summer fellowship program, for assistance in  graph-
 ics  and compulations. We also  appreciate the comments of two
 anonymous reviewers.

                        REFERENCES
 Alexander Grant and Company, Report  to the Chemical Manufac-
  turers Association, Washington, D. C, 1985.
 Bemand, P.  P., M A. A. Clyne, and R.  T  Watson, Atomic resonance
  fluorescence and mass  spectrometry  for measurements of the rate
  constant for elementary reactions- O(3Pt) + NO2-> NO + O2 and
  NO + O3 -> NO2 + O2, J.  Chem. Soc.  Faraday Trans. 2, 70, 564-
  576, 1974.
 Blake, D.  R , E. W. Mayer, S. C Tyler, Y. Makide, D.  C. Montague,
  and F S. Rowland, Global increase  in  atmospheric  methane con-
  centrations between 1978 and 1980, Geophys. Res. Lett., 9, 471-4&0,
  1982.
 Butler, D.  M., The uncertainty in  ozone calculations by a  strato-
  spheric  photochemistry model, Geophys.  Res.  Lett., 5,  769-772
  1978.
Cicerone,  R.  J.,  S. Walters, and  S. C. Liu, Nonlinear response of
  stratospheric ozone column  to chlorine injections, J. Geophys. Res,
  SS, 3647-3661, 1983.
Crutzen, P.  J., and L. T. Gidel, A two-dimensional photochemical

-------
7864
                                        STOLARSKI AND DOUGLASS: CHLORINE PERTURBATIONS
  model of the atmosphere, 2, The  troposphenc budgets  of the an-
  thropogenic chlorocarbons, CO, CH4 CHj  Cl, and  the effect of
  various NOX sources on  troposphenc ozone, J  Geophys Res, 88,
  6641-6661, 1983.
Davis, D  D., 3. T Herron, and R. E. Huie, Absolute rate constants
  for the reaction O(3/>) + NO2 —> NO + O2 over the temperature
  range 230-339K, J. Chem. Phyl, 58, 530-535,  1973.
DeMore, W.  B., D  M  Golden, R F  Hampson,  M  J  Kurylo, C. J.
  Howard, J J.  Margitan,  M J.  Molina,  and A. R. Ravishankara,
  Chemical kinetics and photochemical data for use in stratospheric
  modeling, Evaluation Number 7, JPL Pubi, 85-37, 1985.
Froidevaux, L., M. Allen, and Y  L.  Yung, A critical analysis of CIO
   + O3 in the mid-latitude  stratosphere, J. Geophys Re:,., 90, 12,999-
  13,029, 1985.
Keeling, C C., A  F. Carter, and W  G Mook, Seasonal, latitudinal,
  and secular variations in  the abundance and isotopic ratios of at-
  mospheric  CO2,  2,  Results From  oceanographic  cruises  in the
  tropical Pacific Ocean, J.  Geophys.  Res., 89, 4615-4628, 1984.
Khahl,  M  A. K ,  and R. A. Rasmussen,  Carbon monoxide in the
  earth's atmosphere. Increasing trend, Silence, 224, 54-56, 1984.
Krueger, A J., and R A Minzner, A mid-latitude ozone model for the
  1976 U.S  standard atmosphere, J  Geophys. Re?, 81, 4477-4481,
  1976.
Logan, J  A., Troposphenc ozone:  Seasonal  behavior, trends,  and
  anthropogenic influence, J Geophys. Res., 90, 10,463-10,482,  1985.
McPeters, R  D,  D. F Heath, and  P K Bhartia, Average  ozone
  profiles in 1979 from the  Nimbus 7  SBUV instrument, J. Geophys.
  Res, 89, 5199-5214, 1984.
Molina,  M. J,  and  F. S.  Rowland, Stratospheric sink for chloro-
  fluoromethanes. Chlorine atom catalysed  destruction  of  ozone,
  Nature, 249, 810, 1974
National Academy  of  Sciences, National   Research Council,  Causes
  and Effects of Changes in Stratospheric  Ozone-  Update  1983,  Na-
  tional Academy Press, Washington, D C., 1984.
Prather, M. J., M  B. McElroy, and S C. Wofsy, Reductions in ozone
  at high concentrations of stratospheric halogens, Nature, 312, 227-
  231, 1984.
Qumn, T.  H.,  K. A.  Wolf, W  E.  Mooz, J.  K. Hammm,  T W.
  Chesnutt, and S. Sarma, Projected use, emissions and banks of
  potential ozone depleting substances, draft rep. N-2282-EPA, Rand
  Corp., Santa Monica, Calif., 1985.
Rasmussen, R A., and  M A K Khalil, Atmospheric methane in the
  recent and ancient atmospheres Concentrations, trends, and inter-
  hemisphenc gradient, J Geophys. Res, 89, 11,599-11,605, 1984
Rundel, R  D, D M. Butler, and R  S. Stolarski, Uncertainty propa-
  gation in a stratospheric model, 1, Development of a concise strato-
  spheric model, J. Geophy s Res, 83, 3063-3073, 1978
Slanger,  T. G., B J. Wood, and G. Black, investigation of the  rate
  coefficient for O(3P) + NO2  > 02 + NO, Int.  J. Chem  Kmet.,  5,
  615-620, 1973
Solomon, S,  D  W  Rusch, R  J. Thomas,  and R  S Eckman, Com-
  parison of mesosphenc ozone  abundances measured  by  the Solar
  Mesosphere Explorer and model calculations, Geophys. Res Lett.,
  10,249-252, 1983.
Stolarski, R S., and A  R. Douglass, Parameterization of the photo-
  chemistry of stratospheric ozone  including catalytic loss processes,
  J. Geophys Res, 90, 10,709-10,718, 1985.
Stolarski, R S, D. M  Butler, and R D  Rundel, Uncertainty propa-
  gation in a stratospheric model, 2, Monte Carlo analysis of impre-
  cisions due to reaction rates, J. Geophys Res., 83, 3074-3078, 1978
Thompson, A M , and R. J Cicerone, Troposphenc CH4, CO,  and
  OH from 1850 to  1980, Nature, in press, 1986.
Wemslock, E  M , M J Phillips, and J. G  Anderson, In  situ observa-
  tions  of CIO  in  the stratosphere  A  review  of recent results,  J.
  Geophys Res, 86,  7273-7278, 1981.
Weiss,  R. F, The temporal and  spatial distribution of  troposphenc
  nitrous oxide, J Geophys. Res., 86, 7185-7196, 1981.
World  Meteorological Organization, The stratosphere  1981: Theory
  and measurement, W MO Rep 11, Geneva, Switzerland, 1982.
Wuebbles, D  J, A  theoretical analysis of past  variations  in global
  atmospheric composition and temperature structure, Lawrence Li-
  vermore Lab Rep, UCRL 53423,  1983a
Wuebbles, D J , Chlorocarbon emission  scenarios  Potential impact
  on stratospheric ozone, J. Geophys Res , 88, 1433-1443, 1983b.
  A. R  Douglass, Applied Research Corporation, 8201  Corporate
Drive, Landover, MD 20785.
  R  S  Stolarski, Atmospheric  Chemistry and  Dynamics Branch,
Code 616,  NASA Goddard  Space Flight  Center, Greenbelt, MD
20771

                  (Received November 30,  1985,
                     revised March 18, 1986;
                    accepted March 19, 1986.)

-------
                                                                         UCRL- 95548
                                                                         PREPRINT
                           OZONE PERTURBATIONS IN THE
                          LLNL  ONE-DIMENSIONAL MODEL -
             CALCULATED EFFECTS  OF  PROJECTED TRENDS IN
              CFC's,  CH4, C02, N20 AND HALONS OVER 9O YEARS
                                            Peter S. Connell
                                                  and
                                          Donald J. Wuebbles
                                  Prepared as a special report for the
                                   Environmental Protection Agency
                                              March 1986
This is a preprint of a paper intended for publication in a journal or proceedings. Since
changes may be made before publication, this preprint is made available with the
understanding that it will not be cited or reproduced without the permission of the
author.

-------
                               DISCLAIMER

This document was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor the University
of California nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or useful-
ness of any information, apparatus, product, or process disclosed, or represents that
its use would not infringe privately  owned  rights. Reference herein to any specific
commercial products, process, or service by trade name, trademark, manufacturer, or
otherwise, does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States Government or the University of California. The
views and opinions of authors expressed herein do not  necessarily  state or reflect
(hose of the United States Government or the  University of California, and shall not
be used for advertising or product endorsement purposes.

-------
 I. INTRODUCTION

 LA The Nature of the Problem

    Many processes  in the earth's atmosphere and the earth's biosphere are closely con-
 nected. Mutual dependence is demonstrated in the biosphere's contribution to atmospheric
 composition (oxygen, carbon dioxide, nitrous oxide, methane, and methyl chloroform) as
 well as in the vital roles that atmospheric properties such as climate, weather, ultraviolet
 and infrared opacity and abundances of oxygen and carbon dioxide play in the maintenance
 of the biosphere. These interconnections would exist in the absence of man's intervention.
 In addition to 'natural' interactions of the atmosphere and biosphere, human activities can
 significantly alter properties of the atmosphere and by extension, conditions for life in the
 biosphere.
    Human impacts on the atmosphere are  of two major types.  First, trace  species of
 many types are directly emitted into the atmosphere as by-products of various activities,
 including transportation, combustion, energy production, refrigeration, plastic and poly-
 mer production and  use, high-technology industry and personal hygiene. Second, methods
 of land development, agriculture and animal husbandry can  affect the fluxes of naturally
 occurring atmospheric trace species. Atmospheric observations over the last  20 years have
 revealed dramatic increases in the trace  abundances of man-made perhalogenated chloro-
 carbons and chlorofluorocarbons (CFC's).  Observational evidence also shows enhance-
 ment in the abundances of naturally occurring species which are also directly emitted by
 society (e.g., CO2, CH4, CO), over the century time  scale of industrialization and popu-
 lation growth. Enhancement (e.g., tropospheric ozone) or suppression (e.g., tropospheric
 HO, stratospheric ozone) of the abundance of naturally occurring atmospheric constituents
 that are photochemically produced may also be occurring (WMO, 1986). The atmospheric
 changes that can ensue from these perturbations include increased ultraviolet irradiance at
 the earth's surface, photochemical smog, reduced  visibility, acid precipitation and  perhaps
 changing climate.

    An atmospheric  response to anthropogenic emissions that was unexpected as  recently
 as 1970  is  the potential perturbation of the  stratospheric ozone abundance. The diffuse
 layer  of ozone (O3)  in the stratosphere (approximately 12-55 km above the surface) is
 chiefly responsible for controlling the intensity of solar ultraviolet radiation between the
 wavelengths of 210 and 320 nm penetrating  the  stratosphere and in particular the 280-
 320 nm radiation that reaches the ground. Such radiation is responsible for skeletal growth,
 vitamin  D  production, tanning and sunburn in man and exposure to it is correlated with
 various types of skin cancers. There are also diverse and often deleterious effects on plants
 and animals.  Ozone, as well  as several of the species with  which it is photochemically
 connected, is also involved in the earth's radiation balance and thus climate.
    The study reported here investigates the potential perturbation of stratospheric ozone
 on a global basis, in view of current and projected trends of increase in several  anthro-
 pogenically-related atmospheric species. The authors of a recent review of the  current
state  of knowledge of the stratosphere (WMO,  1986)  have  emphasized the wide  range
of uncertainties in both the projections  of trends of trace species abundances  and their
potential effect on ozone in the troposphere and  stratosphere.  Several previous studies

-------
(Brasseur et al., 1985; Callis et al., 1983; Owens et al., 1985; WMO, 1986; Wang et
al., 1986; Wuebbles et al., 1983) have considered the effects on ozone of simultaneous
trends of several important trace gases and have shown the inherent nonlinearity of ozone-
controlling atmospheric dynamics and  photochemistry. But these studies have not, in
general,  attempted to address the full range of trend possibilities. Our primary  interest
in this report is to investigate the extent of the effects of projection uncertainties  and the
nature of their coupling, chiefly with respect to relative changes in the vertically integrated
ozone column.  Special attention is  given to considering trend projections over a range of
uncertainty and to employing emission trends with economic bases, where appropriate.
The choice of CFC trend projections is based on information provided by the EPA and
the current study is also intended to support  the possible development of an international
protocol governing CFC production and release.

I.B Background

    The  existence of a layer of ozone in the upper atmosphere was first demonstrated by
Hartley in 1881. To explain its origin, Chapman (1929) theorized that a region of enhanced
ozone abundance would occur in an atmosphere of oxygen and nitrogen as a result of fission
of the  O=O bond in molecular oxygen by solar radiation. Observation  of the abundance
and behavior of stratospheric ozone continued in the  early part of this century.   As the
kinetic parameters of the  Chapman reactions were measured in the laboratory,  it  became
clear that the actual situation was not completely explained by the reactions  Chapman
postulated to occur in an atmosphere in which atomic and molecular oxygen and ozone were
the only reactive participants. In particular, the observed abundance was considerably less
than predicted by the oxygen reactions, given the rate parameters for these reactions as
measured in the laboratory. Additional ozone loss processes were inferred to exist.
    Bates and Nicolet suggested in 1950 that cyclic processes were involved in the pho-
tochemistry of atmospheric water vapor.  One generic example of a stratospheric cycle
is
                                 X + 03 =  XO + 02                              (1)

                                 XO + O = X + O2
The catalyst X is regenerated by the pair of reactions, while the process operates effectively
to facilitate the net reaction
                                    O + 03 = 202                                 (2)
Catalytic cycles of this type were discussed as important odd oxygen (O and ©3) loss mech-
anisms by Hampson (1964) and  Crutzen (1970), who proposed HO and  NO, respectively,
as the catalyst X in the reactions above.
    With this background, the potential for perturbation of the ozone amount by anthro-
pogenic emissions was quickly realized (Johnston, 1971 and Crutzen, 1971).  The emissions
receiving the most attention from the standpoint of impacts on ozone have been:  (1) in
situ  injection of NOX and HOZ species by aircraft engines (CIAP, 1975; NEC, 1973), (2)
increases in stratospheric  NOZ abundance following indirectly from nitrogenous fertilizer
production and usage (McElroy et al., 1977), (3) increases in the abundance of chlorine-
containing radicals directly through solid rocket motor emissions or indirectly through

-------
production and atmospheric release of chlorinated and brominated alkanes (Stolarski and
Cicerone, 1974; Molina and Rowland, 1974; NASA, 1977; Wofsy et al., 1975), and (4) NOZ
injections caused by large-weapon nuclear fireballs (NRC, 1975 and 1985).
    Folk knowledge of the beneficial eifects of human exposure to sunlight, and by infer-
ence the short wavelength ultraviolet tail of solar radiation, must predate modern science.
Medical study of both beneficial and deleterious effects of sunlight in man began more than
a century ago  and the effects of exposure to ultraviolet radiation have been  studied for
more than fifty years. But it appears that the connection of varying effects of ultraviolet
exposure specifically to consideration of ozone variability, rather than for example general
latitude dependence or cloud and haze obscuration,  arose only with the suspicion that
anthropogenically-caused variability in  ozone was possible. Knowledge  of the effects of
UVA (320-350 nm) and UVB (280-320  nm) exposure on man, vegetation and susceptible
fauna has greatly increased in the last 15 years, although uncertainties in this area are still
large.

    While the potential for depletion of stratospheric  ozone has been the major theme of
stratospheric studies, it is now also understood that increased emissions of certain species
could result in an increase in ozone abundai.ce.  Carbon dioxide, an active infrared  ab-
sorber and emitter in the atmosphere, shows a well-documented  increase in atmospheric
concentration (Keeling et al., 1982). The stratospheric counterpart to warming of  the
surface/troposphere system by CO2's infrared absorption properties is local stratospheric
cooling by emission of infrared radiation  to space. As increasing CO2 cools  the strato-
sphere, the efficiency of ozone-destroying catalytic cycles may be reduced (Luther et  al.,
1977) producing an increase  in ozone. This effect connects the problem of the potential
human perturbation of the climate (surface temperatures and tropospheric  weather) with
that of human perturbation of stratospheric ozone.
    This connection is strengthened by the infrared radiative (climate) importance of ozone
and several of the trace species mentioned above in consideration of ozone photochemistry.
An increase in atmospheric CH4, which appears  well-documented over the last decade
(WMO,  1986)  and may date back a century (StaufFer et al., 1985),  is expected to  affect
ozone in several ways. For example, stratospheric CH4 is oxidized by HO producing H2O
and radicals  (HOO and  CH3OO), which can produce ozone via  reaction with NO. The
water produced in CH4 oxidation remains available as a source species for HOZ radicals.
Methane also provides a  stratospheric sink for Cl atoms, forming  the less reactive species
HC1.  In addition, CH4 has infrared  radiative properties which make it and its observed
increase currently important  in the study of the perturbation of the earth's climate.

I.C Current Status of Research on Stratospheric Ozone

    A continuing aspect of stratospheric ozone research is the identification of the complete
set of participating species and significant kinetic processes. Recent advances in techniques
for the study  of reactions  in the laboratory and improved detection and  monitoring of
stratospheric species from surface, balloon and satellite platforms have greatly expanded
the set to be considered.  More than 50 trace species have been identified in the laboratory
as reactants and products of reactions of stratospheric importance and many of these
species have been observed in the atmosphere.

-------
    The current picture of stratospheric photochemistry is conveniently discussed in terms
of groupings or families of similar species, such as NOZ (NO + NO2), NOy (NOZ + HNO3 +
HO2NO2 + NO3 + 2*N2O5 + HNO2 + C1ONO2)  and C1OZ (C1+C1O), which may include
short-lived highly reactive radical molecules (those with unpaired electrons) or both radi-
cals and longer-lived molecules such as HNO3 or HC1. Other species such as HO2NO2 and
C1ONO2 have been identified in the laboratory and have recently been tentatively observed
in the stratosphere. These species fall between family groupings and represent important
interconnections between catalytic cycles.  Coupling the species involved in competing cat-
alytic cycles  can result in ozone perturbations that depend nonlinearly on emissions of
anthropogenic trace species.  Additionally, as  a  result of the range of temperature and
ultraviolet intensity encountered as  a function of altitude,  different altitude regions are
dominated by  differing regimes of photochemical processes. Large sets of reactions are
thus required to completely describe the processes controlling ozone chemistry as they are
currently understood.
    Two major current emphases in developing a coherent  picture of stratospheric vari-
ables and properties are, first, understanding in  terms of transport and photochemistry
the observed distributions in space and time of ozone, temperature and the suite of other
species and physical quantities that can be observed and, second, the detection of long-
term  trends in ozone abundance, temperature,  source species such as chlorofluorocarbons
and methane and perhaps even active radical species. One method of synthesizing the lab-
oratory photochemical knowledge that has proved advantageous for comparison of theory
to the atmosphere is the comprehensive photochemical/dynamics numerical model. Such
models represent stratospheric processes in one, two or three spatial dimensions in either
an equilibrium or time-dependent fashion.  One- and two-dimensional models can contain
essentially complete, known, laboratory-based photochemical reaction parameter sets. The
results of these models (for example, vertical profiles of species concentrations or seasonal
and latitudinal fields of abundance integrated vertically) can be compared to observations,
with  due  regard given to the effects of assumptions made in reducing the dimensional-
ity  of the model.  This diagnostic use of models can reveal inadequacies  in the current
understanding of photochemistry in the present atmosphere.  Three-dimensional models,
structured to give the best representation of atmospheric motions, are currently unable to
include full elementary photochemistry because of available computational capabilities.
    The effects of past or projected  atmospheric  perturbations can also be studied with
these models.  If all pertinent physical processes and photochemical kinetic parameters were
known and included in the model, then prognoses of potential atmospheric perturbations
could be produced by changing model boundary conditions, for example those representing
projected emission trends. While comparison to the current atmosphere remains an  impor-
tant method of testing  the correctness and completeness of models, it should be noted that
prognostic applications of models can not necessarily be verified by acceptable diagnostic
performance. That  is, two models (differing for example in photochemical kinetic parame-
ters) can achieve equivalent success in modeling current atmospheric observables but may
predict widely diverging future changes in  ozone for a  proposed perturbation.  These un-
certainties  arise from the restricted,  if currently growing, cluster of stratospheric species
that have  actually been observed, the uncertainties in their  quantitative observation, the
lack of complete spatial and temporal coverage of the 'observations and the complexity of

-------
the kinetic system. The confidence level for model predictions, then, is based somewhat
subjectively on completeness and applicability of kinetic, radiative  and dynamic model
components as well as on diagnostic capabilities.

I.D The One-Dimensional Model

    The chief variable in studying the biospheric impact of altered surface ultraviolet radi-
ation in the UVB region is the optical thickness of the vertically integrated ozone column.
As mentioned above, diverse photochemical behavior occurs in the atmosphere as  a result
of the altitude dependence of solar UV flux, temperature and distributions of long-lived
source species.  A  model with at least one spatial dimension, the vertical,  is required to
investigate the  dependence of the ozone column  abundance  on  the  various atmospheric
variables.
    Various one-dimensional (1-D) models have been used extensively since  1971 to study
the photochemistry of the stratosphere. In general, these models account for homogeneous
mono-, bi- and termolecular reactions, photolytic reactions, atmospheric mixing in the
vertical, and various other physical processes. Current  1-D models usually include the
complete set of significant stratospheric species and the kinetic parameters of their reac-
tions, to the extent that laboratory and observational evidence is available. The intensity
of visible and ultraviolet solar radiation is calculated as a function of altitude based on
the model  computed vertical profiles of the important absorbers O3, O2 and NO2.  The
individual  photolytic constants for each species are then calculated  from the laboratory
absorption spectra. Atmospheric dynamics and processes such as removal of species from
the troposphere by precipitation and dry deposition are included as empirical parameter-
izations constrained by observations and understanding of the associated physics in the
atmosphere. Some models also include the coupled calculation of the stratospheric tem-
perature profile, which depends on solar absorption by ozone  and infrared absorption and
emission by CO2, H2O and O3.
    Because the concentrations of many source species are changing and future projection of
source strengths are uncertain, even for industrially produced species like CFC-11 (CFCla),
predictions for future changes in ozone can not at present be closely constrained. Investigat-
ing the possibilities requires model parameter  studies with boundary conditions specified
over the anticipated range of emission scenarios of the source species.  One-dimensional
models have been used for this purpose because they are much more conservative  of com-
puter resources than 2-D models, although attempts are being made at time-dependent
integrations in two dimensions over several decades of model time.  The results of a 1-D
model can  be considered to represent global or hemispheric average behavior for consid-
erations of ozone change. They can give a global indication of the effects of perturbations
taking into account the fine detail of photochemical interactions  but are incapable of rep-
resenting potentially significant variations with latitude or season. The results presented
in this report will  show the  importance of including simultaneous perturbations of the
various trace gas emissions in predicting future ozone change.
    Some limitations and uncertainties are inherent in the 1-D model formulation. Latitu-
dinal variation in the changing ozone column can not, of course, be considered.  Preliminary
study of CFC perturbations  in 2-D models have  shown a strong latitudinal  dependence,

-------
 with ozone depletion occurring preferentially at middle and high latitudes.  In addition,
 rates of reactions computed with global annual average concentrations will differ from the
 actual global average reaction rate for species that are correlated or anti-correlated in space
 or time. The effects of uncertainties in the numerical parameters of processes which are
 included can be estimated by various approaches, including Monte Carlo studies in which
 the  parameters are allowed to  vary  according to the appropriate uncertainties (Stolarski
 and Douglass, 1986; Grant et al., 1986). Uncertainty caused by the neglect of important
 processes not yet identified can not be estimated, but has had a substantial impact in the
 past. The predicted response of the ozone total column to the given standard perturba-
 tion (usually constant CFC emissions at late 1970's levels) has varied significantly as new
 species and reactions have been introduced.

 I.E Focus of This Report

    We report here on the  application of one numerical atmospheric model, the LLNL
 1-D troposphere-stratosphere model, to the investigation of future response of ozone to
 anthropogenic emissions of chlorine- and bromine-containing alkanes (CFC's and Halons)
 against the background of simultaneously changing levels of CO2, CH4 and  N^.1 The
 atmospheric abundances of some industrially produced halocarbons are currently increas-
 ing at measurable rates, which,  if projected to continue for  decades, may reach a point
 of significant impact on ozone. Knowledge of the magnitude of the impact as a function
 of emissions projections is required to evaluate the effectiveness of measures which might
 be implemented to control the rate of release of these species to the atmosphere.  The
 problem is considered in a globally  averaged sense,  with  detailed consideration given to
 photochemical kinetics and radiative processes in the stratosphere. Lesser consideration is
 given to the model treatment of stratospheric motions and no attempt is  made to address
 ozone changes over other dimensions and smaller spatial scales, i.e. latitudinal behavior.
    Many of the trace atmospheric species that are changing in these scenarios,  CFC-11
 (CFC13),  CFC-12  (CF2C12), N2O, CH4, CO2, stratospheric  O3 and  stratospheric water
 vapor, also contribute to the infrared radiative balance in the earth's atmosphere.  The
 atmosphere can be assumed  well-mixed from the surface over the altitude range  that in-
 frared absorption and emission by the long-lived species is important in determining surface
 temperature.  Lacis and coworkers, using the NASA/Goddard Institute for Space Studies
 1-D  radiative convective model (Lacis et al., 1981 and Hansen et al.,  1981), have param-
 eterized the radiative contribution to the global average equilibrium change in surface air
 temperature arising from specified changes in surface mole fractions of CO2, CH4, N2O,
 CFC-11 and CFC-12 and in  the vertical profiles of ozone and stratospheric water vapor.
 The relationships derived do not  include the effects of climate feedbacks on, for example,
 the hydrologic cycle.  Rather the effects  on  surface temperature predicted by these rela-
 tions represent only the pure radiative effect at equilibrium of the particular trace species
 considered. The effect of climate feedbacks can then be estimated from GCM simulations
 in which these are included.  A recent NAS panel study (NRC, 1983), summarizing the
  1  The term halocarbon (or Halon) refers to organic species (in the chemical sense) con-
taining one or more halogen atoms (F-, C1-, Br- or I-). The term chloroflurocarbon (or
CFC) usually refers to halogenated alkanes containing fluorine and chlorine atoms.

-------
current state of knowledge, predicts a feedback amplification factor of 1.25 to 3.75 for the
equilibrium doubled CO2 radiatively forced surface temperature change of 1.2°C  (with-
out feedbacks), corresponding to a total temperature change of 1.5 to 4.5°C. Using these
relationships and estimates of the climatic feedback factors, the emission scenarios, model-
derived CFC and CH4 mole fractions and calculated ozone and water vapor profiles, surface
air temperature changes have been calculated for several of the scenarios. The timing  of
temperature increases depends on many  factors whose discussion is beyond the scope  of
this report.  As with the case of ozone perturbations, major uncertainties, such as the effect
of the thermal inertia of the  oceans, exist in applying both 1-D  radiative convective and
GCM models prognostically.
    The connected issues of stratospheric photochemistry and global climate are considered
as separate pieces in this report. A discussion of the overall impact of increasing trace gas
abundances on climate and the biosphere is beyond the scope of this report.
    The estimates  of future surface emissions of the CFC's and other industrial halocar-
bons (sometimes termed a scenario) are pivotal to the prediction of ozone changes.  Since
this report focusses on ozone  changes in the near future (defined in this report as the next
90 years) rather than the elucidation of stratospheric response to individual perturbants,
we have adopted CFC scenarios based on the work of Quinn et al.  (1985) and ICF, Inc.
as communicated by the U.S. Environmental Protection  Agency (S. Seidel,  private com-
munication). Scenarios with restricted CFC emissions representing the effects of possible
policy decisions  are also investigated to study the possibility of mitigating CFC impacts.
    The projections of the various CFC and other halocarbon emissions are based on indus-
try  figures  and estimates and can be checked against measured atmospheric  abundance if
the atmospheric lifetimes are known (Prinn et al., 1983).  Projections of the future surface
fluxes of other trace species which have a positive trend (CO2, CH4 and N2O) are more dif-
ficult to define. First, the measurements  of current trends can be subject to  uncertainties
resulting from a short time series of observations, problems of global representativeness in
sampling, unusual events such as El Chichon/El Nino which perturb a monotonic  trend
or simply a small trend relative to the average atmospheric abundance. In addition, these
species have multiple sources, both anthropogenic and natural, and the relationship of the
fluxes to their controlling variables is in many cases uncertain. Finally, the magnitude or
capacity of atmospheric sinks, and thus the atmospheric lifetime, can be uncertain.
    For example in the case of CC>2, predictions of future abundance depend on our under-
standing of biospheric and oceanic uptake  as well as factors determining energy demand and
fossil fuel consumption. Trends in the rates of both surface emission and atmospheric de-
struction may be instrumental in affecting methane abundance. Projecting future methane
concentrations depends on understanding surface emissions of CH4 and CO and the at-
mospheric interactions of CH4, CO and the observationally elusive tropospheric  hydroxyl
radical, HO, each of which affects the others. Assumptions made in predicting future levels
of these species  can impact the overall model behavior significantly, as will be shown  in
the discussion of results.
    Several previous studies of ozone perturbation in response to trends in several species
occurring simultaneously have been  reported (Brasseur  et al., 1985;  Callis et al.,  1983;
Owens et al., 1985; WMO, 1986; Wang et al., 1986; Wuebbles et al., 1983).  These studies

-------
have been mainly directed toward understanding atmospheric processes and have employed
simple constant or smooth growth projections for species such as the CFC's. In this study
we consider the time evolution of  the  ozone response to econometrically derived CFG
projections which may or may not be smooth and monotonic.  We consider a wider range
of possibilities, including large increases to the stratospheric C12 (=  Cl +  CIO -f  HC1
+ C1ONO2 + HOC1) abundance in the so-called chlorine catastrophe regime (Prather et
al., 1984). We also investigate how individual uncertainties in the projection scenarios for
each important trace species or species group (e.g., CFC's) combine to affect the calculated
ozone change.
    In  the next section a brief description of the structure, geophysics and photochemical
kinetics in the model is given, followed by a discussion of ozone-controlling mechanisms
in the  model stratosphere and previous results of ozone perturbations. Historical trends
in important trace species  concentrations and emissions and the model-derived impacts
on ozone and temperature  are discussed in Section  III. Section IV presents the emissions
projections used in the study, and Section V presents a detailed discussion of the model
results for an intermediate  reference case scenario.  The sensitivity of the calculated ozone
and temperature perturbations to scenario assumptions and regulatory possibilities is con-
sidered in Section  VI.

II. THE LLNL ONE-DIMENSIONAL MODEL

    The LLNL one-dimensional atmospheric model, used for this study of stratospheric
ozone perturbation resulting from the changing abundances of several trace constituents,
has been developed in over a decade  of research. Improved solution techniques and new
information arising from both atmospheric observation and laboratory investigations have
been evaluated and incorporated into the model originally described by Chang et al. (1974).
The emphasis in development has been on comprehensive and detailed representation of
photochemical kinetic processes and on numerical accuracy, within the limitations of the
reduced dimensionality for the treatment of atmospheric dynamics.  More  discussion can
be found in Luther et al. (1979) and Wuebbles et al.  (1983).

II.A Model Structure

    The model atmosphere is divided into 44 vertical layers, extending from the surface
to just above the stratopause (at about 56 km). The spatial (vertical) coordinate metric
is the Naperian logarithm of the pressure, relative to the model's surface pressure. Some
results in this report are presented on an approximate effective altitude  scale, mapped
from the pressure levels.  The advantage of the  pressure coordinate is simplification of the
model  coupling to the radiative submodel.
    The chemical and physical processes determining the temporal variation in the con-
centration of the ith atmospheric constituent, ct, can be represented in mathematical form
by a differential (continuity) equation,

                  ^ = ~ [K^p^/p)] + P,(c) - L,(c)c, + St                (3)

-------
where t is time, z is altitude, K^(z) is the one-dimensional vertical diffusion coefficient, p is
air density, Pt(c) and L,(c) c, are photochemical production and loss terms of constituent
i, c, represents the concentration and S, represents any other sources or sinks. The vertical
diffusion term is assumed to represent the global average of the net vertical transport flux.
   Transport of atmospheric trace constituents in the 1-D model is an empirical repre-
sentation which is not derived directly from observed atmospheric motions.  Instead, the
temporal and spatial distribution of various tracers is matched by adjusting the form of
the effective  diffusion constant K2  as a function of altitude.  The  tracers considered in
establishing the K2 profile used here (Figure 1) include long-lived species (e.g., N2O, CH4,
CFC-11, CFC-12 and CHsCl) and excess radioactive 14C injected into the atmosphere by
atmospheric nuclear testing.
   The individual species conservation equations are solved simultaneously by means of a
variable order multistep implicit method  (Hindmarsh, 1976).  The main advantage of this
method is its ability to solve the coupled set of differential equations containing a wide
range of characteristic times (stiffness) encountered  in the chemical kinetic terms. The
model  time step  is variable over orders of magnitude in order to assure both numerical
accuracy and computational efficiency.
   The  boundary condition for individual species can be either fixed  concentration or
fixed 
-------
 discussed above, the surface temperature changes calculated in this report are based on
 the parameterized results of such a model, and are not coupled to the transport-kinetics
 model.

 II.B Model Photochemical Kinetics

    Fifty-two species are included in the model. Oxygen (O2) and CC>2 are assumed well-
 mixed throughout the atmosphere at specified abundances. Three of the minor constituents
 (O(1D), H and N)  are assumed to be in instantaneous photochemical equilibrium, so that
 the effects of transport on their concentration profiles can be ignored. The remaining 47
 species,

 0(3P), 03,
 NO, N02, N20, HN03, HNO4, HNO2, NO3, N2O5,
 HO, H02, H202,H2, H20,
 Cl, CIO, HC1, HOC1, C1ONO2, C1NO2,
 CH4, HCO, H2CO, CH3, CH3O2, CH3O, CO, CH3OOH,
 Br, BrO,  HBr, HOBr, BrONO2,
 CH3C1, CC14, CF3C1 (CFC-11), CF2C12  (CFC-12), CHF2C1  (CFC-22), CFC12CF2C1
  (CFC-113), (CF2C1)2 (CFC-114), CF2C1CF3  (CFC-115), CH3CC13,
 CH3Br, CF3Br (Halon 1301), CF2BrCl (Halon 1211), and C2H4Br2 (EDB),

 are solved using Equation (3) for each species.
    The 165 chemical and photolytic reactions contained in the model for this study are
 listed in Table 1.  The kinetic  rate parameters are based on the recommendations of the
 NASA Panel for Data Evaluation (JPL  83-62, 1983).  Literature values are used for the
 few reactions  for which recommendations were not made.  We have chosen the  recent
 experimental results of Smith et al. (1984) for the rate parameters of the reaction

                         HO + HN04 = H20 + N02 + 02,                      (4)

 while the NASA recommendation is derived from an average of this and earlier reported
 results.  By analogy to the current recommendation for  HO+HNO3, we feel the Smith et
 al. value is more likely to be correct.
    The solar flux, incident at  the top of the atmosphere,  used  in the  model is consistent
 with the recommendations made in WMO  (1982). The solar zenith angle in the model is
 fixed at 30 degrees representing approximate global average conditions at equinox.
    Photodissociation rate coefficients are computed at each vertical level and time step
 consistent with the specified solar conditions including the effect of  multiple scattering
 and cloud cover (Luther et al., 1978) and the calculated distributions of  the significant
 absorbers O2,  O3 and NO2. The absorption coefficients and product quantum yields used
 are  discussed in Connell and Wuebbles (1983) for most of the species.
    Many species  concentrations as well as photodissociation rate constants have a di-
urnally varying component.  The  average rates for reactions involving these species are
computed from consideration of the overlap of the curves of diurnal variability, which are

                                        11

-------
computed in a model including the full diurnal variation of insolation and are assumed to
remain constant throughout a run.

II. C Ozone- Controlling Photochemistry in the Model

    Ozone is the dominant component of the odd oxygen family, Oz, in the stratosphere,
which also includes oxygen atoms in ground and excited  states.  Ozone and O atoms are
rapidly interconverted by the reactions

                           03 + hv = O2 + O                                    (5)
and can be considered equivalent when calculating production and loss rates for ozone.
Odd oxygen is formed primarily in the photolytic decomposition of molecular oxygen by
solar radiation of wavelengths less than 242 nm,

                                  02 + hv = 2 O                                (6)

with a maximum production rate around 40 km in the 1-D model, representing the maxi-
mum in the product of solar short wavelength UV flux and O2 concentration. Ozone pro-
duction in this region is largely balanced  by local destruction processes, but some ozone
is transported both up and down. The  net source region for odd oxygen  in the model
atmosphere is between 25 and 35 km.  The maximum ozone number density occurs  at
23 km, where both local production and transport from above contribute. The maximum
mole fraction of ozone occurs at 31 km.
   Odd oxygen  production by another  path, requiring only near UV radiation which
penetrates to the surface, also contributes to ozone abundance in the model's troposphere
and lower stratosphere (0-16 km).  In this  region,  the oxidation of methane produces
peroxy radicals,  CH3OO and HOO, which can oxidize NO  to NO2.  When the NO2 is
photolyzed
                                   + hv = NO + O                             (7)
the oxygen atom produced combines with O2 to form 03.  Nitric oxide is catalytic in this
process but the CH4 is consumed.
    In contrast to the simple picture for odd oxygen production, odd oxygen loss occurs
by many paths. The direct reaction of O with O3, first proposed by Chapman (1930)
to be important in the  stratosphere, proceeds in the absence of trace catalytic species
and  increases  in rate with altitude, as the O atom  abundance increases.  Other more
important loss processes involve members of the NOZ, C1OI5 BrOj and HOZ families in
several catalytic cycles.  Two major cycles are, schematically,

                                X + 03 = XO + 02                             (8)

                               XO + O3 = X + 2O2
                                        12

-------
    and
                                 X + 03 = XO + 02                              (9)

                                 XO + O = X + O2
 where X in the first cycle can be HO and in the second cycle NO, H, HO, Cl or Br. Fluorine
 does not participate in catalytic photochemistry. Unlike the C-C1 or C-Br bonds, the C-F
 bond is sufficiently strong that  F atoms can not be liberated by radiation reaching the
 stratosphere.  Also the HF molecule is more stable than HC1 or HBr, so that any atomic
 fluorine present is quickly tied up in the form HF. Other cycles of lesser importance can
 also be postulated involving species such as NO3 and HOC1.  The relative importance of
 these cycles in the model of the  current atmosphere is discussed below.
    The sources of these catalytically reactive trace species are the various long-lived sur-
 face emitted gases, N2O,  H2O,  CH4 and perhalogenated Cl-  and Br-containing alkanes.
 NO and HO are liberated from N2O and H2O, respectively, in the stratosphere by the
 reactions
                               N2O + O(1D]  =2 NO                           (10)
 and
                               H20 + 0(1D) = 2 HO.                           (11)

 Methane can also react directly  with O(*D) to form HO, as well as producing H2O as an
 oxidation product. O(JD) is produced by photolysis of ozone at wavelengths less than about
 315 nm.  Chlorine and bromine are liberated from their respective sources by photolysis at
 mid and short UV wavelengths modeled as, for example

                           CFCl3(CFC-ll) + hv = 3Cl.                       (12)

 Reaction (12) does not represent a true elementary photolytic  process, but the set of
 reactions that liberate each Cl atom are assumed to be simple and rapid. The remaining
 fluorine-containing fragments are ignored for the reason mentioned above.
    NOZ, ClOj and BrO2 are  eventually lost from the stratosphere  by conversion to  less
 reactive species which  diffuse down into the troposphere where they are removed by  wet
 and dry deposition processes.  For NOZ, loss occurs chiefly through HNO3, formed by the
 reaction
                           HO + NO2 + M = HNO3 + M.                       (13)
 For C1OZ and BrOz, loss occurs through formation of HC1 and HBr,

                              Cl + CH4 = HCl + CH3                           (14)

 and
                              Br + HOO = HBr + O2.                           (15)
 These loss  processes depend on  interactions of the various families and are an example
of how changes in a given source species can affect several aspects of the photochemistry
 (e.g.,  increasing CH4 will increase HOZ and decrease NOZ by producing more HNO3).
Bromine has been identified as a more efficient catalyst for odd oxygen loss than chlorine,
even though the reaction of BrO with O atoms, the rate-determining step in the cycle,

                                        13

-------
is slower than the corresponding reaction of CIO. The greater efficiency for BrOj cycles
results from the inability of Br atoms to abstract hydrogen from CH4 as well as the lesser
stability of HBr than HC1 with respect to photolysis and reaction with HO. BrO makes
up a larger fraction of the total bromine available than does CIO of the total Clz.  Future
increase in the stratospheric bromine radical abundance will affect the chlorine chemistry
through the reaction
                            BrO + CIO = Br + Cl + O2.                         (16)
    Other interactions among families can also have significant impacts  on the ozone-
related photochemistry.  Three examples of  reactions that  have caused changes in the
theory of ozone depletion are
                             HOO + N0 =

                              ClO + NO = Cl + NO2,                          (17)

and
                              CIO + NO2 = CIONO2.
Consideration of the first two reactions shows that they short circuit catalytic cycles for
odd oxygen destruction for both families involved, effectively reducing the catalytic effi-
ciencies. The discovery that the reaction of HOO with NO was substantially more rapid
than previously thought (Howard and Evenson, 1977) greatly reduced the ozone depletion
expected from supersonic transport NOZ emissions in models at that time. The third re-
action serves to sequester some of the active radicals in both the NOZ and C1OZ into a less
reactive form, making them unavailable for the catalytic destrcution of ozone. The reac-
tion of BrO with CIO, cited earlier, completes an odd-oxygen catalytic loss cycle for both
bromine and chlorine, increasing the efficiency of both in the lower stratosphere where O
atom concentrations are low.

II.D  Previous Ozone Perturbation Results

   While the overall model structure has remained fairly constant since 1975, the set of
species included and the kinetic parameters have changed  continually as new information
has become available. The result has been great variation in the predicted CFC impact
at steady state on ozone column abundance. Figure 2 shows the evolution of computed
results from a standard CFC surface flux assumed constant at levels characteristic of the
mid  or late 1970's.  The ozone depletion calculations in this figure were made with CH4,
COg and NjO fixed at present day surface mole fractions  and usually do not include the
effects of temperature feedback.
   From 1975 to 1981, calculated changes in the total ozone column resulting from the
CFC perturbation at steady state varied from about -20%  to about -4%. These variations
in estimates can be  traced primarily to three related sources. First, rate parameters for
the reactions involving chlorine  chemistry  have changed,  with conflicting impact on the
sensitivity of ozone to chlorine perturbations. The experimental evidence for the higher of
two measured rates of C1ONO2 formation had the greatest impact, reducing the efficiency
of both  Cl and NOZ perturbations and significantly  reducing the depletion estimates.
Second, the computed abundance of HO and the relative importance of NOZ cycles to odd

                                         14

-------
 oxygen loss also contributed to the variation in the CFC-caused ozone perturbation. When
 higher HO abundance is calculated, Cl is freed more rapidly from the inactive form, HC1,
 so that changes that promote higher HO abundance increase CFC perturbations. Current
 models of photochemistry tend to produce lower HO concentrations than prior calculations,
 which are also consistent with available observational evidence. Third, models in which
 the importance of NOz-based cycles is greater lend to exhibit smaller effects on ozone
 from CFC perturbations. The photochemical interactions of C1OZ and NOZ compensate
 increased ozone loss from C1OZ by reducing the efficiency of ozone loss through NOZ.
    Perhaps the most significant recent change in stratospheric photochemical modelling
 incorporated into this study concerns the  optical absorption properties of  oxygen.  The
 absorption cross sections of O2 in the Herzberg continuum region between 195 and 242 nm
 have been  reduced over the values previously employed in 1-D models, as  a result of in
 situ stratospheric measurements of solar flux (Herman and Mentall, 1982; Frederick and
 Mentall, 1982) and subsequent laboratory reevaluation of 03 absorption (Johnston et al.,
 1984; Cheung et al., 1984). Radiation in this wavelength region is thus allowed to penetrate
 more deeply, increasing the rate of photodissociation of species such as the CFC's, N2O
 and HNOs- The role of upper stratospheric ozone in controlling mid-stratospheric UV flux
 at these important wavelengths is also increased as the contribution of oxygen is decreased.
 The coupling of upper stratospheric ozone  depletion to mid stratospheric photochemistry
 is thereby increased.
    If CFC-11  and CFC-12 emissions are  projected to  be constant in the  future, with
 other trace species boundary values fixed  at current levels, the present model (Table  l)
 computes a 5.8% depletion of the vertically integrated  ozone column abundance after 400
 years (essentially the steady state value). Assumed emission rates are 264.5 million kg per
 year for CFC-11 and 412.2 million kg per  year for CFC-12.  Updating the  model to the
 recommendations of JPL 85-37, for which the kinetic parameter changes are listed in Table
 2, we calculate a depletion of 7% for slightly different constant emission rates  of 309 for
 CFC-11 and 433 for CFC-12 in kg per year. (See WMO, 1986, for comparisons with other
 current models). The effect of these updated recommendations on the results of a multiple
 species  scenario are discussed below.  The  calculated depletion is increased  compared  to
 the model,used in this study.

 III. THE CURRENT ATMOSPHERE

    Any human activities that would significantly increase stratospheric  abundances  of
 HOX, NOZ, C1OZ or BrOz have the potential to bring about perturbation of  stratospheric
 ozone.  Various sources (e.g., stratospheric aircraft emissions of water and  NOZ, use  of
 fertilizers and solid propellant booster rockets) have  been evaluated for their potential  to
 increase the rate of the corresponding  catalytic ozone destruction cycles (WMO, 1986;
 NRC, 1982).  Increases in halocarbon emissions and  the resultant stratospheric C1OZ en-
 hancement are the subject of this report. However, given the photochemical complexities
 outlined above, even relatively small trends in  source species, such as N2O  or CH4, that
 may or  may not be directly or indirectly anthropogenic, can  affect ozone response to di-
rect perturbations of other species (e.g., CFC's). For this reason, projected emissions and
trends of all important source species must be considered as coupled in studying the ozone
impact.

                                         15

-------
    In this study we have investigated the effects of nonzero trends in CC>2, N2O, CH4 and
Halons  1211 and 1301 (CF2BrCl and CFSBr) on the ozone response to the central CFG
perturbation. The halocarbons CFC-11, CFC-12, CC14, CH3CC13, CFC-22 and CFC-113
are considered in projecting future Cl emissions. There is strong observational evidence
for continuing increasing trends for each of these species at present.
    Two other species, CO and tropospheric NOZ, may be changing as a result of human
activities with  impacts on ozone  and other species, but the observational cases for their
trends are much weaker as is knowledge of source  arid sink strengths.  There is some
evidence that CO has shown an  average annual increase of a few per cent  over the last
several years at Cape Meares, Oregon (Khalil and Rasmussen, 1984)  and over 30 years in
the Swiss Alps (Rinsland and Levine, 1985).  Increases in tropospheric NOZ  are probable
as a result of combustion  of fossil fuels in the developed world  (Logan, 1983), biomass
burning  in tropical regions (Crutzen et al., 1979) and increasing air  traffic (Liu  et  al.,
1980; Wuebbles et al., 1983).
    The reaction of CO with tropospheric HO is an important loss term for HO, which plays
an important role in determining the lifetimes of many trace tropospheric constituents.  An
increase  in CO leads directly to a decrease in tropospheric HO, all else being equal. CO
oxidation can also contribute to ozone formation in the troposphere and the lowest part of
the stratosphere, if sufficient NOZ is available. In the troposphere NOZ acts  as  a catalyst
for ozone production, so that increases in  NOZ emissions could increase the  tropospheric
contribution to the global ozone column, if the NOX is dispersed into the global troposphere
away from the localized source regions.
    These species, CO and tropospheric NOZ,  have been assumed  in  the present study
to have no trends in  source strengths; however, tropospheric concentration changes can
occur as a result of changes following from other perturbers.  Both  of  these species have
relatively short  atmospheric lifetimes and high variability, severely limiting the applica-
bility of 1-D models in simulating their behavior. Additionally, for long-term projections
only the effects of emissions occurring  within the previous few years  of the time of in-
terest in the future would be significant.  Anthropogenic emissions of  CO and NOZ are,
however, closely connected to processes, such as fossil fuel combustion,  with evident long-
term trends. Projecting the effects  of these emissions ahead 90 years in a 1-D  model on
the basis of current knowledge would be very difficult, so the fluxes have been fixed  at
current  levels (that is, the surface flux boundary condition that  reproduces the current
measured atmospheric abundance).  Tropospheric ozone constitutes only about 15% of the
total column, so that any errors introduced should be minor, at  least for cases of signif-
icant stratospheric ozone perturbation.  However, effects on surface air temperature from
changing tropospheric ozone may be significant as a result of tropospheric ozone's greater
role in infrared radiative transfer.
    The potential for substantial stratospheric NOZ emissions from large commercial fleets
of supersonic transports (SST's) operating at 17-20  km altitudes has also not  been con-
sidered in this study.  Because of the strong coupling of the NO, and C1OZ cycles, SST
emissions could significantly alter the response of ozone to CFC increases. The few Con-
corde SST's now flying should have an insignificant effect on the scenarios considered here.
                                         16

-------
III.A Historical Emissions

    Scenarios projecting emissions into the future are started from a common initial con-
dition of species concentration profiles, representing the 1985 atmosphere. These profiles
are generated by integrating the time-dependent model over the past  135 years with his-
torically  based emission or concentration boundary conditions for the trace species CO2,
CH4, N2O, CFC-11, CFC-12, CC14,  CH3CC13, CFC-22, CFC-113, and Halons 1211  and
1301.  The initial conditions for the model integration of historical emissions are obtained
from integrating the model to steady  state with trace species concentrations fixed at prein-
dustrial (1850)  values, estimated either from direct or  indirect observational  evidence or
by backward extrapolation of observed trends with knowledge of source  and sink  terms.
The evidence, interpretations and current understanding of the reasons for trace species
trends are  developed more fully in WMO (1986).  (See also Wuebbles et al., 1984, for
specific background material used in  the present study). The choice of 1850 as the start of
a 135 year period to represent the period of anthropogenic perturbation of the atmosphere
is loosely based on population  growth and industrialization. Earlier start times for this
period could  have been chosen with little effect on the model results.
    A good historical record is also available for CO2  atmospheric abundances. CO2 is es-
sentially inert photochemically  (though radiatively active) in the troposphere and strato-
sphere and the  resulting long lifetime produces a well-mixed distribution with altitude
over the  domain of the model.  It is  cycled through the natural biosphere by respiration,
biological decay and plant biomass production and is taken up by solution in the oceans,
where  it  is  used to produce skeletal materials by various organisms. A continuous record
of the atmospheric mole fraction of CO2 at Mauna Loa, Hawaii beginning in 1958 (Keeling
et al.,  1982) shows a continuing increase which appears to be the result of anthropogenic
inputs from fossil fuel combustion and changes in land use (e.g., extensive biomass burning
in clearing forest).  Air occluded in the icepack in Greenland and at the South Pole  during
the 19th century has been obtained from stratigraphically dated ice cores and analyzed to
contain around 280±10 ppm CO2 (Oeschger et al., 1982). The current value of 345 ppm
CO2 in 1985 represents an increase  of about 25% over the apparent preindustrial value.
These  data sets were combined with  simple exponential best fits to produce the following
piecewise continuous function:

                               \ 270. exp[0.00141(< - 1850)],  1850 < t  < 1958     1
 C02(mole fraction in ppmv) =  j ^ + ^ exp[QQig{t _ 195g)]>  1Q58 < , < 19g5 )
                                                                                (18)
    Methane surface emissions  appear to be primarily of biological origin, both natural
and anthropogenically influenced (WMO, 1986).  The relative importance of identified
sources,  including  rice paddies, bogs, termites, biomass burning and cows, is in dispute,
but the observed trend of increase appears  largely reconcilable with population increase
and accompanying changes in land use (Thompson and Cicerone, 1985; Levine et al., 1985).
Continuous monitoring of CH4 extends back to only 1977 and shows an increase of around
1% per year, averaged over the various available data sets (Blake et al., 1982; Rasmussen
and Khalil, 1981).  Sporadic  direct  CH4 observations  extend back to 1948 and ice  core
data are available  for the pre-  and early industrial period.  A constant  value of around

                                         17

-------
0.7 ppm on a volume basis is indicated by the ice core data that predates population and
industry growth.  Methane measured in more recently occluded air  and a reanalysis of
1951 atmospheric spectra, giving a value of 1.14±0.08  ppm (Rinsland et al., 1985),  are
consistent with a long-term increase, reaching the present value of about 1.7 ppm.  The
historical trend used in this study, the following simple exponential fit for the surface CH4
concentration based on the approximate average of the various  data  sets, was developed
before firm ice core evidence was available and before the work of Rinsland and coworkers
(1985):
            CH4(mo\e fraction in ppmv) = 1.0 + 0.65 exp[0.035(* -  1980)].         (19)

The historical CH4 changes used in this study (from about 1 ppm in 1850 to 1.28 ppm
in 1951 to the present value) are smaller than those given in the more recent studies, but
the effects on the 1985 initial condition atmosphere used for the projected emission studies
should be small.  Results on calculated surface temperature effects in the past  reported
below will, however, be somewhat underestimated.
    Nitrous oxide is emitted as a byproduct of nitrification and denitrification by soil
bacteria, but combustion processes appear  also to contribute to  global emissions (WMO,
1986).  An extensive data set of surface tropospheric N2O observations exists covering the
last decade, showing an average annual increase of about 0.25% (Weiss, 1981). Air samples
from,,1961 and 1964, analyzed for N2O, support this trend level back  to that time. Weiss
(1981) has proposed that the trend of increase results from fossil fuel combustion, such that
the preindustrial concentration, about 285  ppb by volume, was about 5% lower  than the
current level of about 305 ppb.  A preliminary analysis of ice core samples is now  available
which  indicates a preindustrial value of around 275 ppb.  The historical trend used here
was developed on the basis of the discussion by Weiss, and is again represented by the
following simple exponential expression for the  surface concentration (or mixing  ratio).

            JV2O(mole fraction in ppbv) = 285. + 14.0  exp  [0.04(* - 1978)]        (20)

    The historical trends and model boundary  conditions for the species discussed above
are based on observations of surface atmospheric concentration or mixing ratio.  While
evidence of this  sort is available for the industrially produced halocarbons (continuously
since 1978  with  various  earlier measurements), estimates of historical  surface emissions
based on production data and estimates  are also possible. These historical emission esti-
mates  (CMA, 1982 and OECD, 1981) are used in the model as boundary conditions for
the chlorine-containing halocarbons  other  than the naturally produced species, CH3C1.
CC14 emissions commence in about  1910, CFC-12 in the early 1930's, CFC-11 in the mid
1940's, CFC-22 and CH3CC13  in about 1950 and CFC-113 in 1960.  The emission profiles
are characterized by initially steep rates of increase that decline as significant emissions
levels are reached, roughly over the last 25 years. CFC-11 and CFC-12 are commercially
attractive as refrigerants, blowing agents in foam and aerosol spray propellants. By the
mid 1970's, use as aerosol propellants constituted about 60% of the total emissions (CMA,
1982),  but since  the restrictions on aerosol uses in 1978 and as a result of slowing global
economic growth, net emission of chlorine in the form  of long-lived industrial halocarbons
has remained relatively constant. Nonaerosol uses have increased to 65% of total emissions,
currently growing at an annual rate of about 6%. The observed atmospheric increases in

                                         18

-------
the chlorine-containing haloalkanes, with the possible exceptions of CFC-12 and CC14, can
be more or less reconciled with industrial production inventories, considering the appropri-
ate conditions for release of the individual species produced into the atmosphere.  Eastern
bloc emissions could account for the apparent discrepancies of production and atmospheric
abundance increases.
    Only recent  atmospheric observations are available for the long-lived bromine species
and industrial production in quantity has also commenced only recently. The small atmo-
spheric mole fractions that are observed at present were included as boundary conditions
for the last 35 years of the historical model integration to present. This assumption prob-
ably results in an overestimation of bromine abundance, but the present levels produce an
insignificant impact on ozone.

III.B Effects of Historical Emission Trends and the 1985 Model Atmosphere

III.B.I Historical ozone trends

    Figure 3 shows the relative change in the vertically integrated column of ozone over
the 135 year historical period considered in this study, an increase of 1.4%. The  increase
is mostly  the result of the increases in CC>2 and CH^  abundances. The actual long-term
trend in the globally averaged vertically  integrated ozone column is unknown; that is, the
uncertainties in the accuracy of ozone monitoring over this period are much larger than any
observed trend.  The actual long-term behavior of stratospheric ozone has also been affected
by factors and emission sources not included in this model (e.g., solar variability and the
timing and magnitude of atmospheric nuclear testing). These effects and their impact are
discussed  in detail  by Wuebbles (1983).  The purpose here in the historical integration is
to produce internally consistent initial conditions for the emissions projection model.
    The evolution of the ozone profile referenced to the preindustrial (1850) profile over
the historical integration is shown in Figures 4a and 4b.  The small wiggles most evident
in Figure  4a and other figures in  this report at 13 and 34 km are computational artifacts
of the temperature calculation and should be ignored.   Absolute  changes in the model
ozone  column are  seen to be dominated by the methane-related  ozone increase in the
lower stratosphere  (Figure 4a).  A  computed decreasing trend in the upper stratosphere
appears in about 1960 as the C\z abundance first  reaches a level of  significance (Figure
4b). Whether these calculated upper stratospheric decreases have been observed is a topic
of current study. A statistical study of  the observations of the network of ground-based
Umkehr ozone profile  observing  stations (Reinsel  et  al.,  1984) has  derived statistically
significant linear trends in ozone abundance between 30 and 40 km. The derived negative
trends of  about  3% over the period 1970-1980 are  consistent with the  1-D model results
of Wang  et al.  (1986),  which include factors  such as the 11-year solar variability and
atmospheric nuclear test series (Figure 5).  Whether the short time record of the ozone-
observing satellites is capable of yielding a statistically significant trend over their period
of operation is a question complicated by instrument performance considerations.
                                         19

-------
III.B.2.  Historical temperature trends

    The radiative impacts at equilibrium that the changing trace gas abundances between
preindustrial times and 1985  would have on surface temperature, as computed by the
NASA/Goddard  1-D radiative-convective model, can be computed using the parameter-
izations briefly discussed  above.  The increase  in CO2 is responsible for around 70% of
the calculated equilibrium surface temperature  increase without feedbacks (referenced to
1985 in  Figure 6) of 0.63°C between 1850 and 1985.  The equilibrium surface air temper-
ature increase of  0.79°C calculated by Ramanathan et al. (1985) for the preindustrial to
1980 time period  with a full 1-D radiative-convective model is consistent with this value,
if allowance is made for their  treatment of relative humidity and surface air vs.  surface
temperature. Other trace  species trends and the model calculated ozone and stratospheric
water vapor profile changes contribute the remaining 30%, with the largest contribution
from CH4 (16%).  These values are not intended  to represent historical reality, since the re-
alization of temperature change depends on many factors other than equilibrium radiative
forcing,  such as the thermal lag time of the oceans.  At  present, considering the current
trends in the trace species, differential increase in CC>2 contributes about 50% to the total
calculated surface temperature increase, CH4 
-------
although it seems to point to defects in the photochemical kinetics (Froidevaux et al., 1986).
However, a Monte Carlo analysis of kinetic uncertainties in the model (Grant et al., 1986)
generates a standard deviation of around 40% for upper stratospheric ozone abundance
from the known kinetic uncertainties, so  the  model is not necessarily inconsistent with
observed 40-50 km ozone abundances. In a similar study, Callis et al.  (1985) noted that
modifying the model chemistry within its uncertainties to better match satellite ozone
observations produced other difficulties in comparison with the available satellite data.
    The calculated profiles of N2O, CH4, H2O, CFC-11 and CFC-12 are shown in Figures
8a-c.  Also indicated are representative observational ranges from WMO (1982 and 1986)
for N2O, CH4 and H2O.  Observational data  for CFC-11 and CFC-12 were taken  from
Fabian (1986). All but CH4 have relatively long lifetimes, 50 years or greater, and are well-
mixed in the model troposphere. The calculated methane lifetime, 10  years, is also long
enough to allow CH4  to become well-mixed in the troposphere. Again with the exception
of CH4,  the shapes  of the curves in the  mid- and upper  stratosphere (the  decrease in
mole fraction) are controlled by the  magnitude of the eddy diffusion coefficient assumed
in the lower stratosphere and by the depth of penetration of short wavelength UV flux.
Reaction with HO and Cl provide the stratospheric sink for  CH4.  The Kz (diffusion or
vertical transport  coefficient)  profile used is partly based on comparisons with observed
N2O and CH4 profiles, so the  agreement of the profiles of these species to observations is
not completely independent of the model assumptions.
    Model profiles of the abundances of several NOa species are shown in Figures 9a and
b. Profiles of CIO and HC1 are shown in Figures lOa and b.  In each case, the model values
are not  inconsistent with the  available observations.  As will be  shown later, the ozone
response to high C1OZ levels is sensitive to the mid-stratospheric total  NO^ predicted by
the model. The model employed here develops a maximum value of about 20 ppb of NO2 at
35 km. Several contributors to NO2, including HNO4, and N2Os, have  not been observed
quantitatively, although N2Os was recently  identified in spectra obtained from a shuttle
experiment  (Toon et al.,  1986).  A  value around 18-23 ppb for  total  mid-stratospheric
NO2 can be inferred from the measured species and calculated relative abundances for the
others (WMO, 1986 and  Callis et al., 1985 and 1986).
    The computed 1985 stratospheric temperature profile is compared to the U.S. Standard
Atmosphere (1976) profile in Figure 11.  The increase in temperature  with increasing
altitude in the stratosphere results largely from the greater Os^solar heating. Many of the
reactions controlling ozone abundance in the stratosphere exhibit a temperature dependent
rate constant, so that changes in the ozone  profile and thus local temperature feed back
into the loss terms in the continuity equation for O3.
    In the model of the 1985  atmosphere, NOZ is the dominant catalyst for odd oxygen
destruction, accounting for 36% of stratospheric loss (Figure 12). Catalytic cycles involving
HOZ and Ox reactions account for 27 and 14%, respectively, of stratospheric odd oxygen
loss, occurring mostly in the upper stratosphere. HO2 increases in the lower stratosphere
(for example as a result of CH4 increase) actually increase  ozone through interference
with the distribution of  species in the NO2 family.  By shifting NOa into the  inactive
HNO3 form, NOZ  catalytic rates are reduced. A similar effect follows from small C1OX
increases in the lower stratosphere, through formation of C1ONO2. HOO and CIO also

                                        21

-------
oxidize NO, interrupting the NOZ catalytic cycle and reducing its efficiency.  This negative
feedback depends on the net relative importance of NOx-catalyzed ozone destruction.  If
stratospheric NO2 abundance were smaller, the compensating effects would also be smaller.
This behavior can be seen in some of the results discussed below.
         has a share of 22% in ozone destruction in the 1985 model atmosphere. The total
abundance of Cl-containing species in the model's upper stratosphere is 2.3 ppb. Bromine
based catalytic cycles account for only 0.2% of the odd oxygen  loss in  1985, occurring
mostly between 20 and 30 km. Finally, about 1% of the stratospheric ozone production is
transported downward across  the tropopause, eventually to  be lost by dry deposition at
the model surface.

IV. PROJECTED EMISSIONS

    In the discussion above, CFC-11 (CFC13), CFC-12 (CF2C12), CC14, CH3CC13, CFC-
22 (CHFC12), CFC-113 (CF2CFC12), Halon 1211 (CF2ClBr), Halon 1301 (CF3Br), CH4
and N2O have been identified as significant or potentially significant sources for reactive
photochemically-generated stratospheric species responsible for determining the loss rates
for stratospheric odd oxygen and several are important in infrared radiative transfer. CO2
is  a direct participant in establishing  both tropospheric and  stratospheric temperature
profiles and indirectly affects the kinetic rates of important stratospheric reactions. Addi-
tionally, atmospheric observations of the abundances of these compounds, available over
periods  of varying length, have demonstrated upward trends continuing to the  present.
Continuation of these observed trends for a period on the order of a century would  re-
sult in a substantial increase in the abundance of stratospheric chlorine and increases in
stratospheric water vapor, HOy and NOy. Future increases in source species are here pro-
jected for periods up to 90 years, using the LLNL time-dependent 1-D model to evaluate
concomitant ozone and temperature changes.

IV. A Halocarbons

    Projected production and release values are based on the analysis of Quinn et al. (1985)
for most of the fluorocarbons and the Halons and work by  IGF, Inc. (S. Seidel, private
communication) for CFC-22 and in some cases for CFC-11 and CFC-12. The U.S. Envi-
ronmental Protection Agency supplied a series of halocarbon scenarios interpolated across
the range of possibilities developed in  the original studies.  It is  important to note that
RAND (Quinn et al., 1985) claims no  a priori economic reasons to rank the RAND sce-
narios among themselves  with respect  to likelihood. The presumption is, however, that
the range of scenarios covers a significant portion of the range of probable future emission
curves, developed as they are from economic growth assumptions varying from purpose-
fully pessimistic (slow growth of world economy) to purposefully optimistic  (rapid growth
of world economy). The range of halocarbon scenarios investigated in this study actually
exceeds the range of the seven RAND scenarios. The authors  of WMO (1986) rightly point
out the extreme uncertainties incumbent on such predictions over periods extending to 90
years. A particular RAND scenario (Later Market Maturation Scenario VI)  was picked by
the EPA, in conjunction with projections for other species discussed below, as a reference
scenario. This designation is not intended to imply anything about  its probability within
the full range of RAND scenarios or within the range of scenarios in this report.

                                         22

-------
    For the major chlorofluorocarbons, CFC-11 and  CFC-12, projections  over 90 years
 were broken into three segments. For the period 1985 to 1990, extensive surveys, current
 market trends, industry projections and industry experts were consulted, with a concensus
 choice of an average annual trend of about 5%.  Econometric analysis was applied to
 the  period  1990  to 2040.  From the historical behavior of the US CFG market, Quinn
 et al.  (1985) derived elasticities  as a function of stage of market growth.  The elasticity
 is denned as the relative growth in CFC use for a 1% increase in GNP per capita.  For
 example, the elasticity in the nonaerosol US market over recent years is about 3; that is, the
 production  of CFC's to meet nonaerosol  needs increased three times as fast as per capita
 GNP. Estimates  of population and GNP growth for the various world regions were then
 combined with the life cycle model, involving assumptions concerning the stages of market
 maturity and  the elasticities to produce  estimates of annual growth worldwide.  For the
 designated reference scenario, Quinn et al. (1985) estimated worldwide growth in the period
 1990-2000 to average 5.0% annually for CFC-11 and 3.2% for CFC-12. Between 2000 and
 2040 emissions were projected to increase 2.4%  annually for both species. Emissions after
 2040 were assumed to increase at an annual rate of 1.7%, based simply on projected growth
 in GNP per capita (unit elasticity).  This reference case (Figure 13) is intended to represent
 healthy development of the CFC markets and approximates a 2.5%  compounded annual
 increase in CFC-11 and CFC-12 emissions over the 90 year period considered, an increase
 of a factor  of 9.2 over 1985 fluxes in  90 years. The  exact figures used are compiled in
 Appendix A.
    Information on CCL, (the CFC feedstock), the solvents CH3CC13 and  CFC-113, the
 mobile refrigerant CFC-22 and the fire extinguishents  Halons  1211 and 1301 is scarcer
 and  projections more difficult. Where sufficient information was available, elasticities were
 evaluated, otherwise unit elasticity was assumed. The annual percentage rates of increase
 used in the  reference case were:

                         Global Annual Rates of Change (%)
                                  Period of Analysis
 Species             1985-1990       1990-2000        2000-2040       2040-2075
 CFC-113                3.9              6.7              1.9              1.6
 CFC-22                 9.5              6.6              3.5              1.9
 CC14                   4.1              1.9              1.7              1.6
 CH3CC13                2.2              2.2              1.9              1.6
 1301                   5.9              8.4              3.1              1.7
 1211                   6                8               2.8              1.7

   Curves representing the assumptipns for growth over the 90 year period are  shown in
 Figure 14 for each of these  species, and the emission rates are tabulated in Appendix A.
   Among other scenarios, Quinn et al. (1985)  also developed an alternative growth sce-
nario for CFC-11 and CFC-12  (Figure 15a) using elasticities  at the lower end of the ex-
pected range and using conservative assumptions  about future CFC use in the  various
global regions. Elasticities of 0.5 for CFC-11 and CFC-12 in the developed nations and 1.0
in the Eastern Bloc and  developing nations combine to  produce a global average annual
                                        23

-------
growth in emissions that remains at about 1% over the,entire 90 year period.  Correspond-
ingly conservative  (slow growth) assumptions were made for the minor halocarbons; for
example, CFC-113 was assumed replaced as a defluxing agent in electronics with other
methods.  Halon emissions in 1985 were arbitrarily halved, while the growth rates were
assumed equal to those in the reference case. The global annual percentage rates of change
were:
                                        24

-------
                         Global Annual Rates of Change (%)
Species              1985-1990       1990-2000       2000-2040       2040-2075
CFC-113               3.9              2.4               1.9              1.6
CFC-22                5.5              4.2               1.3              1.0
CCL,                 -15.0              1.9               1.7              1.6
CH3CC13               2.2              2.2               1.9              1.6
1301                   6.0              8.0               3.0              1.7
1211                   6.0              8.0               3.0              1.7

These projections are shown in Figure 15b and tabulated in  Appendix A.
    An alternative high growth rate scenario  (Figures 16a and b and Appendix A) was
supplied by the EPA, adapted from exponential  econometric fits for CFC-11 and CFC-
12 elasticities, developed by IGF. Growth patterns for CFC-113 were based on projected
expansion in degreasing and defluxing and exponential elasticity expressions for CFC-22.
The rates  of increase in this scenario are:

                         Global Annual Rates of Change (%)
Species              1985-1990       1990-2000      2000-2040       2040-2075
CFt-113               3.9              6.7               1.9              1.6
CFC-11                14.0              7.2               7.6              9.4
CFC-12                7.6              6.0               4.0              6.0
CFC-113               13.0             10.0               5.0              1.6
CFC-22                11.0              8.1               4.0              5.3
CC14                   4.1              1.9               1.7              1.6
CH3CC13               2.2              2.2               1.9              1.6
1301                   6.0              8.0               3.1              1.7
1211                   6.0              8.0               2.8              1.7
II.B Other Trace Species

    Past increases in CO2, CH4 and N2O, while generally correlated to increasing popula-
tion and industrialization, involve biospheric and geophysical processes in major roles. For
CO2, emission increases have been projected based on predicted demand and availability
of fossil fuels (Edmonds et al., 1984).  An analytic fit to Edmond's midrange values was
used in the reference scenario and for most other runs, as described by:

    CO2(mole fraction in ppmv) = 341.4 + 1.539(<  - 1983)exp[0.009173(* - 1983)]    (21)

and is approximately equivalent to a 0.74% annual growth rate, producing a 96% increase
over 1985 levels in 2075. An alternate scenario expression was based on the Edmond's low
growth case:

   CO2(mole fraction in ppmv) = 341.4 + 1.82(* -  1983)exp[8.34 x 10~5(* - 1983)]   (22)

with a nearly linear increase of 1.82 ppm annually, giving 50% growth by 2075.

                                         25

-------
    The justification for projecting a continuing increase in N2O comes from the observation
that the  current annual abundance increase of 0.25% represents more than 30% of the
current atmospheric sink for N2O (WMO, 1986), so that the current source is substantially
out of balance with the sink.  The additional source is presumably of anthropogenic origin.
A gradual increase in N2O to a value around 30% higher than present levels is expected,
even if the current source strength stays constant. The observational data do not show
whether the current  small abundance increase is linear or geometric, but the differences
over the next 90 years between the two formulations are not great. Both approaches give
around a 20% increase in 2075. We have projected an annual increase of 0.25% in the N2O
surface concentration.
    Although no alternative scenarios were promulgated in this study for N2O, several runs
were made in which the stratospheric NOa abundance, for which surface emissions of N2O
are the chief source,  was artificially enhanced or suppressed.  This  was accomplished by
altering the branching ratio for the reaction of N2O with O(1D),

                              N2O + O(1D) = 2 NO

                             N20 + 0(l D) = N2 + 02.                           (23)

    Observational indications are  that the NOZ value is known  to perhaps  ± 10%.  The
behavior  of the model was investigated at values of 17.8 ppb and 22.2 ppb, compared to
the value of 20 ppb generated by the standard model.  An extreme  value of 12.4 ppb of
stratospheric NO2 was also investigated, representing a value in the range of the lowest of
current theoretical models (WMO, 1986).
    An argument similar to the N2O discussion above can be made for CH4  surface emis-
sions. The current observed increase in CH4 abundance exceeds the  model's atmospheric
sink term by about 20%, so that  a continued increase over a few decades to a level 20%
above current values would occur even if the surface flux were  held constant at current
levels.  Several additional considerations are important in the case of CH4, however. The
1951 to 1985 observations can be equally well reconciled with a linear increase as with a ge-
ometric increase, although for CH4 the  growth rate (between 1 and  2% annually) is large
enough that substantial differences arise between the  two extrapolations after 90 years.
Also, while the lifetime of N2O  exceeds the 90 year period of interest, many lifetimes of
CH4 pass during the  model simulation. Methane abundance toward the end of the run is
almost entirely  dependent on emissions occurring several decades into  the future, greatly
increasing potential uncertainties.
    Finally, projections can be made on the basis of surface flux  or surface concentration.
Since the abundances of CH4, CO and tropospheric HO are closely and mutually  inter-
dependent, specifying the CH4 surface concentration trend effectively removes a feedback
mechanism from the model, whereby an increasing CH4 concentration will increase the
CO concentration. In the absence of sufficient tropospheric NOZ, tropospheric ozone will
be decreased. All of these factors  tend to diminish tropospheric  HO, which lengthens the
lifetimes of CH4 and CO. Atmospheric  concentrations of CH4 and CO would then be in-
creased even with constant surface emissions, in a system of positive feedback. However,
if sufficient NOX is present in the troposphere, O3 can be increased.  This can increase

                                        26

-------
 HO, which is also increased as a byproduct of CH4  oxidation in the presence of NOZ.
 This potential negative feedback shortens the CH4  lifetime and opposes the effect of in-
 creasing surface emissions on the CH4 concentration.  Other processes,  such as the CO
 increase, would compete by tending to decrease HO. Such feedbacks are allowed to occur
 in the model if the CH4 surface flux rather than the surface concentration is specified
 as a boundary condition.  However, a complete and proper treatment of these feedbacks
 requires a representation of tropospheric processes that includes the effects of spatial and
 temporal variability (i.e., a 3-D model).
    A  difficulty arises, however, in  determining how the surface  emissions must have
 changed with time before the  present and therefore in justifying any projection for the
 future. While it is clear that CH4 emissions have increased from preindustrial times, the
 rate of increase that reproduces the observed concentration increase depends on CO, which
 is also affected by direct emissions and emissions of higher hydrocarbons (e.g., plant hy-
 drocarbons), and surface emitted NOZ, for which trends are not available. In this report,
 we employed a variety of CH4 boundary conditions, specifying annual  percentage changes
 in either surface concentration or surface flux, which are summarized in Appendix A. Until
 better understanding of the CH4 source distribution and sensitivities is developed, none
 of the  scenarios developed can be considered more or less likely than the others. We note
 here that a compounded 1% annual increase, used as a concentration boundary condition
 in the  reference case, produces a doubling in 70 years (2055)  and an increase by a factor
 of 2.44 in 90 years (2075).

 V. REFERENCE CASE

    The development of the projected future emissions or abundances of the various trace
 source species has been discussed above. For the purposes  of detailed discussion of model
 results, a reference case or multispecies scenario has been selected from the range of pro-
jected  possibilities, as they are understood at present.  Results  of other scenarios and
 studies of model sensitivity to projections for individual species in the multispecies context
 are also discussed below and in the Appendices.  The reference case should  not be con-
 sidered as  "most likely" or a "best  guess," but since it falls in the mid-range of scenarios
 considered, it is used  below as a basis for  a detailed  analysis of model  results.

 V.A Ozone Changes

    The change  in the total column of ozone relative to the 1985 column for the reference
scenario is given in Appendix B and Figure 17. It is striking that although this global and
annual average estimate  of ozone depletion reaches about 20% after 90 years, half of the
depletion  occurs in the last 15  years  of the simulation. Depletion over the first 50 years
is less than 5%.  This  very slow initial decrease results  from both the mitigating effects of
the concomitant increases in CH4 and CO2 and nonlinearity in the ozone response to C12
increase.

   A more detailed view of the various factors affecting  stratospheric ozone in the reference
scenario is gained by considering changes  in the ozone  concentration profile as a function
of time, as shown in Figures 18a and b.  For approximately the first 30 years,  ozone
depletion above  30 km from the direct effect of increased C1OZ catalytic cycle activity is

                                         27

-------
roughly balanced by an ozone increase below 30 km. This increase results largely from
three sources: increased odd oxygen production or 'self-healing', C1OZ interference with
the NOZ catalytic processes arid CH4 oxidation, which produces ozone and  HOr, which
interferes with the efficiency of NOZ cycles. The increased odd-oxygen production follows
from the deeper penetration of solar UV at wavelengths below 242 nm, which is capable
of dissociating O2.  Atmospheric transmittance  is increased in this wavelength region by
the decreased overhead burden of O3, the chief atmospheric absorber at wavelengths above
about 207 nm.  The increase in CIO that follows increasing CFC emissions increases the
rates of reaction with both NO and NO2. The reaction with NO2 to form C1ONO2 reduces
the fraction of total stratospheric NO2 that  is present in catalytically active forms.  The
reaction with NO reduces the efficiency of the chlorine cycle. Similarly, increases in  HO
and HO2 reduce the catalytically active fraction of NOZ by forming HNO3 and HNO4.
    At later times,  as the C\z' abundance equals and exceeds the NOZ abundance in the
25 km region, the ability of interactions among the  radical families to mitigate the impact
on ozone diminishes.  Chlorine catalytic cycles  displace NOZ cycles as the primary odd
oxygen loss mechanism in the altitude region of  the ozone peak and in the stratosphere as
a whole.
    The relative change in ozone with time  is seen to reach a maximum around 40 km.
Because the fraction of the ozone column in this region is small, the contribution of the
change in this area to the total column change  is also small.  But the  possibility of early
detection of concentration changes at 40 km is much better than for detecting the expected
changes in the total column. Expected capabilities  in satellite observing platforms should
be able to reveal ozone changes of 0.2% per year at 40 km over 5 to 10 years of operation
at the 95% confidence level (WMO, 1986).  Changes of this magnitude are predicted by
the model within the next 20 years for most  of the  scenarios investigated.
    The dashed line shown in Figure 17 is the  result of  a model run  with the reference
case scenario but with the kinetic and photochemical input  parameters  updated to the
recommendations of JPL 85-37 (1985). The rate constants that were updated are listed
in Table 2.  Very small changes were also made to the incoming solar flux and to ozone
absorption coefficients. The Herzberg continuum absorption  of oxygen was also reduced
slightly. The net effect of these changes was to significantly increase the magnitude of the
calculated ozone column depletion throughout  the 90 year integration for the reference
case scenario, increasing the relative ozone change value from 20 to 32% after 90 years.
    In general, the ozone column change in the model updated to JPL 85-37 recommenda-
tions is more linear with respect to time and increase in stratospheric C12 abundance than
is the model used in this study.  The importance  of interactive effects between families
of odd-oxygen loss catalysts seems to be reduced.  Certainly  this is the case for the odd
hydrogen loss rate, which depends chiefly on reactions of HO with HNO3 and HNO4 in
the pre-update model and more on the reaction of HO with HO2 in the updated model.
Changes were also made in the rates of important reactions in C1OZ chemistry, but a full
interpretation of these results is beyond the scope of this study. While the  quantitative
aspects of this study are affected to some extent, the overall conclusions on the magnitude
of uncertainties in the effects of scenario assumptions should not be heavily dependent on
the use of the earlier model version.

                                        28

-------
 V.B Temperature Changes

    From the surface mole fraction values of CO2, CH4, N2O, CFC-11 and CFC-12 as func-
 tions of time (Appendix C), model-derived profiles of 03 and H2O and the parameterized
 radiative response expressions, the change in equilibrium surface air temperature from the
 changed radiative forcing can be estimated. The calculated equilibrium increase, without
 feedbacks, reaches 1°C, for reference case trace species abundances in 2035, and 2.08°C for
 abundances projected for 2065.  The calculated equilibrium surface temperature change
 is more linear with time and with increase in the radiative species abundances (Figure
 19) than is the ozone response. About half of the increase is caused by CO2 and 40% by
 the combined effects  of increases in the other long-lived infrared absorbing species.  The
 remaining 10% results from alteration of the O3 and stratospheric water vapor profiles dur-
 ing the model calculation.  The order of importance of the direct radiative effects, relative
 to the CO2 effect, among the trace species other than CO2 after 80 years is CFC-12 (35%),
 CFC-11 (22%), CH4 (18%) and N2O (7%). The change in stratospheric water vapor profile
 that results from the CH4 increase accounts for 6% and the change in ozone profile that
 results from the combined effects of all the trace species trends contributes 10%.
    The coefficient of sensitivity for surface temperature change at equilibrium to the
 radiative forcing of CO2 assumed in the parameterization used here is very close to the
 1.2°C  change for doubled  CO2, quoted as the best representation of current knowledge
 (NRC, 1983). Considering the range, 1.25 to 3.75, of the feedback amplification factor for
 temperature changes from CO2 radiative forcing in GCM models, the equilibrium surface
 temperature change could vary from 2.5 to 7°C in for trace species abundances reached
 in 2065 (dashed  lines in Figure 19), if the climate system were  at equilibrium  with the
 radiative forcing. These estimates are based on  simple multipliers which roughly represent
 the results of current climate models. Actual future changes in temperature will depend
 on the time response of the climate system, which is not considered here.
    Again, as for the changes in ozone abundance, the upper stratospheric temperature
 changes are larger, with a decrease of 16°C at 40 km after 40 years and 37°C after 80 years
 (Figure 20).

 VI. OTHER  SCENARIOS AND  SENSITIVITY TO ASSUMPTIONS

    The results of the reference case scenario represent only one of the possible outcomes
 that could arise from different  combinations of plausible assumptions of trends in source
 species. The sensitivity of the predicted ozone depletion (and surface temperature change)
 to the emissions assumptions must also be known to establish the uncertainty in the result,
 the range of possible outcomes and the feasibility  of regulation.  Several approaches to
 investigation of the model's sensitivity to the choice of scenario are possible. First, various
 alternative scenario assumptions can be made to test the effects of, for example, greater
 CFC production, regulatory production caps or declining methane, alone or in combination
on ozone and temperature. For analyzing ozone change, a second approach is to calculate
the ozone response to trends in individual perturbers and then to evaluate the magnitude
of cross terms between simultaneous trends in a  number of species.  Results of applications
of both approaches  are discussed below.
                                         29

-------
VI.A Alternate Scenario Results

VI.A.I Ozone

    For a given multiple species coupled scenario, the nature of the ozone response will
depend on the details of specified emission increases for the various source species, which
determine chiefly whether odd oxygen loss will remain dominated by NOZ reactions or
shift to control by C1OZ. In this study, 36 individual scenarios were investigated with the
model, covering a range of time-dependent CFG growth rates in combination with several
assumptions on CH4,  CO2 and Brz emission increases. The change in total column ozone
for each of the 36 scenarios investigated is tabulated in Appendix B.
    The ozone column responses to the  EPA-supplied "high" and "low" CFC emission
scenarios, described previously in the discussion of emission projections, are depicted in
Figure 21; compare to the reference scenario case, Figure 17, which is intermediate in terms
of CFC emissions. Assumed N2O, CO2 and CH4 trends for Figure  21 are identical to the
reference scenario. In the  "high" case, CFC emissions increase so  rapidly that trends in
other species play a minor role in affecting the ozone change. The ozone column response
with time is strongly nonlinear as a result of the exponential nature of the CFC increases.
A depletion of 20% is reached after 35 years  (2020)  and 60% after 40 years (2025), by
which time the validity of the model  has probably broken down.  The  "low" case ozone
response is also somewhat nonlinear with time, with an ozone depletion of 1.5% over the
first 50 years and an additional 2.3% over the subsequent 40 years.  Interaction among
families and increases in N2O, CO2 and CH4 with time are  relatively more important in
the "low" case, given the gradual CFC increase.
    The range of calculated change in the ozone column for the full range of multispecies
scenarios, with the exception of the  run with very low NO.,, is  indicated by the area
between the dashed lines in Figure 21.  For about the first 25 years, the spread, or the
uncertainty arising from uncertain scenario projections is distributed about the reference
case value by a few per cent, skewed somewhat toward larger depletion. Between 25 and
40 years from present,  the scenarios with the  largest increases in  halocarbon emissions
(chiefly emissions of CFC-11 and CFC-12) diverge quickly from the reference case  to very
large ozone depletions.  The spread of results at 40 years varies from +0.46% to -66%,
demonstrating the great importance  of uncertainties in projections of future emissions
and concentrations, as well as the potential for alleviation of ozone depletion by  control
measures.
    The dotted curves in Figure 21 show the total column ozone changes calculated using
the updated rate  parameter recommendations of JPL 85-37  (1985)  for scenarios 4 and 5,
the "low" and "high" cases, respectively. The  results  are in  both cases more  negative, as
was the reference case result. Because trace species other than the CFC's increase relatively
more rapidly in the low CFC scenario 4 than in the high CFC sceanrio 5, interactions among
catalytic cycles are more important in the low case than the high case, as mentioned above.
The rate parameters recommended in JPL 85-37 tend collectively to reduce the efficiency
of catalytic cycle  interactions, so that the change in chemistry affects the low case more
than the high case. The reference case, discussed above, falls in the middle in this respect.
                                         30

-------
    Brasseur et al. (1985) used a 1-D radiative-convective model with photochemistry to
study the change in the ozone column caused by a multispecies scenario with constant
CFC emissions at current levels. They obtained an increase in the ozone column of about
3% over 90 years from the present. Although the details of the scenario assumptions differ,
a similar run with the model used here produced a 1.5% increase. The scenario differences
in the Brasseur et al. calculation, faster CH4 growth, smaller CFC emissions and projected
increase in tropospheric NO^, would be expected to favor a larger ozone increase, so there
appears to be  essential agreement between the two models.
    Calculated changes  in local ozone concentration in the upper stratosphere are more
closely grouped for the range of scenarios than  are the  changes in vertically integrated
ozone column. All scenarios predict  a significant decrease in upper stratospheric ozone
(Figure 22), resulting from increased C\z abundance.  Near 40 km after 20 years, the
calculated decrease in local ozone abundance ranges from 10 to 30%. After 90 years, the
calculated decrease at this pressure level varies from 26 to > 80%.
    The impact of policy alternatives  for the regulation of CFC emissions was also inves-
tigated using  several CFC emissions growth scenarios supplied by EPA. The results are
presented in Figure 23.  Issues related  to regulation include sensitivity of ozone changes to
the timing, method and degree of regulation and to the nature of the unregulated scenario
to which the results of regulation are compared.  In the cases investigated in this report,
regulatory delays of the  order of 10 years had little impact on the final (90 year) outcomes.
The choice  of unregulated scenario and the degree of CFC emission restriction were of
greater significance.

    VI.A.I.a Effects of methane projections
    Since the proximate causes of the observed methane increase are at present uncertain,
several assumptions other than a continued 1% annual increase in CH4 surface abundance
were investigated for common CFC growth scenarios (the reference case already discussed
and a case with 2.6% average annual  CFC-11 and CFC-12 increase). Three  scenarios, in
which the CH4 surface  flux rather than the surface abundance was assumed to increase
at various compound  growth rates, were compared to the reference case assumption of a
1% annual concentration increase (Figure 24). Two additional scenarios, one assuming a
geometric (compound annual) increase in CH4 flux and  one assuming a linear increase,
but both  corresponding to  an  initial annual concentration increase of 1.5%, were also
investigated.  Over the  90 year period the geometric scenario continues to  produce the
same CH4 concentration growth rate, resulting in a 12% 03 depletion after 90 years, while
the linear case drops to an average 0.9% annual rate of  CH4 concentration  increase and
produces a 21% ozone column depletion. A third case of monotonic geometric flux increase
that gives only a 0.4% concentration growth  in the first few years results in an 18% ozone
column depletion and a 90 year average annual CH4 growth of 1.1%. These  results show
that substantial uncertainty in long term ozone depletion can arise even among monotonic
CH4 growth projections  that would reproduce the currently observed surface concentration
growth rate of 1%.
    If the future of CH4 surface emission is not monotonic growth, the uncertainty in ozone
change from CH4 projection assumptions increases. Four cases were compared, with a 1%

                                        31

-------
annual flux increase to 2010 followed by either (1) 1% annual decrease to 2030, then zero
trend, (2) zero trend from 2010, (3) 1% annual increase to 2075 or (4) 2% annual increase
to 2075.  The ozone column depletion values range from 8% to 62% in 2075 (Figure 25).
As expected, higher  levels of methane corresponding to faster emission growth  rates are
associated with smaller ozone column depletions.
   Model runs with "high" case CFC projections show that when C\z gets  large rapidly,
uncertainties arising from CH4 projections shrink. Cases with methane increasing at 0.5%,
1% and 1.5% annually (Figure 26) differ only by a few years in reaching extremely large
ozone depletions. Conversely, for the "low" CFC growth case, assuming a methane trend of
1.5% annually rather than 1%  is sufficient to change the sign of the ozone column change.
After 90 years, the ozone column has increased by 3.4% (Figure 27).

   VI.A.l.b Effects of Halon emissions
   The effect of the projected reference case growth in Halon emissions  was investigated
relative to scenario 1A, a case with mid-range CFC growth. Fixing the abundances of the
bromine source species, Halons 1301 and 1211, at the current very small levels reduced
the ozone impact of  scenario 1A by a factor of about 0.8. The increases in the bromine
source species projected here contribute between 4 and 5% to the ozone  column decrease
after 90 years (Figure 28).  This nonnegligible contribution,  despite the small projected
emissions  compared to CFC-11 and CFC-12  (about 1%),  demonstrates  the efficiency of
BrOz in catalyzing odd oxygen loss in the model.

   VI.A.l.c Effect of CO2 projection
   The difference between the reference and alternative lower rates of CO2 increase dis-
cussed previously is  small enough  over  90 years and the effects  of other simultaneously
varying species are large enough that the  sensitivity of the ozone column change to  the
CO2 scenario choice is small. The calculated depletion in the ozone column for both alter-
natives in a mid-range CFC scenario differs by only 1% in the 25% depletion calculated at
90 years from present.

   VI.A.2 Effects of stratospheric NOZ abundance
   The importance of the interactions between the various families of reactive particularly
NOZ and  C1OZ,  has  been discussed previously. For two model runs the branching ratio
for NO production in the reaction of N2O with O(1D) was modified to artificially enhance
or suppress the mid-stratospheric abundance of NO2 by about 10%.  This change affected
the initial conditions generated by  the integration of historical emissions as well  as  the
integration of projected emissions, so that the change in ozone column from  the CFC and
other species perturbations can be directly compared. The expected behavior, NO2's miti-
gating role in determining the efficiency of C1OZ catalytic ozone destruction,  was  observed
in the results (Figure 29). At an NO2 value of 22.2 ppb, compared to  the standard model's
20 ppb, the impact of a mid-range CFC emission case (2.6% average annual increase) on
ozone is reduced from a depletion of 26%  at 90 years to 15%. When NO2  is reduced to
17.8 ppb,  the 2075 depletion is more than doubled to 42%.

                                        32

-------
    No cases were run in which surface N2O concentrations increased at other than 0.25%
 annually, but the effect of a larger positive trend would be to flatten the total column ozone
 depletion curve with time, depending on the rate of CFG increase. A faster initial N2O
 increase would cause faster initial ozone depletion, through the direct effects of the NOZ
 catalytic cycle. Further into the scenario, when C1OZ catalytic destruction is dominant,
 increased N2O produces  increased stratospheric NOX, with the effect,  shown above, of
 reducing the ozone decrease  for scenarios with CFC annual increases of around 2%.

 VI.A.2 Temperature

    The calculated changes in  stratospheric temperature show a smaller spread with sce-
 nario than the calculated ozone changes.  This follows from the dependence of the strato-
 spheric radiative equilibrium temperature mostly on local  ozone abundance and the CO2
 atmospheric mixing ratio.  The narrower spread of the calculated upper stratospheric ozone
 change relative to the calculated total ozone column change and the relative certainty of
 CO2  increase (0.5-0.6% per year) tend to produce a restricted range of temperature change
 (Figure 30). The alternate CO2 scenarios differ by less than 5% in the calculated strato-
 spheric temperature decrease at about 45 km in the case of mid-range growth in halocarbon
 emissions. The variation  in  local upper stratospheric ozone depletion among the various
 halocarbon scenarios has  a larger effect on the calculated temperature change, producing
 a range of 18 to >  40° C decreases after 80 years at 45 km. Around 10° C of this decrease
 results from the CO2 increase, ozone decrease accounting for the rest.
    Calculated surface temperature changes also do not vary as widely with scenario as for
 the ozone column. Where the effect on the ozone column represents a precarious balance of
 competing effects (e.g., CH4 increase and CFC increase) increases in the various radiatively
 active species all serve to  warm the surface. Mid and upper stratospheric ozone, which is
 of major importance in controlling the ozone column change, plays only a minor role in
 affecting the surface temperature. The equilibrium temperature increases for the radiative
 forcings of several scenarios range from around 1.3 to 2.2°C for trace species abundances
 reached  in 80  years and  are shown in Figure 31.  Incorporating the climate feedback
 amplification discussed above increases the surface temperature change and its range from
 1.6 to 8.2°C.

    The increase in CO2 accounts for between 50 and 70% of the calculated temperature
 increase in all cases. The combined direct radiative effects of increases in CH4, N2O, CFC-
 11 and CFC-12 contribute between 25 and 40%, with an average over the scenarios of about
 30%.  The direct effects of CFC-12 and CH4 usually dominate over those of N2O and CFC-
 11. The changes in  the ozone vertical profile  and stratospheric water vapor profiles that
 are produced by the trends in the various source species account for the remaining 10% of
 the calculated equilibrium surface temperature increase.

 VLB  Analysis of Ozone Sensitivities

    In the model runs previously discussed, various combinations of two or  three trend or
abundance assumptions for CFC's,  CH4 and mid-stratospheric NO., produced a range of
ozone change from a small increase over the 90-year projection to remarkable decreases after
only 35 years. The question therefore arises whether these results can be systematized, for

                                         33

-------
example by identification of a small number of controlling variables. Although it is just the
purpose of the model to include the nonlinearities and couplings inherent in this complex
problem, it may be possible to explain the major characteristics of the ozone predictions
based on a small number of factors involving the scenario assumptions and model behavior.
    As discussed previously,  there are two major regimes  in the  response of the ozone
column to coupled perturbations involving large increases in stratospheric chlorine abun-
dance. For conditions in which the family of nitrogen oxide-based catalytic reaction cycles
dominates photochemical destruction of odd oxygen, the terms  important to consider in-
clude the  direct effect of chlorine increases on  upper  stratospheric ozone, the effect of
chlorine increases on the nitrogen oxide reactions and the effect of methane increases and
temperature decreases on the NOZ and HOZ catalytic cycles. For the alternate regime in
which C\OX  cycle rates exceed the NOZ rates, the effects of NOZ  and HOZ (produced by
CH4) increases on the chlorine species  and cycles are most important.  The sensitivity of
the ozone column to increases in given species can vary between the two regimes and can
even change sign.
    Figure 32 shows the time  dependence of the ozone depletion for the reference case CFC
trend projections, but with CO2, N2O and CH4 held fixed at  1985 surface abundances
A calculated ozone depletion of 54% is obtained at 90 years 2 compared to 20% in the
multiple scenario reference case.  However, ozone depletion at 50 years is only increased
from 4.3% to about 7%, demonstrating the strong nonlinearity  of the sensitivity  of ozone
to Clz abundance.  In Figure 33, the percentage ozone depletion in the CFC-only case is
plotted against  the total abundance of chlorine-containing species at 55 km.  A  knee in
sensitivity is observed, occurring at around 20 ppb and 20% column ozone depletion. For
C12 increases from 1985 of less than 15 to 20 ppb (by volume)  the relative ozone column
depletion  for a unit absolute  increase in C\z is less than a fifth of the sensitivity for a unit
C12 increase when the total C\z concentration exceeds about 20 ppb. A very similar curve,
also shown in Figure 33, is obtained when the ozone depletion for the case of "high" CFC
emission projections is plotted against absolute change in  C\z  rather than time.  In this
case, a change of 30 ppb in C\z is reached after only 40 years rather than the 90 years of
the reference case, and the trends in the other species have not yet resulted in increases
which could effectively mitigate a portion of the C\z impact.
    The absolute ozone profile changes in the  CFC-only reference case (Figure 34 compare
to Figure 18a)  show that  the source of the  nonlinearity lies in the region of the ozone
maximum, 20 to 25 km, and is the result of the reactions
                               CIO + NO = Cl + NO2                           (24)

 and
                          CIO + N02 + M = ClONOt + M.                      (25)
 When the CIO abundance in the mid stratosphere reaches the point at which C1ONO2 is
 the dominant NO? species and NO is substantially suppressed, significant amounts of NO
 and NO2 are no longer available for reaction, and additional CIO introduced by increasing
  2 This depletion is past the point where the validity of the model structure, lacking
 feedback to the transport parameterization, is expected to hold.
                                         34

-------
emissions participates in catalytic cycles with enhanced efficiency in the region with the
greatest potential for ozone loss.  This effect has also been discussed by Stolarski and
Douglass (1986), who term it a "titration" of NO by the reagent CIO.
    From the above discussion, it appears that the degree of ozone depletion produced by
increased CFG fluxes, in the absence of complicating nonzero trends for other Species, does
not depend on the rate of increase of the CFC flux (the scenario) within reasonable limits,
but only on the absolute increase. This result is expected for the 1-D model for situations
in which the odd-oxygen controlling photochemistry can come into equilibrium with slowly
varying long-lived source species.
    The time scales of photochemical processes vary  widely for different constituents and
altitudes and in comparison to transport time scales. For the CFC perturbations of interest
the relevant time scale for the possible occurrence of biospheric effects is several  decades
up  to about  100 years.  A few  decades  are necessary to increase stratospheric chlorine
levels to the point of observable  impact on the total column  abundance given current and
projected emissions. Around a century is necessary to remove 63% of a chlorofluorocarbon
such as CFC-11, once it  has been released in the atmosphere. The time constants in the
one-dimensional model range from the order of seconds for the fast upper stratospheric
photochemistry to several years for establishing an equilibrium profile between the surface
and the stratopause for  a long-lived species like CFC-12, given a constant surface flux
matched to the current atmospheric burden. The photochemical lifetime of ozone in the
lower stratosphere can range up to a few months, comparable to transport time scales.
Thus the 1-D ozone profile accommodates rapidly to C12 abundance changes at the possible
rates of CFC increase.
    The position of the knee in the 6 (O3)/6 (C12) relationship as a function of 6  (Clz) is
clearly one important aspect of analyzing ozone  depletion.  The C12 value at which this
transition in sensitivity occurs depends on both the abundance of NO2 and the fractions of
NOZ and C12 that are present as active NOZ and C10Z radicals, respectively. The  model's
NO2 abundance, 20 ppb  at 35 km, is a function  of the distribution of solar UV radiation
of wavelengths capable of photolyzing ^O and the choice of the profile of the eddy diffu-
sion constant. (A  smaller contribution arises from in situ cosmic ray production of NO.)
Observations also indicate a value around 20-25 ppb of NO2 in the mid stratosphere.
    While the transition  region for sensitivity  to  C12 perturbations in the CFC-only case
occurs near 20 ppb, which is also the NO2 maximum abundance in the current ambient
model, the position of the sensitivity transition as a function of C12 will vary with scenario.
In the multiple species reference  case (Figure 35), the position of the knee appears to have
moved to a higher value of 6 (C12), somewhere around the 27 ppb endpoint of the scenario
at 90 years. Also shown  are the results of runs made with the reference CFC projections
in conjunction with a 0.25% annual ^O trend  (and fixed methane flux)  and separately
with a 0.5% annual CH4  trend (fixed ^O abundance).
    In the CFC/methane case, the  proportion of C1OZ to total C12 has been reduced  in
comparison to the CFC-only cases (Figure 36) by the higher rate for the reaction

                             Cl + CH4 = HCl + CH3.                          (26)
                                         35

-------
As the proportion of CIO to total Clz is reduced, a higher value of total Clz must be reached
before the mid-stratospheric interaction with NO and NO2 is saturated. This accounts for
most of the mitigating effect of CH4 increase  on  Clr-related ozone depletion, with some
contribution coming from increased HOZ reducing the efficiency of NOZ photochemistry.
The knee in sensitivity appears at about 23 ppb C\z.
    In the CFC/N2O case, the effect of the increasing N2O on ozone actually  reverses at
about 55 years from present. Added N2O increases mid-stratospheric NOX and for small
values of 6 (Clg) NOZ remains the primary catalyst for odd  oxygen loss, so the increased
NOZ increases the calculated depletion. For larger values of 6 (Cl.,), the interference of
NOj with C1O2 catalyzed ozone destruction is a bigger effect than the direct  interaction
of NOZ  with O and O3.  Calculated  ozone depletion is reduced from the  CFC-only case
because more NOZ is available to interfere. The position of the knee is extended to about
24 ppb in this scenario.
    Figure 37 shows the relatively small effects of temperature feedback on the calculated
ozone column change for the CFC-only reference  case. As stratospheric ozone decreases,
the local heating rate, dependent on absorption of solar radiation by ozone, also decreases.
The radiative balance between local heating and local infrared cooling is tilted in favor of
cooling, so the temperature drops. Catalytic cycles for odd oxygen loss become less effec-
tive at lower temperatures so the calculated depletion is somewhat smaller than in a fixed
temperature stratosphere. For the reference case including the projected CO2 increase, as
well as the nonzero trends in the other source species, the resulting stratospheric temper-
ature decreases appear more important in their effect on odd-oxygen loss rates. When  the
temperature profile is  fixed, ozone column depletion increases by a factor of 1.5 compared
to the case  in which temperature feedback is  allowed (also  shown in Figure 37).  While
upper stratospheric temperature decrease represents a damped negative  feedback when
the temperature drop results from local ozone decrease,  temperature decreases resulting
from increasing CO2 are effective in decreasing the local ozone loss rate irrespective of
local changes in ozone or other species.
    The combination  of the  CFC-only ozone  response,  the effect  of CH4 on C1OZ/C12
and the dependence of NOZ on  N2O emissions appears capable of explaining the overall
response of the model to the variety of multiple species scenarios investigated in this study.
Smaller contributions to ozone sensitivity are also made by CO2 and BrOz.

VII. SUMMARY AND CONCLUSIONS

    A detailed description has been given above of the method and assumptions that have
been used to generate the model-predicted time  dependent ozone changes. Inference of
actual changes in atmospheric ozone, trace species and temperature from the results of the
present  study must include proper treatment of the future behavior of the troposphere in
view of  possible trends in CO and NOZ, and the effects of uncertainties and imperfections
in the model structure and assumptions. Uncertainties and errors in the model formulation
include  gross simplification of atmospheric dynamics and possible neglected or incorrect
photochemistry.
    We  have shown that  current uncertainties in the projection  of trends of emissions
of both  the  biogenic and industrially produced source species are sufficient to produce a

                                         36

-------
dramatic range of model-predicted ozone column changes over the next 30 to 90 years. The
variables that appear to contribute most to the uncertainty in the calculated effect on ozone
are the magnitude of the projected change in C\z abundance (projections  of halocarbon
emissions), the current stratospheric NOZ abundance and the lack  of knowledge of the
causes of the observed change in methane abundance, and thus the difficulty in predicting
future changes.  The timing of projected changes in the  abundance or emissions of the
various source species is also important.  For example, rapid increases in CFC-11  and
CFC-12 could overwhelm the mitigating effects of CH4 and ^O trends.
    Another source of significant uncertainty is the model itself. Improvements in knowl-
edge of photochemical kinetics  and spectroscopy have resulted in the  past in dramatic
swings of predicted ozone change for a given simple CFC only scenario.   Whether such
major  oscillations are now or will be damped in the future by the achievement of com-
pleteness and accuracy  in the stratospheric kinetic data set is not clear,  although the
current situation is a substantial improvement over the state of knowledge only 10 years
ago. Recent work with techniques such as Monte Carlo sampling uses kinetic uncertainties,
based on the evaluation of random and systematic errors in laboratory studies, to estimate
the approximate range of uncertainties in predictions of ozone depletion obtained within
the model's existing framework. Two studies of this type show that the uncertainty limits
are on the order of the ozone depletion estimates for Clz abundances reached by constant
production at current levels.
    Finally, the biospheric impacts of stratospheric ozone depletion  and surface UV en-
hancement will depend strongly on latitude. The  geometry of the  earth's orbital incli-
nation produces much smaller polar surface UV fluxes than are present in equatorial re-
gions.  The 1-D model results are quasi-global in nature, as has been  discussed above, but
the biospheric impact of ozone depletion clearly depends on latitude. Steady-state two-
dimensional model results indicate substantial latitudinal dependence of the ozone change.
The relative, but not absolute, impact in surface ultraviolet radiation is  enhanced at high
latitudes over the equatorial change both because of the greater ozone depletion calculated
for high latitudes and because of the very small initial surface UV flux.
    In summary, we have not attempted in this study to define the most probable future for
the integrated ozone column. We have shown that many areas remain in which uncertain-
ties could be reduced through better knowledge of atmospheric and biospheric processes as
well as econometric forecasting.  Improvement in the kinetic and photochemical data base
and in model dimensionality and treatment of transport can also be profitably pursued.
While  the uncertainty of predictions plays a central role in this report, given the recent
progress in all the areas of study listed above, a narrowing of uncertainties can confidently
predicted.

                             ACKNOWLEDGMENTS
    This work was performed under the auspices of the U.S. Department of Energy by
Lawrence Livermore National Laboratory under Contract W-7405-Eng-48.
                                        37

-------
                                  REFERENCES
Bates, D.R. and M. Nicolet, The photochemistry of atmospheric water vapor, J. Geophys.
    Res., 55,  301- , 1950.

Blake, D.R.,  E.W. Mayer, S.C.  Tyler, Y. Makida,  B.C. Montague and F.S. Rowland,
    Global increase in atmospheric methane concentrations between 1978 and 1980, Geo-
    phys. Res. Lett.,  9,  477-480, 1982.

Brasseur, G., A. De Rudder and C.  Tricot, Stratospheric response to chemical perturba-
    tions, J.  Atmos.  Chem., 3,  261-288, 1985.

CIAP (Climatic Impact  Assessment  Program), The stratosphere perturbed by propulsion
    effluents, CIAP Monograph III, U.S. Dept. of Transportation Report DOT-TST-75-53,
    1975.

CMA (Chemical Manufacturers Association), World production and release of chlorofluo-
    rocarbons 11 and 12 through 1981, Report FPP 83-F, Washington, D.C., 1982.
Callis, L.B., M.  Natarajan and R.E. Boughner, On the relationship between the greenhouse
    effect, atmospheric photochemistry and species distribution,  J. Geophys. Res.,  88,
    1401-1426,  1983.

Callis, L.B., M.  Natarajan and J.M. Russell III, Estimates of the stratospheric distributions
    of odd nitrogen from LIMS data, Geophys. Res. Lett., 12,  259-262, 1985.
Callis, L.B., M. Natarajan, R.E. Boughner, J.M. Russell III and  J.D. Lambeth, Strato-
    spheric photochemical studies using Nimbus 7 data 2. Development of inferred trace
    specie distributions,  J.  Geophys. Res., 91, 167-1197,  1986.
Chang, J.S., A.C. Hindmarsh and N.K. Madsen, Simulation of chemical kinetics transport
    in the stratosphere,  Stiff Differential Systems, R.A.  Willoughby, ed., Plenum, New
    York, 51-65, 1974.

Chapman, S., A theory of upper-atmospheric ozone,  Mem. Roy. Meteorol. Soc., 8, 103-
    , 1930.

Cheung,  A.S.C., K. Yoshino, W.H.  Parkinson and D.E. Freeman,  Herzberg continuum
    cross section of oxygen in the wavelength region 193.5-204.0 nm:  New laboratory
    measurements and stratospheric implications,  Geophys.  Res. Lett., 11, 580-582, 1984.
Connell,  P.S. and D.J. Wuebbles, "Summary of photochemical and radiative  data used
    in the LLNL one-dimensional transport-kinetics model of the troposphere and strato-
    sphere: 1982,"Lawrence Livermore National Laboratory Report UCID-19557-82,1983.
Crutzen, P.J., The influence of nitrogen oxides on the atmospheric ozone content, Quart.
    J. Roy. Met. Soc., 96,  320-325,  1970.
Crutzen, P.J., Ozone  production rates in an oxygen-hydrogen-nitrogen oxide atmosphere,
    J. Geophys.  Res., 76, 7311-7327, 1971.
Crutzen, P.J., L.E. Heidt, J.P. Krasnec, W.H. Pollock and W. Seller, Biomass burning as
    a source of the atmospheric gases CO, H2, N2O,  NO CH3C1 and COS,  Nature, 282,
    253-256, 1979.
                                        38

-------
Delany, A.C., P.J. Crutzen, P. Haagensen, S. Walters and A.F. Warthourg, Photochemi-
   cally produced ozone in the emission from large-scale tropical vegetation fires, J. Geo-
   phys.  Res., 90, 2425-2429, 1985.
Edmonds, J.A., J. Reilly, J.R. Trabalka and D.E. Reichle, An analysis of possible future
   atmospheric retention of fossil fuel CO2, U.S. Dept. of Energy Carbon Dioxide Research
   Division Technical Report , 1984.
Fabian, P., Halogenated hydrocarbons in  the atmosphere, in  The  Handbook of Environ-
   mental Chemistry, Volume 4/Part  A,  O. Hutzinger, ed., Springer-Verlag, Berlin, pp.
   24-51, 1986.
Frederick, J.E. and J.E. Mentall, Solar  irradiance in the stratosphere: Implications for the
   Herzberg continuum absorption of  O2, Geophys. Res.  Lett., 9,  461-464, 1982.
Grant, K.E., P.S. Connell and D.J. Wuebbles, "Monte Carlo uncertainty analysis of changes
   in atmospheric ozone concentrations from large trace gas perturbations,"  Lawrence
   Livermore National Laboratory Report UCRL-93375, 1986.
Hampson, J., Photochemical behavior of the ozone layer, Can. Armament Res. and Devel-
   opment Establishment, TN1627/64, 1964.
Hansen, J., D. Johnson, A. Lacis, S. Lebedeff,  P. Lee, D. Rind and G. Russell, Climate
   impact of increasing atmospheric carbon dioxide, Science, 213,  957-966, 1981.
Herman,  J.R. and J.E. Mentall, O2  absorption cross sections (187-225 nm) from strato-
   spheric solar flux measurements, J. Geophys.  Res., 87,  8967-8975, 1982.
Hindmarsh, A.C., "Preliminary documentation of GEARBI: Solution of ODE systems with
   block-iterative treatment of the Jacobian," Lawrence Livermore National Laboratory
   Report UCID-30149, 1976.
Howard, C.J. and K.M. Evenson, Kinetics  of the reaction of HO2 with NO,  Geophys. Res.
   Lett., 4, 437-440, 1977.
JPL  (Jet  Propulsion Laboratory) Publication 83-62, Chemical  Kinetics and Photochemical
   Data  for Use in  Stratospheric Modeling, W.B. DeMore,  ed.,  Pasadena, California,
   1983.
JPL  (Jet  Propulsion Laboratory) Publication 85-37, Chemical  Kinetics and Photochemical
   Data  for Use in  Stratospheric Modeling, W.B. DeMore,  ed.,  Pasadena, California,
   1985.
Johnston, H.S., Reduction of stratospheric ozone by  nitrogen oxide catalysts from SST
   exhaust, Science, 173, 517-522, 1971.
Johnston, H.S., M. Paige and F. Yao,  Oxygen absorption cross sections in the Herzberg
   continuum and  between 206 and 327 K,  J.  Geophys.  Res., 89,  11661-11665,  1984.
Keeling,  C.D., R.B.  Bacastow and  T.P.  Whorf,  Measurements of the concentration  of
   carbon dioxide at Mauna Loa Observatory, Hawaii, in Carbon Dioxide Review:  1982,
   W.C.  Clark, ed., New York, 377-385, 1982.
Khalil, M.A.K. and R.A. Rasmussen,  Carbon monoxide in the earth's atmosphere: In-
   creasing trend,  Science, 224, 54-56, 1984.

                                       39

-------
Kiehl, J.T. and V. Ramanathan, CO2 radiative parameterization used in climate models:
    Comparison with narrow band models and with laboratory data, National Center for
    Atmospheric Research Report NCRA/0304/82-9, 1983.
Lacis, A., J. Hansen, P. Lee, T. Mitchell and S. Lebedeff, Greenhouse effect of trace gases,
    1970-1980,  Geophys. Res. Lett., 8,  1035-1038, 1981.
Levine, J.S., C.P. Rinsland and G.M. Tennille, The photochemistry of methane and carbon
    monoxide in the troposphere in 1950 and 1985, Nature, 318,  245-249, 1985.
Liu, S.C., D. Kley, M. McFarland, J.D. Mahlman and H. Levy II, On the origin of tropo-
    spheric ozone,,  J. Geophys. Res., 85, 7546-7552, 1980.
Logan, J.A., Nitrogen oxides in the troposphere: global and regional budgets,  J.  Geo-
    phys. Res., 88, 10785-10807, 1983.

Luther,  F.M., D.J. Wuebbles and J.S. Chang,  Temperature feedback in a stratospheric
    model, J. Geophys. Res., 82,  4935-4942, 1977.
Luther, F.M., D.J. Wuebbles, W.H. Duewer and J.S.  Chang, Effect of multiple scattering
    on species concentrations and model sensitivity,   J. Geophys. Res., 83,  3563-3570,
    1978.

Luther, P.M., J.S. Chang, W.H. Duewer, J.E. Penner, R.L. Tarp and D.J. Wuebbles, Po-
    tential environmental effects of aircraft emissions, U.S. Dept. of Transportation Report
    FAA-EE-79-32, 1979.
McElroy, M.B.,  S.C. Wofsy and Y.L. Yung,  The nitrogen cycle: Perturbations due to man
    and their impact on  atmospheric N2O and O3, Phil. Trans. Roy. Soc.  London, B277,
    159-181, 1977.

Molina, M. and  F.S. Rowland, Stratospheric sink for chlorofluoromethanes: Chlorine atom
    catalyzed destruction of ozone,  Nature, 249, 810-812,1974.
NASA (National Aeronautic and Space Administration) Reference Publication 1010, Chlo-
    rofluoromethanes and the Stratosphere, R.D. Hudson, ed., Washington, D.C., 1977.
NRG (National Research Council), Climatic Effects of Supersonic Flight, National Academy|
    of Sciences,  Washington, D.C., 1973.
NRC (National Research  Council), Long-term  Worldwide Effects  of Multiple Nuclear
    Weapons Detonations, National Academy of Sciences, Washington, D.C., 1973.
NRC (National  Research Council), Causes and Effects of Stratospheric Ozone Reduction.
    An Update,  National Academy Press, Washington, D.C., 1982.
NRC (National  Research Council), The Effects on the Atmoshere of a Major Nuclear
    Exchange, National Academy Press, Washington, D.C., 1985.
NRC (National  Research Council), Changing Climate, Report of the Carbon Dioxide As-
    sessment Committee, National Academy Press, Washington, D.C., 1983.
OECD (Organization for Economic Cooperation and Development), Scenarios for chlo-
    rofluorocarbons, ENV/CHEM/PJC/81.93, Paris,  1981.
                                       40

-------
Oeschger, H., B. Stauffer, A. Neftel, J. Schwander and R. Zumbrunn, Atmospheric CC>2
    content in the past deduced from ice-core analyses, Annals of Glaciology, 3, 227-232,
    1982.
Owens, A.J., C.H. Hales, D.L. Filkin, C. Miller and M.C. McFarland, Multiple scenario
    ozone change calculations:  The subtractive perturbation approach, in Atmospheric
    Ozone, Proceedings of the Quadrennial Ozone Symposium,  Halkidiki, Greece, C.S.
    Zerefos and A. Ghazi, eds., D. Reidel, Dordrecht, 82-86, 1985.
Prather, M.J., M.B. McElroy and S.C. Wofsy, Reductions in ozone at high concentrations
    of stratospheric halogens, Nature, 312,  227-231,1984.
Prinn, R.G., P. Simmonds, R. Rasmussen, R. Rosen, F. Alyea, C. Cardelino, A. Crawford,
    D.  Cunnold, P. Fraser and J. Lovelock, The Atmospheric Lifetime Experiment,  1:
    Introduction, instrumentation and overview,  J. Geopkys. Res., 88,  8353-8368, 1983.
Quinn, T.H., K.A. Wolf, W.E. Mooz, J.K. Hammitt, T.W. Chesnutt and S. Sarma, Pro-
    jected use, emissions and banks of potential ozone depleting substances, RAND Report
    N-2282-EPA, 1986.
Ramanathan, V., Radiative transfer within the earth's troposphere and stratosphere:  A
   „simplified radiative-convective model, J. Atmos. Set., S3,  1330-1346, 1976.
Ramanathan, V., R.J. Cicerone, H.B. Singh and J.T. Kiehl, Trace gas trends and their
    potential role in climate change,  J. Geophys. Res.,  5547-5566, 1985.
Rasmussen, R.A. and M.A.K. Khalil, Atmospheric methane (CH4): Trends and seasonal
    cycles, J.  Geophys. Res., 86,  9826-9832, 1981.
Reinsel, G.C., G.C. Tiao, A.J. Miller, C.L. Mateer, J.J. Deluisi and J.E. Frederick, Anal-
    ysis of upper stratospheric Umkehr ozone profile data for trends  and the effects  of
    stratospheric aerosols,  J. Geophys. Res., 89,  4833-4840, 1984.
Rinsland, C.P. and J.S. Levine, Free tropospheric carbon monoxide concentrations in 1950
    and 1951 deduced from infrared total column amount measurements,  Nature, 818,
    250-254, 1985.
Rinsland, C.P., J.S. Levine  and  T.  Miles, Concentration of methane in the troposphere
    deduced from 1951 infrared solar spectra,  Nature,  318,  245-249, 1985.
Smith, C.A., L.T. Molina, J.J. Lamb and M.J. Molina, ,  Int. J. of Chem. Kin., 16,  41- ,
    1984.
Stauffer, B.,  G. Fischer, A. Neftel  and H.  Oeschger, Increase of atmospheric methane
    recorded in Antarctic ice core,  Science,  229,  1386-1388, 1985.
Stolarski, R.S.  and R.J. Cicerone,  Stratospheric chlorine:  A possible sink  for ozone,
    Can. J. of Chem., 52,  1610-1615, 1974.
Stolarski, R.S. and A.R. Douglass,  Sensitivity of an atmospheric  photochemistry model
    to chlorine perturbations including consideration of uncertainty propagation, J. Geo-
    phys. Res.,  (in press),  1986.
Thompson, A.M. and R.J. Cicerone, Atmospheric CH4, CO and OH from 1860 to 1985,
    Nature, 321, 148-150, 1986.

                                       41

-------
Toon, G.C., C.B. Farmer and R.H. Norton, Detection of stratospheric N2O5 by infrared
   remote sounding,  Nature, 319,  570-571, 1986.
U.S.  Standard Atmosphere, 1976, NOAA-S/T76-1562, Supt. of Documents, U.S. Govern-
   ment Printing Office, Washington, D.C., 1977.
WMO  (World Meteorological Organization), The Stratosphere 1981:  Theory and Mea-
   surements, WMO Global Ozone Research and  Monitoring Project Report No.  11,
   1982.

WMO (World Meteorological Organization), Atmospheric Ozone: Assessment of Our Un-
   derstanding of the Processes Controlling its Present Distribution and Change,  , 1986.
Wang, P.H., S.S. Hong, M.F. Wu and A. Deepak, A model study of the temporal and spatial
   variations of the zonally-averaged ozone heating rate,  J. Atmos. Sci., 39, 1398-1409,
   1982.

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. of Geophys., 24,  110-140,
   1986.
Weiss, R.W., The temporal and spatial distribution of tropospheric nitrous oxide,  J. Geo-
   phys. Res., 86,  7185-7195, 1981.
     (.
Wofsy,  S.C., M.B. McElroy and Y.L. Yung, The chemistry of atmospheric bromine, Geo-
   phys. Res. Lett., 2, 215-218, 1975.
Wuebbles, D.J., "A theoretical analysis of the past variations in global atmospheric com-
   position and temperature structure," Lawrence Livermore National Laboratory Report
   UCRL-53423, 1983.
Wuebbles, D.J., P.M. Luther and J.E. Penner, Effect of coupled anthropogenic perturba-
   tions on stratospheric ozone, J. Geophys. Res., 88,  1444-1456, 1983.
Wuebbles, D.J., M.C. MacCracken and F.M. Luther, A proposed reference set of scenarios
   for  radiatively active atmospheric constituents, U.S. Dept. of Energy Carbon Dioxide
   Research Division Technical Report DOE/NBB-0066, 1984.
                                       42

-------
                          TABLE 1
                Chemical kinetic rate coefficients1
                    k = A * exp ( B / T )
2
3
4
5

6
7
8
9
10
11
12
13
14
15
16
172
18
19
20

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
O + O3 = 2 O2
O(1D)+M =  O + M
O(*D) + O3 = 2 O2
CfD) + 03 = 02 + 2 O

03 + NO = NO2 + 02
O + NO2 = NO + O2
N2O + 0(1D)  = N2 + O2
N2O + 0(1D)  = 2 NO
N + O2 = NO + O
N + NO = N2 + O
N + NO2 = N2O -1- O
N2 + O(1D) + M = N2O + M
NO + O + M  =  NO2 + M
NO2 + O3 = NO3 + O2
NO + NO3 = 2 N02
N02 + O = NO + 02
NO2 + NO3 = N2O5
N205 = N02 + N03
O + NO3 = O2 + NO2
      + H2O = 2 OH
H2 + O^D) = OH + H
O3 + OH = HO2 + O2
O + OH = O2 + H
O3 + HO2 = OH + 2 O2
O + HO2 = OH + O2
H + O2 = HO2
O3 + H = OH + O2
H02 + H02 = H202 + 02
HO2 + HO2 + H2O = H2O2
HO2 + OH = H2O + O2
H2O2 + OH = H2O + HO2
OH + OH = H2O -f O
OH + OH = H202
H2O2 + O = OH + HO2
H2 + OH = H20 + H
H + HO2 = H2 + O2
H -(- HO2 = 2 OH
O
                                HO
                                                         B
                                                   SI
8.000E-12
2.100E-11
1.200E-10
1.200E-10
1.800E-12
9.300E-12
4.900E-11
6.700E-11
4.400E-12
3.700E-11
3.000E-12
S4
S5
1.200E-13
3.000E-12
S10
Sll
S12
-2.060E+03
9.500E+01
0.
0.
-1.370E+03
0.
0.
0.
-3.220E+03
0.
0.


-2.450E+03
0.



              l.OOOE-11

              2.200E-10
              l.OOOE-10
              1.600E-12
              2.200E-11
              1.400E-14
              3.000E-11

              1.400E-10
                                         3.100E-12
                                         4.200E-12

                                         1.400E-12
                                         6.100E-12
                                         4.200E-11
                                         7.400E-11
S2

S27
S13
S28
                        S6
0.

0.
0.
-9.400E+02
1.170E+02
-5.800E+02
2.000E4-02

-4.700E+02
     -1.900E+02
     -2.420E+02

     -2.000E+03
     -2.030E+03
     -3.500E+02
     0.
                             43

-------
39
402
412
42
43
44
45
46
47
48
49

50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69

70
71
72
73

74
75
762
772
782
792
802
812
82
OH + NO2 = HNO3
OH + HNO3 = H2O + NO + O2
OH + HN03 = H20 + N02 + O
NO + H02 = N02 + OH
N2O5 + H2O = 2 HNO3
HO2 + NO2 = HNO3
HNO4 = HO2 + N02
OH + HNO4 = H2O + NO2 + O2
O + HNO4 = OH + NO2 + O2
HONO + OH = H2O + NO2
OH + NO = HONO

CO + OH = H  + CO2
OH + CH4 = CH3 + H2O
O + CH4 = CH3 + OH
O(1D) + CH4 = CH2O + H2
OJ'D) + CH4 = CH3 + OH
Cl + CH4 = CH3 -I- HC1
CH3 +  O = CH2O + H
CH3 +  O2 = CH3O2
CH3O2 + HO2 = CH3OOH + O2
CH302 + O = CH30 + 02
CH3O2 + NO = CH3O + NO2
CH3O + O2 = CH2O + HO2
OH + CH2O =  HCO + H2O
O + CH20 = HCO + OH
Cl + CH2O  = HC1 + HCO
HCO + O2 = CO + HO2
OH + CH3OOH = CH3O2 + H2O
OH + CH3OOH = CH2O -f H2O + OH
CH3C1 + Cl = H02 + CO + 2 HC1
CH3C1 + OH = Cl + H20 + H02

Cl + 03 = CIO -f 02
CIO + O = Cl + O2
HC1 + 0(1D) = Cl + OH
HC1 + O = Cl + OH

NO + CIO = NO2 + Cl
CIO + NO2  = C1ONO2
C1ONO2 + O = CIO + NO + O2
C10NO2 + O = CIO + NO2 + O
OH + C1ONO2 = HOCI + NO2 + O
OH + C1ONO2 = HOCI + NO + O2
Cl + C10NO2 = 2 Cl + NO2 + O
Cl + C1ONO2 = 2 Cl + NO + O2
Cl + HNO4 = HC1 + NO2 + O2
          S3
9.400E-15
9.400E-15
3.700E-12
5.000E-22
5.100E-13
7.000E-11
6.600E-12
          S14
          S15
          S18

          S7
      7.780E+02
      7.780E+02
      2.400E+02
      0.
     6.900E+02
     -3.370E+03
     0.
2.400E-12
3.500E-11
1.400E-11
1.400E-10
9.600E-12
1.100E-10
S16
7.700E-14
3.000E-11
4.200E-12
1.200E-13
l.OOOE-11
3.000E-11
8.200E-11
3.500E-12
5.900E-12
4.100E-12
3.400E-11
1.800E-12
2.800E-11
6.000E-11
1.400E-10
l.OOOE-11
-1.710E+03
-4.550E+03
0.
0.
-1.350E+03
0.

1.300E+03
0.
1.800E+02
-1.350E+03
0.
-1.550E+03
-3.400E+01
1.400E+02
0.
0.
-1.260E+03
-1.112E+03
-2.570E+02
-l.OOOE+02
0.
-3.340E+03
6.200E-12

3.000E-12
3.000E-12
1.200E-12
1.200E-12
6.800E-12
6.800E-12
3.000E-12
S9
2.940E+02

-8.080E+02
-8.080E+02
-3.330E+02
-3.330E+02
1.690E+02
1.690E+02
-3.000E+02
                                   44

-------
83

84
85
86
87
88
89
90
91
92
93

94
95
96
97
98
99
100
101
102
103
104
105
106
107

1083
1093
HO3
1113
1123
1133
1143
1153
1163
1173
1183
1193
Cl + NO2 = C1NO2

OH + HC1 = H20 + Cl
Cl + H02 = HC1 + 02
Cl + H02 = OH + CIO
Cl + H2 = HC1 + H
Cl + H202 = HC1 + HO2
Cl + HOC1 = OH + 2 Cl
CIO + OH = HO2 + Cl
CIO + H02 = O2 + HOC1
O -f HOC1 + OH + CIO
OH + HOC1 = H20 + CIO

CH3Br + OH = Br + H2O
C2H4Br2 + OH = 2 Br  + H2O
Br + O3 = BrO + O2
BrO + O = Br + O2
BrO + BrO = 2 Br + O2
BrO + CIO = Br + Cl + O2
BrO + NO = Br + NO2
BrO + HO2 = HOBr +  O2
BrO + OH = Br + HO2
Br + HO2 = HBr + O2
Br + CH2O = HBr + HCO
OH + HBr = Br + H2O
O + HBr = Br + OH
BrO + NO2 = BrONO2

CFC13 -f O(1D) = 3 Cl
CF2C12 + O^D) = 2 Cl
F13 + O(JD)  = Cl
F112 + O(1D) = 4 Cl
F113 + O^D) = 3 Cl
F114 + O(1D) = 2 Cl
F115 + O^D) = Cl
CC14 -(- 0(JD) = 4 Cl
OH + F142B  =  Cl
OH + CH3CC13 = H2O  + 3 Cl
OH + F21 = 2 Cl
OH + F22 = Cl + H2O
                                                  S8
3.100E-12
1.800E-11
4.100E-11
3.700E-11
1.100E-11
3.000E-12
9.200E-12
4.600E-13
l.OOOE-11
3.000E-12
6.100E-13
7.300E-12
1.400E-11
3.000E-11
1.140E-12
6.700E-12
8.700E-12
5.000E-12
1.200E-11
2.200E-13
1.700E-11
8.000E-12
6.600E-12
S29
2.300E-10
1.400E-10
l.OOOE-10
3.000E-10
2.750E-10
1.620E-10
l.OOOE-10
3.300E-10
1.500E-12
5.400E-12
8.900E-13
7.800E-13
-4.000E+02
1.700E+02
-4.500E+02
-2.300E+03
-9.800E+02
-1.300E+02
6.600E+01
7.100E+02
-2.200E+03
-1.500E+02
-8.250E+02
-l.OOOE+03
-7.550E-I-02
0.
2.550E+02
0.
2.650E+02
0.
0.
0.
-8.000E+02
0.
-1.540E+03

0.
0.
0.
0.
0.
0.
0.
0.
-1.800E+03
-1.820E+03
-1.013E+03
-1.530E+03
Photolytic processes:
1
2
3
O2 +hv = O + O
O3 + hv = O + O2
03 + hv = 0(!D)
                  0
                                    45

-------
4        NO2 + hv = NO + O
5        N2O + hv = N2 + O(1D)
6        NO + hv = N + O
7        HNO3 + hv = OH + NO2
82       N2O5 + hv = 2 NO2 + O
92       N2O5 + hv = NO2 + NO + O2
10       NO3 + hv = NO + O2
11       NO3 + hv = NO2 + O
122      HNO4 -(- hv = OH + NO -I- O2
132      HNO4 + hv = OH + NO2 + O
14       HONO + hv = OH + NO

15       H202 + hv = 2 OH
16       H02 + hv = OH + O
17       H2O + hv = H +  OH

182      C1ONO2 + hv = Cl + NO + O2
192      C1ONO2 + hv = Cl + NO2 + O
20       HC1 + hv = H + Cl
21       CIO + hv = Cl +  O
22       CIO + hv = Cl +  O(1D)
23       C1NO2  + hv = Cl + NO2
24       HOC1 + hv - OH + Cl

25       CH3OOH + hv =  CH3O + OH
26       CH20 + hv = HCO + H
27       CH2O + hv = CO + H2

28       CH3C1 + hv = CH3 + Cl
293      CFC13 + hv = 3 Cl
303      CF2C12 + hv = 2 Cl
313      CC14 + hv = 4 Cl
323      CH3CC13  + hv = 3 Cl
333      F21 + hv = 2 Cl
343      F13 + hv = Cl
353      F112 + hv = 4 Cl
363      F142B + hv = Cl
373      F113 + hv = 3 Cl
383      F114 -f hv = 2 Cl
393      F115 + hv = Cl
403      F22 + hv = Cl

41       CFSBr  + hv = Br
423      CF2ClBr + hv = Br + Cl
43       BrO + hv  = Br -f O
44       HOBr + hv - Br + OH
452      BrONO2 + hv = Br + NO2 + O
462      BrONO2 + hv = Br + NO + O2
                                    46

-------
Special kinetic rate expressions:

SI: O + O2 = O3
    k=6.0E-34*M* (300/T) * *2.3

S2: H + O2 = H02
    k=5.5E-32*M* (300/T)** 1.6

S3: OH + NO2 = HNO3
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=2.6E-30*M'i (300/T) * *3.2
    B=2.4E-11*(300/T)**1.3

S4: N2 + O^D) = N2O
    k=3.5E-37*M*(300/T)**0.6

S5: NO + O = NO2
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=1.2E-31*M*(300/T)**1.8
    B=3.0E-11*(300/T)**0.

S6: OH + OH = H2O2
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=6.9E-31*M* (300/T) * *0.8
    B=1.0E-11*(300/T)**1.0

S7: CO + OH = H + CO2
    k=1.50E-13*(l+M*2.40E-20)

S8: Cl +  NO2 = C1NO2
    k=(A/(H-A/B))*0.6**(l/(H-[loglO(A/B)]**2))
    A=1.6E-30*M* (300/T) **2.0
    B=1.0E-10*(300/T)**1.0

S9: CIO + NO2 = C1ONO2
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=1.8E-31*M*(300/T)**3.4
    B=1.5E-11*(300/T)**1.9

S10: NO2 + O  =  NO + O2 (actual product NO3, see note 2 above)
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=9.0E-32*M* (300/T) * *2.0
    B=2.2E-11*(300/T)**0.

Sll: N02 + N03  = N205
    k=(A/(l+A/B))*0.6**(l/(H-[loglO(A/B)]**2))
    A=2.0E-30*M* (300/T) * *4.4
    B=1.4E-12*(300/T)**0.5

S12: N205 = N02 +  NO3
                                     47

-------
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
A=1.88E-3*exp(-11001/T)*M*(300/T)**4.4
B=1.32E+15*exp(-11001/T)*(300/T)**0.5

S13: H02 + H02 + H2O = H2O2 + O2 + H2O
     k=1.77E-36*exp(4320/T)+M*1.31E-56*exp(4730/T)

S14: HO2 + NO2 = HNO4
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**?))
     A=2.3E-31*M*(300/T)**4.6
     B=4.2E-12*(300/T)**0.

S15: HNO4 = HO2 + NO2
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
     A=9.87E-5*exp(-10870/T)*M*(300/T)**4.6
     B=1.8E+15*exp(-10870/T)*(300/T)**0.

S16: CH3 + 02 = CH302
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)J**2))
     A=2.2E-31*M*(300/T)**2.2
     B=2.0E-12*(300/T)**1.7

S18: OH + NO = HONO
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
     A=7.0E-31*M*(300/T)**2.6
     B=1.5E-11*(300/T)**0.5

S27: HO2 + HO2 = H2O2 + O2
     k=2.3E-13*exp(590./T)+1.7E-33*M*exp(1000./T)

S28: HO2 + OH = H2O + O2
     k=7.0E-ll+(1.57E-30)*M

S29: BrO + NO2 = BrONO2
     k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
     A=5.0E-31*M*(300/T)**2.0
     B=1.0E-11*(300/T)**1.0

1. JPL 83-82.
2. The primary reaction product NO3 photolyzes quickly to form the products
     NO2 + O or NO -f O2. These products are included as primary products for
     computational reasons involving proper diurnal averaging and the
     branching ratio for NO3 photolysis is included with the rate constant.
3. Not an elementary process.
                                      48

-------
                                  TABLE 2
                   Changes to kinetic data set in 1985 update1
11   N + NO = N2 + O
14   NO + O + M = NO2 + M
16   NO + NO3 = 2 NO2
18   N02 + N03 = N205

31   HO2 + OH = H2O -I- O2
38   H + HO2 = 2 OH

402  OH + HN03 = H20 + NO + O2
412  OH + HNO3 = H2O + NO2 + O
44   HO2 + NO2 = HNO4
46   OH + HNO4 = H20 + NO2 + O2
                            k = A * exp ( B / T )
                                                A
                                            3.400E-11

                                            1.300E-11
57   CH3 -f O2 = CH3O2
61   CH30 + O2 = CH2O
                          HO2
71   CIO + O = Cl + O2
72   HC1 + O(1D) = Cl + OH

84   OH + HC1 = H2O + Cl
90   CIO + OH = HO2 + Cl

99   BrO + CIO = Br + Cl -I- O2
103  Br + HO2 =  HBr + O2
105  OH -I- HBr = Br + H2O
                                            5.200E-11
                                            1.300E-12
8.400E-14
           S5
           Sll

           S28


           S30
           S30
           S14
                                                       S16
B
0.

2.500E+02
5.000E+01
                  3.800E+02
-1.200E+03
4.700E-11
1.500E-10
2.600E-12
l.OOOE-11
1.340E-11
8.000E-13
1.100E-11
-5.000E+01
0.
-3.500E+02
1.200E+02
0.
0.
0.
1. JPL 85-37.
2. See note 2 in Table 1.

Special kinetic expressions:

S5:  NO + O  = N02
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=9.0E-32*M*(300/T)**1.5
    B=3.0E-11*(300/T)**0.

Sll: N02 + N03 = N205
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=2.2E-30*M*(300/T)**4.4
    B=1.5E-12* (300/T) * *0.5

S14: H02 + N02 = HNO4
    k=(A/(l+A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=2.0E-31*M*(300/T)**2.7
    B=4.2E-12*(300/T)**2.0
                                     49

-------
S16: CH3 + O2 = CH3O2
    k=(A/(H-A/B))*0.6**(l/(l+[loglO(A/B)]**2))
    A=4.5E-31*M*(300/T)**2.0
    B=1.8E-12*(300/T)**1.7

S28: H02 + OH = H2O + O2
    k=1.7E-ll*exp(416/T)-|-3.0E-31*M*exp(500/T)

S30: OH -f HNO3 = H2O + N63
    k=7.2E-15*exp(785/T)+A/(l+A/B)
    A=1.9E-33*M*exp(725/T)
    B=4.lE-16*exp(1440/T)
                                     50

-------
                                 APPENDIX A
EMISSION PROJECTIONS 1985-2075

CFC11 (CFC13) 106 kg/year

YEAR
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075

 a. Also used in scenarios 2A, 6, 7, 8, 9, 14, 21, 22, 24, 25 and 27.
 b. Also used in scenario 11.
 c. Also used in scenarios 10 and 20.
 d. Also used in scenario 26.
A
325
426
544
699
842
986
1122
1263
1411
1559
1730
1900
2106
2313
2528
2744
2976
3207
3456
1A°
325
428
553
717
870
1024
1173
1322
1480
1637
1826
2006
2237
2468
2702
2937
3188
3440
3711
4"
324
422
476
522
558
594
626
657
691
725
762
798
837
876
918
960
1006
1053
1104
5C
594
1184
1486
2428
5456
8485
17422
26358
35841
45323
4766L
50009
51790
53570
54795
56020
57044
58068
69555
B
354
464
559
673
761
849
931
1012
1104
1197
1305
1413
1544
1674
1804
1935
2077
2219
2373
C
490
845
L064
1626
3210
4794
9328
13862
18679
23496
24757
26018
27020
28021
28743
29464
30095
30726
36598
12d
325
348
446
598
738
878
1009
1141
1276
1410
1565
1720
1918
2115
2320
2524
2741
2958
3190
13
325
427
553
684
743
883
1014
1146
1280
1415
1570
1725
1922
2120
2324
2529
2746
2962
3194
CFC11 (CFC13) 106 kg/year

YEAR        18      19      28
29
30e
31
D
E/
                                       17
                                      325
                                      426
                                      514
                                      578
                                      624
                                      670
                                      689
                                      708
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
                                      717
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
324
395
443
494
532
570
588
606
610
615
325
426
540
649
728
808
856
904
928
955
326
427
545
671
765
858
915
972
1014
1055
326
357
444
560
642
723
838
852
901
950
283
333
400
445
454
464
464
464
464
464
217
250
283
333
379
425
500
575
672
770
325
426
544
230
282
334
384
434
486
537
107
142
181
230
282
334
384
434
486
537
107
142
164
181
194
208
219
230
247
254
                                      51

-------
2035
2040
2045
2050
2055
2060
2065
2070
2075         616    1069    1432    1391      464     2990    1203    1203     387

 e. Also used in scenario 32.
 f. Also used in scenario F.
616
616
616
616
616
616
616
616
986
1018
1039
1060
1064
1069
1069
1069
1094
1133
1181
1229
1274
1320
1358
1397
994
1038
1095
1152
1203
1258
1304
1350
464
464
464
464
464
464
464
464
895
1019
1264
1410
1658
1905
2238
2570
596
654
726
797
874
952
1034
1116
596
654
726
797
874
952
1034
1116
267
280
294
308
322
337
354
370
                                        52

-------
CFC12 (CF2C12) 106 kg/year
YEAR         A     1A°      46      5C      B       C     12d      13      17
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
449
528
623
734
844
954
1074
1193
1331
1468
1634
1799
1988
2177
2368
2559
2766
2972
3198
449
528
625
746
870
995
1130
1266
1415
1564
1742
1919
2131
2342
2556
2770
2998
3226
3472
446
519
573
611
647
683
723
763
808
853
902
951
1004
1058
1118
1177
1245
1312
1386
527
772
1042
1406
1997
2587
3483
4378
5219
6058
6499
6940
7299
7658
7890
8122
8278
8434
8547
468
547
623
701
774
847
927
1007
1100
1192
1304
1416
1540
1663
1789
1914
2054
2193
2346
509
674
857
1099
1349
1799
2306
2814
3305
3796
4104
4411
4687
4963
5175
5387
5571
5754
5927
449
387
427
516
606
697
792
887
990
1094
1212
1331
1478
1624
1776
1927
2090
2253
2427
449
528
625
660
619
710
805
900
1004
1107
1226
1344
1491
1637
1788
1940
2103
2266
2440
449
528
594
617
631
645
650
656
656
657
657
657
657
657
657
657
657
657
657
 a. Also used in scenarios 2A, 6, 7, 8, 9, 14, 21, 22, 24, 25 and 27.
 b. Also used in scenario 11.
 c. Also used in scenarios 10 and 20.
 d. Also used in scenario 26.
CFC12 (CF2C12) 106 kg/year
YEAR         18      19      28      29      30e      31       D      E'       G
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
446
491
519
534
544
554
559
564
564
565
565
565
565
449
528
622
716
787
858
915
972
1009
1046
1100
1153
1183
445
529
624
729
808
887
956
1024
1096
1167
1232
1296
1350
450
396
430
504
550
597
636
676
718
761
802
856
908
421
490
638
638
638
638
638
638
638
638
638
638
638
330
360
421
490
564
638
724
810
955
1100
1278
1447
1664
449
528
623
367
406
445
498
551
614
678
754
831
918
227
248
290
340
391
442
496
551
614
678
754
831
918
226
244
267
283
300
316
334
352
373
394
416
439
464
                                       53

-------
2050         565    1213    1404     959     638     1880    1005    1005     488
2055         565    1215    1458    1009     638     2170    1094    1094     514
2060         565    1217    1512    1059     638     2460    1184    1184     543
2065         565    1217    1555    1110     638     2843    1280    1280     574
2070         565    1217    1598    1161     638     3225    1376    1376     605
2075         565    1217    1643    1213     638     3690    1479    1479     639

 e. Also used in scenario 32.
 f. Also used in scenario F.
                                       54

-------
CC14 106 kg/year
YEAR        A"     1A6      4C     30d
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
88
90
105
125
149
173
196
219
243
260
277
295
312
329
379
429
479
529
579
153
188
206
226
250
275
300
325
350
384
419
454
489
524
576
628
680
732
784
131
41
45
49
54
59
64
70
76
83
91
99
107
115
126
137
148
160
172
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
 a.  Also used in scenarios B, C, D, E and F.
 b.  Also used in scenarios 2A, 5-10, 12-14, 17-22 and 24-29.
 c.  Also used in scenarios 11 and G.
 d.  Also used in scenarios 31 and 32.
CH3CC13 (methyl chloroform) 106 kg/year
YEAR        Aa     1A6     30C
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
510
569
635
708
793
878
964
1050
1136
1257
1378
1500
1622
510
569
634
708
793
878
963
1049
1135
1256
1377
1499
1621
500
500
500
500
500
500
500
500
500
500
500
500
500
                                       55

-------
2050        1744    1743     500
2055        1901    1922     500
2060        2058    2101     500
2065        2215    2280     500
2070        2372    2459     500
2075        2531    2638     500

 a. Also used in scenarios B-F.
 b. Also used in scenarios 2A, 4-14, 17-22, 24-29 and G.
 c. Also used in scenarios 31 and 32.
                                        56

-------
CFC113 (CF2C1CFC12) 106 kg/year
YEAR        A°     1A6      4C     30d       D      Ee       G
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
102
131
189
274
307
341
374
408
441
489
536
584
631
679
740
800
861
921
982
102
142
210
277
311
344
378
411
445
493
540
588
635
683
753
823
893
963
1033
102
142
157
180
202
224
245
267
289
320
351
381
412
443
489
534
580
625
671
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
102
131
189
104
117
129
142
154
167
185
203
220
238
256
282
309
335
362
3388
43
53
79
104
117
129
142
154
167
185
203
220
238
256
282
309
335
362
252
43
53
60
68
76
84
92
100
108
120
131
143
154
166
183
200
218
235

 a.  Also used in scenarios B and C.
 b.  Also used in scenarios 2A, 5-10, 12-14, 17-22 and 24-29.
 c.  Also used in scenario 11.
 d.  Also used in scenarios 31 and 32.
 e.  Also used in scenario F.
CFC22 (CHF2C1) 106 kg/year
YEAR       1A°       46       5C      30d
1984
1989
1994
1999
2004
2009
2014
2019
2024
2029
2034
2039
52
84
122
167
221
273
332
394
463
541
626
715
54
71
89
107
127
137
148
155
163
170
178
187
59
101
156
226
321
430
564
719
900
1028
1130
1218
0
0
0
0
0
0
0
0
0
0
0
0
                                        57

-------
2044
2049
2054
2059
2064
2069
2074
2079
807
900
998
1097
1195
1292
1387
1495
196
205
216
226
238
250
263
276
1287
1339
1380
1412
1438
1461
1480
1493
0
0
0
0
0
0
0
0
a.  Also used in scenarios 2A, 6-9, 12-14, 17-19, 21, 22, 24-29 and A-F.
b.  Also used in scenarios 11 and G.
c.  Also used in scenarios 10 and 20.
d.  Also used in scenarios 31 and 32.
                                         58

-------
HALON 1301 (CF3Br) 106 kg/year
YEAR       A°     1A6    2AC
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
3
4
6
9
12
15
17
20
23
25
28
30
33
35
39
43
46
50
54
2
3
4
6
9
12
16
19
22
25
29
32
36
39
43
47
51
55
59
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
2
2
3
4
5
7
9
11
12
14
16
18
20
21
23
25
27
29
HALON 1211 (CF2BrCl) 106 kg/year
YEAR      Aa    1A6     2AC     4d
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
1
1
1
2
2
3
3
4
5
5
5
6
6
7
7
8
9
10
11
0
1
1
1
1
2
3
4
5
5
6
7
8
9
9
10
11
12
13
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
2
2
2
2
3
3
4
4
4
5
5
6
7
                                   59

-------
a. Also used in scenarios B-F.
b. Also used in scenarios 5-10, 12-14, 17-22 and 26-29.
c. Also used in scenarios 24, 25 and 30-32.
d. Also used in scenarios 11 and G.
                                         60

-------
  CO2 - surface mole fraction in ppm
                            CH4 - surface mole fraction in ppm
YEAR
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
   Aa
344.5
352.9
362.0
372.0
382.8
394.6
407.4
421.4
436.4
452.7
470.3
489.4
509.9
532.0
555.9
581.5
609.1
638.8
670.7
   14
345.0
354.1
363.3
372.4
381.5
390.7
399.8
408.9
418.1
427.3
436.5
445.6
454.8
464.0
473.2
482.4
491.7
500.9
510.1
  306
348.6
357.4
366.4
375.7
385.2
394.9
404.8
415.1
425.6
436.3
447.3
458.6
470.2
482.1
494.2
506.7
519.5
532.6
546.1
A6,c
1.756
1.845
1.939
2.038
2.142
2.251
2.366
2.487
2.614
2.747
2.887
3.035
3.189
3.352
3.523
3.703
3.892
4.090
4.299
 6
1.756
1.845
1.892
1.939
1.988
2.039
2.090
2.143
2.197
2.252
2.309
2.368
2.427
2.489
2.551
2.616
2.682
2.750
2.819
 a.  Also used in all scenarios other than 14 and 30-32.
 b.  Also used in scenarios 1A, 2A, 4, 5, 12-14, 17-19, 21, 22, 24-26, 28-32, B-E and G.
 c.  Other scenarios used prescribed flux boundary conditions.
    6: 1% increase per year to 1990, then 0.5% increase per year to 2075
    7: 1% increase per year to 2010, then constant surface flux to 2075
    8: 1% increase per year to 2010, then 2% increase per year to 2075
    9 and 27: 1% increase per year to 2010, then -1% per year to
         2030, then constant surface flux to 2075.
    10: 0.5% increase per year
    11 and 20: 1.5% increase per year
    F: 1.25% increase per year.
                                         61

-------
N2O - surface mole fraction in ppb.
YEAR     Afl 306
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
303.1
306.9
310.8
314.7
318.6
322.6
326.7
330.8
334.9
339.2
343.4
347.7
352.1
356.5
361.0
365.5
370.1
374.8
379.5
304.0
307.1
310.1
313.3
316.4
319.6
322.8
326.0
329.3
332.6
336.0
339.3
342.7
346.2
349.6
353.2
356.7
360.3
363.9
 a. Used in all scenarios other than 30-32.
 b. Also used in scenarios 31 and 32.
                                          62

-------
REFERENCE CASE AND SCENARIO SENSITIVITY RUNS
YEAR
A1
                     1A
41
                         B
   SCENARIO IDENTIFIER
C     5     51     2A     6
                                                                                              10
                                                                                      11
                                                                          14
20
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
0.0
-0.07
-0.18
-0.35
-0.61
-0.96
-1.42
-1.99
-2.65
-3.42
-4.30
-5.29
-6.41
-7.67
-9.13
-10.84
-12.90
-15.58
-19.94
0.0
-0.16
-0.37
-0.66
-1.06
-1.61
-2.31
-3.17
-4.18
-5.35
-6.70
-8.23
-9.99
-12.01
-14.35
-17.19
-20.75
-25.47
-31.79
0.0
-0.06
-0.21
-0.41
-0.69
-1.08
-1.57
-2.19
-2.91
-3.75
-4.73
-5.85
-7.15
-8.69
-10.48
-12.67
-15.52
-19.64
-26.15
0.0
-0.07
-0.16
-0.27
-0.40
-0.55
-0.71
-0.88
-1.07
-1.26
-1.47
-1.69
-1.94
-2.19
-2.46
-2.76
-3.09
-3.43
-3.76
0.0
-0.16
-0.34
-0.55
-0.78
-1.03
-1.32
-1.62
-1.95
-2.31
-2.69
-3.11
-3.57
-4.06
-4.59
-5.17
-5.81
-6.52
-7.28
0.0
-0.09
-0.28
-0.62
-1.18
-2.21
-4.05
-7.76
-16.91
-51.48
-67.19
-71.64
-73.69
-74.72
-75.35
-75.78
-
-
-
0.0
-0.07
-0.18
-0.35
-0.50
-0.61
-0.73
-0.87
-1.05
-1.27
-1.53
-1.84
-2.19
-2.58
-3.02
-3.52
-4.09
-4.75
-5.47
0.0
-0.10
-0.38
-0.91
-1.82
-3.66
-7.47
-19.54
-60.98
-
-
-
-
-
-
-
-
-
_
0.0
-0.20
-0.66
-1.46
-2.80
-5.44
-10.86
-29.53
-62.75
-
-
-
-
-
-
-
-
-
—
0.0
-0.06
-0.20
-0.37
-0.63
-0.95
-1.35
-1.84
-2.39
-3.01
-3.72
-4.51
-5.42
-6.46
-7.68
-9.18
-11.26
-14.70
-22.62
0.0
-0.06
-0.22
-0.51
-0.92
-1.44
-2.10
-2.89
-3.82
-4.89
-6.13
-7.59
-9.27
-11.27
-13.73
-16.94
-21.57
-29.20
-39.80
0.0
-0.04
-0.11
-0.24
-0.45
-0.75
-1.19
-1.81
-2.63
-3.66
-4.91
-6.41
-8.22
-10.42
-13.20
-16.96
-22.78
-33.53
-48.63
0.0
-0.04
-0.11
-0.24
-0.45
-0.75
-1.14
-1.54
-1.96
-2.39
-2.83
-3.30
-3.77
-4.24
-4.79
-5.44
-6.22
-7.17
-8.34
0.0
-0.04
-0.11
-0.24
-0.45
-0.75
-1.21
-1.93
-2.94
-4.24
-5.83
-7.70
-9.89
-12.56
-16.00
-20.92
-29.62
-45.79
-61.86
0.0
-0.07
-0.33
-0.87
-1.82
-3.74
-7.77
-21.25
-65.71
-79.06
-85.68
-
-
-
-
-
-
-
_
0.0
-0.02
-0.01
0.01
0.06
0.14
0.23
0.35
0.48
0.65
0.83
.1.03
1.28
1.59
1.92
2.27
2.62
2.98
3.35
0.0
-0.06
-0.19
-0.40
-0.70
-1.12
-1.66
-2.32
-3.11
-4.01
-5.05
-6.23
-7.59
-9.16
-10;98
-13.17
-15.88
-19.89
-25.08
0.0
-0.05
-0.23
-0.61
-1.32
-2.87
-6.20
16.07
-55.82
-72.44
-79.00
-82.85
-
-
-
-
-
-
-
1. Scenario run with kinetic and spectral parameters updated to recommendations of JPL 85-37.

-------
MODEL PARAMETER SENSITIVITY RUNS AND BRASSEUR SCENARIOS
          A1
21
       22
24
       25
26
 SCENARIO IDENTIFIER
27     28      29     30
                                     31
                                                                         32
A3
A4
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
0.0
-0.23
-0.54
-0.93
-1.43
-2.06
-2.82
-3.74
-4.80
-6.02
-7.41
-8.98
-10.76
-12.79
-15.12
-17.84
-21.09
-24.94
-29.42
0.0
-0.01
-0.08
-0.20
-0.40
-0.67
-1.03
-1.49
-2.03
-2.66
-3.37
-4.18
-5.10
-6.14
-7.33
-8.66
-10.26
-12.23
-14.81
0.0
-0.13
-0.33
-0.60
-0.96
-1.43
-2.00
-2.67
-3.44
-4.32
-5.31
-6.47
-7.83
-9.49
-11.61
-14.69
-20.85
-31.37
—
0.0
-0.44
-0.99
-1.67
-2.54
-3.61
-4.91
-6.45
-8.29
-10.49
-13.26
-17.07
-23.46
-33.93
-41.62
-45.88
-
-
_
0.0
-0.14
-0.34
-0.58
-0.90
-1.32
-1.86
-2.52
-3.32
-4.24
-5.31
-6.56
-8.01
-9.70
-11.72
-14.21
-17.40
-21.81
-28.23
0.0
-0.08
-0.22
-0.44
-0.75
-1.19
-1.82
-2.75
-4.00
-5.61
-7.60
-10.02
-13.02
-17.00
-23.04
-35.26
-
-
—
0.0
-0.06
-0.19
-0.39
-0.67
-1.03
-1.47
-1.98
-2.55
-3.18
-3.86
-4.59
-5.38
-6.23
-7.13
-8.11
-9.16
-10.29
-11.53
0.0
-0.06
-0.18
-0.33
-0.53
-0.80
-1.13
-1.52
-1.97
-2.47
-3.02
-3.62
-4.28
-4.99
-5.75
-6.59
-7.51
-8.48
-9.52
0.0
0.01
-0.03
-0.06
-0.09
-0.12
-0.12
-0.10
-0.06
0.00
0.08
0.19
0.33
0.48
0.65
0.85
1.06
1.29
1.54
0.0
0.01
0.01
0.03
0.07
0.11
0.14
0.15
0.15
0.12
0.06
-0.05
-0.20
-0.41
-0.69
-1.06
-1.53
-2.09
-2.76
0.0
-0.00
-0.08
-0.18
-0.31
-0.46
-0.64
-0.83
-1.03
-1.24
-1.47
-1.70
-1.95
-2.21
-2.48
-2.76
-3.06
-3.37
-3.70
0.0
-0.07
-0.18
-0.35
-0.61
-0.96
-1.42
-1.99
-2.65
-3.42
-4.30
-5.29
-6.41
-7.67
-9.13
-10.84
-12.90
-15.58
-19.28
0.0
1.21
0.78
0.30
-0.27
-0.98
-1.84
-2.87
-4.07
-5.47
-7.12
-9.07
-11.44
-14.43
-18.36
-23.96
-32.38
-43.16
-51.95
0.0
-0.27
-0.59
-0.97
-1.45
-2.05
-2.79
-3.67
-4.69
-5.87
-7.23
-8.82
-10.72
-13.07
-16.17
-20.99
-31.66
-46.61
-54.36
0.0
-0.13
-0.28
-0.51
-0.86
-1.36
-2.01
-2.82
-3.80
-4.95
-6.30
-7.89
-9.81
-12.21
-15.42
-20.47
-32.17
-50.02
-60.54
0.0
-0.12
-0.23
-0.39
-0.65
-1.02
-1.52
-2.15
-2.92
-3.81
-4.86
-6.09
-7.54
-9.30
-11.50
-14.52
-19.46
-30.38
-43.80
0.0
-0.28
-0.69
-1.18
-1.76
-2.45
-3.25
-4.14
-5.14
-6.24
-7.44
-8.76
-10.22
-11.88
-13.80
-16.14
-19.27
-24.33
-35.83
1. Scenario A run without temperature feedback calculation.
2. Scenario A including increasing trends only in CFC compounds, fixed CO2, CH«, NjO and temperature profile.
3. As in note 2, but with temperature feedback.
4. As in note 3, but with fixed CH< surface flux rather than fixed surface concentration.
5. As in note 3, but with 0.5% annual increase in CH4 surface concentration.
6. As in note 3, but with 0.25% annual increase in N2O surface  concentration.

-------
REGULATORY SENSITIVITY RUNS

                                          SCENARIO IDENTIFIER
YEAR        12       13       17       18        19       28        29       D
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
ON 2040
Ul
2045
2050
2055
2060
2065
2070
2075
0.0
-0.08
-0.19
-0.35
-0.57
-0.86
-1.25
-1.74
-2.33
-3.01
-3.79
-4.70
-5.73
-6.90
-8.26
-9.86
-11.71
-13.80
-17.17
0.0
-0.08
-0.20
-0.40
-0.68
-1.02
-1.44
-1.94
-2.52
-3.21
-3.99
-4.90
-5.92
-7.11
-8.47
-10.07
-11.95
-14.29
-17.33
0.0
-0.06
-0.19
-0.38
-0.63
-0.93
-1.27
-1.66
-2.07
-2.50
-2.95
-3.42
-3.92
-4.43
-4.97
-5.53
-6.14
-6.78
-7.43
0.0
-0.06
-0.18
-0.35
-0.56
-0.82
-1.10
-1.43
-1.78
-2.15
-2.55
-2.97
-3.41
-3.88
-4.37
-4.89
-5.45
-6.06
-6.67
0.0
-0.06
-0.19
-0.39
-0.66
-1.01
-1.43
-1.91
-2.45
-3.04
-3.66
-4.32
-5.04
-5.79
-6.58
-7.41
-8.29
-9.18
-10.13
0.0
-0.06
-0.18
-0.33
-0.53
-0.80
-1.13
-1.52
-1.97
-2.47
-3.02
-3.62
-4.28
-4.99
-5.75
-6,59
-7.51
-8.48
-9.52
0.0
0.01
-0.03
-0.06
-0.09
-0.12
-0.12
-0.10
-0.06
0.00
0.08
0.19
0.33
0.48
0.65
0.85
1.06
1.29
1.54
0.0
-0.07
-0.18
-0.35
-0.50
-0.61
-0.73
-0.87
-1.05
-1.27
-1.53
-1.84
-2.19
-2.58
-3.02
-3.52
-4.09
-4.75
-5.47
0.0
-0.06
-0.07
-0.07
-0.10
-0.16
-0.27
-0.42
-0.61
-0.84
-1.12
-1.45
-1.82
-2.23
-2.68
-3.20
-3.79
-4.46
-5.20
0.0
-0.01
0.05
0.15
0.25
0.34
0.40
0.44
0.46
0.46
0.44
0.40
0.33
0.24
0.17
0.09
-0.03
-0.21
-0.43
0.0
-0.06
-0.05
-0.01
0.04
0.10
0.16
0.21
0.26
0.31
0.35
0.37
0.39
0.40
0.40
0.39
0.36
0.31
0.24

-------
                                 APPENDIX C

Surface mole fractions in Reference Run (Scenario A)
YEAR

1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
Fll

0.209
0.274
0.358
0.465
0.596
0.749
0.922
1.113
1.322
1.547
1.790
2.051
2.332
2.636
2.961
3.304
3.665
4.040
4.425
F12

0.378
0.481
0.601
0.741
0.904
1.071
1.291
1.517
1.765
2.037
2.335
2.662
3.019
3.407
3.826
4.272
4.751
5.253
5.785
CC14 CH3CC13F113

0.161
0.166
0.173
0.182
0.194
0.211
0.230
0.252
0.277
0.304
0.331
0.360
0.389
0.419
0.452
0.490
0.534
0.583
0.634
ppb
0.137
0.168
0.197
0.226
0.257
0.290
0.325
0.361
0.398
0.438
0.483
0.532
0.582
0.633
0.688
0.746
0.807
0.866
0.919

0.032
0.048
0.070
0.101
0.140
0.182
0.228
0.275
0.325
0.379
0.437
0.498
0.563
0.631
0.703
0.779
0.859
0.943
1.027
F22

0.044
0.059
0.080
0.109
0.148
0.196
0.251
0.314
0.386
0.467
0.559
0.660
0.770
0.888
1.013
1.145
1.282
1.421
1.558
1301

0.
1.40
3.23
5.93
9.63
14.26
19.74
26.1
33.2
41.0
49.2
58.1
67.3
77.0
87.3
98.5
110.4
123.0
136.1
1211
ppt
0.
0.32
0.54
0.84
1.31
1.83
2.38
2.95
3.53
4.09
4.59
5.04
5.46
5.85
6.26
6.74
7.26
7.77
8.20
CH4 NO2
ppm
1
1
1
2
2
2
2
2
2
2
2
3
3
3
3
3
3
4
4
.755
.84
.940
.038
.143
.251
.367
.487
.614
.747
.888
.035
.189
.351
.523
.703
.892
.091
.298
ppb@35
19.9








21.4








23.6
Note: Surface mole fractions for N2O and CO2 are specified as boundary conditions and
are found in Appendix A.
                                      66

-------
Figures
 1. LLNL 1-D eddy diffusion coefficient profile.
 2. Historical record  of calculated ozone depletion at steady state for constant current
    CFG emission fltlxes.
 3. Calculated ozone  column change, 1850-1985, referenced to 1985 model calculated col-
    umn.
 4. Selected vertical profiles of local ozone change, 1850-1985: (a) Absolute concentration
    change, (b) Relative change.
 5. Comparison of model results of Wang et al (1986) to statistically derived decadal trend
    of local ozone from the Umkehr observational network, 1970-1980 (from WMO, 1986).
 6. Calculated direct  radiative equilibrium surface temperature change, 1850-1985. Short
    dashed line is  CO2 contribution, long dashed line is CO2+CH4 contribution.
 7. 1985 ozone vertical profile compared to observations.
 8. 1985 vertical mole fraction profiles compared to observations:  (a) NaO, (b) CH4 and
    H2O, (c) CFC11 and CFC12.
 9. 1985 vertical mole fraction profiles compared to observations: (a)  NO and NC>2, (b)
    HNO3, C1ONO2 and HNO4.
10. 1985 vertical mole fraction profiles compared to observations: (a) CIO, (b) HC1.
11. 1985 vertical temperature profile compared to US Standard Atmosphere (1976).
12. Rates of odd oxygen production and loss terms by family as a function of altitude.
13. Reference case CFCll and CFC12 projected emissions.
14. Reference case projected emissions of other halocarbons.
15. Low case projected halocarbon emissions: (a) CFCll and CFC12 and (b) other halo-
    carbons.
16. High case projected halocarbon emissions: (a) CFCll and CFC12 and (b) other halo-
    carbons.
17. Reference case column ozone change vs.  time (solid line)  and reference case emissions
    with updated chemistry (dashed line).
18. Selected profiles of local ozone concentration change from reference case: (a) absolute
    concentration change and  (b) relative change.
19. Increase in equilibrium radiative surface temperature with time for the reference case
    (solid line) and including climate feedback amplification (dashed lines).
20. Selected profiles of stratospheric temperature change from reference case.
21. High and low case column ozone changes with time (solid lines) and range of ozone
    change over  all scenarios (dashed lines).
22. Relative change in local ozone abundance at 40 km from selected scenarios.
                                         67

-------
23. Total column ozone change with time for selected regulatory scenarios.
24. Total column ozone change with time for scenarios with monotonic methane surface
    flux boundary conditions (dashed lines) compared to the reference case (solid line).
25. Total column ozone change with time for scenarios with variable  methane surface
    flux boundary conditions  (dashed lines) compared to reference cases with specified
    concentration CH4 boundary conditions (solid line).
26. Total column ozone change with time for high case  CFC emissions and alternative
    methane scenarios 5, 10, 20.
27. Total column ozone change with time for low case  CFC emissions and alternative
    methane scenarios 4 and 11.
28. Total column ozone change with time for alternative Halon emission scenarios 1A and
    2A.
29. Total column ozone change with time for scenarios with varying NOy abundances (1A,
    21, 22).
30. Changes in calculated local temperature at 45 km for several scenarios.
31. Radiative calculated changes in equilibrium surface temperature for several scenarios
    (solid lines) and maximum and minimum changes with  time  in surface temperature
    including climate feedback amplification.
32. Total column ozone change with time for reference case CFC-only scenario compared
    to full reference case.
33. Relative change in total column ozone vs. change in  C\z mole fraction for CFC-only
    reference case (solid line) and high case (scenario 5, dashed line).
34. Profiles of local ozone concentration change at selected times for CFC-only case.
                                                  %
35. Total column ozone change vs.  upper stratospheric C\z abundance for reference case
    and CFC-only case (solid lines) and for a CFC/CH4  (long dash) and CFC/N2O case
    (short dash).
36. Vertical profiles of C1O/C1Z for 1985 atmosphere (dotted line) CFC-only reference case
    (solid line)  and CFC/methane case (dashed line).
37. Total column ozone change with time for  CFC-only reference case  with and without
    temperature feedback and complete reference case with and without  temperature feed-
    back.
                                        68

-------
                              Figure 1
LLJ
Q
      55
      50
      45
      40
      35
      30
      25
      20
      15
     10
103
                .   .... ..I
                                                      .  .I.
                        104               105               108


               EDDY DIFFUSION COEFFICIENT Kz   ( cm2  s'1 )
                               69

-------
LLJ


-------
                         IL
OZONE COLUMN CHANGE RELATIVE TO 1 985   ( per cent )

-------
                                                   ALTITUDE    ( km )
SJ
        5
        o
        o
        M
        O
        2
        m

        O
        O
        m
        o.

        8
        o
         b>
                                                                                                           00

-------
                              Figure 4b
UJ
O




1
     30  -
25 -
     20  -
      15  -
     10  -
       5  -
       -15    -10
                   LOCAL OZONE CHANGE   ( per cent )
                                73

-------
                            UMKEHR LEVEL
to
->i
o
OB
o
o
m
o
>

§
3D
rn
2
D
o
CD

-------

-------
                               Figure 7
Ul
Q




§
    U.S. STANDARD ATMOSPHERE


+  WMO (1 982)
                        1011              1012


                  OZONE CONCENTRATION   ( molec cm'3)
                                76

-------
                             Figure 8a
      55
      50
      45
      40
      35
~    30
LLI

O
H

b
<
      25
      20
      15
      10
       10
         -9
                                   I  i  I  i i i
                          WMO (1 982)
                 i  i  i i  i i
                                   i  i  i i  i
                         10
                           -8
10
  -7
                                                         i  i i 11
10
  -6
                     N20 MOLE FRACTION   ( by volume )
                                   77

-------
                              Figure 8b
      55
     50
      45
      40
      35
-     30
LLJ
O
      25
      20
      15
      10
WMO (1 982)
                                                   i  i  i  i i
       10
         -7
              10
                -6
10
  -5
                    CH4 MOLE FRACTION   ( by volume )
                                78

-------
HI
Q

H


5
      50
                             Figure 8c
      55 r—i	1 i i 	1	1  IVI 	1	1  I I     I	1  I I |	1	T-TT
      45
      40
      35
-    30
      25
      20
      15
      10
                                      CFC-12
                     FABIAN (1 986)
            i   i  i i I   i    i  i i I   i   i  i  1
       ID'14      10-13      10-12       10'11      10-10       10'9



               CFC-11, CFC-1 2 MOLE FRACTIONS   ( by volume )
                                  79

-------
                             Figure 9a
      55
      50
      45
      40
      35
~    30
in
o
ID

F
25
      20
      15
      10
      10
        -12
    WMO (1 982)


N02 (sunset) WMO
                                               LOCAL NOON
                                              I	I
10
   -11
                      10
                        -10
10
  -9
10
  -8
10
  -7
                  NO, N02 MOLE FRACTIONS   ( by volume )
                                 80

-------
                            Figure 9b
      55
      50
      45
      40
      35
_*:
~     30
UJ
Q

I
25
      20
      15
      10
HN03 - WMO (1 982)
                                 ,  I
                                         , I
      10
        -12
              10
                ,-11
             10
               -10
10
  -9
10
  -8
                  HNO* CION02 MOLE FRACTIONS   ( by volume )
                                81

-------
                              Figure lOa
      55
      50
      45
      40
      35
~    30
UJ
o
      25
      20
      15
      10
      10
        -12
                    • I
                            WMO (1 986)

                            REEL-DOWN PROFILE  WMO (1 986)
              , I
10
  -11
10
  -10
10
  -9
10
  -8
                    CIO MOLE FRACTION   ( by volume )
                                  82

-------
                              Figure lOb
      55
      50
      45
      40
      35
~    30
                               i   i  i  i i i i i
UJ
Q
ID
      25
      20
      15
      10
       0
      10
                   	WMO (1 986)
                                                       i  i i i 11
-11
                10
-10
                10
-9
                10
-8
                     HCI MOLE FRACTION   ( by volume )
                                 83

-------
                             Figure 11
      55
     50
     45
     40
     35
-    30
LU
O




1
25
     20
     15
     10
      QUO.
                        	U.S. STANDARD ATMOSPHERE

                              (1 976)
        I  .   I  .   i  .   I  .   I  .   I
I
       200  210  220  230  240  250  260  270  280  290  300
                         TEMPERATURE   ( K )
                                 84

-------
00
    UJ
    o

    1
                                      \ LOSS  /ODD OXYGEN PRODUCTION
                                         X J
                                                                                        OQ
                                  REACTION RATE   ( molec ernes'1)

-------
                                98
                PROJECTED EMISSIONS   ( million kg )
  o
   to
         O
          CO
                        Y  '
                                                   I    I
(O
<0
O
ro
o
o
o
ro
o
ro
o
ro
o
ro
o
(*>
o
ro
o
ro
o
Ol
o
ro
o
o>
o
ro
o
\
                            A
                                . . 1

-------
PROJECTED ANNUAL EMISSIONS   ( million kg )

-------
                                88
           PROJECTED ANNUAL EMISSIONS   ( million kg )
  o
   ro

o
-  II
ro
o
ro
o
01
o
ro
o
O)
o
ro
o

-------
                                 68
            PROJECTED ANNUAL EMISSIONS    ( million kg )
  o
   i
o
 o
O
 G>
to
O
ro
o
o
o
IN)
o
ro
o
ro
O
ro
O
ro
o
ro
o
0>
o
ro
o

-------
                                    06
               PROJECTED ANNUAL EMISSIONS   ( million kg )
rn
>
33

-------
                                   T6
            PROJECTED ANNUAL EMISSIONS    ( million kg )
  o
   o
O
 o>
(O

O
ro
o
ro
o
01
o
ro
o
o>
o
ro
o
-j
o
                                .1

-------
8
o
o
§
d
o
     -5
    -10
    -15
    -20
    -25
    -30
                                      JPL 85-37^^
              REFERENCE CASE
1
ro

h-i
^J
          1990   2000   2010    2020    2030   2040    2050   2060   2070
                                      YEAR

-------
                             Figure 18a
LU
Q
     55
     50
     45
     40
     35
     30
     25
     20
      15
     10
              I  '   I   '   I   '   I   '  I
              80
        '  .   I   ,  I   ,   I   .      .      .
     i—i—r
.   I.I.
       -12  -10    -8   -6-4-20     2     4


            LOCAL CONCENTRATION CHANGE   1 0 11 ( molec cm"3)
                               93

-------
111
Q




1
      55
     50
      45
     40
      35
     30
25
     20
      15
     10
                              Figure 18b
          .  I  •  I  .  I  .  ]  1- p .  |  .  |  .   |  11  i  f
          .1.1.1  .  I  i 1  I  I  i  I  I   I  .   II.  M .   I  .
       -90 -80 -70  -60 -50 -40  -30 -20-10   0   10   20  30
                  LOCAL RELATIVE CHANGE   ( per cent )
                                 94

-------
    UJ
    CD
    O
Ln   III
    oc
    13
    (O
    a:
    CD
    13
                  REFERENCE CASE
                             FEEDBACK AMPUFICATION FACTOR - * 3.75
                                                                                      •S
              1990    2000   2010   2020    2030    2040    2050    2060   2070
                                            YEAR

-------
                                Figure 20
111
0




I
       -50     -40
-30
                   LOCAL TEMPERATURE CHANGE   ( K )
                               96

-------
8
0>
CL
LLJ

CD
O

LU


8
O
O
O
SCENARIOS 4 AND 5 - JPL 83-62

SCENARIOS 4 AND 5 - JPL 85-37

FULL RANGE OF RESULTS - JPL 83-62
                                                                                 99

                                                                                 i-t
                                                                                 n
          1990    2000   2010    2020   2030    2040    2050   2060   2070
                                       YEAR

-------
00
    UJ
    o
    o
    LU


    I

    5!
    O
    Q
                                                       G - LOWEST CASE
                                                    30 - CONSTANT PRO
                                                       4 - LOW CASE
                 A - REFERENCE CASE
5 - HIGH CASE
                                              H-
                                              JO

                                              It
                                              re

                                              NJ
                                              NJ
              1990    2000   2010    2020    2030   2040    2050   2060   2070
                                           YEAR

-------
                               66
                 COLUMN OZONE CHANGE   ( per cent )
rn

-------
o
o
     8
     0>
     ex
     LU
     o
     o
     §
     O
     O
    -15X INITIAL ANNUAL CH4 INCREASE

    *" ^ ^ ^GEOMETRIC FLUX TREND
0.4% INITIAL ANNUAL INCREASE
 GEOMETRIC CH4 FLUX TREND
   REFERENCE CASE

1 % ANNUAL INCREASE
                               V    \
                                                                      UNEAR FLUX TRENtf'\
          -20 -
                                                                            H-
                                                                           OQ

                                                                            >1
                                                                            n

                                                                            N9
                1990    2000    2010    2020    2030    2040    2050    2060    2070
                                                YEAR

-------
                                 TOT
                 COLUMN OZONE CHANGE   ( per cent )
m

-------
o
N5
    LLJ

    CD
    O
    ai


    I
    O
    O
          -5
    <   -10
        -15
        -20
         -25
                                   11.5% ANNUAL CH4 FLUX INCREASE
               0.5% ANNUAL FLUX INCREASE
                                                                                     09
                                          1 % ANNUAL CONCENTRATION INCREASE
                                                            5 - HIGH CASE
              1990    2000   2010    2020    2030   2040    2050    2060   2070
                                           YEAR

-------
o
u>
    
    O
    O
         -6
         -8
                 LOW CASE CFC GROWTH
        -10
              1990    2000   2010   2020    2030   2040    2050    2060   2070
                                          YEAR

-------
                   COLUMN OZONE CHANGE    ( per cent )
       i
       ro
       01
 I
ro
o
01
 I
_».
o
 I
C71
         I  i
m
     
     o
     ro
     o
     ro
     o
     en
     o
     ro
     o
     o>
     o
     ro
     o
                                i  i  i  I  i  i  i  i  I  i  i  i
                                  82

-------
                                  SOT
               COLUMN OZONE CHANGE    ( per cent )
   i
  4»
  O
 I
CO
en
 I
CO
O
 I
ro
en
 I
ro
O
                                    en
 I
en
                                                               en
(O
to
O
ro
o
o
o
ro
o
O
ro
o
ro
o
co
o
ro
o
ro
O
en
o
ro
o
O)
o
ro
o

-------
                                    30 - CONSTANT CFC PRODUCTION
                                                                     H-
                                                                     09
                                                                     U)
                                                                     o
1990   2000    2010   2020    2030   2040    2050   2060   2070
                            YEAR

-------
     10
o

I-

UJ
or

eo
cc
CD
3


o
UJ
 9



 6



 7



 6



 5



 4



 3



 2



 1



 0



-1



-2
                    RANGE OF RESULTS W/0 FEEDBACK AMPUFICATION


                    RANGE WITH FEEDBACK AMPUFICATION


                                                           <
                                                             I
                                                               «



          1990    2000   2010    2020    2030   2040    2050    2060   2070
                                       YEAR

-------
o
00
    Q>
    O
    111

    O
    O

    LLJ
    2
    O

    s
    o
    o
 -5




-10




-15




-20




-25




-30




-35




-40'
                                                          REFERENCE CASE
REFERENCE CASE CFC TRENDS

    CONSTANT CO* CH4
                                            u>
                                            NJ
              1990   2000    2010   2020   2030    2040   2050    2060   2070
                                          YEAR

-------
                                   601
                  COLUMN OZONE CHANGE   ( per cent )
1
0»
o
1
en
C71
I
01
o
i
^.
01
1
-N
0
1
c*>
U1
1
CO
o
1
ro
01
l
ro
o
i

01
i

o

1
Ol


o
O
O
m
i
o
>
CD
o
rn
Ol
     ro
     o
     r>o
     01

-------
                             Figure 34
LU
O
i
     55
     50
     45
     40
     35
     30
E    25
     20
     15
     10
         .  ... I  .... I  . ,, .  I . .  , .  I .
       -25    -20    -15    -10
                                    -5
10
               LOCAL OZONE CHANGE   1 0 11 ( molec cm"3)
                                no

-------
                        I  1  I
                                            I   I  I  I
                                                          III
8
0>
Oi
LU
CD

<

O
LU



1
O
O
 -5


-10


-15


-20


-25


-30


-35


-40


-45


-50


-55


-60
                                               REFERENCE CASE
                                                     \      y-..CFC/N20
                    \   •

               \    Y
CFC-ONLY
                         \ CFC/CH4
                      \
                                                                 \
                                                                  \
                                                                    \
                   i   .  .
                                                        i	j  	i   i
                                        m
                                        u>
                              10
15
                                                20
               25
30
                         CHANGE IN TOTAL Cl ABUNDANCE   ( ppb )

-------
                            Figure 36
LU
Q
ID
                      I    I	1  I  I
1 985 ATMOSPHERE
                                •^  2065 - CFC-ONLY CASE


                                2065 - CFC/CH4 CASE
                                                            10V
                      RATIO OF CIO TO TOTAL Cl
                                112

-------
8

-------
                                                                        UCRL- 93375
                                                                        PREPRINT
                                 MONTE  CARLO UNCERTAINTY ANALYSIS OF
                            STRATOSPHERIC OZONE IN AMBIENT AND PERTURBED
                                              ATMOSPHERES
                                            Keith E.  Grant
                                           Peter S. Council
                                          Donald J. Wuebbles
                             This paper was  prepared for submittal  to
                                   Journal  of Geophysical  Research
                                               July  1986
This is a preprint of a paper intended for publication in a journal or proceedings. Since
changes may be made before publication, this preprint is made available with the
understanding that it will not be cited or reproduced without the permission of the
author.

-------
                               DISCLAIMER

This document was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor the University
of California nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or useful-
ness of any information, apparatus, product, or process disclosed, or represents that
its use would not infringe privately  owned rights. Reference herein to any  specific
commercial products, process, or service by trade name, trademark, manufacturer, or
otherwise, does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States Government or the University of California. The
views and opinions of authors expressed herein do  not necessarily  state or reflect
those of the United States Government or the University of California, and shall not
be used for advertising or product endorsement purposes.

-------
                                                                       UCRL-93375
               Monte Carlo Uncertainty Analysis of Stratospheric Ozone
                       in Ambient and Perturbed Atmospheres
                                   Keith E. Grant
                                  Peter S.  Connell
                                 Donald J.  Wuebbles
                      Lawrence Livermore National Laboratory,
                                Livermore, Ca 94550
                                      Abstract
    In recent years, numerous modeling studies have been directed toward assessment of
the potential threat to the stratospheric ozone layer from anthropogenic trace gas pertur-
bations.  Interest is growing in the application of models as tools in assessing the potential
biological,  economic,  and climatic consequences  of possible future perturbations in the
concentrations of trace gases. Uncertainties in model predictions arise from the inherent
uncertainties in  model representations of various physical and  chemical processes, includ-
ing kinetic rate  coefficients, photolysis rates, transport, boundary conditions, and other
physical parameters as well as uncertainties in the future rates of emissions of trace gases.
In this paper we apply the Monte Carlo uncertainty analysis approach to this problem to
examine the effects of these uncertainties on model predictions of trace gas concentrations.
    Using the LLNL 1-D transport-kinetics model of the troposphere and stratosphere, we
examine the extent to which uncertainties in rate parameters might be sufficient to account
for the well-known problem of model underprediction of O3 in the upper stratosphere.  No
single mechanism  for increasing  O3 was  apparent.  Eight cases (of a set  of 100 runs)
were considered  that  satisfy the criteria that the profiles  of C£O, NO,  NO2, and HNO3
be consistent with observations.  For each of these  cases, the O2 photolysis rate at 45
km was greater  than  that for the baseline case and  significant decreases in effectiveness
occurred in one  or more of the major odd-nitrogen, odd-chlorine, or odd-hydrogen loss
mechanisms. We also investigated the probability distribution of the change in  O3 for a

-------
possible future chemically perturbed atmosphere.  Using best-estimate rate parameters,
a total column 03 change of -7.7% was determined.  With recommended uncertainties
applied to all rate and photolysis parameters,  we obtained a standard deviation about
the baseline case of ±6.4% for  100 Monte  Carlo  runs.  When the uncertainty of each
of the parameters was set to 10%, this standard deviation was reduced to ±3.8%.  We
conclude that, without addition of other criteria, substantial prediction uncertainties will
remain even when random and systematic errors in rate constant measurements approach
their practical minimums. Because of significant skewness, baseline values with standard
deviations are insufficient to adequately characterize the probability distributions of trace
gas concentrations that we obtained via the Monte Carlo uncertainty technique.

-------
Introduction

    Numerous modeling studies have been  directed toward assessment of the potential
impact on ozone from anthropogenic perturbations to trace gas concentrations. For the last
decade, the greatest concern has been focused on effects of increases in the concentrations
of chlorofluorocarbons (CFCs)  (NRG, 1979; WMO, 1981, 1986; Wuebbles, 1983a). More
recently, the importance of coupling mechanisms between increases in CFC concentrations
and perturbations to other traces gases such as ,CO2, CH4, and ^O has been recognized
(Wuebbles et al., 1983).
    Considerable use is being made of models as tools in assessing the chemical and cli-
matic  consequences of possible future perturbations of trace gas  concentrations.  These
models generally use as input scenarios for the projected changes in trace species concen-
trations or emissions. Such model  applications are often precursors to the assessment of
potential biological and  economic consequences resulting from these scenarios. They are
also important input to the development of corresponding regulatory and other planning
strategies.  It is crucial that these model results include analyses of uncertainty limits and
probability distributions for predicted trace gas  concentrations.  This study is a response
to this need.  We use the LLNL 1-D transport-kinetics model combined with Monte Carlo
parameter  variation to obtain  information on the uncertainties  in model predictions for
concentrations of ozone and other trace atmospheric species.
    Uncertainties in model predictions for  concentrations of ozone and other chemical
species arise from the inherent uncertainties in measurements of chemical rate coefficients
and photolysis rates, representation of transport processes, boundary conditions, and other
physical parameters used within a  model.  Most past model uncertainty studies have in-
dividually  varied each of a limited set  of such parameters. This produced  an estimate
of the  sensitivity of the model  to each  parameter varied. These sensitivities, along with
the actual  uncertainties of the  parameters, were then combined into a cumulative model
uncertainty (Butler, 1978a, 1978b;  Smith, 1978; NRC, 1979). This approach has the ad-
vantage of producing uncertainties for a small, carefully selected set of parameters without
requiring great expenditures of computer time. Moreover, the sensitivity of a  model to an
individual parameter is immediately and uniquely available, without the use  of elaborate
retrieval techniques or additional a priori information.  On the other hand, this approach
unavoidably neglects contributions  to the predicted uncertainties from parameters not in
the (necessarily small) selected  set,  does not allow for coupling between parameter uncer-
tainties, and yields little information about  nonlinearity in model response to changes in
parameters.
    A number of these shortcomings can be overcome by the use of the Monte Carlo
sampling approach to model uncertainty analysis. Monte Carlo sampling, when applied
to atmospheric transport-kinetics models, allows many kinetic rate constants, photolysis
rates, boundary conditions, and transport parameters to be varied simultaneously. Specif-
ically, for each Monte Carlo run,  each parameter of interest  is varied  as a  function of
both the estimated uncertainty of the parameter and of a random number drawn from a
pre-determined probability distribution. This technique has the immediate advantage, as
compared with individual parameter variation,  of automatically  including effects of cou-
pling between parameters. Moreover, by sampling each parameter over a wide range of

-------
values, effects of nonlinear model response are also included. Use of systematic sampling
techniques in the choice of random numbers combined with a sufficiently large number of
model executions assures that each model parameter is sampled over its entire physically
reasonable range. This, in turn, assures that a well-defined distribution of values for each
model output is produced.
    The power of this method does not come free of practical difficulties. As with other
techniques for uncertainty analysis, implementation of the Monte Carlo sampling technique
initially requires determining both physically reasonable parameter uncertainty magni-
tudes and functional forms. Moreover, with Monte Carlo techniques, one or more suitable
probability distributions must be selected as a basis for  random sampling. Additionally,
because multiple parameters are being simultaneously varied, it can be desirable or neces-
sary to specify covariances between selected uncertainties (Iman et al., 1981,1984; Gardner
et al., 1981).
    Application of the Monte Carlo technique requires a number of model runs sufficient
to obtain adequate statistical precision in the probability distributions of trace gas concen-
trations. The specific number of runs required depends greatly on the model, the sampling
technique, and the statistical acceptance test used.  For instance,  we used the relatively
sophisticated Latin Hypercube Sampling (LHS) method to reduce the number of runs re-
quired to obtain a given variance.  Even with this method, the LLNL 1-D  model required
a minimum of 100 runs to reduce the estimated error in the mean variance (about the
baseline value) of the percent change in total column ozone to 26%. Without the use of
LHS, ten times as many runs could have been required (Gardner and Trabalka, 1985).
    Even using LHS, this represents  a  considerable expenditure of computer time even
for simple well-behaved scenarios run without time-dependence or  temperature feedback.
The addition of time-dependence and temperature feedback or the selection of especially
difficult chemical scenarios would further increase modelling costs. For a time-dependent
run, results would be produced at a number of prechosen times during the run, rather than
just at the end of the run  after species concentrations have reached steadystate. This will
greatly increase the quantity of raw data produced and the associated computer costs. For
a run with temperature feedback, the effects of changing concentrations of radiatively active
trace species are used to calculate a new temperature profile.  The changing temperature
profile changes kinetic rate coefficients which in turn changes species concentrations. This
iterative calculation would directly increase the computer time required.
    In analyzing Monte Carlo  uncertainty runs, we also face the ancillary problems of
storing and processing the large quantities of data produced. These problems would also be
proportionately greater for time-dependent Monte Carlo analyses.  To avoid the necessity
of redoing entire sets of runs, careful prior consideration is required  in selecting which
data from each separate model run are  to be saved.  This must usually be a small subset
of the normal model results and diagnostics produced for single runs.  The selection and
extraction of desired information from this quantity of data  is generally nontrivial.
    Because of the costs of computer time and difficulties of analysis, previous applications
of Monte Carlo sampling to atmospheric transport-kinetics problems have utilized models
in which the number of independent species, reactions, and  layers  has been substantially
reduced from state-of-the-art models (Stolarski et al., 1978; Stolarski and Douglass, 1986;

-------
 Natarajan et al., 1986).  By contrast, the LLNL 1-D model used for this study has not
 undergone such simplification. However, to minimize costs and data handling difficulties
 in doing our initial Monte Carlo studies described below, we  have chosen to investigate
 the behavior of the LLNL 1-D model for calculations of the atmosphere  under steady
 state conditions.  We used the U.S. Standard  Atmosphere temperature profile  (1976).
 Model determined atmospheres were calculated for three different conditions: a reference
 atmosphere  without chlorofluorocarbons  (CFCs), an ambient atmosphere with current
 levels of CFCs, and a perturbed atmosphere with larger trace gas concentrations. The
 information gain per model run was increased  by use of a variance reducing  sampling
 scheme known as Latin Hypercube Sampling which is also described below.

 Model Description

    The LLNL  1-D model (Wuebbles, 1983b) is a coupled transport and kinetics model of
 the troposphere and stratosphere.  It calculates  the concentration profiles of 40 chemical
 species using 130 chemical reactions and  46 photodissociation processes. The model at-
 mosphere, extending from the ground to the stratopause, is divided into 44 vertical levels.
 The vertical coordinate system used within the model is the natural logarithm of the pres-
 sure relative to the surface pressure. The boundary conditions for individual species are
 either  fixed  concentrations or fluxes at the surface and prescribed flux conditions at the
 upper  boundary.
    Three of the minor constituents, O^D), H, and N, are assumed to be in instantaneous
 equilibrium. Each of the other 37 species [O(3P), O3, NO, NO2, N2O, HNO3, OH, HO2,
 H202,  C£, C£ON02, C£0, C£N02, HC£, NO3, N2O5, HONO, HNO4, HOC£, HCO, CH2O,
 CH3, CH3OOH, CH30, CH4, H2, CO,  H2O, CH3C£, CC£4,  CH3CC£3, CFC-11  (CFC£3),
 CFC-12 (CF2C£2), CFC-22 (CHF2C£), CFC-113 (CFC£2CF2C£), CFC-114 (CF2C£CF2C£)
 and CFC-115 (CF2C£CF3)] has its concentration calculated  at each of the 44 levels each
 time step using a variable order multistep implicit method suitable for stiff systems of ordi-
 nary differential equations. Photodissociation rate coefficients, computed at each altitude
 and each time  step, are consistent with all species distributions and the specified solar
 conditions. Also included are the effects of multiple scattering (Luther et al.,  1978). The
 chemical rates and photodissociation cross-sections are essentially those recommended by
 DeMore et al.  (1985). The vertical transport of atmospheric  trace constituents in the 1-D
 model does not utilize atmospheric motion directly, but is rather based on an eddy-diffusion
 formulation, with diffusion coefficients, Kz, derived from longitudinally and latitudinally
 averaged observations of the vertical distributions of selected tracers (Wuebbles, 1983b).
    The model normally runs in a time-dependent diurnally-averaged mode, with averaging
 factors predetermined by a special version of the model designed to run over a diurnal cycle.
 With constant surface boundary conditions for long lived gases, the model can be run to
 steady  state  in a computationally efficient manner.

Monte  Carlo Sampling Implementation

    The efficiency of a Monte Carlo  sampling method can be defined as the reduction in
the variance of the mean of an output result for a given number of model runs. This
is essentially a  measure  of the information gain  per run. Under this  definition,  simple

-------
random sampling has been found to be much less efficient than more systematic sam-
pling methods (Hammersly and Handscomb, 1965; Gardner et al., 1981, 1985). Realizing
that cost effective implementation was a highly important consideration with a complex
transport-kinetics model, a particularly effective variance reducing scheme known as Latin
Hypercube Sampling (LHS) was utilized (McKay et al.  1979; Iman et al. 1981, 1984).
Using this method,  normally  distributed random sampling sets based on  the number of
samples to be generated (NS)  and the maximum number of parameters to be varied (NP)
were created prior to model execution. For the studies reported here, values of 50 and 500
were used for NS and NP,  respectively.
    To create a Latin Hypercube Sample set, the cumulative probability distribution be-
tween 0.001 and 0.999 for each parameter was divided into NS equal intervals. For a normal
distribution, this truncated probability range corresponds to limiting sampling to within
±3.09 standard deviations of the baseline case. NP random permutations of the integers
from 1 to NS were then formed, thus creating an NS by NP matrix. Each of the NS rows
of this matrix constitutes a sampling vector that randomly selected a single interval from
the probability distributions of each of the NP parameters to be varied.
    This combination process insures that over the complete set of NS sample vectors, the
total allowed range of each of the NP parameters is sampled, with each of the NS proba-
bility intervals for a given  parameter being used exactly once. Because of these features,
for  relatively small sampling sets (e.g., 50 to 100  samples), LHS yields much more repre-
sentative approximate distributions  than would have been expected with simple random
sampling. This results in reduced variance among the result means and distributions ob-
tained using successive sampling sets. It has been found that using LHS can reduce the
number of runs required to obtain a given variance by a factor of 10 compared with simple
random sampling (Gardner et al., 1985). Moreover, this is accomplished without biasing
the expected means or distributions. In this initial study, specification of covariance be-
tween parameters (e.g.  between the preexponential and exponential factors of a  kinetic
rate constat) was not attempted.  LHS sampling readily lends  itself to such covariance spec-
ifications, however (Iman  and Conover, 1982; Iman et al., 1984), and utilization of this
feature along with implementation of more extensive checking for fortuitous covariances
are planned for future studies.
    To complete the generation of a sample set, a random probability was uniformly chosen
within each interval selected.  Thus,
where P, and P,+i  are the probabilities at the lower and upper limits of a cumulative
probability interval and u is a uniform random number such that  0
-------
    Variation of rate constants was complicated by the existence of several special forms
 involving multiple parameters (DeMore et al. 1985).  In varying chemical rate constants,
 nonexponential factors were generally varied log-normally, while factors within exponents
 were varied normally. For example, for the standard Arrhenius equation,

                                   k0 = Ae~E/RT,

 the varied form became
 where AA and AE are uncertainties in A and E/RT as given in DeMore et al. (1985). a
 and /? are normally distributed random numbers from a LHS set.
    To vary  photolysis rates, absorption cross-sections for C>2,  O3 and  NC>2 were first
 varied in the calculation of atmospheric transmission.  This was implemented using a
 broad wavelength band formulation (Table 1) for these cross-section uncertainties.  We
 felt this to be a reasonable approach since bin-by-bin uncertainties were not available and
 since measurements yield absorption coefficients that are highly correlated within bands.
 These varied cross-sections were then  used in  calculating the  wavelength and  altitude
 dependent solar  flux, thereby affecting all other  photolysis rates.  An additional  non-
 wavelength dependent variation of all other absorption cross-sections (DeMore et al., 1985)
 was then included in  calculating the photolysis rate for individual processes. In all cases,
 variations were selected from a log-normal distribution.

 The Ambient Atmosphere: Model Results versus Observations

    One  dimensional  models utilizing recommended photochemical rate parameters  have
 consistently been found to underpredict O3 in the upper  stratosphere by about 20% to
 50%, in  spite of simultaneously favorable comparisons with in situ  and remote-sensing
 observations  of key radicals thought to catalyze O3 destruction (Butler, 1978a; Wuebbles,
 1983b; Froidevaux et al.,  1985; Natarajan et al., 1986).  This disagreement is  not yet
 understood and continues to be an  incentive for further investigation.
    Incomplete understanding of the present stratospheric photochemistry is related  both
 to uncertainties in laboratory measurements of reaction parameters and in observations of
 concentrations of key  species. There may also be significant species or reactions missing in
 the postulated photochemical mechanism. Ozone concentrations in the upper stratosphere
 are primarily determined by photochemical processes; transport processes  are relatively
 unimportant in this region.  However, O3  concentrations are determined by many com-
 peting photochemical processes and catalytic cycles in the upper stratosphere and are not
 dominated by any single process or species. Important processes include direct production
 and destruction of odd-oxygen (Ox = O3 +  O(3P) + O(1D)), interconversion between  odd-
 oxygen species, and catalytic  destruction of odd-oxygen involving odd-hydrogen (HOZ),
 odd-nitrogen (NOZ), and odd-chlorine (C£OZ).
    Froidevaux et al.   (1985) investigated  the feasibility of correcting the  ozone  under-
prediction problem by varying individual photochemical reaction rates by factors of two.
Their results  appear to rule out the possibility of making one or two large adjustments to
the current photochemical scheme to fit O3 observations in the upper stratosphere.

-------
    Natarajan et al. (1986) used Nimbus 7 data, giving simultaneous measurements of O3,
H2O, HNO3, NO2, CH4, and temperature to check the consistency of the theoretical O3
photochemistry. They utilized a zero-dimensional model incorporating relevant chemistry
simplified to remove pre-specified or unimportant (in the upper stratosphere) reactions.
Using available measurements and recommended photochemical rate  data,  they found
that their model underpredicted upper  stratospheric O3  by 15-32%. Using Monte Carlo
sensitivity analysis on this model, they concluded that the la uncertainty limits of the
observed O3 are outside the range of the uncertainty bands for the calculated O3. However,
by varying six rate coefficients to the limits of their uncertainties, they were able to obtain
much better agreement without creating other discrepancies.
    Stolarski et al.  (1978)  and Stolarski and Douglass (1986) have compared the distri-
butions obtained via Monte Carlo uncertainty techniques with observations for O3, NO,
NO2, C£O, and OH. Stolarski and Douglass (1986) found that, out of 329 Monte Carlo
runs, none would satisfy the la limits of O3 measurements at 20, 30, 40, and 50 km simul-
taneously. For three cases which fell within the 2a limits, they did not find any obvious
single variation in production, loss, or species variation that produces this result.
    To further examine the extent to which uncertainties in current  best estimates of
chemical kinetic and photolysis rate parameters are able to explain the upper stratosphere
O3 discrepancy, we compared the results of our Monte Carlo runs with sample case selection
criteria derived from observations (Table 2). The  set of runs we used for this  comparison
was based on an ambient 1985 atmosphere with 2.5 ppb  C£X. The criteria we used were
chosen to select runs within the 2a limits for O3  (U.S. Std. Atmos., 1976) in the upper
stratosphere that were still consistent with O3 limits at other altitudes and with C£O, NO,
NO2, and HNO3 observations (WMO, 1981; Austin et al., 1986) at several altitudes. The
choice of species for comparison was based on the availability of stratospheric data, as well
as on their importance to determining O3 concentrations.
    Previously, investigations of this discrepancy involved either varying a very limited set
of rate parameters or using simpler models than the LLNL 1-D model. The  few  studies
that have attempted to derive selection criteria based on comparison of model results with
observations have looked at fewer species and/or fewer altitudes.
    Selection criteria were used at 25 km, 35 km, and 45 km  where sufficient data  were
available. For C£O  and NO2 the data used did not extend to 45 km (WMO, 1981),
therefore 40 km values were used instead.  Likewise, for HNO3, only 25 km  and 35 km
values were used.  Except where otherwise noted, the envelopes of the measurements were
used as the constraining criteria. For C£O, the two July profiles of Anderson et al. (1980)
lying outside the envelope of other measurements were not included in determining the
constraints.  To be consistent with measurement data, diurnally averaged concentrations
output from the 1-D model were converted to noon values, except for NO2 concentrations,
which were converted to  sunset values.  Conversion factors were obtained from a  diurnal
baseline case run of the 1-D model.

    Of the 100 runs analyzed, 31 runs satisfied the O3 criteria  at 45 km (Table 2). When
criteria for 25 km  and 35 km were included, this was reduced to 15 runs. Twenty-four
runs were found to satisfy the criteria for C£O, NO, NO2, and  HNO3 without  O3 criteria.
When all the above criteria were combined, only eight runs survived. Thus, while individual

-------
 criteria can be met with relative ease, the number of ways in which multiple criteria can
 be met is much more limited.
    Following initial formulation of observational criteria, several individual criteria were
 reevaluated from the standpoint of the severity of the effect of applying each criterion versus
 its uncertainty. We originally assumed an observed upper limit for HNO3 at 35 km of 2.6
 ppbv (WMO, 1981). This constraint lies near the median of the Monte Carlo distribution
 obtained for HNO3  and will thus very strongly effect case selection.  This situation is
 clearly visible in the scatter diagram of percent change in O3 versus ambient concentration
 of HNO3 presented in Figure 1.  Based on mean seasonal range concentrations from more
 recent midlatitude LIMS (Limb Infrared Monitor of the Stratosphere) satellite data (Austin
 et al., 1986), we felt justified in  adjusting this limit upwards to 3.7 ppbv. The effects are
 minimal of including the  constraints for HNO3 with those for C£O, NO, and NO2. As
 shown in Table 3, the use of HN03 criteria reduces the number of cases surviving criteria,
 other than those  for O3, from 32 to  24. When the O3 criteria are included, the HNO3
 criteria reduce the number of surviving cases from nine  to eight.  In either instance, the
 addition of HNO3 criteria does  not change  the minimums or maximums for percent O3
 depletion and only slightly affects the means and standard deviations.
    Case selection was found to  be highly sensitive to the upper limit C£O constraint at
 25 km. A value of 0.08  ppbv was used for this constraint. If this  criterion were reduced
 (semi-abstractly)  to  as  low as 0.056 ppbv, the baseline case and four of the eight cases
 previously satisfying all  criteria would have been eliminated. This sensitivity can be seen
 in the scatter diagram of percent change in O3 versus ambient concentration of C£O shown
 in Figure 2. As shown in that figure, the C£O concentrations at 25 km are densely scattered
 between 0.03 to 0.10 ppbv. Any  change to a criterion lying within this range thus has the
 potential of strongly affecting the final result of a multiple criteria selection process.  This
 derived sensitivity to C£O upper limits  is  illustrative of the sensitivity produced when
 one attempts to simultaneously apply a number of constraints - we caution that  such
 comparisons require careful analysis of the criteria used. Our judgment, that the potential
 elimination of  half of the otherwise surviving cases plus the baseline  case was a more
 severe result than justified by the relevant measurement uncertainties, is an indication of
 the considerations that arise in choosing constraints.  In this case, we felt the higher value
 was sufficiently consistent with observations  to merit its use.
    It would be desirable to retrieve the sensitivity of the model to variation of individual
 parameters from the  results of a  series of Monte Carlo runs in which multiple parameters
 have been varied  simultaneously. This would allow direct comparison with the results
 of sensitivity studies in  which only one  parameter is  varied at a time.  However, such
 a retrieval will  generally require  sophisticated techniques embodying substantial a priori
 knowledge of the general nature  of the expected results.  This difficulty arises both from
 the nonlinearity of model response to parameter variation and from the ill-posed nature of
 such inversion problems  (Twomey, 1977; Menke, 1984). Without sufficient constraints on
 the general nature of the retrieval, numerically correct but physically meaningless model
sensitivity coefficients will result. These coefficients  would likely show  little  consistency
between different sets of Monte Carlo runs.

-------
    For this initial study, we avoided the difficulties of doing a complete inversion by only
analyzing the production and loss mechanisms for a very small set of reactions already
known to be important  in determining Oa  at 45 km.  We manually compared the im-
portance of each of these mechanisms for each of the eight cases satisfying all trace gas
criteria. In future studies, our intent is to incorporate such  a priori knowledge of reaction
importance into an  inversion scheme that can be applied to a much larger set of Monte
Carlo runs.
    For these eight  cases, an analysis was made of 0% photolysis rates versus major O3
loss rates at 45 km (Table 4). No single mechanism was found that increased O3 at 45 km
while maintaining agreement with other  criteria. This  is consistent with the  sensitivity
studies of Froidevaux et al. (1985)  and the Monte Carlo results of Stolarski and Douglass
(1986).

    In all of the eight retained cases, the O2 photolysis rate  at 45 km (and rate of O3 loss)
and column O3 above 45 km were greater than that for the baseline case.  In seven of these
cases, the relaxation constant for loss of O3 (loss rate divided by the O3 concentration)
summed over all pathways was less than that for the baseline case. Case 11 was unique in
having a larger relaxation constant, but this case was also  found to be the only retained
case to have a total column O3 amount below the 2a limit  of the observed variability for
the 30° latitude zonal mean as reviewed in MAP (1985).  All cases except for case 85
had a reduced relaxation constant for the HOZ cycle. This case displayed compensating
moderate decreases  in both the NOX and CiO cycles.  Case 26 was notable both for having
the C>2 photolysis rate closest to that  for the baseline  case and for having a relaxation
constant for the C£O cycle about half of that for the baseline case. Cases 26 and 85 were
also the only two of the eight retained cases to survive the HC£ criteria discussed below.
    Although models are known to underpredict HC£  compared to observations in the
middle stratosphere (WMO, 1986), the eight cases surviving these previous criteria were
further subjected to observational criteria for HC£ at 25 km  and 35 km. The required data
were  taken from the suggested mean HC£ vertical profile  and  3a error limits presented
in WMO (1981).  A detailed  look  at the results is instructive in interpreting this type
of analysis (Table 5).  While six of the eight cases previously retained  were eliminated,
elimination of case 11 was borderline.  An increase  in the upper  limit at  25 km from
1.23xl09 to 1.25xl09 molecules cm~3 would have retained it.  Cases were  eliminated
either because they were too high  at 25 km or too low at 35 km. None of the  eliminated
cases violated more than one constraint, however. Half of the six eliminated cases would
have been retained if the upper limit at 25 km was increased by  11.4% and the  lower limit
at 35 km was decreased by 6.2%. To have retained all cases would have required changing
the limits by 26.0% and -6.5%, respectively.

Predictions for The Chemically Perturbed Atmosphere

    Comparison of model calculations for the ambient  atmosphere with  actual measure-
ments is extremely important for gaining understanding of model behavior and for further
refining current models. However, substantial motivation exists for modeling the potential
effects of possible future concentrations of trace species. Accordingly, we calculated the
                                         10

-------
distribution of the percent change in O3 for a possible future chemically perturbed atmo-
sphere relative to a reference atmosphere.  Our perturbed atmosphere corresponded to 15
ppbv C£X, 1.25  x the present concentration of N2O, and twice the present concentration
of CH4. The reference atmosphere was without chlorofluorocarbon (CFC) emissions with
present concentrations of NaO and CH4.
    To calculate this distribution, we executed a series of 100 Monte Carlo runs and a
baseline run for each of the reference and perturbed atmospheres, using the same sequence
of random perturbations to model parameters previously used for the ambient atmosphere.
This  sampling produced a series  of paired runs from which O3 probability distributions
were  derived for the column total and several selected altitudes. These distributions are
shown in Figures 3 and  4. The corresponding probability distributions for the percentage
change of 03 were then  produced by calculating the percent changes for each pair of runs
having the same random sampling (Figure 5).
    Using recommended rate parameter constants (DeMore et al. 1985), we determined a
total  column O3 change of -7.7% for the chemically perturbed atmosphere relative to the
reference atmosphere (Table 6). With recommended uncertainties applied to all rate and
photolysis parameters, we obtained a standard deviation about the baseline case of ±6.4%
for 100 Monte Carlo runs.
    The distributions for percent change in O3 obtained when all uncertainties were reduced
to an assumed minimum laboratory measurement error of 10% are shown in Figure 6. To
facilitate comparison, the same set of random samplings  and  the same altitudes were
used  as for Figures 3-5. This  modification reduced the standard deviation for change in
total  column O3 to ±3.8%.  On the positive side, the relative uncertainty overall for the
model is less than that assumed for the individual parameters. However, we conclude that
significant uncertainties  will still remain in model predictions for trace gas concentrations in
the stratosphere, even if minimum likely measurement uncertainties of 10% are eventually
obtained for all kinetic rate constant parameters and photolysis rates.
    For the distributions obtained using both the recommended and 10% uncertainties, the
predicted uncertainties in percent change of ozone were found to be roughly constant with
altitude. This constancy is also evident from the profiles of the baseline value and standard
deviation for percent change of O3 for both sets of uncertainties, as shown in Figure 7. The
baseline concentration and standard deviation profiles for O3 in the reference atmosphere
are shown in Figure 8. It is apparent from this figure that the largest absolute uncertainties
in ozone concentrations were found to be near 25 km where the ozone concentration is
largest. As seen from the altitude  profile of the moment coefficient of skewness for percent
change of O3  in Figure 9, the shape of the these distributions is a strong function of
altitude. At individual  altitudes,  the  distributions were found to  be skewed opposite to
the calculated baseline  changes as can be seen from Figures 7  and 9. Thus, at 25 km
a small baseline increase with negative skewness was calculated, while at 40 km  a large
baseline decrease with positive skewness was obtained. However, the larger absolute O3
concentrations and uncertainties in the lower stratosphere result in a baseline total column
O3 decrease with a distribution skewed toward even greater decreases.
    Probability distributions obtained for the ambient atmosphere concentrations  (at 25
km, 32 km and 40 km) and column total amounts of CIO, OH, and NOy are presented in

                                        11

-------
Figures 10-12.  As was the case for ozone concentrations and percent change, the distri-
butions for these trace species are asymmetrical, with the shape of the each distribution
being a function of altitude.
    We have noted the presence of significant skewness in most of the probability distri-
butions presented in this study. Additionally, the baseline results and  the means of the
distributions obtained are not coincident.  Stolarski and Douglass  (1986) have observed
that, as a function of increasing CFC source fluxes, the mean O3 percent decrease behaves
more linearly than the baseline percent decrease. This implies that the shape of the dis-
tribution is itself a function of the magnitude of the CFC perturbation. Because of these
considerations, we conclude that  baseline values with standard deviations are generally
insufficient to  adequately  characterize the  probability distributions of trace gas concen-
trations. This conclusion would appear to hold both for column total amounts and for
concentrations at individual altitudes in the stratosphere.
    The effects of case selection criteria applied to results for the ambient atmosphere
on the corresponding results for percent change in  O3 were also investigated. The means,
standard deviations, and extrema for this result for sets of Monte Carlo samples obtained by
applying individual and combined selection criteria are shown in Table 3. For the 24 cases
subjected to all criteria except O3 concentrations, the minimum and maximum percentage
O3  changes are nearly equidistant from the mean.  This implies that this subdistribution
is much less skewed than the 100  case parent distribution. A reduction of skewness when
observational constraints are applied was also observed by Stolarski and Douglass (1986).
This effect is mainly attributable to removal of cases with very large percentage decreases
of C*3 by the criteria applied for concentrations of C£O, as seen in Table 3.
    The eight  cases surviving all  criteria given in Table 2 had the  same extrema and
essentially the same mean as the 24 cases subjected to all criteria except O3  criteria. The
standard deviation was larger than for the 24 case set, but not by more than would be
expected by a factor of three reduction m the number of retained cases.
    We also compared predictions  for percent change in total column O3 with mixing ratios
of HNO3, C£O, OH, and NOy for the ambient atmosphere (2.5 ppbv C£X). The comparisons
are presented as scatter diagrams in Figures  1, 2,  13, and 14. No clear relationship was
evident between column O3 and concentration of HNO3 at 35 km (Figure 1) or between
between column O3 and the concentration  of OH at 25 km (Figure 13). Johnston  (1984)
had suggested a correlation between concentration of OH and column O3 percent change
(all else being equal, which it is  not). Our results as presented in Figures 2 and 14 indicate
a strong negative relationship with C £O at 25 km and a weak positive relationship with
NOy at 35 km.

Conclusions

    In  this  study, we  have used the LLNL 1-D  transport-kinetics model combined with
Monte Carlo parameter variation to obtain  information on  the uncertainties in  model
predictions for concentrations of ozone and  other trace atmospheric species.  We have
implemented the Monte Carlo uncertainty  technique using a variance reducing technique
called  Latin Hypercube Sampling.  This technique greatly reduces the number of runs
                                         12

-------
required to obtain a given variance in the mean value of a model prediction such as percent
change in ozone.
    Significant uncertainties in O3 depletion were found in the upper stratosphere where
the chemistry is generally considered to be understood. Current models of stratospheric
photochemistry and transport commonly underpredict O3 abundance by 20% in this re-
gion. The extent  to which this discrepancy is consistent with current uncertainties in rate
parameters  and rates of photolysis processes has been open to question.  Of the Monte
Carlo runs we made for the ambient atmosphere, fully 30% yielded sufficiently large con-
centrations of O3 at 45 km to be within the  2a limits of O3 measurements. However, this
does not immediately imply that a set of rate constants can be found that would increase
upper stratospheric  O3 predictions by 20% without unrealistically perturbing concentra-
tion profiles of Oa or other trace species.
    To further investigate this question, observational constraints for O3, C£O, NO, NO2,
and HNO3 were applied  to 100 Monte Carlo runs based on an ambient atmosphere with
2.5 ppbv C£X. For the 8 cases surviving all criteria, no single mechanism was found that
increased O3 at 45  km while maintaining agreement with  other criteria. In all of the re-
tained cases, however, the O2 photolysis rate at 45 km (and rate of O3 loss) and column O3
above 45 km were greater than that for the baseline case.  Seven of the eight cases had total
relaxation constants for O3 loss less  than that for the baseline case  while simultaneously
having greater O% photolysis rates. The one case where this was not true was sufficiently
low in total  column O3 to have been discarded. These results would indicate that while an
increase in O3 production is helpful in obtaining increased O3 in the upper stratosphere,
it is not sufficient without net decreases in loss mechanisms.
    A total  column  O3 change of  -7.7% was found  for a possible future chemically  per-
turbed atmosphere  relative to a reference atmosphere. The perturbed atmosphere had 15
ppbv C£X, 1.25 x the present concentration of NaO, and twice the present concentration
of CH4.  The reference atmosphere was without  CFC emissions and had present concen-
trations of N2O and CE^. Using recommended values for the uncertainties in  kinetic  rate
constants and photolysis  processes, a standard deviation of ±6.4% was obtained. The pre-
dicted uncertainties were found to be roughly constant with altitude. The largest absolute
uncertainties in O3 were found, however, near 25 km where  the O3 concentration is largest.
    When all uncertainties were reduced to 10%, the standard deviation of column O3
percent change was reduced to ±3.8%. A similar reduction was found for the altitude
profile.  We conclude that,  even if laboratory measurements of rate constants eventually
achieve a practical minimum uncertainty of 10%, significant uncertainty will  still  remain
in model predictions for stratospheric O3 concentrations and total column amounts.
    At individual  altitudes, the distributions for percent O3 change were found to be  skewed
opposite to the calculated baseline change. Thus, at 25 km a small increase with negative
skewness was calculated, while at 40  km a large decrease with positive skewness  was
obtained. However, the much larger absolute  O3 concentration and uncertainty in the lower
stratosphere resulted in a baseline total column O3 decrease with a distribution  skewed
toward even greater decreases. Because of this skewness, it is apparent that baseline values
with simple symmetrical error limits are not sufficient to  categorize the uncertainties in
stratospheric trace gas concentrations.

                                        13

-------
    The effects of case selection criteria applied to results for the ambient atmosphere on
the corresponding results for percent change in O3 were also investigated.  A reduction
of skewness, attributable to removal of cases with very large percentage decreases of O3,
was obtained by the application of criteria for concentrations of CIO. In combination with
other trace gas criteria that slightly truncated the upper limits of predicted O3 change, the
mean change in O3 was reduced from -8.6% to -6.8%. Although the standard deviation per
se did not decrease greatly from that for all cases, this  is mainly due to the large reduction
in the number of cases retained. In actuality, the distribution obtained with all criteria
applied is much narrower.
    We also compared predictions for percent change in total column O3  with mixing
ratios of HNO3, C£O, OH, and NO,,  for the ambient atmosphere (2.5 ppbv C£X). No
clear relationship was evident between column O3 and concentration  of HNO3 at 35 km
or between between column O3 and concentration of  OH at 25 km. Our results indicate
a strong  negative relationship with C£O at 25 km and a weak positive relationship with
NOy at 35 km.

Acknowledgements

    This  research was supported in part by the NASA Upper Atmospheric Research Pro-
gram and by the Department of Energy Carbon Dioxide Research Program.
    This work was performed under the auspices of the U.S. Department of Energy by the
Lawrence Livermore National Laboratory under Contract W-7405-Eng-48.
                                        14

-------
                                  List of Tables
1.  Absorption cross-section uncertainties for C>2, O3, and NC>2 used in solar-flux trans-
   mission calculations
2.  Observational selection criteria for O3, C£O, NO, NO2, and HNO3 used for Monte
   Carlo run comparisons.
3.  Means, standard deviations, and extrema of percent change in  O3 for the chemically
   perturbed  atmosphere relative to the no-CFC reference atmosphere for sample sets
   meeting O3, individual trace gases other than  O3,  and combined selection criteria
   applied to  the ambient (2.5 ppb C£X) atmosphere.
4.  Comparison of relaxation constants for O3 production and major loss mechanisms at
   45 km for random samples meeting combined O3 and other trace gas selection criteria
   (Table 2).
5.  Comparison of HC£ 3a criteria (WMO, 1981) with model baseline run and cases sat-
   isfying previous observation criteria (Tables 2 and 3).
6.  Statistical  moments for percent change in O3 (perturbed atmosphere relative to  the
   reference atmosphere) obtained for 100 model runs (10% uncertainty results in paren-
   thesis) . The standard deviations and moments of skewness are taken about the baseline
   values, not about the mean values.
                                       15

-------
Table 1: Absorption cross-section uncertainty factors for O2, O3, and NO2 used in solar
flux calculations
O2  (General)
O2  (Schumann-Runge)
O2  (Herzberg)
03
NO2
±1.4
±1.4
-1.1, +1.5
±1.1
±1.25
Table 2: Observational selection criteria used for Monte Carlo run comparisons. All values
are for noon except those for NO2 which are sunset values.  All values are given in ppbv
except those  for O3 which are in molecules  cm~3.  Altitudes shown are on the model
log-pressure grid and differ slightly from actual  physical altitudes which are shown in
parentheses.
Species

03
C£O
NO
NO2
HNO3
25 km
(24.2)
3.29-5.69(12)
0.008-0.08
0.74-2.9
0.66-5.8
2.7-12
35 km
(33.6)
1.11-2.23(12)
0.15-7.0
3.7-21
2.8-22
0.31-3.7
40 km
(38.6)
—
0.4-1.5
—
2.6-18
—
45 km
(43.9)
1.80- 3.76(11)
—
4.4-31
—
—
Table 3:  Means, standard deviations, and extrema of percent change in O3 for the chem-
ically perturbed atmosphere relative to the no-CFC reference atmosphere for sample sets
meeting O3, individual trace gases other than O3, and combined selection criteria applied
to the ambient (2.5 ppb C£X) atmosphere. Except where noted, selection criteria were
applied at all altitudes for which concentrations were given in Table 2.
Set Criteria
None
O3 at 45 km
03
CIO
NO
N02
HN03
All but HNO3
and O3
All but O3
O3 at 45 km &
other trace gases
All but HNO3
All criteria
N
100
31
15
53
84
81
64
32

24
13

9
8
Mean
-8.6
-10.9
-10.9
-5.5
-8.1
-9.8
-9.1
-6.4

-6.9
-7.0

-7.0
-6.8
Std. Dev.
6.4
8.0
9.0
3.4
6.0
6.4
6.2
3.2

3.3
4.0

4.9
5.2
Minimum
-31.6
-31.6
-29.9
-13.9
-31.6
-31.6
-31.6
-13.9

-13.9
-13.9

-13.9
-13.9
Maximum
1.56
0.35
0.35
1.56
1.56
0.35
1.29
0.35

0.35
0.35

0.35
0.35
                                        16

-------
Table 4: Comparison of relaxation constants for O3 production and major loss mechanisms
at 45 km for random samples meeting combined 63 and other trace gas selection criteria.
O2 photodissociation rates  are given in units of 106 molecules cm"3 sec"1.  All other
quantities are O3 loss relaxation constants given as percentages of the total loss relaxation
constant of 2.87xl(T5  for the basecase. Each loss cycle is listed by its rate limiting step.
Case Number
Reaction
O2 + hv
Total loss
Base
2.23
100
11
3.49
111
26
2.25
82.6
55
2.35
83.9
75
2.72
96.2
77
3.20
88.6
78
2.57
94.0
85
2.60
97.1
95
3.87
84.4
O + O3
N02 + O
H02 + O
H + 03
C£O + O
15.7
19.0
26.9
 2.8
34.8
28.9
23.1
24.2
 2.2
32.1
20.3
20.7
19.1
 3.2
17.7
18.2
11.8
23.9
 2.9
26.4
15.7
24.3
19.4
 2.3
33.6
24.8
16.1
18.8
 3.8
24.5
16.3
19.5
24.6
 2.3
30.4
13.6
15.4
30.5
 7.3
29.7
17.7
14.8
13.9
 2.9
34.4
Table 5: Comparison of HC£ 3er criteria (WMO, 1981) with model baseline run and cases
satisfying previous observation criteria. Values shown (molecules cm~3) that conflict with
the criteria are shown in boldface.
Case No.
                          25 km
                                              35km
3cr range

  base
  11
  26
  55
  75
  77
  78
  85
  95
                        4.9-12.3(8)

                          1.05(9)
                         1.24(9)
                          9.25(8)
                         1.43(9)
                          1.16(9)
                         1.55(9)
                          8.34(8)
                          1.10(9)
                         1.36(9)
                                            2.6-5.8(8)

                                              2.83(8)
                                              3.08(8)
                                              3.10(8)
                                              3.68(8)
                                             2.43(8)
                                              3.99(8)
                                             2.45(8)
                                              3.44(8)
                                              3.53(8)
Table 6: Statistical moments for percent change in 63 (perturbed atmosphere relative to
the reference) obtained for 100 model runs (10% uncertainty results in parenthesis). The
standard deviations and moments of skewness are taken about the baseline values not
about the mean values.
Altitude
km
Column Total
25
32
40
Unvaried
% Change
-7.7
+0.5
-22.1
-69.1
Standard
Deviation
6.4 (3.8)
8.0 (4.3)
10.3 (5.9)
9.3 (3.8)
Moment of
Skewness
-1.8 (-1.4)
-2.5 (-1.3)
+0.005 (-0.4)
+1.7 (+0.8)
                                         17

-------
                                   List of Figures
 1.  Scatter plot of column O3 percent change for the chemically perturbed atmosphere
    relative to the reference atmosphere versus the mixing ratio of HNO3 for the ambient
    atmosphere at 35 km.
 2.  Scatter plot of column O3 percent change for the chemically perturbed atmosphere
    relative to the reference atmosphere versus the mixing ratio of CiO for the ambient
    atmosphere at 25 km.
 3.  The probability distributions, using  best-estimate uncertainties, of reference  atmo-
    sphere concentrations of O3 at (a) 25 km, (b)  32 km,  (c) 40 km, (d) column total.
 4.  The probability distributions, using best-estimate uncertainties, of perturbed  atmo-
    sphere concentrations of O3 at (a) 25 km, (b)  32 km,  (c) 40 km, (d) column total.
 5.  The probability distributions, using best-estimate uncertainties, of O3 percent change
    for (a) 25 km, (b) 32 km, (c) 40 km, (d) column total.
 6.  The probability distributions, using 10% uncertainties, of O3 percent change for  (a) 25
    km, (b) 32 km, (c) 40 km, (d) column total.
 7.  Profile of baseline O3 percent change with standard deviations for best-estimate and
    10% uncertainties.
 8.  Profile of baseline O3 concentration with standard deviations for best-estimate and
    10% uncertainties.
 9.  Profiles of moment coefficient of skewness relative to  the baseline case for O3 percent
    change.
10.  The probability distributions, using  best-estimate uncertainties, for ambient  atmo-
    sphere concentrations for CiO at (a) 25 km, (b) 32 km, (c) 40 km, (d) column total.
11.  The probability distributions, using  best-estimate uncertainties, for ambient  atmo-
    sphere concentrations of OH at  (a) 25 km, (b) 32 km, (c) 40 km, (d) column total.
12.  The probability distributions, using  best-estimate uncertainties, for ambient  atmo-
    sphere concentrations of NOy for (a) 25 km, (b) 32 km, (c) 40 km, (d) column total.
13.  Scatter plot of column O3 percent change  for the chemically perturbed atmosphere
    relative to the reference atmosphere versus the mixing ratio of OH for the ambient
    atmosphere at 25 km.
14.  Scatter plot of column O3 percent change  for the chemically perturbed atmosphere
    relative to the reference atmosphere versus the mixing ratio of NOy for the ambient
    atmosphere at 35 km.
                                         18

-------
                                   Bibliography
 Anderson, J.G., H.J. Grassl, R.E. Shetter, and J.J. Margitan, "Stratospheric Free Chlorine
    Measured by Balloon-Borne in situ Resonance Fluorescence", J.  Geophys.  Res., 85,
    2869-2887, 1980.
 Austin, J., R.R. Garcia, J.M. Russell III, S. Solomon, and A.F. Tuck, "On the Atmospheric
    Photochemistry of Nitric Acid", J. Geophys. Res., 91 (D5), 5477-5485, 1986.
 Butler, D.M., "The Uncertainty in Ozone Calculations by a Stratospheric Photochemical
    Model", Geophys. Res.  Lett., 5, 769-772, 1978a.
 Butler, Dixon, M.,  "Input  Sensitivity Study of a Stratospheric Photochemistry Model",
    Pageoph, 117, 430-435,  1978b.
 DeMore, W.B., J.J. Margitan, M.J. Molina, R.T. Watson, D.M. Golden, R.F. Hampson,
    M.J. Kurylo, C.J. Howard, and A.R. Ravishankara, Chemical Kinetics and  Photo-
    chemical Data for Use in Stratospheric Modeling - Evaluation Number 7, Publ. 85-37,
    Jet Propul. Lab., Pasadena, Calif., 1985.
 Froidevaux, Lucien, Mark Allen, and Yuk L. Yung, "A Critical  Analysis of CIO and O3 in
    the Mid-Latitude Stratosphere", J. Geophys. Res., 90 (D7), 12,999-13,029, 1985.
 Gardner,  R.H., R.V.  O'Neill, J.B.  Mankin, and J.H. Carnery, "A Comparison of Sensi-
    tivity  Analysis and Error  Analysis Based on a Stream Ecosystem Model", Ecological
    Modelling, 12, 173-190,  1981.
 Gardner, R.H. and J.R. Trabalka, Methods of Uncertainty Analysis for a Global Carbon
    Dioxide Model, DOE/OR/21400-4, U.S. Dept. of Energy, 1985, Available from NTIS.
 Hammersley, J.M. and D.C. Handscomb, Monte Carlo Methods, Chapman and Hall, Lon-
    don, 1964.
 Iman, Ronald L., Jon C. Helton, and James E. Campbell, "An Approach to Sensitivity
    Analysis of Computer Models:  Part I - Introduction, Input Variable Selection and
    Preliminary Variable Assessment", J. Quality Tech., IS (8), 174-183, 1981.
 Iman, R.L. and W.J. Conover, "A Distribution-Free Approach to Inducing Rank Correla-
    tion Among Input Variables", Commun. Stat., Bll(3), 311-334, 1982.
 Iman, Ronald L., and Michael J. Shortencarier, A FORTRAN 77 Program and User's
    Guide for the Generation of Latin Hypercube and Random Samples for use with Com-
    puter Models, SAND83-2365, Sandia National Laboratories, Albuquerque, New Mex-
    ico, 1984.
 Johnston, Harold S., "Human Effects on the  Global Atmosphere",  Ann.  Rev.   Phys.
    Chem., 35, 481-505, 1984.

 Luther F.M., D.J. Wuebbles, W.H. Duewer, and J.S. Chang, "Effect of Multiple Scattering
    on Species Concentrations and Model Sensitivity", J. Geophys. Res., 83, 3563-3570,
    1978.

MAP (Middle Atmosphere  Program), "Atmospheric Structure and its Variation in the
    Region 20 to 120 km - Draft of a New Reference Middle Atmosphere", Handbook for
    Map, vol. 16, 1985.

                                       19

-------
McKay, M.D., R.J. Beckman, and W.J. Conover, "A Comparison of Three Methods for
   Selecting Values of Input Variables in the Analysis of Output from a Computer Code",
   Technometrics, 21 (2), 239-245, 1979.
Menke,  William, Geophysical Data Analysis:  Discrete Inverse Theory, Academic Press,
   San  Francisco, 1984.
Natarajan, Murali, Linwood B. Callis, Robert E. Boughner, James M. Russell III, and
   James D.  Lambeth, "Stratospheric Photochemical Studies Using Nimbus 7 Data - 1.
   Ozone Photochemistry", J. Geophys.  Res., 91 (D7), 1153-1166, 1986.
NRC (National Research Council Panel on Stratospheric Chemistry and Transport), Strato-
   spheric Ozone Depletion by Halocarbons: Chemistry and Transport, National Academy
   of Sciences, 1979.
Smith, W.S.,  "Uncertainties in Evaluated Atmospheric Rate Constants",  WMO Sympo-
   sium on Geophysical Aspects and Consequences of Changes in the Composition of the
   Stratosphere, WMO No. 511, Toronto, Canada, 37-46, 1978.
Stolarski,  R.S., D.M. Butler,  and R.D. Rundel, "Uncertainty Propagation in a Strato-
   spheric Model - 2. Monte Carlo Analysis of Imprecisions Due to Reaction Rates", J.
   Geophys.  Res., 83 (C6), 3074- 3078, 1978,
Stolarski, R.S., and A.R. Douglass, "Sensitivity of an  Atmospheric Photochemistry Model
   to Chlorine Perturbations Including Consideration of Uncertainty Propagation", J.
   Geophys.  Res., 1986 (in press).
Twomey, S., Introduction to the Mathematics of Inversion in Remote Sensing and Indirect
   Measurements, Elsevier, New York, 1977.
U.S.  Standard Atmosphere, 1976, NOAA-S/T 76-1562, 1976.
WMO (World Meteorological Organization), The Stratosphere 1981  - Theory and Mea-
   surements, WMO Ozone Research and Monitoring Project Report No.  11, May 1981.
WMO (World Meteorological Organization),  Atmospheric Ozone  1985:  Assessment of
   Our Understanding of the Processes Controlling its Present Distribution and Change,
   WMO Global Ozone Research and Monitoring Project Report No. 16, 1986.
Wuebbles, D.J., "Chlorocarbons Emission Scenarios: Potential Impact on Stratospheric
   Ozone", J. Geophys. Res., 88 (c2),1433-1443, 1983a.
Wuebbles, Donald J., A Theoretical Analysis of the Past Variations in Global Atmospheric
   Composition and Temperature Structure, UCRL-53423 (Ph.D Thesis), Lawrence Liv-
   ermore National Laboratory, 1983b.
Wuebbles, D.J., F.M. Luther, and J.E. Penner, "Effect of Coupled Anthropogenic Pertur-
   bations on Stratospheric Ozone", J. Geophys. Res., 88 (c2), 1444-1456, 1983.
                                        20

-------
i   i   i
Column 03 Percent Change
 i  i   i   i   i   i   i  i
                                     i   i   i
^^

^


ro



0)


x
o "^
O)

I01


O)

^


OD

en

" 4-
* .**'.*
"*" "*• 4-
44- + +++ -H- 4-
_ ^. ^jh
-ff*" •**•"*" 4- "*"
4. + + + 4-
4- 4-
•*" 4- "*" +4. + *
4- 4- + + f +
4- + 4f + *
+ +* 4;
+ ^ * -H.
4* 4*
4-
" 4-
•
4*
4-

4-


_
4- +
4-

I
z
o
U)
Q)
rf

CO
CJ1
7T
3

CD
"3
Cfl
C
en
o
o
c
3
D
O
0)
n
U
Q)
D
(D
CD

-------
CIO at 25 km  Versus Column 03 %  Change




0)
en

ID
x:
u
-P
c
Q>
u
c.
0)
Q_

cn
O

3
M^
I •
O
u






c
0
-2
-4
-6
-8


-10

-12

-14

-16


-18


-20
-22
-24
-26
-28

-30
-32
1 +; +
- +* + + ^ ++ +
" +"tk+* + + + +
1 -H- + Jf ++'IN*|+ + +
— + + + •*• +1.
"" + • *"*" ^
•t •" «fc i *T*
— j.

*
HI* t
+ j. j. +
• — T
— +
^U ^
*
—

* T*
— + * +
- +
—
—

_
- +
•»
• I i I i | i 1 i | i 1 i -f 1 i I i | i

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.2
                CIO  ppbv

-------
Distribution of Ozone for Reference Atmosphere
Distribution of Ozone for Reference Atmosphere


0) 60"
c
tr
r-, 50-
IU
o
•V
* 40-
O
O
»H
E 30-
O
c.
«•-
C. 20-
a
z 10-







0.5 1.6 2.6






1
3.7


[





1
'/,
'/
y
\
\
\



I
1






^
4.7 5.8 6.8
Ozone at 25 km ( cm**


Distribution


to 6°-
c
3
DC
rt 50-
U
T3
O
Z 40
O
O
*-i
E 30-
0
c_
t~
t_ 20
0)
J3
2 10-
0-

























-------
         Distribution of  Ozone  for  Perturbed Atmosphere
 n
 c

 IT

 *  50
 O
?.
O
o
E
o
t.
JD
    40
    30
    20
    10













^
1
1





\
y
X
x
!
i
i
i
i
i
i
i
i
i
i
i
i
i






I
P71
                                                                              Distribution  of Ozone  for Perturbed Atmosphere
                                                                      m
                                                                      c

                                                                      DC
                                                                      0)
                                                                      •a
                                                                      o
                                                                      o
                                                                      o
                                                                      E
                                                                      0
                                                                      t.
                                                                      C.
                                                                      a>
                                                                      13
         0.5  1.6  2.6  3.7  4.B  5.8  6.9 7.9  9.0 10.0 11.1  12.1


              Ozone  at 25  km ( cmxx (-3)  x  10XX12 )

60

SO

40

30-
20-
10-







I
V? V
1
1
1
1
1
1
1

I
i
i
i
i
i
i
i
I








I
                                                                               0.2  0.7  1.2  1.7  2.2  2.6  3.1  3.6  4.1  4.5  5.0  5.5


                                                                                   Ozone at  32 km  (  cmxx (-3)  x  10XX12
        Distribution of Ozone for  Perturbed  Atmosphere

                              I
to
c
cr
0)
•o
o
o
o
E
O
t.
L.
0)
a

3
    60-
    50-
    40- •
    30-
    20- •
    10-
I
                              i
I
         0.2  0.7  1.1  1.6 2.0  2.4  2.9  3.3  3.B  4.2  4.7  5.1


              Ozone  at 40 km  (  cmxx (-3)  x lOxxii  )
                                                                      0)
                                                                      c
                                                                     T3
                                                                     O
                                                                     O
                                                                     O
                                                                      E
                                                                      O
                                                                      L
                                                                      C.
                                                                      0)
                                                                         60- •
                                                                         50
                                                                         40
                                                                         30-
                                                                         20-
                                                                         10-
                                                                              Distribution of  Ozone  for Perturbed Atmosphere







0
P71
y
t
%
6

^^^^^^^^^
i
i
i
i
i
i
i
'^
IX
$
'?
'?
i^



                                                                                                            o

                                                                                                            OL

                                                                                                            •i

                                                                                                            8
                                                                               0.9  2.6  4.3 6.0  7.7


                                                                                    Column Ozone
                                                                          9.4 11.1 12.8 14.5 16.2 17.9 19.6


                                                                         ( cmxx (-2)  x IOXXIB  )

-------
           Number  from  100 Model Runs
                                                                   Number from 100  Model  Runs
                                u
                                n)
  •JA
  O L.
    I
    vl



?  "

"  v,
CD  .
3  IU
n

Ol
3
to
CO
   A
01
Ik
O
3   »
                                                n

                                                01
                                                3
                                                (O
                                                ID
           Number  from  100 Model Runs


                   M.      IU      U      *.
            IB      01      *      n>      o
                                                                   Number from 100  Model  Runs
n

0)
3
10
n
o
-}
o
rr
Ol
O
o
l-<
c
3
    i
    u
    o.
    IU


    £j
    I
    IU
    o.
    IO

    IB
-a
CD

n
(D
3
rt-

O
N
a
3
(D

O
31
O)
3
ia
(D
                                                          git
                                                          •«  I
                                                          o *•
                                                            u
                                                        (0

                                                        n
                                                        n
                                                        3
                                                        rr   o
                                                         0)
                                                         3
                                                        (O
                                                         n
                                                         u
                                                         (U
                                                            S!
                                                        (n
                                                        rr
                                                        ~s
                                                        !-••
                                                        or
                                                        C
                                                        rt
                                                        M*
                                                        O
                                                        3
                                                              3
-a
n
~s
o
n
3
rr

O
N
O
3
(D

n

o>
3
to
(D
      Fernbar 1.1 - 1 Aorll 1988:  Plotted: 04/01/1988  10:11:51

-------
          Number from 100 Model  Runs


                       CO
                                                              Number from  100 Model Runs
C
-1
fr«
Si
1* ^
o .
Ti
O " *
0 N
»*
^
TJ •vl'
(t)
n .
S >•.
Z3 n)
n-
n CD
B^J
fn 1
i£j '
ro ?.
ro
Ol
rr |
S '"
T: i
3 ™
ru

i
.
^
i
ru
• 01 » ru o »





. . -
>V

\^XX\^s,^OOOs^OsA^OOs>^^OO1J

IN^S^^V^V^^

\XXX^
v\^\^
3








a
M
rr
^
a
O
N
O
CD

**
n
3-
01
ca
CD

^
o
C
n
CD
T
ri-
al
'M*


                                                       ^
                                                    T
                                                    0
                                                    (0
                                                    9  -
                                                    7  ni<
                                                    01  U
                                                    ia
                                                    CD
                                                    01
                                                    ro
                                                    ui
                                                       Z
                                                       o
                                 :S££Sj
                                                                                                a

                                                                                                en
                                                                         o
                                                                         N
                                                                         O
                                                                         3
                                                                         CD
                                                                         n
                                                                         a-
                                                                         01
                                                                         3
                                                                         ID
                                                                         CD
                                                                                                C

                                                                                                n
                                                                                                co
                                                                                                at
                                                                                                M-
                                                                                                3
          Number from  100  Model Runs
                                        Number from  100 Model Runs
o   S

0)   ru
3   ,
ta   »-
CD   ui
     s
 ro

H-
 rll
-t-
                      n

                      01
                      3
                      ia
                      CD
                                                      r i
                                                      o *•

                                                      I:
                                                    Tl
                                                    m
                                                    n
                                                    ro
                                                    01
                                                    3
                                                    
H-
                                                                                 u
                                                                                 ni
                                                                          U)
                                                                          rr
                                                                          ~i

                                                                          o
                                                                          -»

                                                                          o
                                                                          N
                                                                          a
                                                                          3
                                                                          CD
                            n

                            ai
                            3
                            ia
                            CD
or
                                                                          o
                                                                          M

                                                                          C

                                                                          n
                                                                          CD
                                                                          T
                                                                          rr
                                                                          ai
                                                                          M-
                                                                          3
                                                                          rr
                                                                          •<
     Fernbmr l.l - l Aon 11 1986: Platted:  04/01/1986  11:13:14

-------
Percent  Change  in  Ozone  Relative  to  No  CFC  Reference
                One Standard  Deviation  Error Limits
 HI
 Q

 r-
 M
 H
55


50


45


40


35


30


25


20


15


10


 5
                 — Basel in* CMC
                 — Racoaawndcd Uncertainty Factor*
                 •- 10X Uncertainty Factors
                                                  A\i
     -80  -70  -60
                -50   -40   -30   -20   -10
                      Ozone    cm** (-3)
10
20   30

-------
UJ
Q
_J
<
           Reference  Atmosphere  Ozone  Profile
               One Standard Deviation  Error Limits
                             Baseline Case
                             Recommended Uncertainty Factors
                             10X Uncertainty Factors
                     20  25  30  35  40   45
                        Ozone   10 " cm** (-3)

-------
Percent  Change  in  Ozone  Relative  to  No  CFC  Reference
         Moment  Coefficient of  Skewness  (Rel.  to Baseline)
     55
 ID
 a
 _j
 <
50

45

40

35

30

25

201

15

10

 5
                  Recommended Uncertainty Factors
            	 10% Uncertainty Factors
    -4.0 -3.5 -
          3.0 -2.5 -2.0 -1.5 -1.0-0.5  0.0
                  Coefficient of Skewness
0.5  1.0  1.5  2.0

-------
           Number  from  100 Model  Runs
                                                                 Number  from  100 Model  Runs
    £•3
o
o  «
.-  -t:
"•  «
£  «•£
*•
a
'•"•&!
*   •:•
 i   -
111   O
X   M.
    ni

s   *

21   ?
-J   N


*~"   f*
    Vr
                o

               -4-
                                 g
                                 -4-
 &
-4-
                                                                                   8     a
        m
        ft

        H»
        a1

        ft
        H-
        O


        o
        •*

        n
        M
        o

        -«
        o
        T
                                                     n  a

                                                     o  r

                                                     a

                                                     ^  N
                                                     ro
                                                     en
                                                         o
                                                         to
                                                     x

                                                     i-k
                                                     o
                                             o
                                             n
                                             •o

                                             (0

                                             n
                                                      x  »
                                                      -J  *.
                   *J

                   2-3
 a
-4-
       o
       •*
       w
       ft
       T
       •->
       er
       c
on
                                                                                                      n
                                                                                                      i->
                                                                                                      o

                                                                                                      -»
                                                                                                      o
        §
        !-•
        a

        ft

        >
        ft

        o
        n
       •a

        IB
        ~)
        n
           Number  from  100 Model  Runs
                                                                Number  from  100 Model  Runs
•» •* 8) n) w w »
ouooeiiSao

n a
a
>-• ^
c r
3 A
a
H .""*.
0 *•
ft ._
Wjr
o
i-i • •
u
O U
•» «•
n H.
i-j n
0 M
"^ 5
n K>'
3
« "
ro B
»*
* .»
o •*
* w
x —
"*. b'
U °
— "
o'



^»^~«
^3
sXi

\\v"vx'v\^ "\\^OJ
x.^^^V^v^VX.^V^^^V.^V ^V 1

^^x^\\^\^^^^^^^^^x^^v^^^

xxVvOvvVxx^vS


^^^^os^v\^^^N
•^x^
\X\j
^- -^> Tf
g
xi









a
M-
u>
rt
i
M-

ft
M-
O
o
-»

I-I
a

-»
o
-}

^
§-
a
rt
ft
o
(0
TJ
a-
o
T
n
                                                         n

                                                         o

                                                         01
                                                         rt


                                                         ro

                                                         *•
                                                         "tc
                                                             vl

                                                             «t^
                                                         X




                                                         U  o
                                                         X  M.


                                                         5  '"•

                                                         X  M
                                                                tr
                                                                c
on
                                                                                                      o
                                                                                                      t-t
                                                                                                      o

                                                                                                      -%
                                                                                                      o
                                                                                                      T

                                                                                                      >

                                                                                                      er
                                                                                                      t^>
                                                                                                      a

                                                                                                      ft
                                                                                                          a
                                                                                                          M
                                                                                                          •o

                                                                                                          a

                                                                                                          it
            i.it - ifl Aarii
                               Plott.rt:  07/li/
                                                 14:
                                                     :43

-------
           Number from 100 Model Huns


                    S    8    8    £   S
                                                                Number  from 100  Model Runs
  o *•
  fll •



  ^
  »» Ol
s  s*
   u

—  vl
ff  III
1  SB
         -f-
              -f-
-f-
•4-
-f-
-f-
-f-
                                              o

                                              M
                                              s
                                              a
                                              n
                                              •a

                                              ID

                                              n
                                                       I~

                                                       " M
                                                       js-s
                                                     ?  -.t,
                                                     a   •
                                                     X  "•k
                                                     3  !±
                                                     ^   •*
                                                     S   ~
                                                     -5-3
                                                         S-3
                                                                                a
                                                                                i-»
                                                                                ca
                                                                                rf
                                                                                T
                                                                                M-
                                                                                a-
                                                                                c
                                                                                rr
                                                                                K»
                                                                                O
                                                                                3

                                                                                O
                                                                                -*
                                                                                -»
                                                                                a
                                                                                T

                                                                                >

                                                                                C7
                                                                                i-»
                                                                                n
                                                                                3
                                                                                rf


                                                                                rf

                                                                                a
                                                                                M
                                                                                T3
                                                                                31
                                                                                ID

                                                                                n
           Number from 100 Model Runs
                                                                Number  from 100  Model Runs
o  •
o
*-"  »
C  'k


   M

51  «
   ni
   «-N
o
S  .-.
§  -
*  *x
T  -ti
Ei
s  Sfl
*   M
*
w
    M
                    5
                   -H
                         8
                        -f-
                    a
                    -r-
                                              o

                                              n

                                              n

                                              a-
                                              c
                                              rf

                                              a
                                              3
                                              o
                                  o

                                  a
                                                         c* ^
                                                     U  »
                                                     ru  •*
                                                          .
                                                      ff  >l
                                       n
                                      -t-
                                            M

                                           -f-
                                            g    s    a    a
                                           _i—i—i—H-
                                                                           o
                                                                           M-
                                                                           n
                                                                           rf
                                                                           T
                                                                           !-*•
                                                                           cr
                                                                           c
                                                                           rf
                                                                           M-
                                                                           0
                                                                           3

                                                                           O
                                                                           -»
                                                                                                   -»
                                                                                                   a
                                                                                §
                                                                                i-»
                                                                                ID
                                                                                3
                                                                                rf

                                                                                >
                                                                                rf

                                                                                §
                                                                                (A
                                                                                •a

                                                                                CD

                                                                                ID
     F«rnb«r i.tl - 18 Aarll UMe  natt«* 07/iB/iM»

-------
          Number  from 100  Model Rum
                                                              Number from  100 Model Rune
          Number from  100 Model  Runs
                                                               Number  from 100 Model Runs
n
o
f
c
a  -<
rr


"  '"^
   N S.
O




a
   "•'V
ro
   »

o  *

*  "ii
K»  :_• J
n  •»  3
                   &
                   -+-
 8
-f-
 K

-r-
 a
-r-
                      a
                      H»
                      n
                      rr

                      i*
                      a-
                      c
                      rr
                      >•
                      o


                      o
                      •»


                      a
                      •<

                      •»
                      o




                      J
                      •-•
                      (D

                      rr




                      I
                      «

                      \
                      «

                      a
                                                        5fl
                                                        S
                   §  5
                   «<


                   rr  «1^

                   u
                                                        «. V
                                                        J-J
                                                              M

                                                             -r-
                                                          S3

                                                          3
 s
-4-
 S
-f-
 8
-4-
                                                                                                6
                                                                                                  or
                                                                                                  c
                                                                                                  O
                                                                                                  o
                                                                                                  I
                                                                                                  «
                                                                rr

                                                                S
                                                                IB


                                                                •

                                                                n
            i.it - u ABTU iMft Plotto* 07/lt/i«M

-------
OH at 25 km Versus Column 03 % Change






01
O)
c
(0
c.
u
4J
c
Q)
U
C.
At
Q)
Q.
*•*
Pi
O

C
3
«^
^^^
0
U






<£
0
-2
-4
-6

-8


-10

-12
-14

-16


-18


-20

-22
-24
-26
-28
-30
-32
+ +
- ++
" ++ "*" + + *.
- + + ++^ + + + +
- +* + ++'f+"f+ +++ +
+ + +++"•""*" + + +
— + -H- A. j. +
* *T ^ «r A
™ * ^ «t *4"
— . T +
+ + +
I ++ + %* *** +
• T "F
— +
*T* ^^
"
—
.
* T
— ^. + 4.
—
*
—
—
+
- +
1 1 1 1 1 + 1 1 1 1 1 1

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.(
OH 10'3 ppbv

-------
           Column 03 Percent  Change
  I    I   I   I    I   I   I   I
  tdoirororororoM.
                            1   1    1
                           M.M.M>i
   ro






   en


   CD


   l-k.
   O


   ro






   en
   ro
   o
•o
TJ  ro
or  ro

   ro
   ro
   en

   ro
   CD

   O)
   o

   O)
   ro

   O)
a)
en

U)
CD
1  I '  I  ' I  r  I '  I ' I  ' I '  I ' F r lr I '  I ' I
                                +   «+ -
                             +     j.
                                      *£
                                        *
                                                           o
                                                           0)
                                                           0)
                                                           ai
                                                               (D
                                                               "3
                                                               0)
                                                               c
                                                               en


                                                               n
                                                               o
                                                               »-«
                                                               c
                                                               3
                                                           O
                                                          U)
                                                           n
                                                           zr
                                                           QJ
                                                           D
                                                           (Q
                                                           (D

-------
  A PARAMETERIZED NUMERICAL FIT TO
     TOTAL COLUMN OZONE CHANGES
   CALCULATED BY THE LLNL 1-D MODEL
OF THE TROPOSPHERE AND STRATOSPHERE
               PETER S. CONNELL

   LAWRENCE LIVERMORE NATIONAL LABORATORY
 ATMOSPHERIC AND GEOPHYSICAL SCIENCES DIVISION
                NOVEMBER, 1986

-------
                                   ABSTRACT
    Economic analysis of the costs and benefits of regulation of chlorofluorocarbon emis-
sions with respect to changes in the total atmospheric ozone column abundance depends
on the availability of an accurate means to predict future ozone change as a function of
CFC emissions and trends in other atmospheric species. Over about the last decade, these
predictive calculations have been made chiefly with one-dimensional numerical models of
the stratosphere.  Although  these models, run time-dependently, are substantially more
computationally efficient than the more desirable multi-dimensional models, they remain
expensive from the viewpoint of the requirements of cost/benefit analysis.
    In this note we present a numerical parameterization of the results of the LLNL one-
dimensional model for total column ozone change in terms of surface emissions or abun-
dances of the various source species. Previous versions of the LLNL model have been used
in earlier studies to evaluate time-dependent scenarios of future trends in the emissions or
atmospheric abundances of CFC-11, CFC-12, CFC-22, CFC-113, CC14, CH3CC13, CO2,
CH4, N2O, Halon-1301 and Halon-1211. The numerical fit  reported here produces calcu-
lated ozone  changes within about 2% (column ozone change relative to the 1985 model
atmosphere) of the current LLNL 1-D  model results, for a range of scenarios and ozone
changes. It therefore can approximately represent the  actual 1-D model for the purposes
of economic  analysis of regulatory strategies.
    *•
    It is, however, crucial to note that the 1-D model results, and consequently this parame-
terization, are subject to substantial uncertainties. Future improvements in our knowledge
of atmospheric photochemical kinetics and transport in the troposphere and stratosphere
should be expected to modify these current results in unpredictable directions.

1. INTRODUCTION

    We have recently reported the results of an extensive model study  (Connell and Wueb-
bles, 1986, hereafter CW) of the sensitivity of stratospheric ozone  to various assumed
simultaneous trends in emissions or  abundances of several long-lived source species; chlo-
rofluorocarbons, Halons romochlorofluorocarbons), carbon dioxide (CO2), methane (CH4)
and nitrous oxide (^O). The model used was the LLNL one-dimensional time-dependent
numerical representation of transport and photochemistry in the troposphere and strato-
sphere.  The stratospheric temperature profile was calculated interactively based on the
assumption of local radiative equilibrium. The study showed that a wide range of out-
comes for change in  the total  column ozone abundance  results  from the  uncertainty in
trend projections for source species and that the  effects of projected  trends in individual
species on ozone are  strongly coupled.  The conclusions in CW  (1986)  and WMO (1986)
emphasized  the uncertainties in predicting the behavior of ozone subject to significant
perturbations from current levels of the various source species. Uncertainties arise in spec-
ifying photochemical-kinetic parameters, -in interpreting the results of a single-dimensional
(vertical) model and in the choice of trend prescriptions for the source species.
    An addendum to the CW study, in which the  currently available recommendations on
kinetic and spectral parameters (JPL, 1985) were implemented in the model, showed that
remaining laboratory kinetic uncertainties can produce a substantial impact on the quanti-
tative results of ozone perturbations from the envisioned scenarios of future emissions. All

-------
of these conclusions have significant implications in the consideration of regulatory actions
and the accompanying cost-benefit analysis of controlling CFC emissions.
   An underlying consideration in studying the effects of CFC emission controls is that the
effect of CFC regulation on the total column of ozone also depends strongly on changing
abundances of several other natural  and anthropogenic trace atmospheric species. Deter-
mining the nonlinear ozone response to changes in many  species practically requires the
use of computationally intensive complete  atmospheric model calculations.  However, the
extensive exercise of current models that would be required in cost-benefit analysis may not
be justified in view of continuing progress  in understanding and modelling  of the strato-
sphere and its photochemistry. Consequently, we present here a parameterization of the
results of the complete 1-D model cited above, for total column ozone depletion in terms
of emissions or abundances of the source species. This is intended as a tool for preliminary
economic and regulatory investigations with the recognition that  both the quantitative
specifics and the form of the parameterization, as well as the 1-D model on which they are
based, may be quickly superseded by the  increase of knowledge.  Additionally, it should
be recognized that this parameterization will be a less accurate indicator of the effects of
trace gas emissions on ozone than the actual model calculations on which it is based.
   A brief description of the numerical fit  and its basis is given below, followed by appli-
cation to a few cases of multi-species scenario  results from the LLNL 1-D model.

2. DESCRIPTION OF PARAMETERIZATION

   The intent is to approximate the predicted total column ozone changes calculated by
the LLNL time-dependent temperature-coupled one-dimensional transport-kinetics model
of the troposphere and stratosphere, for specified changing fluxes of several  trace con-
stituents projected over the period  of 90 years from 1985.  The general inadequacies of
such an approach should be emphasized again.  The highly coupled nature  of the strato-
spheric photochemistry and the many parameters being simultaneously varied in the EPA
scenarios result in a complicated nonlinear multidimensional ozone response surface. The
full domain of response can really only be adequately mapped by a  large number of model
runs. Consequently, any simple set  of fitting equations may be inadequate  when applied
away from  areas actually investigated with the complete model. In addition, changes in
kinetic parameters, as have occurred in the past,  may radically  change the shape of the
ozone response surface in unpredictable ways. Problems of linearity of response are exac-
erbated by  the very large ozone depletions  encountered in higher CFC flux scenarios (e.g.,
annual relative increases exceeding 1-2% per year).
   With this in mind, some features of the problem enhance the possibility of fitting the
model-calculated ozone depletion as a function of the many input parameters  in these CFC-
dominated scenarios. The atmospheric decomposition lifetimes of the important CFC's are
much greater than the maximum relevant ozone photochemical lifetime and characteristic
transport times in the model.  Additionally, the time scale of interest, several decades to
90 years, is also much longer than these model timescales. This means that, at  any given
time, the family of species including the  CFC's and their Cl-containing decomposition
products is essentially well-mixed in the atmosphere. In addition,  the ozone profile (and
thus  the total column abundance of interest  here) responds essentially instantaneously

-------
(i.e.. on the order of a model year) to changing abundance of stratospheric inorganic
chlorine-containing compounds and other trace stratospheric species. The first step in the
parameterization, then, is to calculate the expected changes in stratospheric inorganic Cl
(and Br) abundance as a function of time from  projected CFC (and Halon) emissions.
From this value the combined ozone perturbation  may be estimated, given assumptions of
future abundances of CO2, N2O and CH4 and their corresponding sensitivity coefficients
for ozone column change, which can be estimated from 1-D model runs.
   The trends assumed for CH4, N2O and C02 are projected for changes in surface abun-
dance or mixing ratio,  not source  emissions,  as the independent variables  in the param-
eterization equation. The currently observed increases in the atmospheric  abundances of
N2O and CH4 presumably indicate that the respective source strengths exceed the atmo-
spheric sinks and the observed abundances are therefore functions of the emission histories
over a period of a few of their respective atmospheric lifetimes (>  100 years for N20). By
specifying abundances rather than  source strengths, the effects of N2O and CH4 at a given
time are made essentially independent  of their emission history  and  the  parameterized
representation of results is greatly simplified. The drawback to this choice is that feedback
processes in the troposphere, which are important in determining the methane abundance
given a source strength, are not included.
   The parameterization procedure uses time-dependent historical and projected emis-
sion  fluxes (base year 1985) for the individual CFC and Halon compounds considered and
specifications for the abundances of CO2, N2O and CH4, usually relative  to the current
(1985)  ambient value.  The change in the abundance of upper stratospheric total  inor-
ganic chlorine-containing species (A [Cl,] = A ([C1] + [C1O]+[C1ONO2] + [C1NO2]+[HOC1] +
[HC1])) is the principal variable controlling change in the total ozone column. The A Clz
value is derived from integrating the time history of CFC emissions, taking into account the
model-derived atmospheric lifetimes of each CFC compound.  An ozone column depletion
can be associated with the calculated A Clz, based on a curve representing the complete
model's response to  CFC perturbations. Changes in the abundances of CH4 and N2O can
alter this nonlinear relationship. CO2, CH4, N2O and bromine emissions also have direct
effects, both positive and negative, on the ozone column change, which are taken to  be
additive to the primary CFC perturbation.

2.1.  Projected Abundance of Inorganic Stratospheric Cl

   The CFC perturbation dominates in most of the scenarios discussed in CW (1986) and
is the principal variable in this parameterization. The impact on ozone of increasing Cl
species abundance occurs first and largely in the upper stratosphere, so the quantity of
interest is the abundance of total inorganic Cl around 35 km. Vertical profiles of the abun-
dance of the CFC's of interest here show an essentially constant mole fraction throughout
the troposphere, that is, the CFC's are well-mixed in the bulk of the atmosphere. Vertical
mixing in the lower and mid stratosphere is, however, slow enough that at some altitude,
the increasing flux of solar radiation of the wavelength necessary to induce CFC photode-
composition causes a relatively rapid decline  in the CFC mole fraction. The position and
steepness of the  fall-off of the CFC mole fraction is controlled mostly by  the number of
Cl atoms in the CFC species. For CFC's containing three or  more Cl (CFC-11, CFC-
113,  CC14, and CHaCClj) most of the photodecomposition occurs below 35 km.  CFC-12

-------
and CFC-22 contain only two and one Cl, respectively, and show a slower decrease with
altitude.
   The total abundance of a given CFC in the atmosphere is determined by its emission
history and  atmospheric lifetime.  Here we define lifetime, T,  as  the total  atmospheric
burden divided by the CFC source strength at equilibrium. This 'replacement' time is
equivalent in a single box model of the atmosphere to the exponential decay lifetime of
the atmospheric abundance of the CFC following cessation of emissions.  From the CFC
abundances and the knowledge that most of the CFC's are substantially photodecomposed
at 35 km, the stratospheric abundance of inorganic Cl can be directly estimated from the
assumption of partial (CFC-12 and CFC-22) or complete (CFC-11,  CFC-113, CC14 and
CHsCCls) liberation of Cl. The CFC lifetimes in the model depend  on short wavelength
solar uv flux, singlet atomic oxygen abundance (O(1D)), and hydroxyl radical abundance
in the case of CHaCCla and CFC-22. These quantities all depend directly on the ozone
profile, so that the instantaneous CFC lifetimes will change as ozone is perturbed.
   The following input is needed to calculate the expected stratospheric inorganic chlorine
abundance at 35 km as a function  of time:
— Annual emissions for chlorine-containing species with lifetimes longer than about 10
years (for example, units of 106 kg),
— Conversion factors from 106  kg to ppb Cl for each CFC species  (based on molecular
weight, number of Cl atoms and mass of atmosphere),
and
— Equilibrium atmospheric lifetimes for the Cl source species.
The factor converting mass of emission to ppb emitted Cl is simply given by the product of
the number of Cl atoms .in the CFC molecule and Avogadro's number, divided by the CFC
molecular weight,  the column  number density of the atmosphere and the surface area of
the earth.  With appropriate conversions for units, the individual CFC conversion factors
are given by

           Conversion Factor = (5A77E - 3) * # (Cl)/Molecular Weight (g)        (1)

   The ranges of  CFC lifetimes were evaluated from a series of 1-D  model runs in which
constant source fluxes of each CFC were integrated over model times in excess of three
lifetimes for the CFC considered. The lower limit of the lifetime ranges corresponds to runs
with total column  ozone depletions around 10% between an atmosphere in equilibrium with
a constant specified emission and an atmosphere free of the respective CFC. The model-
derived lifetimes and CFC conversion factors are listed in Table 1.  Cunnold et al (1986)
have reported observed lifetimes for CFC-11 and CFC-12 based on analysis  of data from
the Atmospheric Lifetime Experiment. Their results, appropriate to  January 1, 1981, are
74(+31 -17) years for CFC-11 and lll(+222 -44) years for CFC-12.

-------
                                     TABLE 1
Cl Source Species
CFC-11 (CFC13)
CFC-12 (CF2C12)
CC14
CH3 CC13
CFC-113 (CFC12CF2C1)
CFC-22 (CHF2C1)
Lifetime (years)
76-84
139-150
67-75
8.3-12
92-100
22-33
Conversion Factor
1.196E-4
9.06E-5
1.424E-4
1.232E-4
8.91E-5
6.33E-5
    Two additional factors must be included in the estimation of stratospheric inorganic
Cl abundance. Both the model and the real atmosphere have a time constant for mixing a
surface-released tracer completely in the troposphere and stratosphere. The approximate
mixing time in the model is 3.5 years. Second,  the incomplete dissociation of CFC-22 and
CFC-12 at 35 km must be taken into account. Multiplicative factors of 0.53 for CFC-22
and 0.93 for CFC-12 are included to account for the incomplete photodissociation of these
CFC's within the altitude region of greatest effect on the ozone abundance.

    The contribution  of a given year's emissions td stratospheric inorganic chlorine will
decay exponentially according to the particular CFC species lifetime.  For  example, total
inorganic chlorine arising from 1985 emissions of CFC-11 decay according to  the expression,
       AC/Z CFC-n(t) = 1.196£ - 4 * (1985 CFC - 11 emissions) * e-(«-i9«s)/76.5
                                   * (I- e-(t-1985)/3.5(mixing timeh                 /2)

Summing over all CFC species and all years of emissions,
               AClz(t) = A Clg (emissions prior to t0)
                       +   ^^   {conversion factor (j)
                         j=CFC'»
                         t
                       *  ]T emissions (;,t) * e'C-'W) * (l - e^')/3-5)}            (3)
2.2. Estimating Ozone Sensitivity To Changes in Stratospheric Ol

   In summing Cl emissions from the various CFC species we make the assumption that
the effects on ozone will not depend on the particular mix of CFC emissions that results
in a given stratospheric inorganic Cl species increase. Differing relative efficiencies, based
on mass of emissions, of individual CFC's for perturbing the total  integrated vertical
ozone column have been discussed previously (Wuebbles, 1983). Values quoted ranged
from 0.05 (CFC-22) to 1.11 (CC14) relative to CFC-11 as unity, at steady state with
constant emissions. Relative efficiencies using the current chemistry calculated by mass
of emission, by mole of Cl emitted both unweighted and weighted by atmospheric lifetime

-------
and by inorganic Cl produced at 35 km, are given in Table 2. These were calculated from
1-D model results  for constant emissions  of a single CFC species to steady state, with
emissions  adjusted to produce an ozone depletion  of around 7.5%.  Most of the spread
in relative efficiencies by mass results from differences in the  number of Cl atoms  per
molecule, molecular weight and atmospheric lifetime.

                                    TABLE 2
Compound
Relative Efficiencies
      Mass
Mole Cl
Mole C1H
[Cl,] @ 35 km
CFC-11 (CFC13) 1.0
CFC-12 (CF2C12) 1.0
CFC-113 (CF2C1CFC12) 0.78
CFC-22 (CHF2C1) 0.05
CC14 1.06
CH3CC13 0.10
+ Normalized by lifetime.
1.0
1.32
1.05
0.08
0.89
0.10

1.0
0.73
0.88
0.30
1.01
0.91

1.0
0.73
0.90
0.68
1.14
0.94

    Vertical profiles of local ozone change, normalized to 7.5% total column ozone depletion,
for  the six CFC's considered are shown in Figure 1. The profiles for CFC-11, CFC-113,
CC14 and CH3CC13 are similar, while the profiles for CFC-12 and CFC-22 show an ozone
increase in the lower stratosphere and larger ozone depletions around 35 km than the other
CFC's. Since photodecomposition occurs at higher altitudes for CFC-12 and CFC-22 than
for  the other species, the lower stratosphere contains relatively smaller inorganic chlorine
abundances.  The ozone increase for CFC-12 and CFC-22 results from the interference of
CIO with the NOX catalytic cycle  dominant in the lower stratosphere.  This  interference
has been saturated by the larger inorganic Cl abundance in the lower stratosphere for the
other CFC's.  Comparison  of vertical profiles of local ozone change from model results
with smaller specified individual CFC emissions (normalized to a 2% total column ozone
decrease) show positive ozone changes in the lower stratosphere for CFC-11 and CFC-113
as well (Figure 2).
    Figure 3 shows local ozone change profiles normalized to equal abundance of inorganic
Cl  at 35 km for  the steady state  individual  CFC  emission model  runs.  On  the basis
of inorganic Cl abundance at  35 km the upper stratospheric ozone decrease  is  now very
similar for all CFC's.  Differences remain in  the lower stratosphere for CFC-12 and CFC-
22,  resulting from the lower inorganic Cl abundance around 20-25 km. But in multi-CFC
scenarios in which the inorganic Cl abundance grows sufficiently large, Cl contributed by
CFC's other than CFC-12  and CFC-22 would be expected to contribute to saturating
the CIO  interference with the  NOZ catalytic cycle  and lead to ozone decreases even in
the lower stratosphere.  The relative efficiencies for CFC-12 and CFC-22 would thus be
enhanced compared to the single CFC steady state calculations. Based on the CFC-11 low
emission level results, the sign of ozone change in the lower stratosphere reverses when the
inorganic Cl abundance reaches about 3.8 ppb in the upper stratosphere, somewhat higher

-------
than observed current levels but much smaller than most predictions for Cl levels  in the
next century.
    Contributions to the total  Cl abundance by the' individual CFC's in a multi-CFC
scenario, with respect to column ozone depletion and based on the evidence above, can be
assumed additive and the change in ozone column can be functionally related to the total
upper stratospheric inorganic Cl species abundance.  This assumption will produce errors
in estimation for scenarios with relatively small emissions including contributions largely
from CFC-22 or CFC-12, only.
    Figure 4 shows the model-calculated total column ozone response to increasing strato-
spheric inorganic Cl abundance from a multi-CFC run in which other trace species (CO2,
CH4, N2O and Halons) were held fixed. The response is nonlinear, showing three regimes.
In the first regime, with total column ozone depletion less than roughly 20%, the total col-
umn ozone change is composed of local decrease in the upper stratosphere and either local
increases in the lower stratosphere or  decreases mitigated by the interaction of increasing
CIO with the dominant NOZ catalytic ozone destruction cycle. When stratospheric in-
organic  Cl has increased to  about 20 ppb and total  column ozone depletion has reached
about 25%, the slope of the sensitivity  curve steepens  as the mitigating interaction of
CIO with NOZ is saturated and the C1OX catalytic loss cycle for ozone becomes dominant
throughout the stratosphere. The tail of the sensitivity curve at  very large ozone depletion
again becomes less steep. The shape of this curve is the  primary representation of the
behavior of the 1-D model with respect to column ozone depletion caused by increases in
stratospheric inorganic chlorine.  Interactions of the chlorine-ozone photochemistry with
changes in N2O and CH4, which will be discussed in slightly more detail below, are treated
as modifications to this curve.
    The  inverse hyperbolic  sine function provides a convenient basis  function  with  the
necessary S-curve shape.  An approximate  unweighted  nonlinear least-squares fit to  the
curve in Figure 4 gives the expression
            AO3(%) = 14.64{asinh [0.337(20.82 - AC!2)
                             -asinh [0.337*20.82]},         AC/2in ppb.         (4)

Note that the expression is simply a numerical interpolation of the model-derived curve,
with no theoretical  basis for its form.  Although it is defined for all values of Clz, it is
applicable only within the  limits of the validity of the 1-D model. Clearly the expression
fails for very large increases in A  Clz since it is unbounded negative for arbitrarily large
A Clz.  The complete 1-D model should also be expected to be incorrect for very large ozone
depletions (say >  30%), to the extent that the structure of the atmosphere, implicit in the
specified eddy diffusion coefficient profile, changes  as the stratospheric heating (through
ozone solar absorption) changes.

-------
2.3.  Effects of CH< and N2O on 63 Response to Cl Increases

    Both NjO and CH4 affect ozone through indirect as well as direct processes.  The direct
effects (e.g.  ozone catalytic depletion and photochemical ozone production, respectively)
are discussed below. The principal indirect effects result from interaction with the C1OZ
(= Cl -f CIO) catalytic cycles. The formation of chlorine nitrate (C1ONO2) in the reaction
                          CIO + NO2 +M = CIONO2 + M

reduces the efficiency of both NOZ and ClOj odd oxygen depletion cycles. Past the point
at which all available NOZ is effectively converted to C1ONO2, additional increase in C1OZ
is significantly more effective in reducing ozone. Increases in N2O, the precursor of strato-
spheric NOX, would tie up C10X  as C1ONO2 and extend to higher levels of stratospheric
inorganic chlorine abundance the transition from smaller to larger C10X sensitivity.
    The reaction of atomic chlorine with methane

                              Cl + C#4 = HCl + CH3
is a major sink for the reactive ClOr radicals,  Cl and CIO. For higher projected methane
abundances, a smaller fraction of the available inorganic chlorine is present in the active
radical forms and the impact  on ozone is diminished. Several of the pertinent reactions
for methane photochemistry have significant activation energies,  so that the  inclusion of
stratospheric temperature feedback has a significant effect on the sensitivity of ozone per-
turbations  to CH4 increases.
    In equation 4, the interaction of N2O with the effect on ozone of  inorganic chlorine
increases should appear as a change in the coefficient representing the  position of the in-
flection point on the abscissa in Figure 4. A methane change, which affects the  ClOr/(total
Cl) ratio, should appear in equation 4 as a scaling factor for the inorganic Cl abundance
term. From the results of runs  (Figure 5) that included either N2O or CH4 changes coupled
with the CFC perturbation depicted in Figure 4, the following expressions were determined

                 AO3(%)  = 14.64 { asinh [0.337 (20.82 RN,O - AC/,)
                            -asinh [0.337 * 20.82 RNaO]  }                          (5)
and
         AO3(%) =  14.64 { asinh [0.337 (20.82 - AC7Z exp(-0.2 (RCHt - 1))]
                               -asinh [0.337 * 20.82] }                             (6)

where
                                ,O = N2O(i)/N2O(1985)
and                                                                              (7)

                             RCH4=CH4(0/CH4(1985).

-------
The differences between the solid and dashed lines represent the direct effects of N2O and
CH4  on ozone, which will be accounted for below.
   Methane increases  also affect  the  abundance distribution among the nitrogen  oxide
species (NO, NO2, N03, N2O5, HNO3, HNO4, HNO2), so that the effects of CH4 and N2O
increases are coupled directly, as well as  through photochemistry involving Cl species.
The magnitude of this coupling can be estimated from combined multiple species scenario
model  runs (taking into account the small direct effects on ozone of the various trends
discussed below). The following expression appears to work fairly well for a range of CFC
scenarios:
A03 (%)  = 14.64 {asinh [0.337 (20.82 (1 +      - ) _ &ciz * exp (-0.2 (RCH4 -
                  - asinh [0.337 * 20.82 (l + R™~1)]}                          (8)
    One aspect of CFC-related ozone perturbation that is not directly included in the
expression above is sensitivity of the result to the model's ambient stratospheric abundance
of nitrogen oxide species. If the ambient level were actually higher (lower) than the model
value, the onset of the steeper portion of the  inorganic C1/O3 sensitivity curve would be
pushed to higher (lower) abundances of stratospheric inorganic chlorine. However, since
the total nitrogen oxide abundance is not an external variable in the model but is a derived
quantity, we will here consider changes in this value only if they result from change in the
N2O source emission.
    A slightly different implementation of equations (3) and (8) has been used in the draft
EPA document, "An Assessement of the Risks of Stratospheric Modification" (Hoffman,
ed., 1986).  The calculated 1-D total column ozone changes in the single CFC steady state
runs were compared to the ozone change predicted by equation (8a), a variant of equation
(8)  using total  inorganic chlorine abundance  at the top of the model atmosphere.  Total
dissociation of all CFC compounds, including CFC-22 and CFC-12, was assumed.
A03 (%) = 14.58 {asinh [0.332 (21.05 (l +      - -) - AC7Z * exp (-0.15(RCH< ~
                  - asinh [0.332 * 21.of (l +   %P ~  )]}                         (8a)
Efficiencies for perturbing the ozone column were thus produced for the various  CFC
compunds relative to the parameterization result.  These relative efficiencies were than
used as multiplicative corrections in equation (3), where they can be subsumed into the
conversion factors. These alternatives to the values given in Table 1 are given in Table la
below.
                                         10

-------

Cl Source Species
CFC-11
CFC-12
CC14
CH3CC13
CFC-113
CFC-22
Table
Lifetime (years)
76.5
138.8
67.1
8.26
91.7
22.
la
Relative Efficiency
1.14
0.84
1.20
1.19
1.06
0.45

Conversion Factor
1.363E- 4
7.61E- 5
1.719E- 4
1.46E- 4
3.44E- 5
2.06E- 5
2.4. Estimating Direct Ozone Sensitivities to Non-CFC Species

    While the scenarios of interest can involve order of magnitude increases in the total
stratospheric inorganic chlorine abundance over the next 90 years, increases in other trace
species over this period, currently projected, are usually less than a factor of two. The
direct effects on the total ozone column of these increases, individually, are on the order of
a few  per cent, compared to the tens of per cent changes possible for large CFC increases.
These smaller direct effects of the other trace species are treated additively in this note.

2.4.1.  Carbon dioxide (CO2)

    Changes in C02 affect the ozone abundance through the  decrease  in stratospheric
temperature that accompanies increased infrared emission to space by CC»2.  The rates
of many kinetic processes in  the stratosphere  are sensitive to temperature, so that C02
increases affect most  of the  ozone-controlling photochemical processes to some degree.
Atmospheric CC>2 may as much as double over the next century, but for changes of this
magnitude or smaller, the effects  of CC>2 on  ozone are small  enough to be considered
additive to the ozone perturbations of CFC's and other source species.  The expression

                     A0?°' (%)  = 3.6 In [CO2(t)/CO2(lQ85)}.                   (9)
adequately explains the  differences among multiple scenario 1-D model runs with CC>2
trends varying from zero trend to a  doubling of 1985 abundance in 2077, as well as step
function increases  of  10 and 100% in  the CW  (1986) 1-D model representation  of the
current atmosphere.

2.4.2.  Methane (CH4)

    The direct effect on ozone  of methane increase arises  from the oxidation of CH4  by
hydroxyl radical in  the presence of NO, in a process analogous  to photochemical oxidant
(smog) production in  urban areas.  Fitting the ozone change in multiple-species scenario
model runs as a linear additive term to the CFC-caused ozone perturbation gives a sensi-
tivity  of about 3.75. The expression  is


                                « (%)  = 3.75 (Ren. ~  1).                      (10)

                                         11

-------
2.4.3. Nitrous oxide (N2O)

   The direct effect of N2O increases on ozone results from increasing the NOZ abundance
and the rate of odd oxygen loss through NOZ catalytic cycles in the mid stratosphere. Thus
the direct effect  of N2O increases is reduction  in ozone while the  indirect effect through
the C1OZ interaction is opposing the chlorine-caused ozone reduction.  From the results of
the CFC/N2O 1-D model run and other scenario runs, the direct ozone/^O sensitivity
in the linear regime of small changes is about 7. The corresponding expression is
                                  (%)  =  -7.  (RN,0 - 1).                       (11)
2.4.4. Halons (bromocarbons)
    Although the bromine-catalyzed stratospheric odd-oxygen loss processes are substan-
tially more efficient per atomic bromine input than the C1OZ cycles, the bromine pertur-
bations at currently envisioned emission rates remain relatively small. The stratospheric
abundance of the various inorganic bromine-containing species can be estimated in the
same way as that outlined for stratospheric inorganic  Cl above, ignoring in the case of
Halons the very small pre-1985 historical emissions.
         ABrr 1301(f) = .0368 {  (1301 emissions (»)) e-(«-»)/ioi (i _ e-(«-t)/3.5jj

                                                                               (12)

        ASr2 1211 (0 = .0331 {(1211 emissions (i)) e^-)/12-9 (l - e-('"')/3-5)},

                       emissions in 106 kg/year, ABrz in ppt.

    The 101 year model-derived atmospheric lifetime for Halon 1301 is in reasonable agree-
ment with the value of 110 years estimated by Prather et al. (1984). For Halon 1211, the
LLNL  1-D model-derived atmospheric lifetime, for a run during which ozone decreased
about  10%, is 12.9 years, or about half the 25 years lifetime estimated by Prather and
coworkers.  For the expected small stratospheric abundances,  direct catalysis of ozone
loss should dominate the  effects of any ClOz-like interaction with NOZ photochemistry.
Based  on individual runs  containing either 1301 or 1211 emissions and on the difference
between multiple scenarios runs (using mid-range CFC emissions) with and without in-
creasing emissions of 1301 and 1211, a linear expression of the relationship of stratospheric
bromine increase and column ozone decrease is given by the expression

              AO3 Br' (%} = -0.0302 ABrz  (1301) - 0.0618 ABr2  (1211).          (13)
2.5 Summary

    The sum of the total column ozone perturbation values calculated in equations (8)
through  (11) and (13) is the parameterized approximation to the total column ozone
depletion that would be calculated by the actual 1-D model run.
                                         12

-------
3.  APPLICATION OF PARAMETERIZATION

   The contributions of historical (pre-1986) CFC emissions to the future change in strato-
spheric inorganic Cl abundance have been calculated using expression (2) (to=1911 with
no prior emissions) and historical emission data taken from Wuebbles et al. (1984).  The
data tabulated below (Table 3) for 5-year intervals from 1985 are also plotted in Figure 6.
These changes must be added to the Clz introduced by the projected future emissions.
                                        13

-------
YEAR
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075
C12 ppb
2.259
2.304
2.180
2.053
1.943
1.848
1.764
1.688
1.618
1.553
1.492
1.436
1.383
1.333
1.286
1.242
1.200
1.161
1.124
TABLE 3
Cl, ppb
0.000
0.045
-0.079
-0.206
-0.316
-0.411
-0.495
-0.571
-0.641
-0.706
-0.767
-0.823
-0.876
-0.926
-0.973
-1.017
-1.059
-1.098
-1.135
    The calculated changes in total stratospheric inorganic Cl from equation (3) for three
CFC scenarios are compared below (Table 4)  to the 1-D model calculated abundance
changes (t0=1985).
                                    TABLE 4
YEAR
1985
1995
2000
2015
2025
2035
2045
2055
2065
2075
LOW
1-D
0.
1.04
2.22
3.53
4.97
6.52
8.22
10.07
12.09
14.34
Fit
0.
0.96
2.11
3.38
4.75
6.21
7.82
9.57
11.52
13.67
MID
1-D
0.
1.05
2.41
4.33
6.78
9.71
13.2
17.2
21.7
26.8
Fit
0.
1.01
2.59
4.62
7.07
9.94
13.3
17.2
21.6
26.8
HIGH
1-D
0.
1.36
4.27
11.5
32.5





Fit
0.
1.54
4.96
14.1
38.0





   The method works reasonably well when the upper limits of calculated lifetimes are
used in the low emission case and the lower limits in the middle and high emission cases,
in which ozone depletion was substantial. Total inorganic chlorine abundance is somewhat
underestimated in the low case and overestimated in the high case, as a result of the
dependence of CFC lifetime 6n the ozone profile.
                                       14

-------
   The C02, NaO- CH4 and total Br abundances at 10 year intervals from 1985 in the
low, mid and high case scenarios are listed in Table 5.

                                   TABLE 5

YEAR         CO2 (ppm)   N20 (ppb)   CH4 (ppm)     Br (ppt)    Br (low case)
1985
1995
2005
2015
2025
2035
2045
2055
2065
2075
344.5
362.0
382.8
407.4
436.4
470.3
509.9
555.9
609.1
670.7
303.1
310.8
318.6
326.7
334.9
343.4
352.1
361.0
370.1
379.5
1.756
1.939
2.142
2.366
2.614
2.887
3.189
3.523
3.892
4.299
0.0
3.8
10.9
22.1
36.7
53.8
72.8
93.9
117.7
144.3
0.0
1.9
5.4
11.1
18.3
26.9
36.4
46.9
58.8
72.2
    Applying  equations (4)  and (5), using the change in total inorganic Cl abundance
calculated above with the prescribed changes in CH4 and ^O gives the following estimated
CFC contributions to total column ozone depletion (Table 6):

                                    TABLE 6

                                O3 % (Equation 8)
YEAR                      LOW               MID	HIGH
1985
1995
2005
2015
2025
2035
2045
2055
2065
2075
0.0
-0.69
-1.48
-2.32
-3.21
-4.17
-5.25
-6.43
-7.75
-9.20
0.0
-0.75
-1.87
-3.29
-5.08
-7.38
-10.5
-14.7
-21.0
-31.5
0.0
-1.13
-3.71
-13.4
-39.8 (@2020)





                                       15

-------
   The estimated magnitudes of the direct effect of the other trace species are (Table 7
                                   TABLE 7
YEAR            CO2          N2O          CH4          Br         Br (low)
1985
1995
2005
2015
2025
2035
2045
2055
2065
2075
0.
.0.18
0.38
0.60
0.85
1.12
1.41
1.72
2.05
2.40
0.
-0.18
-0.36
-0.54
-0.74
-0.93
-1.13
-1.34
-1.55
-1.76
0.
0.39
0.83
1.30
1.83
2.42
3.06
3.78
4.56
5.43
0.
-0.15
-0.44
-0.88
-1.47
-2.15
-2.91
-3.76
-4.71
-5.77
0.
- 0.08
- 0.22
- 0.44
- 0.73
- 1.08
- 1.46
- 1.38
- 2.35
- 2.38
   The calculated time profiles of the total column ozone change are the sums of Table 6
and Table 7 and are given in Table 8.
                                   TABLE 8
                      Total Column Ozone Change in per cent
YEAR
1985
1995
2005
2015
2025
2035
2045
2055
2065
2075
LOW
PAR
0.0
-0.37
-0.85
-1.40
-1.99
-2.64
-3.36
-4.15
-5.04
-6.02
1-D
0.0
-0.34
-0.78
-1.32
-1.95
-2.69
-3.57
-4.59
-5.81
-7.28
MID
PAR
0.0
-0.51
-1.46
-2.82
-4.60
-6.92
-10.02
-14.3
-20.7
-31.3
1-D
0.0
-0.37
-1.06
-2.32
-4.18
-6.70
-9.99
-14.4
-20.8
-31.8
HIGH
PAR
0.0
- 0.89
-3.30
- 12.9
- 39.4





1-D
0.0
-0.66
-2.80
-10.9
-29.5 @2020





                                       16

-------
    The time profiles of total column ozone change calculated from the expressions devel-
oped here are also plotted in Figure 7. The parameterization in the mid-range case has an
absolute rms deviation over 90 years from the full 1-D model calculation of .41% and the
differences are always within about 0.5% in absolute calculated ozone column depletion
relative to 1985.  Differences between the parameterization and the full  1-D model in the
other cases are generally  less than 2% except at very large ozone depletions. Figure 8
shows the comparison of the alternative implementation of equation (8a) and Table la to
the 1-D model results. Agreement between the, parameterization (dashed line) and  1-D
model (solid line) is about the same. It is not possible, however, to assume that differences
over the whole surface.of ozone column change as a function of source species scenarios
will remain within these limits. We want to reiterate  that these expressions simply provide
a means of estimating the 1-D model behavior to allow preliminary investigations of pos-
sible regulatory efforts. We do not claim that the column ozone changes calculated from
these expressions are accurate or reliable indications of what could actually arise in the
atmosphere over the next  90 years.

4. CONCLUSIONS

    We have shown here that it is possible to predict the 1-D model-derived total inorganic
stratospheric Cl  abundance that would arise from a given CFC emission projection in the
LLNL 1-D model. In the absence of changes in other species and for total chlorine increases
less than about 20 ppb, it  is also fairly straightforward to parameterize the resulting ozone
column depletions calculated in the 1-D model.  Properly treating the effects of coupling
substantial changes in methane with major CFC increases or accounting for the possibility
of greater or lesser NOZ abundance is more difficult.
    The parameterization of the ozone column sensitivity of a 1-D model presented here can
be used to estimate within about  2% in units of total column ozone change from t0=1985
the values that  would be calculated by the updated CW 1-D model.  But continuing
improvement in  our understanding and numerical representations of the troposphere and
stratosphere may well be expected quickly to supersede the  quantitative details  of this
study.
                                         17

-------
BIBLIOGRAPHY

Connell, P.S. and D.J. Wuebbles. "Ozone  perturbations  in the LLNL  one-dimensional
   model - Calculated effects of projected trends in CFC's. CH4, C02. N2O and Halons
   over 90 years," UCRL-95548, 1986.

Cunnold, D.M., R.G..Prinn, R.A. Rasmussen,P.G. Simmonds, F.N. Alyea. C.A. Cardelino,
   A.J. Crawford, P.J. Fraser and R.D. Rosen, "Atmospheric lifetime and annual release
   estimates for CFC13 and  CF2C12 from 5 years of ALE data," J.  Geophys. Res., 91,
   10797-10817, 1986.

EPA, -'An Assessment of the Risks of Stratospheric Modification." J.S. Hoffman, ed.. U.S.
   EPA, October, 1986.

JPL  (Jet Propulsion Laboratory) Publication 83-62, Chemical Kinetics and Photochemical
   Data for  Use in Stratospheric Modeling, W.B.  DeMore, ed.,  Pasadena, California,
   1983.

JPL  (Jet Propulsion Laboratory) Publication 85-37, Chemical Kinetics and Photochemical
   Data for  Use in Stratospheric Modeling, W.B.  DeMore, ed.,  Pasadena, California,
   1985.

Quina, T.H., K.A. Wolf, W.E. Mooz, J.K. Hammitt, T.W. Chesnutt and S. Sarma, "Pro-
   jected use, emissions, and banks of potential ozone-depleting substances, RAND Note
   N-2282-EPA, 1986.
Prather, M.J., M.B. McElroy and S.C. Wofsy, "Reductions in ozone at high concentrations
   of stratospheric halogens," Nature, 312, 227-231, 1984.
WMO (World Meteorological Organization), "Atmospheric Ozone 1985: Assessment  of
   Our Understanding of the Processes Controlling its Present Distribution and Change",
   Global Ozone Research and Monitoring Project — Report No.  16, 1986.
Wuebbles, D.J.,  "Chlorocarbon emission scenarios:  Potential impact  on  stratospheric
   ozone,"  J. Geophys. Res., 88, 1433-1443, 1983.
Wuebbles, D.J., M.C. MacCracken and F.M. Luther, "A proposed reference set of scenarios
   for radiatively active atmospheric constituents," U.S. Dept. of Energy Carbon Dioxide
   Research Division Technical Report DOE/NBB- 0066, 1984.
                                       18

-------
   FIGURE CAPTIONS
1.  Vertical profiles of ozone abundance change at steady state for a series of 1-D model
   runs with fixed constant emissions of individual C'FC compounds. Total column ozone
   change normalized to -7.5%.
2.  Vertical profiles of ozone abundance change at steady state for a series of 1-D model
   runs with fixed constant emissions of individual CFC compounds. Total column ozone
   change normalized to -2.0%.
3.  Vertical profiles of ozone abundance change at steady state for a series of 1-D model
   runs with fixed constant emissions of individual CFC compounds. Profiles normalized
   to equal increase in startospheric inorganic Cl abundance at 35 km.
4.  Total column ozone change in per cent as a function of change in stratospheric inorganic
   chlorine mixing ratio for the CFC-only reference scenario. Solid line is updated CW
   (1986) model, dashed line is parameterization fit.
5.  Total column ozone change in per cent as a function of change in stratospheric inorganic
   chlorine mixing ratio for combined CFC/CH4 and CFC/N^O scenarios.  Solid lines are
   updated CW (1986) model, dashed  line are parameterization fit.
6.  Contribution of historical (pre-1985) CFC emissions to future stratospheric inorganic
   Cl mole fraction.
7.  Total column ozone change in per  cent as a function of time.  Solid lines  are  time-
   dependent 1-D  multiple scenario calculations (CFC emissions  as in Figure 4, N2O
   @0.25%/year, CH4 @0.5%/year, C02 mid-range scenario from Wuebbles et al. (1984)).
   Long dash lines are parameterization fit using 1-D calculated changes in stratospheric
   inorganic Cl, short dash lines are parameterization fit using stratospheric Cl estimated
   from CFC emissions.
8.  Total column ozone changes in per cent as a function of time for alternative formulation
   (equation 8a and Table la).
                                       19

-------
E
-X
LU
Q
      55
     50
     45
     40
     35
-    30
     25
     20
     15
     1 0
      0
      -1 0
CFC-11
CFC-12
CCI4
CH3CCI3
CFC-1 1 3
CFC-22
  -6
-4
0
                CONCENTRATION CHANGE   1 0 11 ( mol cm"3)

-------
     55
     50
     45
     40
     35
-    30
LLJ
Q
     25
     20
     15
     I 0
CFC-11
CFC-12
CFC-113
CFC-22
      0
      -30    -25   -20    -15    -10    -5
                                   10
               CONCENTRATION CHANGE   1 0 10 ( mol cm"3)

-------
     55
LLJ
Q
     50 -
     45 -
     40 -
     35
     30
      25
      20
      1 5
      10
    CFC-11
    CFC-12
    CCI +
    CH3CCI3
    CFG-113
    CFC-22
       -8
-6
-4
-2
0
                CONCENTRATION CHANGE   10 11 ( mol cm"3)
                                ,-3

-------
cr
o>
o
CD
ex
CD
I
o
O
M
O
O
O
o
  0



 -5



-10



-15



-20



-25



-30



-35



-40



-45



-50



-55



-60
               1  i  '   I  '  I   '  I  '   I  '  I  '   I  '  I   '  I  '   I  '  i   '  i  '   i
        0
                      8
10   12   14   16   18   20   22  24   26   28   30
                    STRATOSPHERIC INORGANIC CHLORINE INCREASE   ( ppb )

-------
c:
CD
O
CD
Q.
LU
CD
X
O
UJ
2
O
M
O
_
O
O
<
O
                 \—•—I—'—I—'—I  '  I
                  CFC/CH4  \
        o
8   10   12   14   16  18   20   22   24   26   28   30
                                 INORGANIC CHLORINE INCHCASIE   (  ppb )

-------
                      CI2 MOLE  FRACTION CHANGE    ( ppb  )
                 I     I     I    I     I     I    I
                o   o   o    o   o   o   o
                                                       P    O
                                                       ro    :_i.
                                                           O   :_•.   ro
      IO
      o
      ro
      O
      o
      o
      ro
      o
      ro
      O
      ro
      O
-c
m
ro
O
GO
O
      ro
      O
      ro
      O
      CJ1
      o
      ro
      O
      05
      O
      ro
      O
      ro
      O
      CD
      O
                   ,   I

-------
                TOTAL  COLUMN OZONE CHANGE    ( per cent )
       OJ
       o
            i
            ro
            en
 I
no
O
 I
tn
m
     CD
     ID
     O
     ro
     o
     o
     o
     o
     o
     ro
     O
     ro
     O
ro
O
UJ
O
     ro
     O
     ro
     O
     en
     O
     ro
     O
     ro
     O
     ro
     O
     CD
     o
                                     I  I  I  I  I  I  I  I

-------
                 TOTAL  COLUMN OZONE  CHANGE    ( per cent )
        i
       CO
       o
             I
            no
            en
 I
r\i
O
                                      cn
 I
cn
m
>
ID
     to
     10
     O
     ro
     O
     o
     o
     ro
     o
     O
     r\3
     O
o
cu
o
     ro
     O
     IN3
     O
     CJ1
     o
     O
     01
     O
     IN3

     O
     O
     OD
     O
                     I    I  I          I     I

-------
GLOBAL MODELING OF THE ULTRAVIOLET SOLAR FLUX INCIDENT ON




                      THE BIOSPHERE
                    George N. Serafino




               Applied Research Corporation




                   8201 Corporate Drive




                 Landover» Maryland 20785








                           and








                    John E. Frederick




          Department of the Geophymical Scienci




                The University of Chicago




                 5734 South Ellis Avenue




                 Chicago* Illinois 60637

-------
                          Abstract




     This report summarizes an algorithm designed to




estimate the ultraviolet solar flux that reaches the Earth's




surface at any location on the globe and time of year.




Inputs consist of global ozone abundances*  terrain height,




the distribution of cloudcover, and the albedos of clouds




and the underlying surface.  Intended users of the algorithm




include atmospheric scientists* the photobiology community,




and environmental policymakers.

-------
I. Introduct ion



     The interaction of solar radiation with the Earth and




its atmosphere is closely coupled to the planet's ability to




support life.   Ultraviolet solar radiation likely initiated




the chemical processes which led to formation of the first




organic molecules on the primitive Earth (eg. Ponnamperuma»




1981), while the development of a substantial ozone layer




created a surface environment where complex self-rep 1icating




molecules could evolve.  The decreases shown by both the




absorption cross section of ozone and the DNA action




spectrum at wavelengths between S80 and 320 nm provide




persuasive evidence of the coupling that has existed between




the geophysical and biological realms which ultimately




provided for the evolution of higher life forms.




     Issues of more immediate practical concern center on




the observation that the incidence of various skin cancers




shows latitudinal variations.  This appears related to the




biologically active ultraviolet flux reaching the surface of




the Earth.  While this fact alone is of great significance?




couplings of a more subtle nature apparently exist between




the radiation  environment and biological systems.  A prime




example is the work by DeFabo and Noonan (1983) which




indicates a link between ultraviolet radiation dosage  and




suppression of the immune system in laboratory mice.




Photobiologists have adopted the term UV-A to refer to




radiation over the wavelength range 320-^00 nm, while  UV-B

-------
denotes the region 580-320 nm.  Absorption by ozone and




atmospheric scattering reduce the solar UV-B flux at the




surface of the Earth to a small  fraction of what would




otherwise exist.  The UV-A, being outside the range of




strong absorption by ozone, experiences much less




attenuat ion.




     This report describes the conceptual formulation of an




algorithm designed to predict the UV-B and UV-A radiation




fluxes as functions of wavelength at any point on the Earth




for any time of year.  Papers which summarize the




mathematical methods used in the code already exist in the




published literature.  We make reference to these rather




than presenting details here.  The algorithm utilizes global




scale ozone measurements obtained by the Solar Backscattered




Ultraviolet (SBUV) Spectral Radiometer carried on the Nimbus




7 satellite.  We combine this data set with additional




information on cloudcover and cloud transmission obtained




from independent sources.  While the development of the




algorithm is an exercise in radiative transfer and




atmospheric science, we intend the final product to be a




tool for use by the photobiology community and environmental




po1icymakers.

-------
 II.  The Radiative  Transfer  Formulation



      W« divide the atmosphere  into  two parts*  (1)  the clear



 atmosphere above any  cloudtops  and  <2> the cloud  layer,  the



 atmosphere beneath the cloud  (the "sub-cloud  layer"), and



 the  ground.   In the absence of  clouds only case 1  is



 required.   When clouds are  present  we merge a  model of this



 portion of the atmosphere onto  the  base of the clear sky



 calculation.   We assume that the lower boundary of the clear



 atmosphere* being  cloudtops or  ground, is a Lambertian



 surface of known albedo.  A radiative transfer calculation



 which  includes all  orders of multiple scattering  and



 absorption by  ozone then gives  the  direct and  diffuse



 components of  solar flux incident on the cloudtops or, for



 clear  skies* on the ground.



     The downward  diffuse flux, F <2 ,6,T), for a wavelength



 "X i  solar  zenith angle 6» and optical depth T, can be



 expressed  as the sum of an  atmospheric scattering component



 f\ (^ »6»T> and a contribution arising from downward



 scattering of  radiation that has already been  reflected from



 the  lower  boundary  (Dave and Furukawa* 1966).
                 *                      *
         e,T> = F. < A »e»T> +• o<;\ ,R,T->F ( A , e, T >
where:



         Q(A ,R,f> - R / C1-RS(A ,T->]                (2)

-------
Here T- is the atmospheric optical depth above the




reflecting surface (ground or cloudtop) of albedo R,




S represents the backscattering power of the




atmosphere, and F (A ,6»T) measures the contribution from




flux that has already reflected from the lower boundary and




is then scattered back into the lower hemisphere.  The




advantage of the formulation in equations 1 and 2 is that




the quantities Fa » P3  » and s can be computed without




knowledge of the surface albedo.  In practice* we calculate




these terms using the Herman and Browning (1965) clear sky,




multiple Rayleigh scattering model.




     For clear sky conditions the formulation summarized




above produces the UV-B and UV-A flux at the ground as a




function of wavelength, solar zenith angle (local time), and




ozone amount.  Under  cloudy sky conditions, however, we must




define the transmission and reflectivity of the cloud-




subcloud-ground layer.   For this we use a two stream




radiative transfer model coupled with the "adding method"




for a multi-layer atmosphere developed by Lac is and Hansen




(1974).  This approach divides the atmosphere into a series




of homogeneous layers where each layer has a known




reflectivity and transmission.  Clouds occupy the uppermost




layers, while the bottom layer is the ground with a




transmission of zero.  The composite reflectivity and




transmission of the multilayer system is determined by




combining the reflectivities and transmissions of the




individual layers with proper account taken of multiple

-------
 reflections  of upward and downward directed fluxes.  Lacic




 and  Hansen  <197<»> have presented quantitative details of the




 technique.   The model adopts fractional cloudcover as a




 function of  latitude from Hughes <198<+).  We assume a




 mixture of thick  low clouds, with an optical depth for




 scattering of 30, and middle level clouds of optical depth




 15 (Stephens, 1978).  We assume cloud drops to be non-




 absorbing in the UV-B and UV-A.  However, absorption of




 radiation still occurs in the clouds owing to the




 tropospheric ozone amount included in the model.  The




 calculations assume that 85 percent of the downward




 radiation incident at the cloudtops is scattered into the




 lower hemisphere.  The derived transmission of the cloud-




 subcloud system multiplied by the total (direct plus




 diffuse) flux incident on the cloudtops from equation 1




 gives the flux at the ground.  Note that we assume all




 radiation transmitted through the cloud to be isotropic over




 the  lower hemisphere, consistent with a large optical depth



 for scattering.




     The reflectivity and transmission of atmospheric layers




 beneath the cloud deck are defined by expressions for a two




 stream model  as given by Coakley and Chylek (1975) and




 Joseph et al. (1976).  Each layer has a known optical depth




and an ozone amount based on climatology supplied with the




SBUV data set.   To obtain the flux  at the ground for a




c1imatological  fractional cloudcover,  we simply combine




values derived  separately for clear and cloudy skies using

-------
                                                            8
the weight* 1-f and f respectively, where f is the




fractional cloudcover at the latitude of interest.




      In principle one could compute the UV-B and UV-A fluxes



at the ground using a complete radiative transfer




calculation for any combination of wavelength, ozone




abundance, cloudcover, solar zenith angle,  and ground




reflectivity.  In practice this is not necessary.  Instead




we generated three sets of flux tables, one with the base of




the clear atmosphere at iOOO mb» another at 700 mb,  and the




last with the base at 400 mb.  Each table contains the




radiative transfer quantities of equations i and 2 for 23




wavelength bands which span the wavelength range 290 to 400




nm, 9 total column ozone amounts, and 13 solar zenith




angles.  Surface reflectivities corresponding to the ground




or cloudtops need not enter the tables in view of the form




of equation 1.  Each table allows interpolation to obtain




surface fluxes for any ozone value and local time, while a




combination of all three tables provides fluxes for  varying




terrain heights and cloudcover conditions.   This flexibility




allows the algorithm to predict the ultraviolet radiation




environment at any location on the globe for any time of




year by interpolation based on precomputed radiation tables.




Surface fluxes may refer to specific local  times or  to




averages over the daylight period at any location and date.

-------
III. The Input Data Sets




     The SBUV instrument provides the total column ozone and




vertical ozone profiles needed to evaluate terms in the




radiative transfer calculations.  We use SBUV column ozone




amounts averaged over one month time intervals and over all




longitudes in 10 degree wide latitude bands.  We associate




these means with the center of each month and latitude bin.




Interpolation in latitude and time then provides the ozone




amount for a specific location and day of the year.  Figure




1  illustrates the behavior of column ozone as a function of




latitude and month derived from SBUV.  We note that a very




recent revision in the SBUV data set uses improved




absorption cross sections and yields values approximately 6'/.




greater than those shown in Figure 1.  The current version




of the global radiation algorithm uses the updated ozone




results.  The extraterrestrial solar irradiance, ozone




absorption cross sections* and Rayleigh scattering cross




sections used in the calculations are from Chapter 7 of



WMO/NASA (1986).

-------
                                                            1O
IV.  Algorithm Operation and Sample Results




     The algorithm allows the user a high degree of




flexibility in selecting parameters for a given calculation.




Mandatory inputs supplied by the user aret (1)  latitude and




longitude, <2> day number of the year, 1  through 365,  and




(3) local time.  As an alternative to local time the user




can choose to compute mean fluxes over the daylight portion




of a 2
-------
                                                            11
largest fluxes, <+ watt* per square meter*  reach the ground




in the tropics because the sun is most  nearly overhead here*




and the atmospheric ozone amounts are relatively small.  The




major feature of Figure 2 is the large  variation in




radiation flux with latitude* especially during the winter




season.  In the Northern Hemisphere for December and January




the flux decreases by a factor of 10 between the equator and




50 degrees latitude.  During summer the latitudinal




gradients are much less pronounced than in winter*  and one




must move from the tropics to 60 degrees to experience a




factor of two decrease in flux at the ground.  There is very




little change in the lOiOO A.M. fluxes  in the tropics over




the course of a year.  At middle latitudes* however* the




seasonal cycle can range between a factor of two and ten




depending on location.




     A calculation analogous to that in Figure 2 could be




done for the UV-A spectral region.  Although the contours




would be similar in shape* the gradients would be much less




pronounced because of the greatly reduced absorption by




ozone at wavelengths longward of 320 nm.  Figure 3




illustrates this behavior by giving contours of the ratio of




UV-B to UV-A fluxes as a function of latitude and month at a




local time of 10:00 A.M.  Clearly, the UV-B flux is much




smaller than the UV-A, with  the ratio ranging from 2 to




7.5V..  The most significant  information in Figure 3 is the




differing latitudinal and seasonal gradients shown by  the




UV-B and UV-A.  As one moves from the tropics to 60 degrees

-------
latitude in winter, the UV-B flux decreases more rapidly




than the uV-A by a factor of three to four.  In summer  the




relative variation is much less than a factor  of two.




     The examples presented above illustrate latitudinal and




seasonal variations.   Future updates of the algorithm for




use in truly global studies should include longitudinal




variations in both fractional cloudcover and ozone.  For




many applications* however, the focus is on the radiation




environment at a specific location as well as on changes in




dose rates with parameters such as the ozone amount and




fractional cloudcover.  A separate report by H. Pitcher and




J. Scotto now in preparation will describe such studies.




including the comparison of model predictions with ground-




based measurements from Robertson-Berger meters.

-------
                                                            13
                         References




Coakley, J. A., Jr., and P. Chylek, 1975: The two-stream




approximation in radiative transfer: Including the angle of




incident radiation* J_. A_tmg_». Sc i . , 35, <»09-M8.








Dave, J. V., and P. M. Furukawa, 1966: Scattered Radiat ion




LD. the Ozone Absorption Bands at Selected Levels of a




Terrestr ial Ravleiqh Atmosphere, Meteor.  honogr., Vol. 7,




No.  29.








DeFabo,  E. C., and F. M. Noonan, 1983: Mechanism of immune




suppression by ultraviolet radiation in vivo  I. Evidence




for  the existence of a unique photoreceptor in skin and  its




role in photoimmunnology, J. Exp. Med., 157, 84-98.








Herman,  B. M., and S. R. Browning,  1965:  A numerical




solution to the equation of radiative  transfer, J. Atmos.




Sci.. 23,  559-566.








Hughes,  N. A., 1984: Global cloud climatologies: A




historical review, J. Climate Appl. Meteor., 53, 724-751.








Joseph,  J. H., W. J. Wiscombe, and  J.  A.  Weinman,  1976:  The




delta-Eddington approximation for radiative flux transfer,




J. Atmos.  Sci., 33, 175-204.

-------
Kalnay, E.» et al.» 1983: Documentation of the GLAS Fourth




Order GCM.  Vo lume £:  Model Documentat ion, internal report,




Laboratory for Atmospheric Sciences, NASA/Goddard Space




Flight Center, Greenbelt, MD. 20771.








Lacis, A. A., and J. E. Hansen, 197*»s A parameterization for




the absorption of solar radiation in the earth's atmosphere,




J. Atmos. Set . , 31, 118-133.








Ponnamperuma, C., 1981: The quickening of life, in Fire of




Life, Smithsonian Exposition Books, W. W. Norton and




Company, New York,  118-125.








Stephens, G. L., 1978: Radiative properties of extended




water clouds: Part  II, J. Atmos. Sci.. 35, 2111-2132.








WMO/NASA, 1936: Atmospheric Ozone 1985, World Meteorological




Organization, Geneva,  Global Ozone Research and Monitoring




Project, Report Number 16.

-------
                                                            15
                      List of Figures




Figure 1.  Contours of total column ozone (mi 11i-atmosphere-




centimeters) as a function of latitude and month derived




from the SBUV instrument.








Figure 2.  The latitudinal and monthly distribution of UV-B




radiation at the ground computed for clear sky conditions




and a local time of 10:00 A.M.  Contour values* in watts per




square meter, include all wavelengths between 290 and 320




nm.








Figure 3.  The ratio of solar energy flux in the UV-B from




Figure 2 to that in the UV-A (320-^00 nm) as a function of




month and latitude.  Values refer to radiation reaching the




ground for 10:00 A.M. local time and clear sky conditions.




Contours are in percent (7.5 means that the UV-B energy flux




is 7.5V. of that in the UV-A).

-------
    N
   so - COLUMN OZONE
   GO
   40
   20
o
D
t   o
  -20
   •40
  -60
  -80

    S
       N
D
M
M
                                MONTH
                                            Pltnire 1

-------
                                     .-2,
N
 85
 65
 45
 25
  5
 -5
-25
-45
-65
-85
    Total  UV-B  Flux  at Ground (W-M  )
                                  0.8. -
     N  0  J  F
               M  A  M  J
                  Month
J  A  S  0
                                     Figure

-------
  Ratio  UV-B  UV-A (Units: 1C)2)
N  0  J  F
M  A  M  J
  Month
J  A  S  0
                                Figure 3

-------