EPA/600/A-97/024

For Presentation at the Air & Waste Management Association's 90th Annual
Meeting & Exhibition, June 8-13,1997, Toronto, Ontario, Canada

97-RP94B.01


     Development  and  Testing of an  Improved Photolysis Rate
              Model  for Regional Photochemical Modeling

Shawn J. Roselle*, Jonathan E. Pleim*, Kenneth L.  Schere*

Atmospheric Sciences Modeling Division, Air Resources Laboratory, National Oceanic and
Atmospheric Administration, Research Triangle  Park, North Carolina 27711

*On assignment to the National Exposure Research Laboratory, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.

Adel F. Hanna, and Ji-Chcng C. Jang

MCNC-North Carolina Supercomputing Center,  Research Triangle Park, North Carolina 27709

Introduction

Almost all chemical reactions in the atmosphere are initiated by the photodissociation of a number of
trace gases.  These reactions are responsible for  most of the buildup of smog, which has detrimental
effects on human, animal,  and plant life.  In order to accurately model and predict the effects of air
pollution, good estimates must be made of the rates of  these photochemical reactions.  A  direct
measure of this photodissociation is the photolysis rate.  Many current air quality models  use crude
approximations to calculate the photolysis rates with little consideration of variations in vertical
profiles of temperature, ozone concentrations, aerosol concentrations, cloud parameters, and spectral
surface albedo.  However,  more accurate  estimates of photolysis rates can be produced  with
advanced radiative transfer modeling techniques  and with the utilization of detailed measurements.

An advanced computational model has been developed  to simulate the actinic flux and photolysis
rates of tropospheric species for air quality  and tropospheric photochemical modeling.  The model
combines advanced radiative transfer models with explicit computations of photolysis rates using
detailed  information on prevailing atmospheric conditions during simulation episodes, satellite ozone
data, and detailed characteristics of clouds,  aerosols and surface albedo.  In this paper we discuss the
advantages of using the refined model to simulate photolysis rates over a standard  lookup table
method.  We also discuss the sensitivity of  the photolysis rate simulations to the radiative transfer
scheme used as well as key parameters such as cloud, ozone profiles and surface albedo.

Background

Photodissociation is the conversion of solar radiation into chemical energy  to activate and dissociate
chemical species. Examples of species that photodissociate include many important trace
constituents of the troposphere such as NO2, O3,  HCHO, CH3CHO, HONO, the NO3 radical, and
H2O2. The accuracy of the simulation of  the entire chemical system is highly dependent upon the

                                                                                        1

-------
accuracy of the rates of photolysis which are the primary source of radicals to the system.
Photolysis rates (mirf1), also called j-values, are computed as
                                            K
where, F; is the actinic flux (photons cm'2 mirf1), o, the absorption cross section (cm2 molecule'1),
(J)A the quantum yield (molecules photon '), and A the wavelength (nm).  Absorption cross sections
and quantum yields are molecular properties  that are functions of wavelength and temperature and
are unique to each species.  Laboratory experiments have been conducted for many of the species
that photodissociate in the troposphere to measure the absorption cross sections and quantum yields.
Actinic flux, on the other hand, is a radiometric quantity that is  a measure of the integrated spectral
radiance over all solid angles per unit area. The spherical receiving surface distinguishes the actinic
flux from the more commonly measured irradiance, which is the radiance falling on  a horizontal
surface.  Thus, the actinic flux can be called  spherical spectral irradiance.  The actinic flux changes
with time of day, longitude, latitude, and season, and is governed by the astronomical and
geometrical relationships between  the sun and the earth; it is greatly affected by the  earth's surface as
well as by various atmospheric scatterers and absorbers. All of these factors make quantifying
atmospheric radiative transfer very complicated. Hence, correct model calculation of the temporal
and spatial variation of the actinic flux is critical to obtaining accurate photolysis rates for regional
and mesoscale episodic photochemical modeling.

Model Description

An advanced model (JPROC: J-value PROCessor) has been developed to calculate photolytic rate
constants  for regional air quality models. JPROC is largely based on  the Madronich's model for
Tropospheric Ultraviolet and Visible (TUV) radiation12.  JPROC adds to TUV the ability to calculate
photolysis rates for all grid cells of a specified Eulerian air quality model.  In our case, we are
applying JPROC to EPA's Models-3 Community Multiscale Air Quality Model  (CMAQ)3.  Other
features that have been  added to TUV include the ability to  compute photolysis rates for each
simulation hour, making use of datasets available from the Mesoscale Meteorological Model version
5 (MM5)4 and from other available sources.  JPROC is  flexible  in the specification of wavelength
bands, extraterrestrial irradiance data, ozone profiles, aerosol profiles,  cloud distributions, and
absorption cross section and quantum yield data.

Radiative Transfer
Radiative transfer models, including two-stream and multi-stream discrete ordinate approximations2,
are used for computing  the actinic flux.  Two-stream models make two basic assumptions; (1) the
phase function is completely isotropic; and (2) radiances in the upward and downward  hemispheres
are individually isotropic.  The two-stream approximations are limited in application  to cases where
the scatter is not highly anisotropic. The multi-stream radiative  transfer models  divide the radiation
field into  more than the two described above, and include all orders of multiple scattering.  This is
particularly important when considering radiative effects of clouds and aerosols.  JPROC uses these

-------
models to provide detailed descriptions of the radiative processes in the atmosphere; detailed
descriptions of clouds, aerosols, ozone absorption, oxygen absorption in the Schuman-Runge Bands5,
Rayleigh scattering6 and surface albedo are provided to the radiation models.  One of the most
critical elements in calculating the actinic flux is to describe  the structure of the atmosphere as
accurately as possible, making use of as much data as possible on the episodic meteorological and
chemical conditions.

Vertical Coordinate

The vertical resolution for calculating photolysis rates  in JPROC (i.e. the vertical grid and number of
layers)  is set to correspond with the air quality model's vertical coordinate system.  In the case of
Models-3/CMAQ, a terrain-following cordinate system is used in specifying the vertical  model
structure.  Depending on the model application,  there may be any number of vertical layers, although
typically the troposphere is subdivided into 6 to 30 layers.  Since the radiative transfer calculations
must consider the effects of the stratosphere, the vertical coordinate for JPROC extends above that of
CMAQ, through most of the stratosphere, to a height of 50 km.

Surface Albedo

An important parameter  for radiative transfer calculations is the earth's surface albedo.  The albedo
data given by Demerjian et al.7 have been used extensively in radiative transfer models.  These data
are given as a function of wavelength, but not of landuse type. Current air quality modeling domains
cover most of the United States, including a broad range of land surface types. Therefore, more
detailed spatial data are needed to fully resolve the details within modeling domains.  A  good
candidate for JPROC is MM5's spatially-resolved surface albedo data with solar zenith angle
variance.  However, MMS's data does not have the wavelength distribution needed for the radiative
transfer calculations.  Therefore, a procedure was developed in JPROC to make use of both datasets
to resolve the spectral and spatial distributions of the surface albedo.  MMS's surface albedo data are
treated as broad-band integrated values, which get distributed  across wavelengths using a normalized
profile of the albedo data from Demerjian et al.7.

Temperature Profile
Several factors in JPROC depend on the temperature profile, including ozone absorption, SO2
absorption, and the absorption cross sections and quantum yields for individual photolysis reactions.
To improve the accuracy of these calculations, spatially variant temperature profiles from MM5 are
used in JPROC to describe the temperature profiles within the troposphere.  Above the tropopause,
JPROC uses the U.S. Standard Atmosphere8 temperature profile.

Aerosols
The radiation field, especially in the atmospheric boundary layer, is greatly affected by aerosols.  The
important aerosol properties that affect the atmospheric radiative transfer are size distribution,
number density, and complex refractive index.  Currently, a single profile of aerosol number density9
is being used for all grid cells in JPROC. The utility of using dynamically-calculated aerosol
profiles, as simulated by the  Models-3 particulate model10 will be investigated in the future  as an
input to JPROC,  Such an  approach may result in considerable improvements in the accuracy of
photolysis rates especially  during regional haze episodes in the U.S.11.

-------
Ozone Absorption

Absorption by ozone is calculated using the most recent NASA recommendations12.  Ozone profiles
are set by taking the U.S. Standard Atmosphere8 O3 profile and uniformly rescaling this profile for
all model grid cells to match the integrated total ozone  column as measured by the Total Ozone
Mapping Spectrometer (TOMS) instrument aboard the sun-synchronous polar orbiting Nimbus 7
satellite. TOMS data  (currently Version 7) are archived and available at the National Satellite
Service Data Center (NSSDC)  in the form of digital daily maps with a resolution of 1  degree latitude
by 1.25 degrees longitude, for the satellite's operating period November 1978 - May  1993.  The
TOMS data provide a daily measurement of the total  column ozone as gridded information that feeds
into the radiative transfer model to calculate the actinic flux.   Bilinear interpolation  is used to  map
the TOMS data  to the model grid system.  In the absence of TOMS  data, the U.S. Standard
Atmosphere8 ozone profile is used.
Cloud Droplet Scattering and Absorption

Meteorological modeling includes the use of sophisticated cloud parameterization schemes'3 to
provide pertinent physical characteristics of clouds to air quality models.  These schemes provide
detailed descriptions of cloud dynamics and microphysics that are necessary for chemical and
radiative transfer parameter calculations. The radiative transfer model uses detailed cloud parameters
such as the profiles of the liquid  water content and droplet size distribution to calculate actinic
fluxes.

Cloud coverage is a critical parameter for calculating actinic fluxes and photolysis rates. JPROC
takes cloud cover information to  determine if the radiative transfer calculations will be for  either
clear or cloudy sky conditions.  For grid cells with completely clear skies, only one radiative transfer
calculation is performed using clear sky conditions.  For overcast grid cells, again only  one radiative
transfer calculation is made using the cloudy sky conditions. For cells with fractional cloud
coverage, the radiation model is run twice, once for the clear sky fraction and another time for the
cloudy fraction.  A weighted average is then performed on the actinic flux14 to determine a grid cell
average value.

Species Kinetic Parameters
Periodic updates of the absorption cross sections and quantum yields for different species are
published by different organizations12.  JPROC provides the capability of computing photolysis  rates
for any chemical mechanism, using absorption cross  section and quantum yield data specified by the
user.  Default sets have been set  up for  the Carbon Bond Mechanism  IV15, the Regional Acid
Deposition Model mechanism version 2 (RADM2)16, and the SAPRC  mechanism17. In addition,
users can  deviate from these standard sets to test other data, including the revisions suggested by
NASA12.

Results

Table lookup
One  method being used to specify photolysis rates for air quality models involves the calculation of
rates for various predefined conditions, such as different  zenith angles, altitudes, albedos, total ozone

-------
column values, etc., and the interpolation of photolysis rates from this predescribed table using the
current conditions (e.g. current zenith angle, etc). This method has been employed in the Regional
Acid Deposition Model (RADM)'8. We will use RADM's approach as the basis for comparison.
RADM's table has three dimensions of interpolation, including the hour from local solar noon,
latitudinal band, and altitude.  Values in the table have been computed using the Delta-Eddington
two-stream method19 for calculating the actinic flux.  In addition, the vertical profiles of temperature,
air number density, and ozone vary latitudinally and are interpolated from seasonal data.  Within the
RADM, values are interpolated from the table to each model grid cell and layer, and then a
correction is applied to the interpolated value to take into account the effects of clouds. In this
application, there were 33 columns, 36 rows, and 6 layers in the modeling domain, with a horizontal
grid size of 80  km.

Figure 1 shows a spatial plot of the interpolated table  photolysis rates for NO2 on August  2, 1988 at
18:OQZ for an 80 m layer just above the earth's surface (model layer 1), which is near local noon and
the daily maximum for photolysis rates (j-values).  The photolysis rates for NO2 (or J(NO2) values)
are high across much of the modeling domain, with values ranging from 0.520 min"1 to  0.570 min'1.
A few locations have higher values, but these are very limited in spatial extent.  Several areas have
lower values  (denoted by the darker shading), particularly the "bullseyes" over Mississippi/Alabama
and Georgia/Tennessee, as well as the broad band across most of Ontario, Canada.  All of these
areas have cloud cover which is attenuating the actinic flux and lowering the photolysis rates.   Figure
2 shows the percentage cloud cover over the modeling domain.

Figure 3 shows a vertical cross section of j(NO2) corresponding to the same hour as Figure  1.  Here,
we have selected an x-z cross-section  through northern Georgia.  Higher rates in the upper model
layers  (e.g. layers 5 and 6) appear over columns  16-25. This is the location of the highest cloud
cover fraction and shows the enhancement of photolysis rates by reflections above the cloud top.
Through the cloud layer, the rates decrease linearly to the cloud base.  Below the cloud, the
photolysis rates are much lower than surrounding grid cells.

JPROC
Comparison to  Table lookup plots
Our first simulation with the JPROC model uses  a two-stream radiative transfer model,  and
incorporates TOMS and MM5 data. The same cloud inputs (i.e. cloud cover, cloud base, cloud top,
and liquid water content) and chemical data  (e.g. cross sections and quantum yields) were used in
JPROC as were used  in the interpolated table simulation.  Results  were directly compared with the
RADM interpolated table results.  Similar comparisons were conducted by Hass and Ruggaber20.
Figure 4 shows the percent difference in j(NO2) between the interpolated table values and  the JPROC
calculated values for model layer  1 at 18:OOZ on  August 2,  1988.  The values are similar across the
Mid-Western and Atlantic Coastline States, while values are lower for JPROC over portions of
Georgia, Tennessee, Alabama, Mississippi, Ontario, Canada, and the open waters of the Gulf of
Mexico and Atlantic Ocean.  On average, j(NO2) for JPROC is 4% lower than for the RADM
interpolated table values, with the largest difference of 70% occuring over northern Georgia.

Looking next at the percent difference plot in j(NO2) (Figure 5) along the same X-Z cross section
plot as Figure 3 shows that the differences are small except in the vicinity of clouds.  Values are
lower for JPROC above some clouds,  but are higher by as much as 36% above the clouds in

-------
northern Georgia,  Under the same cloud mass, we find that the NO2 photolysis rates are lower for
JPROC by as much as 42%.  This illustrates that the treatment of clouds can have a significant
impact on calculated photolysis rates.  For JPROC, the cloud parameters are input before the
radiative transfer calculations, whereas for the RADM interpolated table the cloud parameters are
used to adjust precalculated clear sky j-values,

Two-stream vs. Multi-stream

Next, we replaced  the two-stream radiation model with a multi-stream discrete ordinates model in
JPROC.  The multi-stream  code is quite complex and consumed about  150 times more computer
CPU time than the two-stream code.  Because of this, we wanted to check whether the two-stream
code results differed significantly from those of the multi-stream code.   Zeng et al.2 have previously
compared relative changes  in j-values  for various cloud optical depths using the same  two-stream and
multi-stream models.  They found that the two-stream model overestimates j-values  by 2-10% under
the cloud layer.  Figure 6 shows the percent difference in j(NO2) for our simulations with JPROC.
Over most  of the modeling domain, the differences are small between the two  radiation models, with
an average difference of less  than 1%.  Over the northern regions, the multi-stream model gives
slightly higher photolysis rates (less than  5%), while over the Gulf of Mexico and Atlantic Ocean  the
two-stream model gives higher photolysis rates (by at most 13%).  A vertical cross section plot (not
shown) of the differences showed that the multi-stream model led to higher photolysis rates in the
upper model layers.

TOMS data comparison
Another input to the model that can be improved upon is the specification of the total  ozone column
data.  We have compared two runs of JPROC,  one using the U.S. Standard Atmosphere8 profile for
ozone and the other with the  same profile rescaled to the total ozone column as specified in the
TOMS database for each grid cell. Differences in j(NO2) between the two simulations were
examined first, but were found to be very small (< 0.5%).  This was expected because most of the
ozone absorption occurs in  a  different  wavelength  band than the band for NO2  photolysis.  Therefore,
we looked at the O3->O(1D)  photolysis reaction to see the effects that TOMS data can have on other
photolysis rates. The percent difference in the  O3->O('D) photolysis reaction rate is shown in Figure
7. Differences are highest  in the southern part  of  the modeling domain, where the simulation  with
TOMS data had photolysis  rates over 36% higher  than the simulation with the  standard profile.
Photolysis rates were about 20% higher in the TOMS run over most of the modeling domain.
Almost every grid  cell in the  modeling domain had lower total ozone column values than the  320DU
value of the U.S. Standard  Atmosphere.  The average total ozone column in the modeled region was
305 DU. The results shown by these two simulations are expected to be highly variable since the
total ozone column values have considerable  seasonal and latitudinal  variation.   Our result agrees
with the results of  Mass and Ruggaber20 demonstrating that the use of TOMS data can  better resolve
episodic  estimates of photolysis rates.

Albedo sensitivity
Another  tested sensitivity was the specification  of  surface albedo.  We ran JPROC with an albedo of
0.1 for all grid cells and then a second time with MMS's spatially varying albedo data. The percent
difference plot in j(NO2) is shown in Figure 8.  Little differences are noted over the water areas
where MM5's albedo is close to 0.1.  Larger  differences are noted over all land areas,  where the

-------
MM5 data led to more than a 5% increase in NO2 photolysis rates. Over the entire domain, j(NO2)
was 3% higher in the simulation with MM5 albedo data.

Conclusions

An advanced photolysis rate model was developed for use in air quality models.  The model
(JPROC) incorporates information on modeled meteorological profiles and cloud cover, and total
ozone column data, and provides considerable improvements over current methods used to compute
photolysis rates. JPROC has been tested in EPA's Models-3/CMAQ.  The calculated photolysis rates
were similar to rates previously used in the Regional Acid Deposition Model (RADM), with most of
the differences occurring immediately above and below clouds.  Differences in photolysis rates using
either a two-stream or a multi-stream radiative transfer model were smaller than the differences
between the 3-dimensional two-stream model and the lookup table approach used in the RADM.
Results also showed that the specification of total ozone column can significantly impact the rates for
some photolysis reactions. Another sensitivity test was conducted on the use of MMS's
spatially-varying albedo data; over most land areas the NO2 photolysis rate increased by 5-7% when
MMS's albedo was used compared to simulations  using a spatially uniform albedo of 0.1.

Much work is still  needed to fully test the results  of JPROC.  First and foremost, the results must be
evaluated against any available datasets.  Second,  we are in the process of testing different treatments
of clouds, including methods to link JPROC with  MM5, and the utility of satellite cloud coverage
data21.  Another  important area to be improved upon is in the  area of aerosols.  Models-3 will have
detailed aerosol distribution information, and we should be able to utilize this data to improve the
photolysis rate calculations within the boundary layer.

Acknowledgements

We are very grateful to Sasha Madronich for sharing his TUV model with us, upon which our model
is based. This paper has been reviewed in accordance with the U.S. Environmental Protection
Agency's peer and  administrative review policies and approved for presentation and publication.
Mention of trade names or commercial products does not constitute endorsement or recommendation
for use.

References

1.      Madronich,  S. J.  Geophys. Res. 1987, 92, 9740-9752.
2.      Zeng, J.;  Madronich, S.;  Stamnes, K. J, Geophys, Res.  1996, 101, 14525-14530.
3.      Byun, D; Hanna, A.; Coats, C.; Hwang, D. Transactions, Air & Waste Management
       Association's Conference on Regional Photochemical Measurement and Modeling Studies,
       San Diego,  CA, Air & Waste Management Association, Pittsburgh, PA,  1995; pp 197-212.
4.      Grell, G.A.;  Dudhia, J.; Stauffer, D.R. A Description of the Fifth generation Penn State/NCAR
       Meso-scale  Model (MM5); National Center for Atmospheric Research, Boulder, CO, 1993;
       NCAR/TN 398+IA.
5.      Kockarts, G. Ann. Geophys.  1994, 12, pp 1207 ff.
6.      World Meteorological Organization, "Atmospheric Ozone  1985:  Assessment of Our
       Understanding of the Processes Controlling its Present  Distribution and Change"; WMO Rep.
       No. 16; Global Ozone Research and Monitoring Project, Geneva, Switzerland, 1986.

-------
7.     Demerjian, K.L.; Schere, K.L,; Peterson, J.T. In Advances in Environmental Science and
       Technology: Pitts, J.N, Metcalf, R.L., Grosjean, D. Eds,; John Wiley & Sons,  Inc.;  1980; Vol.
       10, pp 369-459.
8.     U.S. Standard Atmosphere, National Oceanic and Atmospheric Administration. U.S.
       Government Printing Office: Washington, DC, 1976; NOAA-S/T76-1562.
9.     Elterman, L.  Visible and IR Attenuation for Altitudes to 50 km; Air Force Cambridge Res.
       Lab.: Bedford, MA, 1968; AFCRL-68-0153, 285.
10.    Binkowski, F.S.; Shankar, U. J. Geophys. Res. 1995, 100, 26191-26209.
11.    Hanna, A.F.; Binkowski, F.S.;  Shankar, U. Proceedings, Air & Waste Management
       Association's Conference on Regional Photochemical Measurement and Modeling Studies,
       San Diego, CA, Air & Waste Management Association, Pittsburgh, PA, 1995.
12.    DeMore, W.B.; Sander, S.P.; Golden, D.M.; Hampson, R.F.; Kurylo, M.J.; Howard, C.J.;
       Ravishankara, A.R.; Kolb, C.E.; Molina, M.J. Chemical Kinetics and Photochemical Data for
       Use in Stratospheric Modeling: Evaluation Number 11; National Aeronautics  and Space
       Administration, Jet Propulsion  Laboratory: Pasadena, CA, 1994; JPL Pub. 94-26.
13.    Kain, J.S.; Fritch, J.M. J. Atmos. Sci. 1990, 47, 2784-2802,
14.    Cotton, W.R.; Anthes, R.A, Storm and Cloud Dynamics; Academic Press, Inc: New York,
       1989; p. 183.
15.    Gery, M.W.;  Whitten, G.Z.;  Killus, J.P.; Dodge, M.C. J. Geophys.  Res. 1989 94,
       12925-12956.
16.    Stockwell, W.R.; Middleton, P.; Chang, J.S. J. Geophys. Res. 1990, 95, 16343-16367.
17.    Carter, W.P.L. Atmos. Environ.  1990, 24A, 481-518.
18.    Chang, J.S.; Brost, R.A.; Isaksen, I.S.A.; Madronich, S,; Middleton, P.; Stockwell, W.R.;
       Walcek, C.J.  J. Geophys. Res.  1987, 92, 14681-14700.
19.    Wiscombe, W.J. The Delta-Eddington Approximation for a Vertically Inhomogeneous
       Atmosphere; National  Center for Atmospheric Research, Boulder, CO, 1977;
       NCAR/TN-121+STR.
20.    Mass, H.;  Ruggaber, A. Meteorol Atmos. Phys. 1995, 57, 87-100.
21.    McNider, D.; Song, A.; Casey, D.; Pleim, J.; Roselle, S. Accepted for presentation  at the
       22nd NATO/CCMS International Technical Meeting on Air Pollution Modeling and Its
       Application, Clermont-Ferrand, France,  June 1997.

-------
Figure 1.  NO2 photolysis rates (min"1) for layer 1 on August 2, 1988 at
1800Z, interpolated from the RADM table with cloud cover corrections
applied.

   0.600      36
   0.480
   0360
   0240
   0.120
   0.0 00
/min
  FAVE
   by
  HCNC
1
   1                                            33
             August 2,198818:00:00
     Min=OJ080 at (13,12), Max=0.660 aft (10,3)

-------
Figure 2.  Cloud cover fraction used in RADM on August 2, 1988 at
1800Z.

   1JDDD      36
   OJ800
   0.600
   0.400
   0200
   0.000
1
                                                      •y  *•••
FRACTION (0-1) 1
  FAVE
   by
  HCNC
                                            33
             August 2,198818:00:00
     Min=0 J)00 at (26,1)i Max=1 JOOO at (16,33)
                                                                       10

-------
Figure 3.  Vertical cross section of the NO2 photolysis rates (mirf1) on
August 2, 1988 at 1800Z, interpolated from the RADM table with cloud
cover corrections applied. Vertical slice through cloud mass over northern
Georgia.
  1D23

  0.7S7

  0.511

  0.256
  0.000
ftnin
                                  August 2,138818:00:00
                             Min=0.115 at (11,1), Max=1378 at (18,8)
                                                                                     11

-------
Figure 4.  Percent difference in NO2 photolysis rates on August 2, 1988
at 1800Z for layer 1 between the interpolated RADM table and the JPROC
calculated values.
                 (RADM-JPROQ/JPROC* 100%
   ZOJJOO      36
   15.0 00
   0.000
   -SJOOO
  PAVE
   by
  MCMC
                 1
                                          33
         August 2,108818:00:00
Min=-8 JOi at (2.321 M«=70J086 at (17,15)
                                                                         12

-------
Figure  5.   Vertical  cross  section of  the  percent  difference in NO2
photolysis rates for layer 1 on  August 2, 1988 at 1800Z between the
interpolated  RADM table and the JPROC calculated values.
                   (RADM-JPROC)/JPROC* 100%
 MCHC
                               August 2,198818:00:00
                         Minr-35518 at (18,4), Max=42.198 at (18,1)
                                                                               13

-------
Figure 6.  Percent difference in the NO2 photolysis rates on August 2,
1988 at 1800Z for layer 1 between the JPROC two-stream calculations and
the multi-stream calculations.
                (T WOstr-MULTstr)/MULTstr * 100 %
   15JDDD     36
   11.0 00
   7JOOO
   3.000
   -1.000
   -5.0 00
  PftME
   by
  HCNC
                  1
                                           33
         August 2,198818:00:00
Min=-4.674at (!36;t Max=13.375 at (16,33)
                                                                           14

-------
Figure 7. Percent difference in the O3->O('D) photolysis rates on August
2, 1988 at 1800Z for layer 1 between the JPROC calculations with TOMS
data and with the U.S. Standard Atmosphere ozone profile.
                  (TOMS-STDO3)/STDO3* 100%
              36 \
   36.000
   27.000
   18JOOO
   0.0 00
CAVE
                 1
                            August 2,198818:00:00
                    Min=6.646 at (1.33). Max=44.12 2 at (16,2)
                                                                          15

-------
Figure 8.  Percent difference in the NO2 photolysis rates on August 2,
1988 at  1800Z for layer 1 between the JPROC calculations with MMS's
albedo data and with spatially uniform albedo data.
                (MM5alb-UNIFalb)/UNIFalb* 100%
   BJDOO      36
   7.000
   5JOOQ
   3.000
   1JOOO
   -1400
                  1
  PAVE
   by
  MCMC
        August 1,188818:00:00
M in=-1.7 20 at (14,1), Max=9.151 it (27,2)
                                          33
                                                                           16

-------
                                   TECHNICAL REPORT DATA
.1.  REPORT NO.
    EPA/600/A-97/024
 4.  TITLE AND SUBTITLE
 Development  and Testing of an Improved Photolysis
 Rate  Model for Regional Photochemical Modeling
                                                                 5.REPORT DATE
                                                                 6.PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)

 Shawn  J.  Roselle*,  Jonathan E.  Pleim*, Kenneth  L.
 Schere*,  Adel F, Hanna,  and Ji-Cheng C, Jang
             8.PERFORMING ORGANIZATION REPORT NO.
   PERFORMING ORGANIZATION NAME AND ADDRESS
                                                                 10.PROGRAM ELEMENT NO.
 *same  as Block 12
 MCNC-North Carolina Supercomputing  Center,  Research
 Triangle Park,  North Carolina  27709
             11.  CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS

 U.S.  Environmental  Protection Agency
 Office  of Research  and Development
 National  Exposure Research Laboratory
 Research  Triangle Park, NC 27711
             13.TYPE  OF REPORT AND PERIOD COVERED
             Proceedings, FY-97
             14.  SPONSORING AGENCY CODE

             EPA/600/9
IS. SUPPLEMENTARY NOTES
16. ABSTRACT

An advanced computational  model has been developed to simulate actinic flux  and photolysis  rates of
chemical species for air quality and tropospheric photochemistry modeling.   The module combines advanced
radiative transfer models  with explicit computations of photolysis rates using detailed information on
prevailing atmospheric conditions during simulation episodes, satellite ozone data, and characteristics
of clouds, aerosols and surface albedo on a  time dependent basis over a 3-dimensional grid.  The
sensitivity of the photolysis rate simulations to different  radiative transfer schemes as well as key
parameters such as ozone profiles and surface albedo is examined.  Photolysis rate results  are similar to
rates previously used in the Regional Acid Deposition Model  (RADM),  with most of the differences
occurring immediately above and below clouds.  Differences in photolysis rates using either a two-stream
or a multi-stream radiative transfer model were smaller than the differences between the 3-dimensional
two-stream model and the lookup table approach use in the RADM.  Results also showed that the
specification of total ozone column can significantly impact the rates for some photolysis  reactions; in
our test case, using TOMS  data to specify the total ozone column for the 3-dimension grid increases the
03->0(1D)  photolysis rate  by -20% over most  of the modeling domain compared  to a simulation using the U.S
Standard Atmosphere ozone  profile.  Our last sensitivity test was the use of MMB's spatially-varying
albedo data; over most land areas the NO2 photolysis rate increased by 5-7% when MMB's albedo is used
compared to simulations using a spatially uniform albedo of 0.1.
17.
                                   KEY WORDS AND DOCUMENT ANALYSIS
                   DESCRIPTORS
b,IDENTIFIERS/ OPEN ENDED
TERMS
                                                                                  C.COSATI
18.  DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19.  SECURITY CLASS (This
Report)

UNCLASSIFIED
                                                                                  21.NO. OF PAGES
                                                    20. SECURITY CLASS  (This
                                                    Page!

                                                    UNCLASSIFIED
                                                                                  22. PRICE

-------