Transactions on Ecology and the Environment vol 28, © 1999 WIT Press, www.witpress.com, ISSN 1743-3541
Evaluation of a Compact Chemical Mechanism
for Photochemical Smog Modelling
A contribution to subproject GLOREAM
P. Viaene, W
Debruyn, C Mensink and J van Rensbergen
A compact chemical mechanism
The compact mechanism evaluated in this study is used in the EUROS model.
EUROS is an Eulerian grid model that was originally developed at the Dutch
National Institute of Public Health (RIVM). The model is intended for long term
simulations and should be applicable on a midrange workstation. The chemical
mechanism that was derived from the larger EMEP/MSC-W scheme (Barrett
and Berge, 1996) is therefore limited to 17 reactions. Nine of these are inorganic
reactions in which only Os, OH, NO, NO] , NO] aerosol, HNOg, SO: and SO4
aerosol are considered. In the EUROS scheme the reactions with the NO]
radical and ^Os from the original EMEP scheme were lumped into one
reaction in which NO: is converted to HNOs. Only CH/^ CzHj, €2%, €3%,
C^Hi o, xylene and isoprene are considered as organic components. Each of these
organic species participates in one reaction of the form:
ORG
-/>,
+ OH
-» a; Og
[1]
where a, = ct.^^e^^ + a ^ in which [NOx] is the sum of the NO and NO%
concentration (ppb) and a,,max« o,,mm and b, are parameters that depend on the
nature of the organic species being considered.
Full details of this chemical mechanism are presented by van Loon (1996). In
the present study we updated the rate coefficients in the mechanism according to
Barrett and Berge (1996).
Evaluation strategy
To evaluate the above mechanism it was compared to the original EMEP
mechanism from which it originated. For the comparison, four of the boxtests
used in the Chemical Mechanism Working Group model intercomparison
fmceed/Mgj ofEUROTTMC S
Editors: P.M. Borrell and P. Borrell
© 1999: WITPRESS, Southampton
Transactions on Ecology and the Environment vol 28, © 1999 WIT Press, www.witpress.com, ISSN 1743-3541
Photochemical Smog Modelling
457
(Poppe et al. 1996) were repeated: LAND, BIO, PLUME/1 and PLUME/2. The
main differences between these different boxtests are summarised in Table 1.
Details can be found in (Poppe et al. 1996).
Table 1:
Main differences between the box tests used
boxtest
initial
concentrations
i:soprene
emissions
photolysis
frequencies
LAND
BIO
rural
rural
no
yes
no
no
specified
specified
PLUME 1
polluted
no
yes
specified
PLUME2
polluted
no
yes
not specified
Input generation for the CMWG boxtests is straightforward except for the
translation of the organic emissions from the inventory used in the PLUME
boxtests to model input. It was decided to take only alkane, alkene and aromatic
emissions into account. The emissions which do not correspond to a model
component were respectively added to CJHU €3% and xylene. In this
aggregation of emissions, differences in OH reactivity and molecular weight
were neglected. Numerical solutions of both these mechanisms were generated
using the Kinetic Preprocessor programme (Damian-Iordache and Sandu, 1995).
Results and discussion
In Fig. 1 the boxtest results for Os are presented. Differences observed between
the results for the EUROS and the EMEP mechanisms are (at least partially) due
to values of the coefficients, a, for Og in the organic component reactions [1]. In
the BIO case, isoprene is quickly converted and after 2 hours less than 1 % of
the initial concentration of 1 ppb remains. For the NO* levels considered in the
BIO boxtest a value of about 7 was found for the coefficient, a^ene which
explains the sudden increase in Og levels with 7 ppb in the first hours of the run.
In the PLUME case one can notice that the day by day increase in O]
concentration is nearly constant for the EMEP scheme while there is an increase
for EUROS. This different behaviour with increasing NOx concentrations can
also be attributed to the values and/or the expression used to calculate the
coefficients. a\.
Transactions on Ecology and the Environment vol 28, © 1999 WIT Press, www.witpress.com, ISSN 1743-3541
458
P. Viaene et al.
LAND: O,
BIO: Oj
x
1
"
*"*
0
"
| — euros!
—
50
hour
100
'
^— ^— s
*• 300 L— -C
50
hour
PLUME2: Oj
250
200
g. 150
100
50
100
PLUME1: 0,
a
50
hour
^ ,. —euros
100 50 -I
0
100
,- y I— euros I
I — emep |
-— s~~^
50
100
hour
i
Fig. 1: 63 concentrations calculated for the different box tests with the EUROS and
EMEP chemical mechanisms.
Uncertainty in the input for air pollution models is significant e.g. in a
comparison of traffic VOC emission inventories in Belgium, De Vlieger (1993)
found differences up to 200 %. From Fig. 1 it is clear that there is a difference in
calculated Og concentrations between the EUROS and the EMEP schemes for
the boxtests considered in this study. However one can wonder if these
differences in calculated values are significant if the uncertainty in model input
is taken into account.
hour
Fig. 2: Results from a Monte Carlo experiment for the PLUME/2 case in which the
organic emissions are assumed to be normally distributed with a variation coefficient of
20 %. The mean calculated O] concentration and an interval of ± 2 standard deviation are
shown for EUROS and EMEP.
Transactions on Ecology and the Environment vol 28, © 1999 WIT Press, www.witpress.com, ISSN 1743-3541
Photochemical Smog Modelling
459
To assess the significance of the differences in model results the boxtest
PLUME2 was used in a Monte Carlo experiment. In this experiment all the
uncertainty was attributed to the emission input which was considered to be
normally distributed with a variation coefficient (i.e. standard deviation/mean)
of 20 %. The results for the Monte Carlo experiment with 500 runs are shown in
Fig. 2. The results indicate a significant difference in the results obtained after
70 hours (significance level, a = 0.05). There is also a difference in model
sensitivity to changes in VOC emission: for EUROS the variation coefficient for
the Og concentration is 11.0 % while it is 2.8 % for EMEP at the end of the
boxtest run.
References
Barrett, K. and Berge, E. (eds), Transboundary Air Pollution in Europe, Part 1: Estimated
dispersion of acidifying agents and of near surface ozone, EMEP MSC-W Status
Report, 1996.
Damian-Iordache,V. and Sandu, A. KPP; A symbolic preprocessor for chemistry kinetics
User's guide. Technical report, University of Iowa, Department of Mathematics, 1995.
De Vlieger, I. Reliability of VOC emissions from road transport in Belgium in
Proceedings of TNO/EURASAP workshop on the reliability of VOC emission
databases, MW-TNO publication p9 3/040, 1993.
Loon, M. van, Numerical methods in smog prediction, PhD thesis, University of
Amsterdam, June 1996.
Poppe, D., Andersson-Skold, Y., Baart, A. Builtjes, P.J.H., Das, M., Fiedler, F., Hov, O.,
Kirchner, F., Kuhn, M., Makar, P.A., Milford, J.B., Roemer, M.G.M., Ruhnke, R.
Simpson, D., Stockwell, W.R., Strand, A., Vogel, B. Vogel, H., Gas-phase Reactions
in Atmospheric Chemistry and Transport Models: A Model Intercomparison,
EUROTRAC ISS, Garmisch-Partenkirchen, Germany 1996.