Ozone is formed in the atmosphere through chemical reactions between nitrogen oxides (NOx) and volatile organic compounds (VOC). Tens of inorganic and hundreds of organic compounds are known to participate in thousands of photochemical reactions. The explicit treatment of all of these compounds and reactions would be prohibitively complex in an Eulerian-based chemical transport model such as LOTOS-EUROS, especially when such a model is used for long-term (multi-annual) calculations in the framework of regulatory purposes. Since condensation of atmospheric chemistry is required to reach a level of simplification imposed by computational constraints, methods for minimizing the size of a chemical mechanism have been proposed.
A possible way of condensing the inorganic chemistry within photochemical mechanisms is through the lumping of species or the lumping of reactions utilising specific assumptions, e.g. steady state for some radicals. In the lumped structure approach, organic compounds are apportioned to one or more species on the basis of carbon-carbon bond type or on basis of a reactive group (Gery, 1989). For example, propane (CH3-CH2-CH3) is represented by three parafinic groups (PAR) since all three carbon atoms have only single bonds: propene (CH2=CH-CH3) is represented as one olefinic group (OLE) representing the carbon-carbon double bond, and one PAR representing the methyl group.
The most widely applied mechanism using the lumped structure approach for representing urban photochemistry is the Carbon Bond-IV (CB-IV) mechanism. The CB-IV mechanism originally consisted of 81 reactions. It is probably the most widely used mechanism due to its good performance in polluted areas and its relative small number of reactions. In LOTOS-EUROS we use two different versions of CB-IV, called CBM-IV and CB99.
The gas phase mechanisms also describe the
photochemical formation of sulphuric acid and nitric acid, which drive the
formation of secondary inorganic aerosol. Below we describe the set-up for CBM-IV
and CB99 schemes as well as the aerosol chemistry.
The gas phase photochemistry CBM-IV module in LOTOS-EUROS is a
modified (condensed) version of the CBM-IV mechanism by Whitten et al. (1980). Characteristic for the Carbon-Bond Mechanism
(CBM) are the structure molecules, such as PAR, ETH, FORM, ALD2, MGLY, XO2,
XO2N, etc. The structure molecules represent parts of the organic molecules,
only ETH has a one-to-one relation with ethane. The
full mechanism including the reaction rate parameterisation is shown in Annex A.
The scheme includes 28 species and 66 reactions, including 12 photolytic reactions.
Compared to the original scheme steady state approximations were used to reduce
the number of reactions. In addition, reaction rates have been updated
regularly. The mechanism was tested against the results of an intercomparison
presented by Poppe et al. (1996) and found to be in good agreement with results
presented for other mechanisms. The chemistry scheme further includes gas phase
and heterogeneous reactions leading to secondary aerosol formation as presented
below. The CBM-IV chemistry
is solved using the QSSA method.
Sulphate production
It is important to give a good representation of sulphate formation, since sulphate is an important aerosol component. In addition, it competes for the ammonia available to combine with nitric acid. Most models that represent a direct coupling of sulphur chemistry with photochemistry underestimate sulphate levels in winter in Europe. This feature can probably be explained by a lack of model calculated oxidants or missing reactions (Khasibatla et al., 1997). Therefore, in addition to the gas phase reaction of OH with SO2 (in CBM-IV) we represent additional oxidation pathways in clouds with a simple first order reaction constant (Rk), which is calculated as function of relative humidity (%) and cloud cover (e):
Rk = 8.3e-5 * (1 + 2*e) (s-1), for RH < 90 %
Rk = 8.3e-5 * (1 + 2*e) * [1.0 + 0.1*(RH-90.0)] (s-1), for RH ³ 90 %
This parameterization is similar to that used by Tarrason and Iversen (1998). It enhances the oxidation rate under cool and humid conditions. With cloud cover and relative humidity of 100 % the associated time scale is approximately two hours. Under humid conditions, the relative humidity in the model is frequently higher than 90 % during the night.
Heterogeneous N2O5
chemistry
The reaction of N2O5
on aerosol surfaces has been proposed to play an important role in tropospheric
chemistry (Dentener and Crutzen, 1993). This reaction is a source for nitric
acid during night time, whereas during the day the NO3 radical is
readily photolysed. We parameterised this reaction following Dentener and
Crutzen (1993). In this parameterisation a Whitby size distribution is assumed
for the dry aerosol. The wet aerosol size distribution is calculated using the
aerosol associated water obtained from the aerosol thermodynamics module (see
below). The reaction probability of N2O5 on the aerosol
surface has been determined for various solutions. Reaction probabilities
between 0.01 and 0.2 were found (Jacob, 2000 and references therein). A study
by Mentel et al. (1999) indicates values at the lower part of this range.
Therefore, we use a probability of g = 0.05, which is somewhat lower than the generally used
recommendation by Jacob (2000). In the polluted lower troposphere of Europe,
however, the hydrolysis on the aerosol surfaces is fast, with lifetimes of N2O5
less than an hour (Dentener and Crutzen, 1993). Therefore the exact value of g does not determine the results strongly. Due to the limited availability of detailed cloud information, we
neglect the role of clouds on the hydrolysis of N2O5,
which may also contribute to nitric acid formation. However, due to the very
fast reaction of N2O5 on aerosol in polluted Europe, the
role of clouds on N2O5 hydrolysis is probably less important.
The second gas phase chemistry mechanism that is included in LOTOS-EUROS, CB99, is the officially documented and vindicated version by Adelman (1999). CB99 is presented as an updated version of the mechanism and is produced through a critical review of the relevant literature. Kinetic and minor mechanistic updates are applied to the mechanism to make it consistent with the currently best available information. Empirical verification for each major change is presented through modeling Outdoor Chamber and Indoor Teflon smog-chamber experiments. Quantitative and qualitative analyses are presented on the performance of the new mechanism and its predications are compared to those of two older versions of CB-IV. Adelman shows that CB99 exhibits extremely good performance in modelling a wide range of experiments in multiple smog chambers. He recommends the new mechanism for future applications of regulatory air quality simulation models and areas for further improvement are discussed.
CB99 includes 42 species and 95 reactions, including 13 photolytic reactions. Major changes comprise the addition of four reactions with sulphur dioxide, methanol and ethanol, see also Carter (1994), and an updated CB-IV isoprene chemistry mechanism based on the work of Carter (1996). The translation of this updated CB-IV isoprene chemistry mechanism into CB-IV components is given in Whitten et al. (1996). The full chemical mechanism is given in Annex B. The CB99 chemistry is solved using a Rosenbrock-3 method.
Semi-volatile aerosol species
are species that maintain equilibrium between the aerosol and gas phase. Ammonium
nitrate is a well known example but also organic species can be described as
semi-volatile components. Below we specify the methods used to calculate the
formation of these components in LOTOS-EUROS.
Secondary biogenic aerosol concentrations may
contribute significantly to the total aerosol mass, especially in remote
regions. There are little to no measurements of these compounds and there is
only very limited experimental knowledge on their formation in the atmosphere.
Moreover, large parts of the SOA arise from condensed biogenic precursors whose
emissions are still not well known. Hence, the model
description and its results are very uncertain. Below we describe the module
that computes the secondary biogenic aerosol concentrations, which can optionally
be turned on during a model run.
Secondary
organic aerosols are computed in a similar way as their inorganic counterparts,
starting with a number of organic precursors, in literature usually called
Reactive Organic Gases (ROG). These organic gases react with OH, the NO3
radical and O3 (or with a subset of these species) resulting into a
number of products (Schell, 2000), schematically represented by
ROG + OH → Σ αi Ci
ROG + NO3
→ Σ αi
Ci
ROG + O3
→ Σ αi Ci
The products Ci are partitioned between the gas-phase and the aerosol-phase through equilibrium. In order to calculate the equilibrium concentrations, the module SORGAM is used. This module takes into account 8 different degradation products (from the reaction of an ROG with OH, NO3 or O3). Mainly the biogenic precursors (isoprene, α-pinene) lead to degradation products that give contributions to the aerosol-phase. Anthropogenic ROGs hardly result into a significant contribution to the SOA concentrations.
Since we
think that the SOA concentrations are small (on average), they are neglected in
most LOTOS-EUROS applications, since they require a disproportional amount of
extra CPU time. Recall that 16 additional species (8 gas phase and 8 aerosol
phase) need to be taken into account.