The HAMMONIA Chemistry Climate Model: Sensitivity ofthe Mesopause Region to the 11-year Solar Cycle and

CO2 Doubling

H. Schmidt∗, G. P. Brasseur, M. Charron†, E. Manzini‡, M. A. Giorgetta, T. Diehl§

Max Planck Institute for Meteorology, Hamburg, GermanyV. I. Fomichev

York University, Toronto, CanadaD. Kinnison, D. Marsh, S. Walters

National Center for Atmospheric Research, Boulder, USA

Accepted for publication in Journal of Climate, 19 December 2005.

Abstract This paper introduces the three-dimensional Hamburg Model of the Neutral and Ionized Atmo-sphere (HAMMONIA) that treats atmospheric dynamics, radiation and chemistry interactively for the heightrange from the Earth’s surface to the thermosphere (approximately 250 km). It is based on the latest versionof the ECHAM atmospheric general circulation model of the Max Planck Institute for Meteorology in Ham-burg, which is extended to include important radiative and dynamical processes of the upper atmosphereand coupled to a chemistry module containing 48 compounds. The model is applied to study the effects ofnatural and anthropogenic climate forcing on the atmosphere, represented on the one hand by the 11-yearsolar cycle and on the other hand by a doubling of the present-day concentration of carbon dioxide. Thenumerical experiments are analyzed with the focus on the effects on temperature and chemical composi-tion in the mesopause region. Results include a temperature response to the solar cycle by 2 to 10 K inthe mesopause region with the largest values occurring slightly above the summer mesopause. Ozone inthe secondary maximum increases by up to 20% for solar maximum conditions. Changes in winds are ingeneral small. In the case of a doubling of carbon dioxide the simulation indicates a cooling of the atmo-sphere everywhere above the tropopause but by the smallest values around the mesopause. It is shownthat the temperature response up to the mesopause is strongly influenced by changes in dynamics. Duringnorthern hemisphere summer, dynamical processes alone would lead to an almost global warming of up to3 K in the uppermost mesosphere.

∗Max Planck Institute for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany, hauke.schmidt@dkrz.de†Meteorological Service of Canada, Dorval, Canada‡National Institute for Geophysics and Volcanology, Bologna, Italy§GEST/UMBC, NASA Goddard Space Flight Center, Greenbelt MD, USA

1

1. Introduction

The mesosphere-lower thermosphere (MLT) region,which covers the approximate altitude range 50-150 km, is one of the most complex, yet least ex-plored layers of the atmosphere. Rocket soundingsand satellite monitoring have highlighted the strongvariations that occur as a function of height in thephysics and in the chemical composition, especiallyin the vicinity of the mesopause between 85 and100 km altitude (Smith 2004). For example, as thealtitude increases through this atmospheric region,the strong vertical mixing conditions that maintainthe mixing ratio of long-lived species relatively con-stant with height are progressively replaced by con-ditions in which molecular diffusion prevails, leadingto gravitational separation of chemical species ac-cording to their respective mass.

The rapid change with height in the amount ofsolar radiation available to photolyze atmosphericmolecules also leads to a rapid chemical transitionbetween the mesosphere and the thermosphere,with molecules dominating the atmospheric compo-sition at the lower levels and atomic species becom-ing most abundant in the upper layers.

The nature of the energy budget is also chang-ing with altitude. Below 90 km, the major contribu-tions are provided by solar heating resulting fromozone absorption, by the emission of infrared radi-ation by CO2, by adiabatic processes, and by thetransport of heat through advection. In the lowerthermosphere, the energy budget is determined pri-marily by the absorption of short-wave solar radi-ation by molecular oxygen, by radiative emissionsby CO2, and by downward molecular conductionof heat. In the MLT region, the pattern is con-siderably more complicated. As a consequenceof the low collision frequency between particles,the thermal radiative transfer starts to be affectedby non-thermodynamic equilibrium (non-LTE) condi-tions. Additionally, processes such as energetic par-ticle input, chemical heating, airglow emissions, andturbulent heat exchanges must be taken into account(Mlynczak 2000; Roble 2000; Fomichev et al. 2002).

The deposition of momentum from upward propa-

gating waves provides the major forcing mechanismfor the meridional circulation of the middle atmo-sphere. The dissipation of gravity waves produces abody force on the zonal flow, leading even to a rever-sal of the zonal wind in the vicinity of the mesopause.Simultaneously, this momentum source produces amesospheric circulation directed from the summerpole to the winter pole, and through mass continu-ity gives rise to ascending and descending motionsin the summer and winter hemispheres, respectively(e.g. Holton and Alexander 2000). Other wave mo-tions have strong effects on the structure of the mid-dle atmosphere, including diurnal and semi-diurnaltides, which produce periodic fluctuations in tem-perature, wind components and species concentra-tions, planetary waves, and equatorial waves whichare (together with gravity waves) believed to drivethe observed semi-annual and quasi-biennial oscil-lations (e.g. Garcia 2000; Giorgetta et al. 2002). Inall cases, the source of the waves is located in thelower atmosphere or at the Earth surface, and theirability to reach the mesopause region depends onfiltering and dissipation mechanisms that take placeat lower altitudes. In the thermosphere, where thedegree of ionization becomes significant, electro-magnetic forces, which tend to align the motions ofcharged particles along magnetic field lines, affectatmospheric motions, so that the resistance effect ofions on the neutral matter (usually named “ion drag”)must be considered in the dynamical equations.

The dynamical structure of the mesopause regiontherefore depends simultaneously on solar energythat is progressively absorbed by the atmosphericmedium as it penetrates downward, and momen-tum sources provided by waves originating from thelower layers of the atmosphere and absorbed as theypropagate upward and dissipate or encounter crit-ical levels. Natural variations and human-inducedchanges to be expected in the MLT region cannotbe assessed without understanding the dynamical,radiative and chemical couplings existing betweenthe different layers of the atmosphere. Therefore,as pointed out by Roble (2000), there is a need toincrease the height of the boundary of current mid-dle atmosphere models well into the thermosphereand include solar and auroral processes as well

2

as other physical and chemical processes such iondrag, molecular diffusion, molecular thermal con-ductivity, electrodynamic interactions with the iono-sphere.

The HAMMONIA model (Hamburg Model for theNeutral and Ionized Atmosphere), which is pre-sented here, has been designed to investigate howthe couplings between atmospheric regions affectthe response of the atmosphere to external pertur-bations, including solar variability and anthropogenicchemical emissions at the Earth surface. A particu-larly challenging question is to determine the depthof penetration of the signal produced by solar vari-ability in the atmosphere. Another interesting ques-tion is to quantify the impact on the upper atmo-sphere of the release at the surface of large quan-tities of carbon dioxide and other radiatively activegases. Addressing these questions requires thatthe chemical processes of importance in the atmo-sphere be treated fully interactively with the dynami-cal, physical and radiative processes included in themodel.

Most contemporary general circulation models(GCMs) extend typically to approximately 2-10 hPawith the stratospheric layers being considered as abuffer between the tropopause and the top of themodel. Some of these GCM have been extended toapproximately 75 to 100 km altitude (e.g. Fels et al.1980; Boville 1995; Hamilton et al. 1995; Manziniet al. 1997; Beagley et al. 1997) or even up to thethermosphere (Miyahara et al. 1993; Fomichev et al.2002; Sassi et al. 2002). Chemical transport mod-els that treat chemical processes up to the meso-sphere “off line” from the dynamics have also beendeveloped (e.g. Chipperfield et al. 1993; Brasseuret al. 1997). Coupled dynamical-chemical modelscovering this altitude range used mostly a mechanis-tic approach (e.g. Rose and Brasseur 1989; Lefevreet al. 1994; Sonnemann et al. 1998) in which thecomplex processes of the troposphere are replacedby boundary conditions (e.g., planetary wave forc-ing) applied in the vicinity of the tropopause. Inmore recent times GCMs have been coupled “online” to chemistry, some of them having the lid inthe mesosphere (e.g. Rasch et al. 1995; Steil et al.2003) or close to the mesopause (de Grandpre et al.

2000). Only very recently, some GCMs with coupledchemistry extending from the surface to the thermo-sphere have been developed or are under develop-ment. This is the case for the Extended CanadianMiddle Atmosphere Model (CMAM, e.g. Fomichevet al. 2002), for the Whole Atmosphere CommunityClimate Model (WACCM, e.g. Sassi et al. 2002) de-veloped at the National Center for Atmospheric Re-search in Boulder, Colorado, and for HAMMONIA,initiated at the Max Planck Institute for Meteorologyin Hamburg, Germany, with the contribution of scien-tists in other research institutions.

The purpose of this paper is to present anddescribe the newly developed HAMMONIA model(Section 2), and to evaluate calculated atmosphericfields (temperature, winds and chemical species) bycomparison with available observational data (Sec-tion 4). The model is then used to calculate the re-sponse of the atmosphere to the 11-year solar cy-cle (Section 5) and to a doubling in the atmosphericconcentration of CO2 (Section 6). A discussion ofthe results and some conclusions are presented inSection 7.

2. Model description

The HAMMONIA model consists of the vertical ex-tension to the thermosphere of the MAECHAM5model (Giorgetta et al. 2005; Manzini et al. 2005, thisissue) which is itself a vertical extension to the meso-sphere of the ECHAM5 atmospheric general circula-tion model (Roeckner et al. 2003, 2005b,a, this is-sue). ECHAM5 is the most recent version in a se-ries of ECHAM models evolving originally from thespectral weather prediction model of the EuropeanCentre for Medium Range Weather Forecasts (Sim-mons et al. 1989, ECMWF), and includes, in partic-ular, comprehensive descriptions of the energy bud-get, the water cycle and land surface processes. TheECHAM model versions have been applied to a largevariety of climate research issues.

HAMMONIA is a spectral model with triangulartruncation at wavenumber 31 (T31) and with 67 ver-tical levels ranging from the surface to 1.7*10−7 hPa(∼ 250 km). Figure 1 shows the distribution of the

3

vertical levels. In the mesosphere and lower thermo-sphere the distance between the levels is constantin log-pressure and corresponds to about 2 to 3 km,depending on temperature. The model includes a fulldynamic and radiative coupling with the MOZART3chemical module (Kinnison et al. 2004). For inte-gration of the dynamical and chemical processes weadopt a time step of 10 min.

The dynamical and radiative processes that havebeen specifically implemented in HAMMONIA in-clude solar heating in the ultraviolet and extreme ul-traviolet wavelength regime (down to 5 nm), a non-LTE radiative scheme, energy deposition and eddydiffusion generated by gravity wave breaking, verti-cal molecular diffusion and conduction, and a sim-ple parameterization of electromagnetic forces in thethermosphere (ion drag and Lorenz forces).

a. Radiative heating and cooling

The parameterizations adopted to quantify radiativeheating and cooling are up to certain pressure levelsand for large parts of the solar spectrum identical tothe ones used in ECHAM5. In particular, the RapidRadiative Transfer Model (RRTM) for long-wave ra-diation (Mlawer et al. 1997) and an updated versionof the short-wave parameterization of Fouquart andBonnel (1980) that divides the spectrum from 250to 4000 nm into four bands are employed. Additionalformulations are introduced to account for processesspecific to the MLT region: solar heating at shorterwavelengths, non-LTE effects in radiative processes,and chemical heating. A similar approach of mergingdifferent heating and cooling schemes as detailedin the following was already employed by Fomichevet al. (2002).

Infrared cooling by O3 and CO2 above 0.02 hPa(∼ 75 km) is calculated from the parameterizationof Fomichev and Blanchet (1995) and of Fomichevet al. (1998), respectively, with non-LTE effects ex-plicitly taken into account for the CO2 cooling. Thelatter parameterization was modified so that the in-terpolation of matrix coefficients is based on CO2

column amounts rather than on local CO2 densities.This modification allows the use of non-constantCO2 profiles also below 95 km (∼ 5.*10−4 hPa). Be-

low 0.1 hPa (∼ 65 km) infrared radiative fluxes arecomputed purely by RRTM. Between 0.1 and 0.02hPa, the infrared radiative fluxes are calculated asa linear combination of both schemes. A parame-terization of solar heating by CO2 absorption in thenear infrared according to the non-LTE scheme ofOgibalov and Fomichev (2003) is utilized. This con-tribution is particularly important at heights of around75 km (∼ 0.02 hPa) where it can account for up to30% of the total radiative heating (Fomichev et al.2004a). The values from the parameterization areignored below 30 hPa (∼ 24 km, because heatingin the near infrared is already considered by thescheme of Fouquart and Bonnel (1980)), fully con-sidered above 3 hPa (∼ 40 km), and merged with thenear infrared CO2 heating provided by the schemeof Fouquart and Bonnel (1980) between these twopressure levels.

Two slightly different combinations of solar heat-ing parameterizations are used in the experiments ofthis study, hereafter called SW1 and SW2. In SW1,solar heating at wavelengths from 680 to 2500 nm iscomputed in three bands using the parameterizationof Fouquart and Bonnel (1980) as in ECHAM5. Thefourth band of the Fouquart and Bonnel (1980) pa-rameterization (250 to 680 nm) is used only up to 70hPa (∼ 18 km). Above 30 hPa (∼ 24 km), heatingby O3 and O2 at wavelengths from 120 to 680 nm iscomputed using the same algorithm as for the calcu-lation of the photodissociation rates (see section 2e),and noting that, if Jλ,x is the photolysis rate for amolecule x and a wavelength λ, the correspondingrate of energy deposition is given by Eλ,x = hνJλ,x,with ν being the frequency of the radiation and hPlanck’s constant. Between 30 and 70 hPa the twoapproaches are merged. In SW2, the fourth band ofthe Fouquart and Bonnel (1980) parameterization isused for all pressure levels. Only between 120 and250 nm, heating and photodissociation rate compu-tation are done consistently. The rationale of theseapproaches is that SW1 with its significantly betterresolution in the UV part of the spectrum is muchbetter suited to represent the spectral dependenceof solar variability and is therefore used in the solarcycle experiments of this study. SW2 on the otherhand is very similar to the approach used in the orig-

4

inal ECHAM5 model. It allows an easier compari-son to earlier ECHAM5 experiments and is used inthe CO2 doubling experiments of this study. At ther-mospheric heights, an extreme ultra-violet (EUV) so-lar forcing based on the model of Richards et al.(1994) is used for wavelengths ranging from 5 to105 nm. Efficiency factors to account for the loss ofinternal energy due to airglow processes are takenfrom Mlynczak and Solomon (1993) for the O3 Hart-ley band and O2 Schumann-Runge continuum, andfrom Roble (1995) for the EUV portion of the spec-trum. The solar EUV heating efficiency also ac-counts for energy losses associated with radiativecooling in the 5.3 µm NO band. Although this is notan uncommon approach, it should be taken into ac-count in the interpretation of the experiments in thisstudy as a potential source of errors in particular dur-ing polar night in the upper atmosphere. With theexception of the parameterization of Fouquart andBonnel (1980), all heating schemes consider heatingalso at solar zenith angles larger than 90o. However,the Fouquart and Bonnel (1980) scheme is impor-tant only in the troposphere and in parts of the strato-sphere, and according to Fomichev et al. (2004b) thesphericity adds no noticeable contribution to the totalheating below the stratopause.

Chemical heating is computed within the chem-istry scheme for 7 exothermic reactions as given byBrasseur and Offermann (1986). The loss of en-ergy by chemiluminescence is taken into account us-ing efficiency factors from Mlynczak and Solomon(1993). Chemical heating is the release of heatby recombination reactions between atoms or radi-cals produced as a result of photolysis. This con-tribution is particularly important at heights wherethe photodissociation products can travel large dis-tances before recombining. In order to avoid doublecounting the incoming radiative energy, in the TUVcode, the energy corresponding to the photodisso-ciation energy threshold EB,x is subtracted from theenergy deposition rate: Eλ,x = (hν −EB,x)Jλ,x, withx being O3 or O2. For the heating computed by thescheme of Fouquart and Bonnel (1980) in SW2, weassume that, in this part of the spectrum, 23% of thetotal incoming energy is converted to chemical po-tential energy.

The parameterizations of radiative processes re-quire that the concentrations of the following com-pounds be known: O3, O2, O(3P) (used in the in-frared cooling scheme in the deactivation process ofCO2), CO2, H2O, CH4, N2, N2O, CFCs and aerosols.The two latter groups are taken from climatologiesrepresenting present-day conditions. For all othercompounds the prognostic model concentrations areused. The specific heat cp and the gas constant Rare computed at each time step according to thecalculated concentrations of chemical compounds.This is of importance in height regions above about90 km (∼ 2*10−3 hPa) where the major atmosphericconstituents are not uniformly mixed.

b. Parameterized gravity wave tendencies

Orographic gravity wave drag and surface blockingare parameterized using Lott and Miller (1997), asin ECHAM5. The momentum flux deposition froma spectrum of non-orographic gravity waves is pa-rameterized as in MAECHAM5. In addition, in HAM-MONIA the parameterizations of the heating andthe eddy diffusion of constituents, momentum, andtemperature generated by the breaking of gravitywaves are taken into account following the param-eterization of Hines (1997a,b). The viscous correc-tion to the eddy diffusion coefficient suggested byAkmaev et al. (1997) is applied. It reduces eddycoefficients computed by the Hines scheme at alti-tudes where molecular mixing is strong. HAMMO-NIA also differs from MAECHAM5 in the specifica-tion of the gravity wave source spectrum. In HAM-MONIA, this spectrum includes waves from tropo-spheric frontal activity (Charron and Manzini 2002)and a lower boundary for the vertical wavenumberspectrum. The parameterized gravity wave tendencyfrom non-orographic sources is strongly sensitive tothe choice of a cutoff vertical wave number. After testsimulations with different choices of this parameter, ithas been decided not to allow vertical wavenumberssmaller than 2π/(12 km) in order to obtain realisticmesospheric winds and temperatures.

5

c. Molecular diffusion and conduction

The radiative balance in the lower thermosphereregion is dominated by the extreme ultraviolet so-lar heating and the downward transport of heat bymolecular processes. Hence, molecular viscosityand conductivity are of primary importance in thisregion of the upper atmosphere. The transport andvertical distribution of constituents are also stronglyaffected by molecular viscosity and diffusion.

Vertical molecular diffusion of constituents (la-beled i) in the vertical direction follows the governingequation (Huang et al. 1998):

∂Xi

∂t=

1

ρ

∂

∂z

(

ρDi∂Xi

∂z

)

−

1

ρ

∂

∂z(ρwDiXi). (1)

Xi is the mass mixing ratio of constituent i, Di isthe respective diffusion coefficient (m2/s), and wDi isthe drift vertical velocity of constituent i which acts toseparate constituents of different molar masses. Us-ing the hydrostatic equation and performing a vari-able change from altitude coordinates to pressurecoordinates, the diffusion equation can be rewrittenas

∂Xi

∂t= g

∂

∂p

(

gρ2Di∂Xi

∂p

)

+ g∂

∂p(ρwDiXi). (2)

A semi-implicit vertical discretization that conservesmass is used to solve equation (2) for each con-stituent i. The semi-implicit factor is chosen to be1.5. The diffusion coefficient (multiplied by den-sity, ρDi) and the vertical drift velocity (multiplied bydensity, ρwDi) are taken from Banks and Kockarts(1973):

ρDi = 4.17 × 10−6

(

T

273.15K

)0.5 (

mA +m2

A

mi

)0.5

(3)

ρwDi =ρDig

R?T

(

mA − mi + αTiρR?

(

∂T

∂p

))

, (4)

where mA is the molar mass of air (g/mol), mi is themolar mass of constituent i (g/mol), R? = 8.31436J/mol/K, and αTi is the expansion coefficient for con-stituent i (-0.38 for H, zero for all other constituents).It has to be noted that after diffusion of the con-stituents, the sum of their mass mixing ratios is nor-malized to unity, because there is no other constraint

to locally conserve this sum in the numerical treat-ment of molecular diffusion. Another option wouldbe to solve the diffusion equations for number den-sities rather than mixing ratios (see Chabrillat et al.2002). Both methods have been tested off-line andresults were almost identical.

The vertical molecular diffusion of momentum isperformed as for constituents, except that:

Xi → ~u = (u, v) (5)

(ρDi) → µ = 1.87 × 10−5

(

T

273.15K

)0.69

(6)

(ρwDi) → 0, (7)

where µ is chosen to be the viscosity of atomic oxy-gen (Banks and Kockarts 1973), since this atom isthe dominant component in the uppermost modellayers where the importance of molecular diffusionbecomes largest.

Diffusion of heat uses similar equations in which

~u → T (8)

µ → µ/Pr. (9)

Pr = 0.72 is the Prandtl number observed for air(Batchelor 1967).

Frictional heating is also taken into account (themolecular diffusion of momentum leads to energydeposition in the form of heat). The temperature ten-dency due to frictional heating is written (Gill 1982)

(

∂T

∂t

)

frict.=

g2ρµ

cp

[

(

∂u

∂p

)2

+

(

∂v

∂p

)2]

. (10)

d. Ion drag and Lorenz force

In the MLT region, the neutral mass flow is affectedby the presence of ionized gases that follow themagnetic field of the Earth. In HAMMONIA, this ef-fect is parameterized following the simple Hong andLindzen (1976) approach, in which the impact onthe neutral flow can be represented by a drag anda force normal to the flow (the Lorenz force). There-fore, the wind tendency caused by ion drag (D) andLorenz force (L) is expressed as:

∂u

∂t= −Du − (L sin β)v; (11)

6

∂v

∂t= −(D sin2 β)v + (L sin β)u. (12)

where tan β = −2 tanφ, if φ is the latitude. Thevalues of D and L are given by Hong and Lindzen(1976) as a function of geopotential height. Theleapfrog time discretization is performed using asemi-implicit scheme for stability reasons, with thesemi-implicit factor chosen to be 1.5.

The ion drag is also producing heating. The kineticenergy dissipated by ion drag is introduced into thediabatic term of the temperature equation.

It should be noted that this parameterization givesonly a relatively crude mean estimation of the ther-mospheric ion drag. It is planned that future modelversions include a detailed treatment of E-regionions to allow for a more realistic response of the iondrag to the variability of solar irradiance and geo-magnetic activity.

e. Chemistry

HAMMONIA uses a condensed version of theMOZART3 chemistry module (Kinnison et al. 2004)with 48 compounds, 46 photolyses (Table 1) and107 bi- and termolecular gas phase reactions (Ta-ble 2). While the adopted tropospheric chemicalscheme is limited to the CH4-NOx-HOx-O3 systemand ignores therefore the effects of non-methane hy-drocarbons, this reaction scheme represents in suit-able detail the neutral chemistry of importance in themiddle and upper atmosphere. Heterogeneous re-actions associated with the formation of the ozonehole in the lower polar stratosphere are not includedin the present study. The chemical system is solvedwith a 10 minute time step, using the fully implicitEuler backward scheme with Newton-Raphson itera-tion. An explicit algorithm is utilized for the long-livedspecies (CH4, N2O, CO, CO2, CH3Cl, CH3Br, CFCl3,CF2Cl2, CFC113, HCFC22, CCl4, CH3CCl3, CF3Br,and CF2ClBr).

The computation of photodissociation rates isbased on the TUV scheme (Madronich and Flocke1998) with the clear sky rates tabulated for 122bands in the wavelength interval extending from 200to 780 nm. The tabulation is performed with re-spect to geopotential height, surface albedo, O3 col-

umn amount, and the solar zenith angle. The partof the spectrum ranging from 120 to 200 nm istreated explicitly with 34 wavelength intervals anduses parameterizations for the photolysis of O2 fromChabrillat and Kockarts (1998, Lyman-α), Brasseurand Solomon (1986, Schumann-Runge continuum),and Koppers and Murtagh (1996, Schumann-Rungebands). The NO photolysis is treated according toMinschwaner and Siskind (1993).

In the thermosphere, NO is produced mainly as aconsequence of the dissociation of molecular nitro-gen involving ambient or energetic electrons. Thelatter can be either of auroral origin or producedthrough photoionization processes. A more detaileddescription of the thermospheric NO production isgiven, e.g., by Bailey et al. (2002). In this first ver-sion of HAMMONIA, which does not treat the iono-sphere explicitly, the thermospheric NO productionis parameterized following the work of Huang et al.(1998) and McEwan and Phillips (1975). The pa-rameters used by Huang et al. (1998) have beenadapted to obtain thermospheric NO concentrationsthat are close to recent observations with the Stu-dent Nitric Oxide Explorer (SNOE) satellite instru-ment (Barth et al. 2003). The auroral production isderived by assuming that the Earth’s magnetic Northpole is located at geographic coordinates (78.5 N,291.0 E). We compute an additional NO productionin the stratosphere as a consequence of ion pro-duction by cosmic rays following Heaps (1978), andassuming that one NO molecule is produced pergenerated ion pair. An option for future model ver-sions would be to apply the more detailed parame-terization of thermospheric NO production providedby Marsh et al. (2004).

For most of the constituents, concentrations atthe lower boundary are specified at the surfaceas monthly mean values as calculated by theMOZART2 model (Horowitz et al. 2003) for presentday conditions. Only CO2 and N2O are not takenfrom MOZART2 but fixed to volume mixing ratios of360 (720) ppmv and 315 pptv, respectively. The up-per boundary is assumed to be a rigid lid for mostcompounds with the exception of atomic hydrogen(H) and oxygen (O3P). For these two constituents,the concentrations in the top layer are relaxed to val-

7

ues provided by the extended Mass-Spectrometer-Incoherent-Scatter (MSISE-90) climatology (Hedin1991).

Chemical species are updated each time step bya sequence of operators: advection, molecular diffu-sion, turbulent diffusion, cloud processes, and chem-istry. Cloud processes include convection and thephase transitions of water. Advection of tracersis performed using the flux form semi-Lagrangianscheme of Lin and Rood (1996).

3. Setup of the simulations

In order to assess the effect of the 11-year solar cy-cle two simulations with a length of 20 years eachhave been performed using the SW1 solar heatingapproach (see section 2a). Both simulations are per-formed for “present-day” conditions of greenhousegas concentrations as described above. The firstone represents conditions typical of a solar cycleminimum (as observed in September 1986). Thissimulation is also used in the validation of the model(chapter 4). The second simulation represents con-ditions typical of solar maximum (November 1989).The following parameters differ in the two runs:

• Solar irradiance from 120 nm to 300 nm asshown in Fig. 2. Differences between 300 and700 nm are also considered but in general arewell below 1%. The values were provided byLean (personal communication, 2004). Data forthe part of the spectrum from 200 to 400 nm aredescribed by Lean et al. (1997).

• The changes in the extreme ultraviolet (EUV) ra-diation are represented according to the param-eterization of Richards et al. (1994) with valuesof the solar F10.7 flux changing from 69 (meanfor September 1986) to 235 (November 1989).

• Upper boundary concentrations for atomic hy-drogen and oxygen are taken from the MSIS(Hedin 1991) climatology for the two values ofthe F10.7 flux mentioned above.

• The thermospheric NO production due to highenergetic solar irradiance is assumed to in-crease by 30% for solar maximum conditions.This value has been chosen in order to repro-duce reasonably the solar induced variation ofthermospheric NO observed by SNOE (Barthet al. 2003). The auroral NO production doesnot differ between the two model runs.

To estimate the sensitivity of the model to doublingof CO2 we performed four simulations over 11 years:

1*CO2 : As the simulation for solar minimum, i.e.with a CO2 mixing ratio of 360 ppmv fixed atthe surface, but with the SW2 approach for solarheating (see section 2a).

2*CO2 : As “1*CO2” but with a CO2 mixing ratio of720 ppmv fixed at the surface, and the sea sur-face temperatures (SSTs) and sea ice modifiedaccording to a 720 ppmv CO2 climate (see be-low).

SST(2*CO2) : As “1*CO2” but with the SSTs andsea ice modified according to a 720 ppmv CO2

climate.

2*CO2 (no SST) : As “1*CO2” but with a CO2 mixingratio of 720 ppmv fixed at the surface.

To avoid transient changes in the 720 ppmv CO2

simulations, the initial values of the CO2 mixing ra-tio are doubled in the entire model domain. The firstsimulated year, however, is not used for the anal-ysis of the runs. SSTs and sea ice cover for the360 ppmv CO2 simulations are taken from the Atmo-spheric Model Intercomparison Project (AMIP2) cli-matology. In the case of the 720 ppmv CO2 simula-tions, we added the differences of SSTs and sea icecover from coupled simulations with the ECHAM5 at-mospheric GCM (Roeckner et al. 2003) and the MaxPlanck Institute Ocean Model (MPI-OM, Marslandet al. 2003) for 720 and 360 ppmv of CO2, respec-tively. This results in an increase of July SSTs fordoubled CO2 between about 0 to 1 K for high south-ern latitudes, 2 to 3 K in tropical and mid-latitudenorthern oceans, and peaks of up to 6 K in singlenorthern polar locations where also the ice cover is

8

reduced significantly. The simulations “SST(2*CO2)”and “2*CO2 (no SST)” were performed to separatelocal radiative and non-local dynamical effects re-sulting from a doubling of the CO2 concentration.

All statistical significances for differences betweensimulations presented in this paper are computedusing Student’s T-test.

The comparison of model results with satellite ob-servations in the following section is done using the20-year solar minimum simulation. However, the dif-ferences between the reference simulations (solarminimum and “1*CO2”, respectively) that are due tothe use of the slightly different approaches for solarheating (SW2 and SW1, respectively) are negligiblefor most parameters.

The observational data used in the comparisonare coming from the following sources: in the caseof temperature and mesospheric ozone, version1.06 data from the Sounding of the Atmosphereusing Broadband Emission Radiometry (SABER)instrument (Mertens et al. 2001, 2004) on theThermosphere, Ionosphere, Mesosphere Ener-getics and Dynamics (TIMED) satellite are used.SABER ozone mixing ratios shown here are thosederived from observations of the O2(

1∆) daytimeemission at 1.27 µm. The data are available athttp://saber.larc.nasa.gov. In the case of zonalwind, methane, water vapor and stratosphericozone, the extended data from the Upper At-mosphere research Satellite (UARS) ReferenceAtmosphere Project (URAP) in the version 1.0 areused. All URAP data are taken from the web sitehttp://code916.gsfc.nasa.gov/Public/Analysis/UARS/urap/home.html. Ozone data are described byWang et al. (1996), methane and water vapor byRandel. et al. (1998). Wind data (Swinbank andOrtland 2003) are a composite of United KingdomMeteorological Office (UKMO) analyses belowabout 0.2 hPa (∼ 60 km) and UARS observationsabove. It should be noted, that the observations aremean values from limited periods: two years (2002and 2005) in the case of SABER, and 4 (winds) or 7years (chemical species) in the case of URAP.

4. Comparison of model resultswith satellite observations

In the following analysis of model results, we will fo-cus on the month of July and on zonal mean valuesfor most of the variables. Only in the case of temper-ature and ozone, the comparison is performed fora fixed local time because the observations are notuniformly distributed in local time and tidal variationsare expected to be large in the mesopause region.All model results are 20-year averages taken fromthe solar minimum simulation (see section 3). Mostdynamical and chemical features presented in thissection for the month of July are simulated symmet-rically in an at least qualitatively very similar way alsofor January.

Fig. 3 shows temperatures from HAMMONIA andthe SABER instrument as zonal means for 16:00local time. In general terms, the temperature dis-tribution is similar in the model and the observa-tions. Important features captured by the model in-clude the cold summer mesopause and the warmwinter stratopause (both caused mainly by the mo-mentum deposition from gravity waves), different al-titudes for summer and winter mesopause, the rapidincrease in temperature in the lower thermosphere,the warm summer stratosphere and the cold equato-rial tropopause. The differences in the temperatureat the equatorial mesopause region may be associ-ated with phase shifts in tides. The temperature min-imum below 180 K at about 30 to 60S and 10−4 hPais a feature of this particular local time and not ap-pearing in the zonal mean temperature. The strongvertical wave pattern at the tropical mesopause issignificantly weaker in the zonal mean than for 16:00LT. Simulated summer mesopause temperatures areabout 10 K warmer in January than in July.

In the case of zonal winds, the model repro-duces the general pattern shown in the compos-ite of UKMO analyses and observations from theHigh Resolution Doppler Imager (HRDI) on boardthe Upper Atmosphere Research Satellite (UARS,Swinbank and Ortland 2003) with some differencesin the positions of the jets and in associated max-imum speeds (Fig. 4). In particular, the reversal

9

from strong easterly winds in the summer meso-sphere to westerlies in the thermosphere is reason-ably well reproduced. The winter westerly jet in thestratopause region is slightly weaker in the modelthan in the observations. In the northern (sum-mer) hemisphere, the maxima of both the meso-spheric easterly and the thermospheric westerly jetsare stronger in the model than in the observations.The simulated mesospheric jet does not extend suf-ficiently poleward. In order to evaluate possiblereasons for the differences it is useful to compareour simulations to those with the extended CMAM(Fomichev et al. 2002) where the mesospheric sum-mer jet seems also to be underestimated at mid tohigh latitudes. Given that the non-orographic grav-ity wave drag is a major forcing term at these alti-tudes and that both models use the same param-eterization by Hines (1997a) this discrepancy be-tween models and observations may be attributedto the treatment of gravity waves. Figure 5 showsa comparison of the annual cycle of monthly andzonal mean zonal winds from URAP and HAMMO-NIA close to the stratopause (1 hPa, ∼ 50 km) andin the mesopause region (0.001 hPa, ∼ 92 km).At both altitudes the simulated and observed an-nual cycles exhibit a qualitatively similar wind pat-tern in both hemispheres with easterlies in summerand westerlies in winter at the stratopause and withsummertime westerly maxima at 0.001 hPa. At thestratopause, maximum wind speeds in the modelare higher than in the URAP data and the jets donot extend sufficiently poleward. However, in bothmodel and data, winds are weaker in the northernthan in the southern hemisphere. A difference in themesopause region are the equatorial easterly windsof the URAP analysis which are not featured in themodel. These easterlies are supposed to be forcedby the momentum deposition of the breaking diur-nal tide (e.g. Lieberman and Hays 1994). An under-estimation of the tidal forcing by the model may bea reason for this discrepancy. However, above andbelow 0.001 hPa (not shown), the model simulatesequatorial easterlies during non-solstice conditionsin a manner qualitatively similar to the URAP obser-vations.

Zonal mean ozone mixing ratios for 15:00 local

time are compared to observations from SABERdaytime data above 1 hPa (∼ 50 km) and the URAPclimatological data below (Fig. 6). Again, the mainobserved features are well reproduced by the model:the primary and secondary ozone maxima, extend-ing from pole to pole respectively in the stratosphereand near the mesopause, as well as the tertiarymaximum (Marsh et al. 2001) in the high latitudewinter mesosphere. Up to the mid-mesosphere,observed and simulated mixing ratios agree well.The slight underestimation of ozone mixing ratiosin the stratospheric maximum may be connected tothe overestimation of water vapor (see below). Themodel mixing ratios in the secondary maximum areapproximately 30% lower than the observations. Thecause of this difference is unclear, but it should benoted that the new SABER observations have notbeen fully validated. However, the spatial patternsof the simulated and observed secondary daytimemaxima are very similar. Both data sets show for ex-ample a remarkable local maximum in the high lat-itude summer mesopause region. This feature is aconsequence of very low temperatures (cf. Fig 3)and can be explained from the ozone equation un-der photochemical equilibrium. Following Allen et al.(1984), in the upper mesosphere the ozone concen-tration is approximately given by

[O3] =kO+O2+M [O][O2][M ]

JO3+ kH+O3

[H ] + kO+O3[O]

, (13)

with the brackets indicating the number densities ofchemical species, k the reaction rate constants, JO3

the photodissociation rate of ozone, and M any re-action partner. kO+O2+M is increasing exponentiallywith decreasing temperature. Additionally, any num-ber density is, given a constant volume mixing ra-tio, inversely proportional to temperature. Therefore,ozone mixing ratios in the polar summer mesopauseregion are significantly higher than at lower latitudesalthough the mixing ratio of atomic oxygen is rela-tively low (see Fig. 12) due to the polar upwelling.

Simulated nighttime ozone mixing ratios (notshown except for polar night values) in the sec-ondary maximum are about one order of magnitudehigher than those from daytime as also observede.g. by the Cryogenic Infrared Spectrometers and

10

Telescopes for the Atmosphere (CRISTA) satelliteobservations (Kaufmann et al. 2003).

The comparison of simulated mixing ratios for wa-ter vapor and methane with URAP data demon-strates the models ability to simulate a realistic lat-itudinal and vertical distribution of important tracegases. In the case of water vapor (Fig. 7), the up-ward tilt of the water vapor isolines from the winter tothe summer hemisphere is indicative of upwelling inthe summer and downwelling in the winter (i.e. theBrewer-Dobson circulation). In addition, in the lowertropical stratosphere close to the tropopause, thewater vapor distribution in the model is asymmetric,with higher water vapor around 30N, in agreementwith a monsoon source of moisture for the strato-sphere (Gettelman et al. 2004). In most of the strato-sphere and mesosphere, simulated water vapor mix-ing ratios are up to 1 ppm higher than in the URAPdata. The water vapor level in the model is very sen-sitive to the temperature in the equatorial tropopauseregion (see below). An error in this parameter andthe relatively coarse vertical model resolution maylead to an overestimation of the water vapor trans-port across the tropopause. The high observed mix-ing ratios in the high latitude upper mesosphere insummer may be a consequence of the downwardtransport and evaporation of icy particles from po-lar mesospheric clouds (von Zahn and Berger 2003)which are not included in our model. Fig. 8 presentsthe time evolution over 10 simulated years of theequatorial water vapor mixing ratio in terms of itsdeviation from the 10-year mean value at heightsfrom the surface to 10−3 hPa (∼ 92 km). The modelreproduces the well known features like the “taperecorder” (Mote et al. 1996) in the lower stratosphereand the semi-annual oscillation (SAO) close to thestratopause (Randel. et al. 1998). The tape recorderis currently slightly too fast because of the missingquasi-biennial oscillation (QBO, see Giorgetta et al.2005, this issue). An SAO with relatively large ampli-tude and almost opposite phase to the stratopauseSAO is simulated for the mesopause region. Thislatter feature has been clearly observed in temper-atures and zonal winds (Garcia et al. 1997) and isalso detected in water vapor time series observedby the Halogen Occultation Experiment (HALOE) in-

strument as presented by Steil et al. (2003). An in-teresting feature is the positive tropopause water va-por anomaly in year 4 of Fig. 8 that is connected toa strong positive temperature anomaly. This event isleading to enhanced water vapor mixing ratios in themiddle atmosphere for several years.

In the case of methane (Fig. 9), the equatorial rateof vertical decrease is well captured by the model,indicating that upwelling is realistic. In addition, inthe winter (southern) hemisphere, the methane iso-lines show a flattening at mid-latitudes in the strato-sphere, indicative of isentropic horizontal mixing.The Brewer-Dobson circulation manifests itself as alatitudinal gradient of methane (as for water vapor) inthe mesosphere with lower values at the winter polewith respect to the summer pole.

5. Sensitivity to the 11-year solarcycle

a. Model results

Fig. 10 shows the solar cycle effect on the main com-ponents of the energy budget in the mesosphere andlower thermosphere: solar heating (including chem-ical heating), infrared cooling (including heating inthe near IR CO2 bands), and molecular heat con-duction. The increased radiation for solar maximumconditions leads to a substantial increase in solarheating above approximately 0.01 hPa (∼ 80 km).Up to 10−4 hPa (∼ 105 km) this increase amountsto about 2 K/day. In this region, about 60% of theincrease (as well as of the absolute heating rate)are due to chemical heating. The increased heat-ing is balanced to a large degree by enhanced in-frared cooling. Above 10−4 hPa the increase in so-lar heating becomes very large. In a small regionaround 5*10−5 hPa (∼ 110 km) the major changeof the energy budget is provided by heating due tomolecular diffusion (conduction), which more thandoubles. In the thermosphere, heat is transporteddownward by conduction, leading to a positive tem-perature effect of conduction in this height range.Above 2*10−5 hPa (∼ 120 km), the conduction ef-fect becomes negative, and it is the main balancing

11

process for solar heating and its increase.This paper’s focus is on the mesosphere and lower

thermosphere. However, because dynamics of theupper atmosphere may be substantially influencedby non-local responses feeding back from the lowerpart of the atmosphere, we present in Fig. 11 thesolar cycle effect on zonal mean temperature, zonalwind, and vertical residual wind from the surface to10−5 hPa (∼ 130 km).

The July zonal mean temperature difference be-tween solar maximum and minimum (Fig. 11)is positive everywhere above approximately thetropopause except for the southern (winter) hemi-sphere poleward of about 40oS. Here, the signalis significantly positive only above about 0.03 hPa(∼ 75 km). The stratospheric temperature response(with values of about 1 K at the tropical stratopause)is well within the range of 0.4 to 2 K from differ-ent observational analyses which are presented e.g.by Rozanov et al. (2004). In the mesopause re-gion, the change lies between 2 and 10 K with val-ues increasing in general with altitude and towardsthe summer pole. This latitudinal gradient is due toa change in the radiative budget. In high summerlatitudes at around 0.001 hPa the net radiative ef-fect is a warming that is mainly due to the strongincrease in chemical heating (which is an indirectradiative effect) caused by the increase in ozone(see below). The absolute minimum temperatureat the summer mesopause changes only by about1.5 K. The larger changes are occurring above themesopause. The general heating leads to an ex-pansion of the atmosphere. Fig. 1 shows an annualand global mean expansion of approximately 1 km at10−4 hPa (∼ 105 km) and 100 km at the model top(1.7*10−7 hPa).

Regarding the July zonal mean zonal wind differ-ence, only relatively small changes can be seen inFig. 11 except close to 10−5 hPa (∼ 130 km, andabove, not shown). Here, the zonal wind is directlyaffected by a change in the ion drag which is a resultof the ion drag increasing with geopotential heightand the expansion of the atmosphere. However,the increase of the ion drag for solar maximum isprobably underestimated in these simulations as weuse the same profile of the ion drag coefficient for

solar maximum and minimum. Hong and Lindzen(1976) suggest larger coefficients for solar maxi-mum. In the summer hemisphere, the change in thezonal wind implies a weakening of the lower ther-mospheric westerlies and the mesospheric easter-lies. Such wind changes are consistent with reducedupwelling between 0.01 and 0.001 hPa (∼ 80 and92 km), and reduced downwelling above this layer.The zonal wind response can be explained with thethermal wind equation. The positive meridional gra-dient in the high latitude temperature change in themesopause region is causing a negative vertical gra-dient in the zonal wind. The solar response ofthe MLT region includes also changes in the grav-ity wave drag (not shown in the figures), but they aremore likely a response to the changes in the windfield than their cause.

Figure 12 shows the solar cycle response in zonalmean July volume mixing ratios for ozone, atomicoxygen, nitric oxide, the hydroxyl radical, water vaporand methane. Mesospheric ozone is characterizedby a substantial diurnal variation with nighttime mix-ing ratios being about an order of magnitude largerthan daytime values above 0.01 hPa. Therefore,considering zonal mean values does not seem to beadequate. However, as the relative solar responsesof day and nighttime ozone exhibit very similar pat-terns, we present zonal mean values also for thisparticular compound. In general, the photochemistryleading to ozone production and destruction is am-plified for enhanced solar irradiance which leads toa strengthening of the ozone maxima and an ozonedecrease in regions with already low mixing ratios.The increase in the secondary maximum of up to20% results from the enhanced mixing ratio of O,which is a consequence of increased photodissoci-ation of O2. The general decrease in a small heightregion below 0.01 hPa is mainly due to the enhancedmixing ratio of OH. For the relation of mesosphericozone and atomic oxygen to odd hydrogen (HOx =H + OH + HO2) see Brasseur and Offermann (1986)and Marsh et al. (2003). The higher temperature atsolar maximum tends to decrease ozone. At the al-titude of the secondary maximum at mid northernlatitudes, this decreasing effect is stronger than theozone increase resulting from higher mixing ratios of

12

atomic oxygen. The complicated structure of ozoneresponse can also be seen in Fig. 13, which showsa horizontal cross section at 2*10−3 hPa (∼ 90 km).The ozone response is significantly negative onlyat nighttime between approximately 45 and 65 de-gree North. This behavior can again be explained bythe low summer temperatures (see section 4) andthe non-linearity of the temperature dependence ofozone. Fig. 13 shows also that for most latitudes thechanges in nighttime and daytime ozone are similar.

Zonal mean mixing ratios of H2O and CH4 ex-hibit a decrease for solar maximum, which is mainlyan effect of increased photodissociation particularlyby the approximately 60% increase of Lyman-α ra-diation. In the case of water vapor, the decreaseis characterized by a strong latitudinal dependenceand reaches down to about 1 hPa (∼ 50 km) inthe winter hemisphere and only to about 0.03 hPa(∼ 75 km) in summer. This tilt is again the resultof the prevailing circulation. The increase in theNO mixing ratio depends strongly on the assumptionmade for the increase in thermospheric NO produc-tion (see section 2e). However, less than half of thethermospheric NO increase can be explained by thisimposed change of external production. The largerpart of the change results from the large temperaturechanges which affect rate constants, and specificallythe rate of the (N + O2 → NO + O) reaction (seee.g. McEwan and Phillips 1975, for a description ofNO chemistry). The downwelling in the winter hemi-sphere leads to a significant NOx increase down tothe upper stratosphere, especially near the pole.

All dynamical and chemical features of the solarresponse described above for the month of July aresimulated symmetrically in a qualitatively very similarway also for January.

b. Comparison with other models

A relatively large number of models has been ap-plied to assess the effect of solar variability ontemperature, dynamics and chemistry of the strato-sphere. See for example a recent study with acoupled 3-dimensional climate-chemistry model byRozanov et al. (2004) and references therein. Theannual mean stratospheric response of temperature

(about 1K at the equatorial stratopause) and ozone(about 3% in the upper stratosphere) simulated withHAMMONIA lies well within the large variety of ob-servational and modeling results listed by Rozanovet al. (2004).

For the mesosphere and lower thermosphere, thenumber of studies is small. Several studies ad-dressed the effect of the 27-day rotational variationwith 1-dimensional (Brasseur et al. 1987; Summerset al. 1990; Chen et al. 1997) or 2-dimensional (Zhuet al. 2003) chemical dynamical models. The longterm (11-year solar cycle) variability was studied withdifferent versions of the NCAR interactive 2-D modelSOCRATES by Huang and Brasseur (1993, here-after called HB1993) and Khosravi et al. (2002, here-after called K2002). Most of their results can qualita-tively be confirmed by our simulations although anaccurate comparison of the studies is difficult be-cause results are given for different seasons. In ad-dition, the top of most models is located at a consid-erably lower altitude than in our case. While HB1993simulate a temperature peak to peak response to so-lar activity in the mesopause region of about 10 K,the value of about 5 K reported by K2002 is closeto our simulations. As in our model, HB1993 de-tect only small changes in mesospheric winds. Thestatement of K2002 that dynamical feedbacks causemost of the change in temperature above 75 km(∼ 0.02 hPa) can not be confirmed by our model. Inthe case of ozone, the response patterns simulatedby K2002 are relatively similar to the patterns simu-lated by HAMMONIA but with slightly higher ampli-tudes. HB1993 simulate a very large response inatomic oxygen and ozone mixing ratios in a smalllayer above 80 km (∼ 0.01 hPa) with increases ofup to 120 and 55%, respectively. Several otherimportant features like the change in sign for theOH response around 0.01 hPa and the strong H2Oresponse reaching down below the stratopause insummer are simulated in all models.

6. Sensitivity to CO2 doubling

The cooling resulting from infrared CO2 emissionsis a major contribution to the energy budget of the

13

middle atmosphere and lower thermosphere. Therapid increase of the atmospheric CO2 concentrationresulting from anthropogenic emissions is thereforeexpected to lead, in general, to a substantial coolingin this height range. Heating due to absorption ofnear infrared radiation by CO2 can potentially coun-teract the cooling. However, according to Fomichevet al. (2004a) this effect is expected to be small.The simulation of the atmosphere for a doubled CO2

concentration with respect to present day conditionshas become a common benchmark experiment forclimate models, and the respective simulation withHAMMONIA not only provides insight in the futureatmosphere but also allows comparisons with a va-riety of similar studies.

a. Model results

Figure 14 shows the change in the main compo-nents of the energy budget in the mesosphere andlower thermosphere induced by a doubling in the at-mospheric CO2 concentration. As can be expected,the infrared cooling (which is mainly due to infraredemission in the 15 µm CO2 band) increases stronglyin the regions where this contribution is of impor-tance: below 0.1 hPa (∼ 65 km) and above 10−4 hPa(∼ 105 km). In the lower mesosphere this increase ispartly balanced by increased solar heating as a con-sequence of the ozone feedback (see below). Be-tween 10−4 and 10−5 hPa (∼ 105 and 130 km) themajor balancing is achieved through an increase inthe heating produced by conduction.

The CO2 doubling effect on zonal mean July tem-perature and winds is given in Fig. 15. As expected,the temperature decreases significantly almost ev-erywhere above the tropopause. The general cool-ing leads to a shrinking of the atmosphere (Fig. 1)that reaches 1 km slightly above the stratopause and10 km around 10−6 hPa (∼ 180 km). While the de-crease in the lower mesosphere reaches 10 K, themesopause region shows the smallest and in someplaces insignificant cooling. This is the region whereradiative transfer in the infrared CO2 bands providesvery little cooling or can even heat the atmospheredue to the energy exchange with warmer underlyinglayers. In addition, in this region, the near-IR CO2

bands provide heating.In the following, it will be shown that dynami-

cal effects can also influence the temperature re-sponse in the middle and upper atmosphere andcontribute to the small effect around the mesopause.In our simulation, the dynamical response in theMLT is characterized by a general increase in theresidual circulation with an increased downwellingin the polar winter mesosphere and increased up-welling around the mesopause near the summerpole (Fig. 15). Consequently, zonal wind and tem-perature changes are less pronounced in the po-lar winter upper mesosphere, where the increaseddownwelling contributes to a warming and weakerzonal winds. Zonal wind and temperature changesare instead more pronounced above the summer po-lar mesopause region, where an upward shift of thesummer easterlies and of the associated wind re-versal, together with increased westerlies close tothe 10−5 hPa (∼ 130 km) level are predicted bythe model. The analysis of two additional simula-tions (“SST(2*CO2)”, Fig. 16, with only the SSTschanged, and “2*CO2 (no SST)”, Fig. 17, with onlythe CO2 mixing ratio increased) allows the sepa-ration of remote dynamical from local radiative ef-fects of the CO2 doubling. Changes in the zonal andthe residual vertical winds with respect to the refer-ence simulation show large similarities between the“2*CO2” and “SST(2*CO2)” simulations (cf. Figs. 15and 16) at least up to the middle mesosphere whilein the thermosphere the responses of the “2*CO2”and the “2*CO2 (no SST)” experiments are similar(cf. Figs. 15 and 17). The sum of the two sepa-rate responses, for both temperature and winds, isalmost equal to the changes observed in the com-bined simulation.

The positive temperature response to changingSSTs in the uppermost mesosphere suggests thatthe limited cooling in this region produced in re-sponse to a doubling of CO2 is partly caused bydynamics. Though it is difficult to separate causesand consequences of the complex dynamical inter-actions in the middle atmosphere we suggest thefollowing mechanism. The temperature increase inthe summer hemisphere in the “SST(2*CO2)” ex-periment is due to the decrease in upwelling lead-

14

ing to less adiabatic and advective cooling. The al-most global positive temperature effect close to themesopause is mainly due to a decrease in resid-ual zonal southward winds between about 0.01 and0.001 hPa leading to less advection of cold air.The reason for the decreased residual circulationat these altitudes is a reduction in the gravity waveforcing which is mainly induced by changed prop-agation conditions with, in particular, the decreaseof the wintertime westerly jet around 30S at thestratopause. In the upper summer mesosphere, thedecrease of upwelling in the “SST(2*CO2)” experi-ment is more than compensated by the increase ob-served in the “2*CO2 (no SST)” experiment. Thetotal effect of CO2 doubling in this height region istherefore an increase in residual circulation. Thereason for the increased upwelling is the latitudinalgradient in the temperature response leading to achanging circulation via the thermal wind relation.In the very cold summer polar mesopause the longwave radiative effect of an increase in CO2 is, incontrast to most other regions in the atmosphere, awarming causing this gradient in the temperature re-sponse.

The upper mesospheric heating in the“SST(2*CO2)” experiment can be observed dur-ing northern hemisphere summer from June toSeptember but has its maximum in July. In thesouthern summer however (see Fig. 18), an almostglobal warming of the mesopause region can notbe observed. Reduced upwelling is leading to awarming of the summer hemisphere. Changes ingravity wave forcing and subsequent mesopausemeridional winds are much less significant thanin July, so that the change in advective heating issmall. The total CO2 doubling effect (“2*CO2” minus“1*CO2”, not shown) shows therefore a less pro-nounced mesopause cooling minimum in Januarythan in July.

As described earlier, the CO2 mixing ratio is fixedat the surface, but calculated in the atmosphere.This is why the actual mixing ratio is not exactly dou-bled at all altitudes. Fig. 19 shows that the CO2

increase above 0.01 hPa (∼ 80 km) is larger than100%. This can partly be explained by a modifica-tion of the ratio between CO2 and CO. The increase

in OH (see below) increases this ratio via the reac-tion CO + OH → CO2 + H. Additionally, the increasedCH4 mixing ratio leads to an increased CO2 produc-tion through photodissociation.

The response of several other chemical com-pounds (displayed in Fig. 20) to the doubling of theCO2 concentration can be explained as a combina-tion of changes in rate constants and atmosphericdensity (due to the cooling), and in the strength ofup- and downwelling. Under enhanced CO2 con-centrations, the ratio between the O3 and O mixingratios is generally shifted towards higher O3, whichis a consequence of the strong temperature depen-dence of the ozone production reaction (O + O2 +M → O3 + M). The large decrease in the atomic oxy-gen mixing ratio at high summer latitudes above 0.01hPa results from increased upwelling and leads alsoto an ozone decrease at this level. The ozone de-crease in the polar winter around 0.1 hPa (∼ 65 km)is mainly caused by the increase of NO and (to alesser extent) Cl mixing ratios due to stronger sub-sidence of NO and Cl-rich air. The methane in-crease can be explained by the temperature depen-dence of the loss reactions with OH and chlorine.The increase in thermospheric water vapor is a con-sequence of the increased production by photodis-sociation of methane. It should be noted that thesometimes large relative changes in ozone, methaneand water vapor in the thermosphere are occur-ring where the absolute mixing ratios of these con-stituents are very low. OH shows a strong responsewith changing sign at ∼ 0.01 and ∼ 3.*10−5 hPa(∼ 115 km). At different heights, different produc-tion reactions are the main source of OH what leadsto the change of signs: The OH increase above∼ 0.01 hPa is caused by the temperature depen-dence of the reaction HO2 + O → OH + O2 and bythe ozone increase leading to more OH via the re-action H + O3 → OH + O2. The reaction rate ofO + H2 → OH + H is decreasing with decreasing tem-perature and therefore leading to the OH decreaseabove ∼ 3.*10−5 hPa.

15

b. Comparison with other models

A variety of different models have been used in thepast to study the effect of CO2 doubling. Theyhave concentrated on temperature not only in thestratosphere but also in the mesosphere and ther-mosphere. Most of them are 2-D (Portmann et al.1995) or 3-D (Rind et al. 1990; Berger and Dameris1993; Akmaev and Fomichev 1998) dynamical mod-els. The feedback on chemistry was included e.g. ina 1-D model study by Roble and Dickinson (1989),in simulations with slightly different versions of the2-D model SOCRATES by K2002 and by Gruzdevand Brasseur (2004), and in a 3-D model study byJonsson et al. (2004) covering heights up to ap-proximately 100 km (∼ 2.*10−4 hPa). Most mod-els derive a cooling of about 10 K (or more) nearthe stratopause and the mesopause, and a weakercooling in the mesosphere. Our results suggest rela-tively low cooling rates around the mesopause prob-ably due to the full inclusion in HAMMONIA of dy-namical and chemical (increase of O3) feedbacks.This is in agreement with the study by Jonsson et al.(2004) that also includes the ozone feedback but noSST changes. The studies by Rind et al. (1990) andPortmann et al. (1995) point out the importance ofthese dynamical feedbacks. The increased residualcirculation reported by Rind et al. (1990) is in generalagreement with our simulations, as are the changesin the residual vertical velocity calculated by K2002.The chemical response to CO2 doubling presentedby K2002 is relatively similar to our simulations forO3, O, and OH. In the case of H2O and CH4, bothmodels show a relatively strong increase of the re-spective mixing ratios in the lower thermosphere.However, in K2002, the water vapor response is pos-itive in the entire mesosphere while, in the presentstudy, it is weakly negative in the lower mesosphere.The CH4 response predicted by HAMMONIA is pos-itive everywhere in the mesosphere, while in K2002it is negative in the lower mesosphere. Jonssonet al. (2004) present similar percentage variationsfor the mesospheric ozone response in January aspredicted by HAMMONIA for July including relativelysmall areas of negative response at high latitudes inthe upper mesosphere.

7. Discussion and Conclusions

The results of this study have been presented us-ing atmospheric pressure as the vertical coordinate.This is a natural choice because model levels (above100 hPa) are levels of constant pressure. In ad-dition, the interpretation of changes in the chemi-cal composition is easier on constant pressure lev-els than in height coordinates. However, as mostobservations are performed with respect to geomet-ric altitude it is useful to present model results aswell for constant height levels. It has been notedearlier in the context of CO2 doubling experiments(e.g. by Akmaev and Fomichev 1998) that atmo-spheric changes can be very different for fixed heightand fixed pressure. This is a consequence of botha temperature (and subsequent density) change atone pressure level and the integrated effect of tem-perature changes below this level. Figure 1 showsthat strong atmospheric expansion and contractionoccurs for the experiments of this study. Figure 21presents the effects of the solar cycle and of CO2

doubling on temperature and ozone mixing ratios inthe mesosphere and lower thermosphere with re-spect to geometric height. For the solar cycle ef-fect with only a moderate atmospheric expansionup to the mesopause, differences between pressureand height coordinates are relatively small. In thecase of the temperature response, a weak verticalwave structure around the summer mesopause witha local minimum at 100 km (∼ 2.*10−4 hPa) occursonly in height coordinates. The latitudinal gradientwith a strong temperature response at high summerlatitudes around 10−3 hPa (∼ 92 km, see Fig. 11)is not visible in height coordinates. In the case ofozone, there is a continuous pole-to-pole region ofpositive response near the mesopause which is in-terrupted when represented in pressure coordinatesas described earlier. The effect of CO2 doubling onozone appears as completely different in pressurecoordinates (positive everywhere except at some al-titudes in high latitude regions) and in height co-ordinates (vertically alternating strong positive andnegative responses). This latter pattern is mainlya consequence of large positive and negative ver-

16

tical gradients in the ozone mixing ratios. The at-mospheric cooling due to the CO2 increase exhibitsa second minimum in a height region located near110 km (∼ 5*10−5 hPa) while it is constantly increas-ing above the mesopause when displayed in pres-sure coordinates. However, the temperature effectremains negative everywhere in the thermosphere,a result that differs from those obtained by Akmaevand Fomichev (1998) and Gruzdev and Brasseur(2004) who present a temperature increase in a partof the lower thermosphere. Reasons for these dis-crepancies to our results may be that the simulationsby Akmaev and Fomichev (1998) do not include in-teractive chemistry so that a possible enhancementof the cooling by the more than doubled CO2 mixingratio in the thermosphere could not be included. Therelatively low upper lid at a log-pressure altitude of120 km may represent an error source in the simu-lations of Gruzdev and Brasseur (2004). However, itshould be noted that none of the models representsexplicitly the NO cooling in the 5.3 µm band so thata potentially important feedback is missing.

Observational studies of mesospheric and ther-mospheric variations in relation with solar variabilityare rare. Beig et al. (2003), in their review paperon mesospheric temperature trends, present a col-lection of different local analyses for the temperatureresponse to the solar cycle. For the layer between85 and 90 km (∼ 3*10−3 hPa) altitude, a relativelylarge number of observations exists. The responseslie between 0 and 5 K normalized to an irradiancechange of 100 solar flux units (sfu; solar radiation ata wavelength of 10.7 cm in 10−22 m−2Hz−1). Thelatitudinal differences in the response suggested bythe observations can not be confirmed by our simu-lations that give 2 to 2.5 K/(100 sfu) at all latitudes.However, the errors associated with the analysis ofthe observations are in general large (see Beig etal. 2003, and references therein). DeLand et al.(2003) show a decrease in the occurrence frequencyof PMCs for solar maximum conditions. This shouldtheoretically either be related to an increase in tem-perature, a decrease in the water vapor concentra-tion or a decrease in the concentration of conden-sation nuclei in the polar summer mesopause region(or a combination of these causes). Our simulation

suggests that at least the first two phenomena areoccurring. Reliable analyses of the solar effect onother dynamical or chemical parameters are to ourknowledge not available. It is difficult to evaluate themodel results for a doubling of CO2 with observa-tional data. However, if one assumes that the tem-perature trends observed for the past decades in themiddle and upper atmosphere are mainly due to theCO2 increase, the lack of a detectable trend in thesummer mesopause region (Beig et al. 2003) is con-sistent with our results for a CO2 doubling.

HAMMONIA represents a new generation of at-mospheric models that try to combine classical gen-eral circulation models with comprehensive chem-istry over a large altitude range. It aims at includingall important atmospheric feedbacks in the model tomake the simulations as realistic as possible. Run-ning such a model has become possible with the in-crease in computer power made available during thelast years. Nevertheless, it is not yet possible to runsuch a large model for very long time spans or makea large number of sensitivity tests. HAMMONIA willmainly be applied for research on vertical couplingphenomena in the atmosphere and for those caseswhere a simulation should be as realistic as possible,e.g. for the explanation of observed mesospherictrends.

Despite the complexity of the model, we still seedeficiencies that may have altered the results of thisstudy. For example, the current model version doesnot simulate the QBO of equatorial stratosphericwinds which is believed to interact with the effectof solar-induced atmospheric variability (e.g. Lab-itzke 2003). This deficiency is expected to be over-come by increasing the vertical model resolution aswas done in the MAECHAM5 model (Giorgetta et al.2002). The present version of HAMMONIA is alsolacking a representation of the role of ions. Theabundance of ions depends on the intensity of so-lar irradiance and energetic particle fluxes and influ-ences neutral chemistry in particular in the thermo-sphere and mesosphere.

This paper is intended to introduce a new modelthat extends from the surface to the thermosphere,evaluate its results by comparing them to availableobservations, and assess the atmospheric response

17

to different types of external forcing which are clas-sic benchmarks for such models. The model simu-lations have produced a huge amount of data. Thebasic analysis of the data has lead to the results pre-sented in the present study. More detailed analysesof e.g. the energy budget or the response of tidesand other waves to the different types of forcing willbe the subject of subsequent papers.

Acknowledgement The authors are grateful to Ju-dith Lean, Naval Research Laboratory, Washing-ton, for providing the solar irradiance spectra. Wealso thank the partners of the Mesospheric Dynam-ics, Energetics and Chemistry (MEDEC) researchproject for help in the evaluation of HAMMONIA. Weare grateful to the TIMED-SABER Science Team forproviding access to data from the SABER instru-ment. The work was financed by the German Min-istry of Education and Research (BMBF) within theframework of the AFO2000 research program un-der contract 07ATF10. We also acknowledge thesupport of the German Climate Computing Cen-ter (DKRZ) where the numerical computations wereperformed. Finally, we would like to thank two anony-mous reviewers for their very helpful comments.

References

Akmaev, R. A. and V. I. Fomichev, 1998: Cooling ofthe mesosphere and lower thermosphere due todoubling of CO2. Ann. Geophys., 16, 1501–1512.

Akmaev, R. A., V. A. Yudin, and D. A. Ortland, 1997:SMLTM simulation of the diurnal tide: Compari-son with UARS observations. Ann. Geophys., 15,1187–1197.

Allen, M., J. I. Lunne, and Y. L. Yung, 1984: The ver-tical distribution of ozone in the mesosphere andlower thermosphere. J. Geophys. Res., 89, 4841–4872.

Bailey, S. M., C. A. Barth, and S. C. Solomon,2002: A model of nitric oxide in the lowerthermosphere. J. Geophys. Res., 107, 1206,doi:10.1029/2001JA000258.

Banks, P. M. and G. Kockarts, 1973: Aeronomy, PartB. Academic Press, New York, 355 pp.

Barth, C. A., K. D. Mankoff, S. M. Bailey, and S. C.Solomon, 2003: Global observations of nitric ox-ide in the thermosphere. J. Geophys. Res., 108,1027, doi:10.1029/2002JA009458.

Batchelor, G. K., 1967: An Introduction to Fluid Dy-namics. Cambridge University Press, Cambridge,UK, 635 pp.

Beagley, S. R., J. de Grandpre, J. N. Koshyk, N. A.McFarlane, and T. G. Shepherd, 1997: Radiative-dynamical climatology of the first-generationCanadian middle atmosphere model. Atmos.Ocean, 35, 293 – 331.

Beig et al., G., 2003: Review of mesospherictemperature trends. Rev. Geophys., 41, 1015,doi:10.1029/2002RG000121.

Berger, U. and M. Dameris, 1993: Cooling of themesosphere due to CO2 increases: a modelstudy. Ann. Geophys., 11, 809–819.

18

Boville, B. A., 1995: Middle atmosphere version ofthe CCM2 (MACCM2): Annual cycle and inter-annual variability. J. Geophys. Res., 100, 9017–9039.

Brasseur, G., A. De Rudder, G. M. Keating, andJ. Nicholson III, 1987: Response of middle atmo-sphere to short-term solar ultraviolet variations: 2.Theory. J. Geophys. Res., 92, 903–914.

Brasseur, G. P. and D. Offermann, 1986: Recombi-nation of atomic oxygen near the mesopause: In-terpretation of rocket data. J. Geophys. Res., 91,10818–10824.

Brasseur, G. P. and S. Solomon, 1986: Aeronomy ofthe middle atmosphere. D. Reidel, Norwell, Mass.,452 pp.

Brasseur, G. P., X. Tie, P. J. Rasch, and F. Lefvre,1997: A three-dimensional model simulation ofthe antarctic ozone hole: Impact of anthropogenicchlorine on the lower stratosphere and upper tro-posphere. J. Geophys. Res., 102, 8909–8930.

Chabrillat, S. and G. Kockarts, 1998: Correction to”Simple parameterization of the absorption of thesolar lyman-alpha line”. Geophys. Res. Lett., 25,79–80.

Chabrillat, S., G. Kockarts, and D. Fonteyn,2002: Impact of molecular diffusion on theCO2 distribution and the temperature in themesosphere. Geophys. Res. Lett., 29, 1–4,doi:10.1029/2002GL015309.

Charron, M. and E. Manzini, 2002: Gravity wavesfrom fronts: Parameterization and middle atmo-sphere response in a general circulation model. J.Atmos. Sci., 59, 923–941.

Chen, L., J. London, and G. Brasseur, 1997: Middleatmospheric ozone and temperature responses tosolar irradiance variations over 27-day periods. J.Geophys. Res., 102, 29957–29979.

Chipperfield, M. P., D. Cariolle, P. Simon, R. Ra-maroson, and D. J. Lary, 1993: A 3-dimensionalmodeling study of trace species in the arctic lower

stratosphere during winter 1989-1990. J. Geo-phys. Res., 98, 7199–7218.

de Grandpre, J., S. R. Beagly, V. I. Fomichev, E. Grif-fioen, J. C. McConnell, and A. S. Medvedev,2000: Ozone climatology using interactive chem-istry: Results from the Canadian middle atmo-sphere model. J. Geophys. Res., 105, 26475–26491.

DeLand, M. T., E. P. Shettle, G. E. Thomas, andJ. J. Olivero, 2003: Solar backscattered ultraviolet(SBUV) observations of polar mesospheric clouds(PMCs) over two solar cycles. J. Geophys. Res.,108, 8445, doi:10.1029/2002JD002398.

Fels, S. B., J. D. Mahlman, M. D. Schwarzkopf, andR. W. Sinclair, 1980: Stratospheric sensitivity toperturbations in ozone and carbon dioxide: Radia-tive and dynamical response. J. Atmos. Sci., 37,2265–2297.

Fomichev, V. I. and J.-P. Blanchet, 1995: Develop-ment of the New CCC/GCM longwave radiationmodel for extension into the middle atmosphere.Atmos. Ocean, 33, 513–531.

Fomichev, V. I., J.-P. Blanchet, and D. S. Turner,1998: Matrix parameterization of the 15 µm CO2

band cooling in the middle and upper atmospherefor variable CO2 concentration. J. Geophys. Res.,103, 11505–11528.

Fomichev, V. I., C. Fu, J. de Grandpre, S. R.Beagley, V. P. Ogibalov, and J. C. McConnell,2004b: Model thermal response to minor ra-diative energy sources and sinks in the mid-dle atmosphere. J. Geophys. Res., 109, D19107,doi:10.1029/2004JD004892.

Fomichev, V. I., V. P. Ogibalov, and S. R. Beagley,2004a: Solar heating by the near-IR bands in themesosphere. Geophys. Res. Lett., 31, L21102,doi:10.1029/2004GL020324.

Fomichev, V. I., W. E. Ward, S. R. Beagley,C. McLandress, J. C. McConnell, N. A. McFarlane,

19

and T. G. Shepherd, 2002: The extended Cana-dian middle atmosphere model: Zonal mean cli-matology and physical parameterizations. J. Geo-phys. Res., 107, doi:10.1029/2001JD000479.

Fouquart, Y. and B. Bonnel, 1980: Computations ofsolar heating of the earth’s atmosphere: A newparameterization. Contr. Atm. Phy., 53, 35–62.

Garcia, R. R., 2000: The role of equatorialwaves in the semiannual oscillation of the mid-dle atmosphere. Atmospheric science across thestratopause, D. E. Siskind, S. D. Eckermann, andM. E. Summers, eds., AGU, Washington, DC,number 123 in Geophys. Monogr. Ser., 161–176.

Garcia, R. R., T. J. Dunkerton, R. S. Lieberman, andR. A. Vincent, 1997: Climatology of the semian-nual oscillation of the tropical middle atmosphere.J. Geophys. Res., 102, 26019–26032.

Gettelman, A., D. E. Kinnison, T. J. Dunkerton, andG. P. Brasseur, 2004: The impact of monsooncirculations on the upper troposphere and lowerstratosphere. J. Geophys. Res., in press.

Gill, A. E., 1982: Atmosphere-ocean dynamics. Aca-demic Press, New York, 662 pp.

Giorgetta, M. A., E. Manzini, and E. Roeckner, 2002:Forcing of the quasi-biennial oscillation from abroad spectrum of atmospheric waves. Geophys.Res. Lett., 29, doi:10.1029/2002GL014756.

Giorgetta, M. A., E. Manzini, E. Roeckner, M. Esch,and L. Bengtsson, 2005: Climatology and Forcingof the QBO in MAECHAM5. J. Climate, this issue.

Gruzdev, A. N. and G. P. Brasseur, 2004: Long-termchanges in the mesosphere calculated by a two-dimensional model. J. Geophys. Res., in press.

Hamilton, K., R. J. Wilson, J. D. Mahlman, and L. J.Umscheid, 1995: Climatology of the SKYHI tropo-sphere stratosphere mesosphere general circula-tion model. J. Atmos. Sci., 52, 5–43.

Heaps, M. G., 1978: A parameterization of cosmicray ionization. Planet. Space. Sci., 26, 513.

Hedin, A., 1991: Extension of the MSIS thermo-sphere model into the middle and lower atmo-sphere. J. Geophys. Res., 96, 1159–1172.

Hines, C. O., 1997a: Doppler-spread parameteriza-tion of gravity wave momentum deposition in themiddle atmosphere. Part 1: Basic formulation. J.Atm. Sol. Terr. Phys., 59, 371–386.

— 1997b: Doppler-spread parameterization of grav-ity wave momentum deposition in the middle at-mosphere. Part 2: Broad and quasi monochro-matic spectra, and implementation. J. Atm. Sol.Terr. Phys., 59, 387–400.

Holton, J. R. and M. J. Alexander, 2000: The roleof waves in the transport circulation of the mid-dle atmosphere. Atmospheric science across thestratopause, D. E. Siskind, S. D. Eckermann, andM. E. Summers, eds., AGU, Washington, DC,number 123 in Geophys. Monogr. Ser., 21–36.

Hong, S. S. and R. S. Lindzen, 1976: Solar semidi-urnal tide in the thermosphere. J. Atmos. Sci., 33,135–153.

Horowitz et al., L. W., 2003: A global simu-lation of tropospheric ozone and related trac-ers: Description and evaluation of MOZART,version 2. J. Geophys. Res., 108, 4784,doi:10.1029/2002JD002853.

Huang, T., S. Walters, G. Brasseur, D. Hauglustaine,W. Wu, S. Chabrillat, X. Tie, C. Granier, A. Smith,S. Madronich, and G. Kockarts, 1998: Descrip-tion of SOCRATES - A chemical dynamical ra-diative two-dimensional model. Technical ReportNCAR/TN-440+EDD, NCAR, Boulder, Co.

Huang, T. Y. W. and G. P. Brasseur, 1993: Effect oflong-term solar variability in a two-dimensional in-teractive model of the middle atmosphere. J. Geo-phys. Res., 98, 20412–20427.

Jonsson, A. I., J. de Grandpre, V. I. Fomichev,J. C. McConnell, and S. R. Beagley, 2004: Dou-bled CO2-induced cooling in the middle atmo-sphere: Photochemical analysis of the ozone ra-diative feedback. J. Geophys. Res., in press.

20

Kaufmann, M., O. A. Gusev, K. U. Grossmann,F. J. Martin-Torres, D. R. Marsh, and A. A.Kutepov, 2003: Satellite observations of day-time and nighttime ozone in the mesosphere andlower thermosphere. J. Geophys. Res., 108, 4272,doi:10.1029/2002JD002800.

Khosravi, R., G. Brasseur, A. Smith, D. Rusch,S. Walters, S. Chabrillat, and G. Kockarts, 2002:Response of the mesosphere to human-inducedperturbations and solar variability calculated bya 2-d model. J. Geophys. Res., 107, 4358,doi:10.1029/2001JD001235.

Kinnison, D. E., G. P. Brasseur, S. Walters, R. R.Garcia, F. Sassi, B. Boville, D. Marsh, L. Harvey,C. Randall, L. Emmons, R. W., and L. Pan, 2004:Sensitivity of Chemical Tracers to MeteorologicalParameters in the MOZART-3 Chemical TransportModel. J. Geophys. Res., in prep.

Koppers, G. A. A. and D. P. Murtagh, 1996: Modelstudies of the influence of O2 photodissociationparameterizations in the Schumann-Runge bandson ozone related photolysis in the upper atmo-sphere. Ann. Geophys., 14, 68–79.

Labitzke, K., 2003: The global signal of the 11-yearsunspot cycle in the atmosphere: When do weneed the QBO? Meteorol. Z., 12, 209–216.

Lean, J., J. Rottman, G. J. Kyle, H. L. Woods, T. N.Hickey, and J. R. Pugga, 1997: Detection andparameterization of variations in solar mid andnear-ultraviolet radiation (200-400 nm). J. Geo-phys. Res., 102, 29939–29956.

Lefevre, F., G. Brasseur, I. Folkins, and A. K.Smith, 1994: Stratospheric chlorine monoxide andozone: Three-dimensional model simulations. J.Geophys. Res., 99, 8183–8195.

Lieberman, R. S. and P. B. Hays, 1994: An esti-mate of the momentum deposition in the lowerthermosphere by the observed diurnal tide. J. At-mos. Sci., 51, 3094–3105.

Lin, S.-J. and R. B. Rood, 1996: Multidimensionalflux-form semi-lagrangian transport schemes.Mon. Wea. Rev., 124, 2046–2070.

Lott, F. and M. J. Miller, 1997: A new subgrid-scaleorographic drag parameterization: Its formulationand testing. Quart. J. Roy. Meteor. Soc., 123, 101–127.

Madronich, S. and S. Flocke, 1998: The role of solarradiation in atmospheric chemistry. Handbook ofEnvironmental Chemistry , P. Boule, ed., Springer,Heidelberg, Germany, 1–26.

Manzini, E., M. A. Giorgetta, M. Esch, L. Kornblueh,and E. Roeckner, 2005: Sensitivity of the North-ern Winter Stratosphere to Sea Surface Temper-ature Variations: Ensemble Simulations with theMAECHAM5 Model. J. Climate, this issue.

Manzini, E., N. A. McFarlane, and C. McLandress,1997: Impact of the Doppler spread parameteri-zation on the simulation of the middle atmospherecirculation using the MA/ECHAM4 general circula-tion model. J. Geophys. Res., 102, 25751–25762.

Marsh, D., A. Smith, G. Brasseur, M. Kaufmann,and K. Grossmann, 2001: The existence of atertiary ozone maximum in the high-latitude mid-dle mesosphere. Geophys. Res. Lett., 28, 4531–4534, doi:10.1029/2001GL013791.

Marsh, D., A. Smith, and E. Noble, 2003:Mesospheric ozone response to changes inwater vapor. J. Geophys. Res., 108, 4109,doi:10.1029/2002JD002705.

Marsh, D. R., S. C. Solomon, and A. E. Reynolds,2004: Empirical model of nitric oxide in the lowerthermosphere. J. Geophys. Res., 109, A07301,doi:10.1029/2003JA010199.

Marsland, S. J., H. Haak, J. H. Jungclaus, M. Latif,and F. Roske, 2003: The Max-Planck-Instituteglobal ocean/sea ice model with orthogonal curvi-linear coordinates. Ocean Modelling, 5, 91–127.

McEwan, M. J. and L. F. Phillips, 1975: Chemistry ofthe atmosphere. Edward Arnold Ltd., London, UK,301 pp.

Mertens, C. J., M. G. Mlynczak, M. Lopez-Puertas,P. P. Wintersteiner, R. H. Picard, J. R. Winick,

21

L. L. Gordley, and J. M. Russell III, 2001: Re-trieval of mesospheric and lower thermospheric ki-netic temperature from measurements of CO2 15µm earth limb emission under non-LTE conditions.Geophys. Res. Lett., 28, 1391–1394.

Mertens, C. J., F. J. Schmidlin, R. A. Goldberg,E. E. Remsberg, W. D. Pesnell, J. M. Russell III,M. G. Mlynczak, M. Lopez-Puertas, P. P. Winter-steiner, R. H. Picard, J. R. Winick, and L. L. Gord-ley, 2004: SABER observations of mesospherictemperatures and comparisons with falling spheremeasurements taken during the 2002 summerMaCWAVE campaign. Geophys. Res. Lett., 31,L03105, doi:10.1029/2003GL018605.

Minschwaner, K. and D. E. Siskind, 1993: A newcalculation of NO photolysis in the stratosphere,mesosphere, and lower thermosphere. J. Geo-phys. Res., 98, 20401–20412.

Miyahara, S., Y. Yoshida, and Y. Miyoshi, 1993: Dy-namic coupling between the lower and upper at-mosphere by tides and gravity waves. J. Atmos.Terr. Phys., 55, 1039–1053.

Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Ia-cono, and S. J. Clough, 1997: Radiative transferfor inhomogeneous atmospheres: RRTM, a vali-dated correlated-k model for the longwave. J. Geo-phys. Res., 102, 16663–16682.

Mlynczak, M. G., 2000: A contemporary assesse-ment of the mesospheric energy budget. Atmo-spheric science across the stratopause, D. E.Siskind, S. D. Eckermann, and M. E. Summers,eds., AGU, Washington, DC, number 123 in Geo-phys. Monogr. Ser., 37–52.

Mlynczak, M. G. and S. Solomon, 1993: A detailedevaluation of the heating efficiency in the middleatmosphere. J. Geophys. Res., 98, 10517–10541.

Mote, P. W., K. H. Rosenlof, M. E. McIntyre, E. S.Carr, J. C. Gille, J. R. Holton, J. S. Kinnersley,H. C. Pumphrey, J. M. Russell III, and J. W. Wa-ters, 1996: An atmospheric tape recorder: Theimprint of tropical tropopause temperatures on

stratospheric water vapor. J. Geophys. Res., 101,3989–4006.

Ogibalov, V. P. and V. I. Fomichev, 2003: Parameteri-zation of solar heating by the near IR CO2 bands inthe mesosphere. Adv. Space Res., 32, 759–764.

Portmann, R. W., G. E. Thomas, S. Solomon, andR. R. Garcia, 1995: The importance of dynami-cal feedbacks on doubled CO2-induced changesin the thermal structure of the mesosphere. Geo-phys. Res. Lett., 13, 1733–1736.

Randel., W., F. Wu, J. M. Russell III, A. Roche, andJ. W. Waters, 1998: Seasonal cycles and QBOvariations in stratospheric CH4 and H2O observedin UARS HALOE data. J. Atmos. Sci., 55, 163–185.

Rasch, P. J., B. A. Boville, and G. P. Brasseur,1995: A three-dimensional general circulationmodel with coupled chemistry for the middle at-mosphere. J. Geophys. Res., 100, 9041–9071.

Richards, P. G., J. A. Fennelly, and D. G. Torr, 1994:A solar EUV flux model for aeronomic calcula-tions. J. Geophys. Res., 99, 8981–8992, (Correc-tion, JGR, 99, 13283, 1994).

Rind, D., R. Suozzo, N. K. Balachandran, and M. J.Prather, 1990: Climate change and the middle at-mosphere. Part I: The doubled CO2 climate. J. At-mos. Sci., 47, 475–494.

Roble, R. G., 1995: Energetics of the mesosphereand thermosphere. The upper mesosphere andlower thermosphere: A review of experiment andtheory , R. M. Johnson and T. L. Killeen, eds.,AGU, Washington, DC, number 87 in Geophys.Monogr. Ser., 1–21.

— 2000: On the feasability of developing a globalatmospheric model extending from the ground tothe exosphere. Atmospheric science across thestratopause, D. E. Siskind, S. D. Eckermann, andM. E. Summers, eds., AGU, Washington, DC,number 123 in Geophys. Monogr. Ser., 53–68.

22

Roble, R. G. and R. E. Dickinson, 1989: How willchanges in carbon dioxide and methane modifythe mean structure of the mesosphere and ther-mosphere? Geophys. Res. Lett., 16, 1441–1444.

Roeckner, E., R. Brokopf, M. Esch, M. Gior-getta, S. Hagemann, L. Kornblueh, E. Manzini,U. Schlese, and U. Schulzweida, 2005a: Sensitiv-ity of simulated climate to horizontal and verticalresolution in the ECHAM5 atmosphere model. J.Climate, this issue.

— 2005b: The atmospheric general circulationmodel ECHAM 5. Part II: Sensitivity of simulatedclimate to horizontal and vertical resolution. Tech-nical Report 354, MPI for Meteorology, Hamburg,Germany.

Roeckner, E., G. Buml, L. Bonaventura, R. Brokopf,M. Esch, M. Giorgetta, S. Hagemann, I. Kirchner,L. Kornblueh, E. Manzini, A. Rhodin, U. Schlese,U. Schulzweida, and T. A., 2003: The atmosphericgeneral circulation model ECHAM 5. Part I: Modeldescription,. Technical Report 349, MPI for Mete-orology, Hamburg, Germany.

Rose, K. and G. Brasseur, 1989: A three-dimensional model of chemically active tracespecies in the middle atmosphere during disturbedwinter conditions. J. Geophys. Res., 94, 16387–16403.

Rozanov, E. V., M. E. Schlesinger, T. A. Egorova,B. Li, N. Andronova, and V. A. Zubov, 2004: At-mospheric response to the observed increase ofsolar UV radiation from solar minimum to solarmaximum simulated by the University of Illinoisat Urbana-Champaign climate-chemistry model. J.Geophys. Res., 109, doi:10.1029/2003JD003796.

Sassi, F., R. R. Garcia, B. A. Boville, and H. Liu,2002: On temperature inversions and the meso-spheric surf zone. J. Geophys. Res., 107, 4380,doi:10.1029/2001JD001525.

Simmons et al., A. J., 1989: The ECMWF medium-range prediction model: Development of the nu-merical formulations and the impact of increasedresolution. Meteorol. Atmos. Phys., 40, 28–60.

Smith, A. K., 2004: Physics and chemistry of themesopause region. J. Atm. Sol. Terr. Phys., 66,839–857.

Sonnemann, G., C. Kremp, A. Ebel, and U. Berger,1998: A three-dimensional dynamic model of theminor constituents of the mesosphere. Atm. Envi-ron., 32, 3157–3172.

Steil, B., C. Bruhl, E. Manzini, P. J. Crutzen,J. Lelieveld, P. J. Rasch, E. Roeckner, andK. Kruger, 2003: A new interactive chemistry-climate model: 1. Present-day climatologyand interannual variability of the middle at-mosphere using the model and 9 years ofHALOE/UARS data. J. Geophys. Res., 108, 4290,doi:10.1029/2002JD002971.

Summers, M. E., D. F. Strobel, R. M. Bevilacqua,X. Zhu, M. T. DeLand, M. Allen, and G. M. Keating,1990: A model study of the response of meso-spheric ozone to short-term solar ultraviolet fluxvariations. J. Geophys. Res., 95, 22523–22538.

Swinbank, R. and D. A. Ortland, 2003: Com-pilation of wind data for the Upper Atmo-sphere Research Satellite (UARS) reference at-mosphere project. J. Geophys. Res., 108, 4615,doi:10.1029/2002JD003135.

von Zahn, U. and U. Berger, 2003: Persis-tent ice cloud in the midsummer upper meso-sphere at high latitudes: Three-dimensionalmodeling and cloud interactions with ambientwater vapor. J. Geophys. Res., 108, 8451,doi:10.1029/2002JD002409.

Wang, H. J., D. M. Cunnold, and X. Bao, 1996: Acritical analysis of stratospheric aerosol and gasexperiment ozone trends. J. Geophys. Res., 101,12495–12514.

Zhu, X., J.-H. Yee, and E. R. Talaat, 2003: Effectof short-term solar ultraviolet flux variability in acoupled model of photochemistry and dynamics.J. Atmos. Sci., 60, 491–509.

23

____________________________________________________________

_____

__

0 50 100 150 200height [km]

1e-06

1e-05

1e-04

1e-03

1e-02

1e-01

1e+00

1e+01

1e+02

1e+03

pres

sure

[hP

a]

a)

model layers

0.001 0.01 0.1 1 10 100height difference [km]

b)

solar min - solar max

0.01 0.1 1 10height difference [km]

c)

2*CO2 - 1*CO2

Figure 1: The annual mean global average relation of pressure and geometrical height in the model. a)Simulated in the solar minimum run. Horizontal dashes indicate the centers of the 67 model layers. b)Simulated height change at given pressure levels for the solar maximum run with respect to solar minimum.c) Simulated height change for the “2*CO2” run with respect to “1*CO2”. The dashed line marks negativevalues.

24

120 160 200 240 280wavelength [nm]

0

10

20

30

40

50

60

irrad

ianc

e ch

ange

[%]

Figure 2: Difference in the solar irradiance input between simulations for solar maximum and minimumwith respect to solar minimum in percent. The step width indicates the wavelength resolution used in thecomputation of photodissociation and heating rates. Changes applied to wavelengths longer than 300 nmare in general smaller than 1%.

25

Figure 3: Observed and simulated July temperatures [K]. Left: Mean values for 1600 LT fromSABER/TIMED observations between 1 and 15 July (mean of 2002 and 2005). Right: Simulated Julymean values for 1600 LT. The dashed box marks the region of available observations.

26

Figure 4: Analyzed and simulated July zonal mean zonal winds [m/s]. Left: URAP analysis which is com-posed of UKMO analyses below the horizontal dashed line and UARS observations within the two dashedboxes in the upper part. Right: Simulated values.

27

Figure 5: Annual cycle of analyzed (left, URAP, see Fig. 4) and simulated (right) monthly and zonal meanzonal winds [m/s] for pressure levels of about 1 (bottom) and 0.001 (top) hPa. In the regions poleward ofthe dashed lines and in the months of May and June at 0.001 hPa the availability of observations is poorsuch that the URAP climatology is derived mainly by interpolation.

28

Figure 6: Observed and simulated July daytime ozone mixing ratios [ppmv]. Left: Mean values for 1500local time (LT) from SABER observations between 1 and 15 July (mean of 2002 and 2005) above 1 hPa,and zonal mean July values from URAP below. Right: Simulated July mean values for 1500 LT. The dashedboxes mark the regions of available observations

29

Figure 7: Observed and simulated July zonal mean mixing ratios [ppmv] of water vapor. Left: Observationsfrom URAP. Right: Simulated values.

30

Figure 8: Simulated deviation of the equatorial water vapor mixing ratio from the 10 year mean values forthe respective altitudes. Mixing ratios are monthly and zonal mean values for 10 years of the solar minimumsimulation. The isoline indicates the zero deviation.

31

Figure 9: Observed and simulated July zonal mean mixing ratios [ppmv] of methane. Left: Observationsfrom URAP. Right: Simulated values.

32

Figure 10: Annual mean global average values for the major components of the atmospheric energy budgetin [K/day]. Solid: solar heating including chemical heating. Dashed: Infrared cooling/heating. Dotted:Molecular heat conduction. Left: Solar minimum simulation. Right: Changes for solar maximum withrespect to the solar minimum simulation.

33

Figure 11: Zonal mean values of temperature [K], zonal wind [m/s] and the residual vertical wind [cm/s] assimulated in the solar minimum run for July, and the changes for solar maximum with respect to the solarminimum run in the respective units. Statistical significance of the changes that is larger than 90% (99%) isindicated by light (dark) gray shading.

34

Figure 12: Zonal mean July volume mixing ratios for O3, O, NO, OH, H2O, and CH4 as simulated in the solarminimum run, and the relative changes [%] for solar maximum with respect to solar minimum. Statisticalsignificance of the changes that is larger than 90% (99%) is indicated by light (dark) gray shading.

35

Figure 13: July mean ozone volume mixing ratios at an altitude of 2*10−3 hPa (∼ 90 km) for given latitudesand local times. Left: Solar minimum simulation. Right: Relative changes [%] for solar maximum withrespect to solar minimum. Statistical significance of the changes that is larger than 90% (99%) is indicatedby light (dark) gray shading.

36

Figure 14: Changes in the annual mean global average values for the major components of the atmosphericenergy budget in [K/day] for the “2*CO2” simulation with respect to “1*CO2”. Solid: solar heating includingchemical heating. Dashed: Infrared cooling/heating. Dotted: Molecular heat conduction.

37

Figure 15: Changes of July zonal mean temperature [K], zonal wind [m/s] and the residual vertical wind[cm/s] for the “2*CO2” simulation with respect to “1*CO2”. Statistical significance of the changes that islarger than 90% (99%) is indicated by light (dark) gray shading.

38

Figure 16: Changes of July zonal mean temperature [K], zonal wind [m/s] and the residual vertical wind[cm/s] for the “SST(2*CO2)” simulation with respect to “1*CO2”. Statistical significance of the changes thatis larger than 90% (99%) is indicated by light (dark) gray shading.

39

Figure 17: Changes of July zonal mean temperature [K], zonal wind [m/s] and the residual vertical wind[cm/s] for the “2*CO2 (no SST)” simulation with respect to “1*CO2” (cf. Fig. 11). Statistical significance ofthe changes that is larger than 90% (99%) is indicated by light (dark) gray shading.

40

Figure 18: Changes of January zonal mean temperature [K], zonal wind [m/s] and the residual vertical wind[cm/s] for the “SST(2*CO2)” simulation with respect to “1*CO2”. Statistical significance of the changes thatis larger than 90% (99%) is indicated by light (dark) gray shading.

Figure 19: Zonal mean July volume mixing ratios for CO2 and CO as simulated in the “1*CO2” run, and therelative changes [%] for the “2*CO2” simulation with respect to “1*CO2”. The statistical significance of thechanges is everywhere larger than 99%.

42

Figure 20: Relative changes [%] of July zonal mean volume mixing ratios for O3, O, NO, OH, H2O, and CH4

for the “2*CO2” simulation with respect to “1*CO2”. Statistical significance of the changes that is larger than90% (99%) is indicated by light (dark) gray shading.

43

Figure 21: Changes of July zonal mean temperature [K] and ozone [%] for the solar maximum simulationwith respect to solar minimum (upper panels) and for the “2*CO2” simulation with respect to “1*CO2” (lowerpanels). Please note: In contrast to all previous figures the changes are shown here with the geometricalheight as vertical coordinate. For the comparisons of the same fields in pressure coordinates see Figs. 11,12, 15, and 20.

44

Table 1: Photolyses included in the modelReaction

( 1) O2 + hν → O + O1D( 2) O2 + hν → 2*O( 3) O3 + hν → O1D + O2

( 4) O3 + hν → O + O2

( 5) N2O + hν → O1D + N2

( 6) NO + hν → N + O( 7) NO2 + hν → NO + O( 8) N2O5 + hν → NO2 + NO3

( 9) N2O5 + hν → NO + O + NO3

( 10) HNO3 + hν → NO2 + OH( 11) NO3 + hν → NO2 + O( 12) NO3 + hν → NO + O2

( 13) HO2NO2 + hν → OH + NO3

( 14) HO2NO2 + hν → NO2 + HO2

( 15) CH3OOH + hν → CH2O + H + OH( 16) CH2O + hν → CO + 2*H( 17) CH2O + hν → CO + H2

( 18) H2O + hν → OH + H( 19) H2O + hν → H2 + O1D( 20) H2O + hν → 2H + O( 21) H2O2 + hν → 2*OH( 22) Cl2 + hν → 2*Cl( 23) OClO + hν → O + ClO( 24) Cl2O2 + hν → 2*Cl( 25) HOCl + hν → OH + Cl( 26) HCl + hν → H + Cl( 27) ClONO2 + hν → Cl + NO3

( 28) ClONO2 + hν → ClO + NO2

( 29) BrCl + hν → Br + Cl( 30) BrO + hν → Br + O( 31) HOBr + hν → Br + OH( 32) BrONO2 + hν → Br + NO3

( 33) BrONO2 + hν → BrO + NO2

( 34) CH3Cl + hν → Cl + CH3O2

( 35) CCl4 + hν → 4*Cl( 36) CH3CCl3 + hν → 3*Cl( 37) CF2Cl3 + hν → 3*Cl( 38) CF22Cl2 + hν → 2*Cl( 39) CF2C113 + hν → 3*Cl( 40) HCF2C22 + hν → Cl( 41) CH3Br + hν → Br + CH3O2

( 42) CF23Br + hν → Br( 43) CF22ClBr + hν → Br + Cl( 44) CO2 + hν → CO + O( 45) CH4 + hν → H + CH3O2

( 46) CH4 + hν → H2 + .18*CH2O + .18*O + .66*OH + .44*CO2 + .44*H245

Table 2: Gas phase chemical reactions included in the model and corresponding reaction ratesReaction Rate coefficient

( 1) O + O2 + M → O3 + M 6.e-34 * (300/T)2.4

( 2) O + O3 → 2*O2 8.00e-12*exp( -2060/T)( 3) O + O + M → O2 + M 4.23e-28 / T2

( 4) O + H2 → OH + H 7.00e-11*exp( -5130/T)( 5) O1D + N2 → O + N2 1.80e-11*exp( 110/T)( 6) O1D + O2 → O + O2 3.20e-11*exp( 70/T)( 7) O1D + H2O → 2*OH 2.20e-10( 8) O1D + N2O → 2*NO 6.70e-11

( 10) O1D + O3 → O2 + O2 1.20e-10( 11) O1D + CF22Cl2 → 2*Cl 1.20e-10( 12) O1D + CF2C113 → 3*Cl 1.50e-10( 13) O1D + HCF2C22 → Cl 7.20e-11( 14) O1D + CH4 → CH3O2 + OH 1.13e-10( 15) O1D + CH4 → CH2O + H + HO2 3.00e-11( 16) O1D + CH4 → CH2O + H2 7.50e-12( 17) O1D + H2 → H + OH 1.10e-10( 18) O1D + HCl → Cl + OH 1.50e-10( 19) N2D + O2 → NO + O 5.00e-12( 20) N2D + O → N + O 4.50e-13( 21) N + O2 → NO + O 1.50e-11*exp( -3600/T)( 22) N + NO → N2 + O 2.10e-11*exp( 100/T)( 23) NO + O + M → NO2 + M troe : ko=9.00e-32*(300/T)1.5

ki=3.00e-11( 24) NO + HO2 → NO2 + OH 3.50e-12*exp( 250/T)( 25) NO + O3 → NO2 + O2 3.00e-12*exp( -1500/T)( 26) NO2 + O → NO + O2 5.60e-12*exp( 180/T)( 27) NO2 + O + M → NO3 + M troe : ko=9.00e-32*(300/T)2

ki=2.20e-11( 28) NO2 + O3 → NO3 + O2 1.20e-13*exp( -2450/T)( 29) NO2 + NO3 + M → N2O5 + M troe : ko=2.00e-30*(300/T)4.4

ki=1.40e-12*(300/T)0.7

( 30) N2O5 + M → NO2 + NO3 + M k29*3.333e26 * exp(-10991/T)( 31) NO2 + OH + M → HNO3 + M troe : ko=2.40e-30*(300/T)3.1

ki=1.70e-11*(300/T)2.1

( 32) HNO3 + OH → NO3 + H2O aux2+2.4e-14*exp( 460./T)aux1=6.5e-34 * exp( 1335/T) * amaux2=aux1/(1+aux1*(2.7e-17*exp( 2199/T))−1)

( 33) NO3 + NO → 2*NO2 1.50e-11*exp( 170/T)( 34) NO3 + O → NO2 + O2 1.00e-11( 35) NO3 + OH → HO2 + NO2 2.20e-11( 36) NO3 + HO2 → OH + NO2 + O2 3.50e-12( 37) NO2 + HO2 + M → HO2NO2 + M troe : ko=1.80e-31*(300/T)3.2

ki=4.70e-12*(300/T)1.4

( 38) HO2NO2 + OH → H2O + NO2 + O2 1.30e-12*exp( 380/T)( 39) HO2NO2 + M → HO2 + NO2 + M k37* exp( -10900/T ) /2.1e-27( 40) CH4 + OH → CH3O2 + H2O 2.45e-12*exp( -1775/T)( 41) CH3O2 + NO → CH2O + NO2 + HO2 4.20e-12*exp( 180/T)( 42) CH3O2 + HO2 → CH3OOH + O2 3.80e-13*exp( 800/T)( 43) CH3OOH + OH → CH3O2 + H2O 3.80e-12*exp( 200/T)( 44) CH2O + NO3 → CO + HO2 + HNO3 6.00e-13*exp( -2058/T)( 45) CH2O + OH → CO + H2O + H 1.00e-11

46

Table 2: (continued)Reaction Rate coefficient

( 46) CH2O + O → HO2 + OH + CO 3.40e-11*exp( -1600/T)( 47) CO + OH → CO2 + H 1.5e-13 * (1. + 6.e-7*kB*am*T)( 48) H + O2 + M → HO2 + M troe : ko=5.70e-32*(300/T)1.6

ki=7.50e-11( 49) H + O3 → OH + O2 1.40e-10*exp( -470/T)( 50) H + HO2 → 2*OH 7.05e-11( 51) H + HO2 → H2 + O2 7.29e-12( 52) H + HO2 → H2O + O 1.62e-12( 53) OH + O → H + O2 2.20e-11*exp( 120/T)( 54) OH + O3 → HO2 + O2 1.50e-12*exp( -880/T)( 55) OH + HO2 → H2O + O2 4.80e-11*exp( 250/T)( 56) OH + OH → H2O + O 4.20e-12*exp( -240/T)( 57) OH + OH + M → H2O2 + M troe : ko=6.20e-31*(300/T)

ki=2.60e-11( 58) OH + H2 → H2O + H 5.50e-12*exp( -2000/T)( 59) OH + H2O2 → H2O + HO2 2.90e-12*exp( -160/T)( 60) HO2 + O → OH + O2 3.00e-11*exp( 200/T)( 61) HO2 + O3 → OH + 2*O2 2.00e-14*exp( -680/T)( 62) HO2 + HO2 → H2O2 + O2 (2.3e-13 * exp( 600/T) + aux1 ) * aux2

aux1=1.7e-33 * am * exp( 1000/T)aux2=1. + 1.4e-21 * am * H2O * exp( 2200/T)

( 63) H2O2 + O → OH + HO2 1.40e-12*exp( -2000/T)( 64) Cl + O3 → ClO + O2 2.30e-11*exp( -200/T)( 65) Cl + H2 → HCl + H 3.70e-11*exp( -2300/T)( 66) Cl + H2O2 → HCl + HO2 1.10e-11*exp( -980/T)( 67) Cl + HO2 → HCl + O2 1.80e-11*exp( 170/T)( 68) Cl + HO2 → OH + ClO 4.10e-11*exp( -450/T)( 69) Cl + CH2O → HCl + HO2 + CO 8.10e-11*exp( -30/T)( 70) Cl + CH4 → CH3O2 + HCl 9.60e-12*exp( -1360/T)( 71) ClO + O → Cl + O2 3.00e-11*exp( 70/T)( 72) ClO + OH → Cl + HO2 7.40e-12*exp( 270/T)( 73) ClO + OH → HCl + O2 3.20e-13*exp( 320/T)( 74) ClO + HO2 → O2 + HOCl 4.80e-13*exp( 700/T)( 75) ClO + NO → NO2 + Cl 6.40e-12*exp( 290/T)( 76) ClO + NO2 + M → ClONO2 + M troe : ko=1.80e-31*(300/T)3.4

ki=1.50e-11*(300/T)1.9

( 77) ClO + ClO → 2*Cl + O2 3.00e-11*exp( -2450/T)( 78) ClO + ClO → Cl2 + O2 1.00e-12*exp( -1590/T)( 79) ClO + ClO → Cl + OClO 3.50e-13*exp( -1370/T)( 80) ClO + ClO + M → Cl2O2 + M troe : ko=2.20e-32*(300/T)3.1

ki=3.50e-12*(300/T)( 81) Cl2O2 + M → ClO + ClO + M k80/(1.3e-27 * exp( 8744/T))( 82) HCl + OH → H2O + Cl 2.60e-12*exp( -350/T)( 83) HCl + O → Cl + OH 1.00e-11*exp( -3300/T)( 84) HOCl + O → ClO + OH 1.70e-13( 85) HOCl + Cl → HCl + ClO 2.50e-12*exp( -130/T)( 86) HOCl + OH → H2O + ClO 3.00e-12*exp( -500/T)( 87) ClONO2 + O → ClO + NO3 2.90e-12*exp( -800/T)( 88) ClONO2 + OH → HOCl + NO3 1.20e-12*exp( -333/T)( 89) ClONO2 + Cl → Cl2 + NO3 6.50e-12*exp( 135/T)( 90) Br + O3 → BrO + O2 1.70e-11*exp( -800/T)

47

Table 2: (continued)Reaction Rate coefficient

( 91) Br + HO2 → HBr + O2 1.50e-11*exp( -600/T)( 92) Br + CH2O → HBr + HO2 + CO 1.70e-11*exp( -800/T)( 93) BrO + O → Br + O2 1.90e-11*exp( 230/T)( 94) BrO + OH → Br + HO2 7.50e-11( 95) BrO + HO2 → HOBr + O2 3.40e-12*exp( 540/T)( 96) BrO + NO → Br + NO2 8.80e-12*exp( 260/T)( 97) BrO + NO2 + M → BrONO2 + M troe : ko=5.20e-31*(300/T)3.2

ki=6.90e-12*(300/T)2.9

( 98) BrO + ClO → Br + OClO 9.50e-13*exp( 550/T)( 99) BrO + ClO → Br + Cl + O2 2.30e-12*exp( 260/T)(100) BrO + ClO → BrCl + O2 4.10e-13*exp( 290/T)(101) BrO + BrO → 2*Br + O2 1.50e-12*exp( 230/T)(102) HBr + OH → Br + H2O 1.10e-11(103) CH3Cl + Cl → HO2 + CO + 2*HCl 3.20e-11*exp( -1250/T)(104) CH3Cl + OH → Cl + H2O + HO2 4.00e-12*exp( -1400/T)(105) CH3CCl3 + OH → H2O + 3.0*Cl 1.80e-12*exp( -1550/T)(106) HCF2C22 + OH → Cl + H2O + CF22O 1.00e-12*exp( -1600/T)(107) CH3Br + OH → Br + H2O + HO2 4.00e-12*exp( -1470/T)

Read 6.e-34 as 6. * 10−34; am: air density (number per cm3); rate coefficients are expressed in cm3/s forbimolecular and cm6/s for termolecular reactions. troe: Rate coefficients for Troe-reactions are given ask = ko

1+(ko∗am)/ki

∗ 0.6(1+(log(ko∗am/ki))2)−1

.

48