The role of cumulus parameterisation in global and regional sulphur transport

НазваниеThe role of cumulus parameterisation in global and regional sulphur transport
Размер65.7 Kb.

Trond Iversen and Øyvind Seland1


In connection with regional acidification studies and investigation of Arctic haze and hemispheric-scale transport, limited-area chemistry-transport models (CTMs) on limited domains horizontally as well as vertically have been used. More recently, global models are being used for oxidized sulphur components for the purpose of calculating of possible impacts of sulphate particles on climate. To some surprise sulphur modeling has proven more difficult when integrated in global circulation models (GCMs) than experience from limited-area models, and to some extent by global CTMs. In particular the vertical distribution appears to be wrong. This also shows up as a considerable mismatch between ground-level measurements and calculations in source-regions. The inter-comparison exercise by Barrie et al. (2001) emphasized this problem, and it was confirmed in experiments with a new scheme by Iversen and Seland (2002). That paper presented results from using an extended version of the NCAR CCM3 atmospheric GCM to calculate sulphate and black carbon (BC). Also the result of the NCAR-group’s own sulphur model produced these biases (Rasch et al., 2000).

Sensitivity tests by Iversen and Seland (2002) revealed that too efficient vertical exchange in deep cumulus clouds in combination with underestimated scavenging by convective precipitation, can explain large portions of these error biases. In that paper some rather arbitrary and radical assumptions were made to avoid the biases. Here we test some more physically sound methods to account for processes involving deep convection. These methods still leave considerable errors, but there are other physico-chemical processes that partly may account for those. Examples include clouds in the boundary layer influencing the efficient oxidation rate and deposition of sulphate. Here we focus on the vertical transport and scavenging in connection with deep convection only, and we use the NCAR CCM3 model as before.


Processes inside and nearby deep cumulus clouds that may influence aerosols are precipitation scavenging, vertical transport in updrafts and downdrafts, aqueous phase oxidation of precursors, and coagulation processes between sub-saturated particles and cloud droplets. The efficiency of the two latter processes depends on availability of oxidizing agents, the relative motion of droplets and interstitial particles, and the effective fraction of time air parcels are in cloudy versus clear air. We do not see an immediate way to improve the handling of these processes in present climate models. For the two first mentioned processes we have some thoughts on how at least parts of the weaknesses can be accounted for.

2.1 Scavenging by cumulus precipitation

Precipitation scavenging by cumulonimbi clouds is generally more vigorous than for stratiform precipitation. Over typical a model time-step (15-20 min.), lower level air parcels may be sucked into cumulonimbi and detrained in the upper free troposphere. In this process contaminants will be exposed to precipitation inside and below the cloud. In-cloud scavenging is the generally much more efficient for aerosol particles. In-cloud processes involve saturation and growth of much of the hygroscopic particles. In particular for cumulus clouds, the realized super-saturation can be sufficient to turn most hygroscopic particle mass into cloud droplets. These cloud droplets may quickly add to the precipitation release by efficient coalescence in updrafts (auto-conversion). Below-cloud scavenging of particles is entirely determined by impaction of falling precipitation, which normally is inefficient for unsaturated particles in the accumulation mode (Seinfeld and Pandis, 1998, pp. 1020-1026).

In a model grid column where conditions for triggering sub-grid cumulus parameterization are present, an ensemble of cumulus towers of varying depth is assumed. In the NCAR model a version of the scheme developed by Zhang and McFarlane (1995) is used. This scheme produces a vertical profile of precipitation intensity depending on the influxes of water vapor and the strength of the vertical mass-fluxes. The vertical mass fluxes are estimated from the convective available potential energy in the height range of conditional stability, which is the closure assumption for the parameterization.

For cumulonimbi we propose to apply in-cloud scavenging efficiencies in all layers below the level of maximum rate of precipitation release. Air in those layers is prone to become cloudy over a significant fraction of the model time-step, and during those time-slots to be exposed to strong turbulent mixing. In the original version of the model, below-cloud scavenging rates are used below the cloud-base and the scavenging efficiency in any layer is multiplied by the minor fraction of cumulus cloudiness.

2.2 Mixing of air between updrafts and downdrafts

The overall effect of cumulus parameterization on the resolved scale is an effective vertical redistribution of quantities over a model time-step. The redistribution is a consequence of the atmosphere being conditionally stable over a layer of air. It takes place in saturated updrafts, in downdrafts caused by falling precipitation, and in the slowly subsiding air between the clouds. Clear air entrains into the up- and downdrafts by mixing and sub-grid convergence, and detrains after dehydration. The scheme of Zhang and McFarlane (1995) assumes an ensemble of clouds-plumes that only detrains air at the level where rising air become negatively saturated buoyant. This may well be an adequate assumption for the water-budget, but it will probably underestimate mixing that exchange other airborne constituents. So far we have no remedy for this.

Possibly more seriously is the lack of exchange between rising air in updrafts and sinking air in downdrafts. These turbulent air masses exist side by side with strong and opposite vertical winds. We assume that the mixing between updrafts and downdrafts are sufficiently efficient so that the updraft contaminant flux is counteracted by the downdraft mass fluxes. This may probably be too efficient, but has the benefit that we do not need to specify an exchange coefficient and solve a diffusion equation.

Consider a model layer no. k, where k=1 for the uppermost and k=K for the lowermost layers. Convective mass fluxes (transport of mass through a horizontal grid square per time unit) are given at interfaces between the layers, and these interface levels are numbered k+1/2 (below) and k-1/2 (above). Mass fluxes are zero for ½ and K+1/2. In a grid-column conditioned for deep moist convection, the parameterization assumes an ensemble of Cu clouds with strong mass-fluxes in the saturated fraction and a slower and exactly compensating subsiding motion in the unsaturated air. Air from the saturated updrafts (u) and downdrafts (d) may detrain into the ambient unsaturated air in layer k with rates dk and dudk respectively. Entrainment rates for ambient air that becomes saturated are euk and edk. If quk-1/2, qdk-1/2, and qk are mixing ratios of a contaminant in the updraft, the downdraft and in the ambient air at the level given with the index, we can write the contaminant budget equation for layer k as follows:

where q is the mass mixing ratio for a contaminant, index t and p means derivative w.r.t. time and pressure, F are vertical contaminant fluxes due to the deep convective motion, and superscripts u, d and a signifies saturated updraft, saturated downdraft, and compensating vertical motion in ambient unsaturated air. Note that the fluxes are positive when directed upwards. Assuming that mass-fluxes satisfy a maximum Courant number of 1, mass consistency for positive definite quantities are secured (P. J. Rasch, NCAR, personal communication) by assuming zero fluxes at k=1/2 and k=K+1/2,

and for the compensating ambient-air currents for k=1,…, K-1:

Contaminant mixing ratios for k=1,…,K in updrafts are determined by closing the flux budget over layer no. k of pressure thickness (flux out + detrainment = flux in + entrainment):

and in downdrafts:

Here we have defined a tracer (a) which is zero in the original model version and 1 in the version which assumes mixing between updrafts and downdrafts.

2.3 Other model adjustments

The basic model version we run is thoroughly described in Iversen and Seland (2002), but with a few adjustments based on results in that paper. Here we only summarize the changes. We no longer keep separate track of accumulation mode SO4 produced by coagulation of nucleation mode particles in dry air. This minor fraction of SO4 particles is merged with the fraction produced by condensation onto pre-existing accumulation mode particles and is denoted SO4(a) (formerly a1 and a2), whilst the portion produced by oxidation in cloud droplets is denoted SO4(aw) (formerly a3). The prognostic components calculated by the model are then: DMS (di-methyl-sulphide); SO2; externally mixed nucleation/Aitken-mode sulphate (SO4(n)) produced in gas phase; internally mixed sulphate produced in gas phase condensed on background accumulation-mode particles or coagulation of nucleation-mode sulphate on the same (SO4(a)); internally mixed sulphate produced by aqueous phase oxidation (SO4(aw)); externally mixed nucleation/Aitken mode black carbon (C(n)); externally mixed and fractal black carbon (Cx(a)); internally mixed black carbon which is either emitted in that phase or produced by coagulation of the externally mixed black carbon with background particles (C(a)).

The below-cloud scavenging efficiencies given in Iversen and Seland (2002) are divided by 10 for accumulation mode particles in closer agreement with Seinfeld and Pandis (1998, pp. 1020-1026): Ebc=0.01 for SO4(aw) and 0.02 for SO4(a), C(a), and Cx(a). For cumulonimbi, the in-cloud scavenging efficiency for SO4(aw) is increased to 1 (from 0.8 in all clouds). Finally, black carbon (BC) emitted from biomass burning is distributed vertically over the three lowermost model-layers in the same way as sulphur from fossil fuel combustion. This is a consequence of deep vertical exchange in rising hot plumes from forest fires (wild or initiated by man).


We have run a set of 4 experiments to demonstrate different aspects of the properties of deep convection on the global distribution of sulphate and black carbon. E1 is the new basic run equivalent to the basic run in Iversen and Seland (2002), except that some parameter values are adjusted as mentioned in the previous paragraph. There is no vertical transport of the contaminants by the cumulus mass fluxes in this test. In E2, vertical transport by the mass fluxes is included with a=0, i.e. the original scheme. In E3 the wet scavenging of cumulonimbus precipitation is enhanced by assuming that the efficient scavenging ratios for soluble gases and particles below the level of maximum rate of convective precipitation generation are those valid for in-cloud conditions. The rational is that air in those levels under influence of deep convection, is saturated over a considerable fraction of the model time-step. In the final experiment, E4, an efficient exchange of air between updrafts and downdrafts is assumed, and in the formula for qu we assume a=1, which reduces the updraft transport of contaminants by the downdraft mass fluxes.

Emissions are taken from IPCC (provided by Dr. J. Penner) and are supposed to be valid for the year 2000. Each experiments is run for 5 years repeating the annual cycle for sea surface temperature. Only results for the three latter years are used for evaluation.


We focus on the results for sulphur but also present some results for black carbon. The space only allows selected results. Table 1 showes global budgets for sulphate and BC calculated for the four experiments and compared to results from a selection of other models. The burden and turnover time for SO2 are quite insensitive to the different ways of treating processes in convective clouds. The reason is that deposition competes with oxidation in determining the fate of SO2. The numbers for sulphate is, on the other hand, very sensitive to the treatment of convective cloud processes. The treatment of vertical transport in deep cumulus clouds and the way wet scavenging by convective precipitation is parameterized, have decisive consequences for the abundance of sulphate. This is also evident by comparing the results from the two papers of Chin and co-workers. The 1996-paper has a statistical treatment of the convective cloud processes, whilst that from 2000 is more direct, as in Rasch et al (2000) and Koch et al (1999). There are other differences, but the influence on the burden and turnover of sulphate is striking.

In Table 2 similar budget numbers are given for black carbon. These budgets split

Table 1. Global budget parameters for the production of airborne particulate sulphate.

SOx SO2 SO4-prod . . SO2 . SO4 SO4 .

source dep. Aq. Gas. burden T source wetdep. burden T

(TgS/a) (%) (%) (%) (TsS) (days) (TgS/a) (%) (TgS) (days)

E1 = no conv. transp. 90.4 43.5 40.9 14.1 0.40 1.6 52.7 79 0.60 4.1

E2 = full conv. transp. 90.4 35.7 45.2 17.3 0.52 2.1 60.0 86 2.40 14.6

E3 = E2 +incr. wet dep. 90.4 43.7 41.4 13.5 0.42 1.7 52.7 93 0.63 4.4

E4 = E3 +decr. conv. transp. 90.4 45.5 40.6 12.4 0.39 1.6 51.0 92 0.44 3.1

Rasch et al.(2000) 81 32a 55a 12 0.4 1.9 55a 93a 0.60 4.0a

Koch et al. (1999) 82.3 43.4a 38.4a 15.9 0.56 2.6 46.6a 80.3a 0.73 5.7a

Chin et al.(2000) 92.5 56.6 26.5 15.1 0.43 1.8 40.7 85.3 0.63 5.8

Chin et al.(1996) 95.6 48.6 43.5 7.8 0.34 1.3 49.1 89 0.53 3.9

Table 2. Global budget numbers for production and turnover of black carbon.

. Hydrophobic BC . Total BC .

source burden T source burden T

(TgC/a) (TgC) (days) (TgC/a) (TgC) (days)

E1 = no conv. transp. 10.5 0.06 2.0 12.4 0.21 6.1

E2 = full conv. transp. 10.5 0.14 4.9 12.4 0.57 16.8

E3 = E2 +incr. wet dep. 10.5 0.09 3.1 12.4 0.22 6.6

E4 = E3 +decr. conv. transp. 10.5 0.08 2.7 12.4 0.18 5.3

Koch (2001) (case S) 12.4 0.04 1.1 12.4 0.15 4.4

Cooke et al.(1999) (fossil fuel) - - - 5.1 0.07 5.3

T is turnover times.

between hydrophobic and hydrophilic BC. In our model, all BC produced by fossil fuel combustion and half of that produced by biomass burning is assumed emitted in hydrophobic mode. The turnover to hydrophilic BC is brought about by coagulation with hydrophilic particles. We have neglected the effect of condensation of hydrophilic material in this connection. The effect of deep convection is profound, and it is crucial to treat scavenging in a consistent way.

Figure 1. Calculated sonally averaged mixing ratios of airborne sulphate over three model years, for the four experiments (E1, E2, E3, and E4).

Figure 2. SO2 (left) and Sulphate (right) measured in a flight campaign in February over Guam (crosses). data from Pacific Exploratory Mission (PEM) (Barth et al., 2000). Calculations averaged over three February months are shown as lines: E1=solid; E2=dashed; E3=dashed-dotted; and E4=dotted line.

Figure 3. Ground-level concentrations calculated sulphate in the experiments E1, E2, and E4 and compared with measurements. Left: Annual sulphate at marine sites in the Atlantic Ocean; Middle: monthly sulphate at Union County, Kentucky, USA.

Figure 1 shows the profound effects of the modeled convective transport on the vertical sulphate distribution. Convective transport in plumes that only detrain at the level of negative buoyancy leads to higher concentrations in the upper troposphere than at the middle levels. Increased scavenging and exchange between updrafts and downdrafts improves the situation. The single comparison with aircraft data in Figure 2 confirms this.

Figure 3 shows that the exaggerated low level tropical concentrations are not fully improved by including convective transport. Increased scavenging is imperative. Results from the North American site indicate that scavenging during summer may then be too efficient at mid-latitudes. The treatment of the effect of convection in and around ITCZ should probably differ from that in local convective storms in the summer extra-tropics.


This work is part of the AerOzClim and RegClim projects financed by the Research Council of Norway. Computational costs are covered by a grant from the Research Council’s Programme for Supercomputing. We are grateful to P. J. Rasch, J. E. Kristjansson, and A. Kirkevåg for co-operation and discussions around the use of the NCAR CCM3 model.


Barth, M. C., Rasch, P. J., Kiehl, J. T., Benkowitz, C. M., and Schwartz, S. E. (2000) Sulfur chemistry in the National Center for Atmospheric Research Community Climate Model: Description, evaluation, features, and sensitivity to aqueous chemistry. J. Geophys. Res., 105, 1387-1415.

Chin, M., Jacob, D. J., Gardner, G. M., Foreman-Fowler, M. S., and Spiro, P. A. (1996) A global three-dimensional model of tropospheric sulfate. J Geophys. Res. 101,18,667-18,690.

Chin, M., Rood, R. B., Lin, S.-J., Müller, J-F., and Thompson, A. M. (2000) Atmospheric sulfur cycle simulated in the global model GOCART: Model description and global properties. J Geophys. Res. 105, 24,671-24,687.

Cooke, W. F., Liousse, C., Cachier, H., and Feichter, J. (1999) Construction of a 1x1 fossil fuel emission data set for carbonaceous aerosols and implementation and radiative impact in the ECHAM4 model. J Geophys. Res. 104, 22,137-22,162.

Barrie, L.A., Yi, Y., Leaitch, W.R., Lohmann, U., Kasibhatla, P., Roelofs, G.-J., Wilson, J., McGovern, F., Benkovitz, C., Melieres, M.A., Law, K., Prospero, J., Kritz, M., Bergmann, D., Bridgeman, C., Chin, M., Christensen, J., Easter, D., Feichter, J., Land, C., Jeuken, A., Kjellstrom, E., Koch, D. and Rasch, P. (2001) A comparison of large-scale atmospheric sulphate aerosol models (COSAM): overview and highlights. Tellus, 53B, 615-645.

Iversen, T., and Seland, Ø., 2002: A scheme for process-tagged SO4 and BC aerosols in NCAR CCM3: Validation and sensitivity to cloud processes. J. Geophys. Res., 107 (D24), 4751, 10.1029/2001JD000885.

Koch, D. (2001) Transport and direct radiative forcing of carbonaceous and sulphate aerosols in the GISS GCM. J Geophys. Res. 106, 20,311-20,332.

Koch, D., Jacob, D., Tegen, I., and Chin, D. (1999) Tropospheric sulfur simulation and sulfate direct radiative forcing in the Goddard Institute for Space Studies general circulation model. J Geophys. Res. 104, 23,799-23,822.

Rasch, P. J., Barth, M. C., Kiehl, J. T., Schwartz, S. E., and Benkovitz, C. M., (2000): A description of the global sulfur cycle and its controlling processes in the National Center for Atmospheric Research Community Climate Model, Version 3. J. Geophys. Res., 105, 1367-1385.

Seinfeld, J.H. and Pandis, S.N. (1998) Atmospheric Chemistry and Physics. From air pollution to climate change. John Wiley and Sons, Inc., USA. 1326 pp.

Zhang, G.J. and McFarlane, N.A. (1995) Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos. Ocean., 33, 407-446.

1 Trond Iversen and Øyvind Seland, Department of Geophysics, University of Oslo, P.O.Box 1022, Blindern, N-0315 Oslo, Norway. (,


The role of cumulus parameterisation in global and regional sulphur transport iconI. Global, regional and national sector issues

The role of cumulus parameterisation in global and regional sulphur transport iconMilitary Power, Grand Strategy, and the eu’s Changing Global Role

The role of cumulus parameterisation in global and regional sulphur transport icon94, 181-203. Aagaard, T. and Greenwood, B. 1994. Suspended sediment transport and the role of infragravity waves in a barred surf zone. Marine Geology, 118

The role of cumulus parameterisation in global and regional sulphur transport iconGeomagnetic observatories in Antarctica; state of the art and a perspective view in the global and regional frameworks

The role of cumulus parameterisation in global and regional sulphur transport iconGlobal Environment Facility The Regional Environmental Center
Усовершенствование оценок выбросов парниковых газов в сельском хозяйстве и внедрение элементов эффективной практики в национальные...
The role of cumulus parameterisation in global and regional sulphur transport iconTaskforce on Accessible Transport to the Australian Transport Council

The role of cumulus parameterisation in global and regional sulphur transport iconIn the Asia Pacific regional context, the understanding of regional and national security has been much covered broader strategic issue areas. The main purpose

The role of cumulus parameterisation in global and regional sulphur transport iconGlobal warming projected water cycle changes over the Mediterranean, East and West: a comparison study of a super-high resolution global model with cmip3

The role of cumulus parameterisation in global and regional sulphur transport iconJan / 07 Feb / 09: Suncor Voyageur Upgrader Sulphur Complex, Fort MacMurray, A. B

The role of cumulus parameterisation in global and regional sulphur transport iconStudents studying for Bio U117 Exam 1 should only review the sections «Introduction and terminology», «Basic chemistry», «Cells and membrane transport
Разместите кнопку на своём сайте:

База данных защищена авторским правом © 2014
обратиться к администрации
Главная страница