GEOPHYSICAL RESEARCHLETTERS, VOL. 25, NO.11, PAGES 1883-1886,JUNE 1, 1998
The impact of subvisible cirrus clouds near the tropical tropopause on stratospheric water vapor JoanE. Rosenfield David B. Considine 2, Mark R. Schoeberl a, and Edward V. Browell 4 Abstract.
The radiative impact of subvisible cirrus ice clouds at and just below the tropical tropopause has been studiedusinga zonally averagedinteractive chemistryradiation-dynamicsmodel. Model runs have been per-
drationof the stratosphere [Jensenet al., 1996b]. In
this paper we usethe Goddardinteractive2D model to study the radiative impact of these thin cirrus. Model runs with and without the radiative heating of these formed with and without the inclusion of the radiative cloudshave been carried out, and the additional heatheating of thesethin ice clouds,and with and without ing is allowedto feed back on both temperaturesand sedimentation. Near-infrared optical depths of 0.005- vertical velocity. The results of this study suggestthat 0.08 werecomputedfor assumedlog-normalsize distri- the absorptionof heat by thesesubvisiblecirrus clouds butions of sphericalparticles having mode radii of 2- may resultin increasedamountsof water vaporreaching 10 pm. Particles with 6 pm mode radii have computed the tropical lower stratosphere. scatteringratios of 3-15 at 603 nm, in good agreement with lidar observations.
The increased radiative
heat-
ing of theseclouds,0.1-0.2 K/day, resultsin tempera- Method ture increasesof 1-2 K and vertical velocity increases
The coupled2D model has been describedin Rosen-
of 0.02-0.04mm/s. As a consequence of the warmer
tropopause, lower stratospherewater vapor increases field et al. [1997]. Sincethat study,variousimproveby as much as i ppmv. The dehydrationresultingfrom ments have been made in the model. The model gensedimentation
was found to be a much smaller
erated water vapor is now usedin the radiative heating calculation,and the radiative heating due to NAT and
effect.
ice clouds,as describedin Rosenfield[1992],has been Introduction Subvisible cirrus clouds have been observed at and
just belowthe tropicaltropopausein both satellitedata
added. The vertical resolutionin the dynamicsmodule has been increasedfrom a grid size of 2.66 km to 2.0 km. Reaction rates and photolysis crosssections have been updated to the Jet Propulsion Laboratory
[DeMoreet al., 1994].The nu[Wanget al., 1994]andin situobservations [Heymsfield,(JPL) recommendation
1986].In Stratospheric Aerosoland Gas ExperimentII (SAGE II) data, zonalmeancloudfrequencies ranged from 20 to 70%, with a meanfrequencyof 45% in the tropicsat 15 km [Wanget al., 1996]. Jensenet al. [19964]inferredopticaldepthsof 0.005-0.1from lidar backscatterdata taken during the Central Equatorial PacificExperiment(CEPEX) in 1993. More recently,
merical advection schemein the chemistry module has
beenchanged fromthe scheme of Prather [1986]to that of Lin and Rood [1996]. The performanceof this algorithm in the GSFC fixed transport model has been
discussed in Jackmanet al. [1996].
The condensedmass of tropical ice cloudswas calculated by integrating over a temperature probability subvisiblecirrus cloudswere observedby the DIAL li- distribution which takes into account the longitudinal dar during the Tropical OzoneTransport Experiment variations in temperature, as describedin Considineet
al. [1994]and Rosenfieldet al. [1997]. The distribuTOTE) andthe VortexOzoneTransportExperiment tion is based on 15 years of NCEP temperature data. VOTE) in 1995-1996.They weredetectedwhentem-
peratureswere below 195 K, and they had maximum In this study, the center of the probability distribu-
tion at each time and tropical latitude has been shifted 603 nm scatteringratios of 10-14. from the zonal mean observedtemperature to the model The radiative heatingof thesecloudshas beencompredicted temperature. In this manner, the additional putedby Jensenet al. [19964],andtheir occurrence has heating of thesecloudswill feed backon modeltemperbeenproposedas providinga mechanism for the dehyatures thus affectingthe total massof ice cloudspredicted. The particles are assumedto obey a specified log-normalsize distribution, and additional condensa•General SciencesCorporation, Laurel, MD 2Joint Center for Earth System Science, University of tion of water resultsin changesin the number of particlesas opposedto a growth in the sizeof particles. The Maryland, College Park, MD size distribution was constrainedby lidar observations SLaboratoryfor Atmospheres,NASA Goddard SpaceFlight as discussed below.
Center, Greenbelt, MD The flux of condensedwater due to particle sedimen4AtmosphericSciencesDivision, NASA Langley Research tation is determined by first calculating the massflux Center, Hampton, VA
expected of a monodisperseparticle size distribution. The particle sedimentationrates are determed using
Copyright1998by the AmericanGeophysical Union.
the formulaof Kasten [1968]. By assumingthe parti-
Papernumber98GL01294.
cle fall speedis proportionalto the squareof its radius
0094-8534/98/98GL-01294505.00
(i.e., Stokes'relationship)a simplecalculationresults 1883
1884
ROSENFIELD, ET AL: IMPACT OF SUBVISIBLE CIRRUS CLOUDS
in a proportionality constantwhichweuseto relatethe
Model Water Vapor
massflux for the monodispersedistributionto that for the log-normalsizedistribution. Results
5O
40: ..... 5 '
and Discussion
The total ice water content predicted by the model
is controlledprimarilyby the modeltemperatures and watervapor. Time- heightcontoursof modeltemperatures,averaged between20Sand 20N, compared with NCEP temperaturesare shownin Figure 1. The annual cycleof the tropopause temperatureis simulated
E
..
.
2O 5
35
4.5 4
3O
3.5
3 correctlyby the model, with the lowestand highesttem25 peratures in northern winter and summer, respectively. 2.5 The model tropopauseis roughly 4 K colder than that 2 of the NCEP data; however, it is known that the vertical resolution of satellite data used in analysesis in1 sufficientto resolvethe sharp gradientsthat occur near 0 the tropopause. In addition, during the TOTE missiontemperatures near the tropopauselower than 190 K 0.0 0.5 1.0 1.5 2.0 were measured by the Microwave Temperature Profiler. In light of these low temperatures observed,the model Years temperatures are not unreasonable. Time- height contoursof the model steady-state wa- Figure 2. Steady state model computedwater vapor between20Sand20N. The secondyear ter vapor, averagedbetween20S and 20N, are shownin (ppmv)averaged a repeat of the first year. Ice particle heatinghasbeen Figure 2. Two years are shown, with the secondyear is included. being a copy of the first year, in order to better illustrate the vertical transport. The "tape recorder" signal is clear in the figure and can be comparedwith Plate 1 H20 (not shown)liesbetweenSAGE and MLS observaof Mote et al. [1996].Highervaluesof H20 ascendfrom tions, while falling within the range of values observed the troposphereduring May through July, while lower in situ during the StratosphericTracers of Atmospheric valuesrise in Decemberthrough February, as expected Transport(STRAT) mission.The agreementbetween from the annual cycle in tropopause temperatures. The the modeledwater and both SAGE and MLS is good rate of ascentinto the stratosphere,however, appearsto above 20 km. be somewhathigher than the observationsindicate. At In the model, ice water condensesnear the tropical and just below the tropical tropopause,where the sub- upper troposphere whenever the specified supersatuvisib• cirrus cloudsform, the annually averagedmodel ration of 40% has been reached. This correspondsto the roughly 2 K supersaturationsuggestedby Peter et
MODEL T, ILATI < 20 24
22
al. [1991]. The ice cloudsare computedmainly in the months Septemberthrough February, during the time when temperaturesare falling and there is sufficientwater vapor to condense. In addition, the temperature
probability distribution used has maximum probabi
20
ties oflow temperatures during those times. Figure 3
. , .. , . ., . .
•22
• 15•• •.•
14' , •4••, 16 I
4
7
••10
MONTH
0 12 • I
'
'
'
4
'
'
'
7
'
'
'
10
'
'
ROSENFIELD, ET AL: IMPACT OF SUBVISIBLE CIRRUS CLOUDS
1885
CHANGE IN T (K)
CHANGE IN HEATING (K/D) 2O
•
18
• 17 •' 16 •-0.0 15
CHANGE IN W (mm/s)
W (mml$)
•' 10 17
• 16 13
.....
I
4
7 MONTH
10
I
• .
4
7 MONTH
.
10
Figure 4. Changesin net heatingand temperatures(top), for the particlesize distributionwith moderadius of 6 ttm. Verticalvelocities(with iceparticleheating)and changes in verticalvelocitiesdueto ice cloudheating (bottom). All quantitieswereaveragedbetween20S and 20N. showsthat the fraction of the grid box saturated with respectto ice computedby the model rangesfrom 10 to 50%. This quantity can be comparedwith the observedfrequencyof occurrenceof the clouds,whichwere found to be roughly20 to 70% by SAGE II. Thus the modeled cloud occurrencerate may be somewhat lower
crosssections,while at the sametime correlatingwith smaller number densities, for a constant mass of con-
densedwater. The net result was a decreasein optical depth with size. The 6 /tm mode radii particles gavescatteringratios of 3 to 15 in goodagreementwith the lidar observations. Optical depths in the near inthan that observed. The maximum ice water content frared for thesesizedparticleswere roughly 0.005-0.02. computed is roughly10-4 g/m3, with tropicalvaluesof It should be noted that the behavior of optical depth
0.2-1.2x 10-4 g/m3. This canbe compared with typi- with particle size characteristic of these model runs difcalvalues observed by Heymsfield [1986]of 10-4 g/m3 fers from that which would be obtained if the total mass andpeakvaluesof 10-4 g/m3 modeled by Jensen et al. of condensedwater was not conservedand the particles [1996a]. The sensitivityof the computedoptical propertiesto the specifiedparticle sizewasstudied. Crosssectionsof the ice particleswere calculatedusingMie theory, averagingoverassumedlog-normalsizedistributionshaving moderadii of 2, 6, and 10/tm with a standarddeviation of 1.5/tm. Particle number densitieswere determined
were allowedto grow in size, while keepingfixed the total number density. In this case, optical depths would increasewith particle size. Steady-state runs were carried out with and without the inclusionof the radiative heating of the ice clouds.
Figure 4 (top left) showsthe changein the net heat-
ing introduced by the ice cloud heating for the 6 /tm mode radiuscase. The increaseddiabatic heating of 0.1from the total mass of condensed water and the assumed particle size distribution. Table 1 shows603 nm 0.2 K/day is mainly an infraredeffect,asradiationcoming from the lower troposphereis absorbedby the cloud. scatteringratios and near-infraredoptical depthscomThe model distributesthis increasedheating into simulputed ibr the three cases.Larger particles have greater taneous increasesin temperature and vertical velocity, as shown in Figure 4. Temperatures at and just below the tropical tropopause are increasedby more than 2 K Table 1. ScatteringRatiosandNear-Ir Optical Depths. from October to February. At the same time vertical upwellingincreasesby roughly 0.02-0.04 mm/s. The Mode radius ScatteringRatio Near-It Optical Depth vertical velocitiesare lower during the northern hemisphere summer and higher during the northern hemi2 /•m 5- 30 .01- .08 sphere winter, in agreement with residual circulation 6 /•m 3- 15 .005- .02 vertical velocities diagnosedusing Upper Atmosphere 10 /•m 2- 10 .005- .015
ResearchSatellite(UARS) data [Rosenlof,1995].
1886
ROSENFIELD,
ET AL: IMPACT OF SUBVISIBLE CIRRUS CLOUDS
MODEL H20, P = 51 mb
': ' ' Clo"dHeaiing, R=2' ' ......... ..... • .......... ....
6 ......
i
5E
Cloud Heating, R=6 Cloud Heating, R=10 No Cloud Heating, R=2 No Cloud Heating, R=õ No Cloud Heating, R=10
No CloudHeating,
larger than we have assumedwould lead to a greater degreeof dehydrationdueto sedimention; however,lidar scatteringratios computedusingtheselarger sized particleswouldbe lessin accordwith observations. References
Considine, D. B., et al., Effectsof a polarstratospheric cloud parameterization on ozonedepletiondueto stratospheric aircraftin a two-dimensional model, J. Geophys.Res., 99, 18,879-18,894, 1994.
DeMore, W. B., Ghemicalkineticsand photochemicaldata
Nø Sedimentatiøn' •..•1
for usein stratospheric modeling,Evaluationnumber11,
JPL Publ., 9•!-œ6,273 pp., 1994. Heymsfield,A. J., Ice particlesobservedin a cirriform cloud
at -83G andimplications for polarstratospheric clouds,J. Atmos. Sci., •!$, 851-855, 1986.
3 I
4
7 MONTH
10
Jackman,G. H., et al., Past, present,and future modeled ozonetrendswith comparisons to observed trends,J. Geophys. Res., 101, 28,753-28,767,1996. Jensen,E. J., et al., On the formationand persistence of subvisiblecirruscloudsnear the tropicaltropopause,J. Geophys.Res., 101, 21,361-21,375,1996a.
•'igure 5. Model water vapor profilesat 51 mb, aver- Jensen,E. J., et al., Dehydrationof the uppertroposphere and lower stratosphere by subvisiblecirrus clouds near agedbetween20S and 20N for differentice particlesize distributions and cases,where R is the mode radius. the tropicaltropopause,Geophys.Res. Lett., œ3,825828, 1996b. The thin solid, thin dot, and thin dash-dot curvesare coincident.
Kasten, F., Falling speed ofaerosol particles, J. Appl.Meteorol., 7, 944-947, 1968.
Lin, S.-J., and R. B. Rood, Multidimensionalflux-formsemi-
Lagrangiantransportschemes, Mort.WeatherRev., 1ϥ, 2046-2070, 1996.
It should be noted that the heating perturbation is Mote, P. W., et al., An atmospherictape recorder:The imlessthan it would be if the temperatures and ozonewere print of tropicaltropopause temperatures onstratospheric not allowedto adjust in the interactive model. Higher water vapor, J. Geophys.Res., 101, 3989-4006,1996. temperatures and lower ozone due to increased up- Peter,Th., et al., Increase in the PSC-formation probability wellingtend to reducenet heatingrates. For this reason causedby high-flyingaircraft, Geophys.Res. Lett., 18, the heating rate perturbation cannot be strictly com1465-1468, 1991. paredto that computedby Jensenet al. [1996a],who Prather, M. J., Numericaladvectionby conservationof calculatedheatingrate perturbationsof 0.5-2.0 K/day second-order moments,J. Geophys.Res.,91, 6671-6681, 1986. for cloud optical depths of 0.01-0.03. Figure 5 showsthe computed annual cycle of 51 mb Rosenfield,J. E., Radiativeeffectsof polar stratospheric water vapor mixing ratios for the caseof no cloud heatcloudsduringthe AirborneAntarcticOzoneExperiment andthe AirborneArcticStratospheric Expedition,J. Geoing and no sedimentation,three casesof no cloudheating with the inclusionof sedimentation,and three cases phys. Res., 97, 7841-7858, 1992. in which both cloud heating and sedimentationare in- Rosenfield,J. E., et al., Stratosphericeffects of Mount Pinatubo aerosolstudiedwith a coupledtwo-dimensional eluded. The effect of sedimentation is apparent only model, J. Geophys.Res., 10œ,3649-3670,1997. for the 10 ttm particles,for which casethe lowerstratospheric water vapor is decreasedby at most 0.1 ppmv. Rosenlof,K.H., Seasonalcycleof the residualmean meridional circulation in the stratosphere, J. Geophys. Res., With the inclusionof cloud heating, the warmer trop100, 5173-5191, 1995. ical tropopauseleads to a greater amount of water vaWang, P.-H., et al., Tropical high cloud characteristicsdepor reachingthe lower stratosphere. The water vapor rived from SAGE II extinction measurements, Atmos. increaseswith decreasingparticle size, in accordance Res., $•, 53-83, 1994. with the model computed behavior of the subvisible Wang, P.-H., et al., A 6-year climatology of cloud occurcirrus optical depth with particle size. For all three rence frequencyfrom StratosphericAerosol and Gas Exsize distributions,the water vapor increasesdue to ice perimentII observations(1985-1990),J. Geophys.Res., cloud heating outweighany decreasesdue to sedimen101, 29,407-29,429, 1996. tation. For the 6 ttm mode radius casethe net result is that as much as 1 ppmv more water vapor reachesthe E. V. Browell, NASA Langley ResearchCenter, MS 401A, stratosphere. Hampton, VA; D. B. Considine, J. E. Rosenfield, M. R. Thus the results of this study suggestthat one role Schoeberl,Code 916, NASA/Goddard SpaceFlight Center, that subvisiblecloudsmay play in the water vapor bud- Greenbelt,MD 20771,(e-mail:
[email protected]) get is to increasewater vapor in the lower stratosphere by reducing the amount of water condensingat the (ReceivedOctober23, 1997; revisedMarch 17, 1998; warmer tropopause. Size distributions with mode radii acceptedApril 15, 1998.)
