Environmental Conditions Required for Contrail Formation and Persistence Eric J

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. D4, PAGES 3929-3936, FEBRUARY 27, 1998

Environmental conditions required for contrail formation and persistence Eric J. Jensen,1 OwenB. Toon,2 StefanKinne, • GlenW. Sachse,4 Bruce E. Anderson,4 K. RolandChan, • CynthiaH. Twohy,s BruceGandrud, s Andrew Heymsfield,s and RichardC. Miake-Lye•

Abstract. The ambient temperatures and humidities required for contrail formation and persistenceare determinedfrom in situ measurementsduring the SubsonicAircraft: Contrail and Cloud EffectsSpecial Study (SUCCESS) experiment. Ambient temperatures and water vapor concentrations were measured with the meteorologicalmeasurement system, a laser hygrometer,and a cryogenic hygrometer(all onboardthe DC-8). The thresholdtemperatures are compared with theoretical estimates based on simple models of plume evolution. Observed contrail onset temperatures for contrail formation are shown to be 0-2 K below the liquid-saturation thresholdtemperature, implying that saturation with respect to liquid water must be reached at some point in the plume evolution. Visible contrailsobserved during SUCCESS persistedlonger than a few minutesonly when substantialambient supersaturationswith respectto ice existedover large regions. On someoccasions, contrails formed at relativelyhigh temperatures(>_ -50øC) due to very high ambientsupersaturations with respectto ice (of the order of 150%). These warm contrails usually formed in the presenceof diffusecirrus. Water vapor from sublimatedice crystalsthat entered the enginewas probably necessaryfor contrail formation in someof these cases. At temperaturesabove about -50øC, contrails can only form if the ambient air is supersaturatedwith respect to ice, so these contrails should persist and grow.

1. Introduction mate in particular regions by compiling climatologiesof During recent years, considerableattention has been contrails based on ground-basedobservations [Liou et focused on the climatic impact of clouds. Of particu- al., 1990]. Evaluationof the contrailinfluence on global climate will require use of satellite observations. Assess- lar interest is the possibility that as anthropogenicin- ment of potential future contrail climate impacts based fluences alter the atmosphere and climate, cloud prop- on projected air traffic requires a knowledgeof the en- erties may change, resulting in a cloud-climate feed- vironmental conditions required for contrail formation. back [Twomey,1974; Ackerman et al., 1997]. Contrails The contrail threshold conditions also provide informa- formed by jet aircraft in the upper troposphere are ice tion about how effective aircraft exhaust particles are cloudsformed directly by anthropogenicinjections into as ice nuclei. the atmosphere. Hence, contrails represent a direct an- thropogenicinfluence on cloudsand possiblyon climate. Numeroustheoretical studiesof the thermodynamics Past studies have analyzed the effect of contrails on cli- of contrail formation have been conductedduring the pastseveral decades (see Schumann [1996] for a detailed review). Appleman[1953] used simple arguments about •NASA Ames ResearchCenter, Moffett Field, California. the evolutionof temperature and water vapor in aircraft 2Universityof Colorado,Department of AtmosphericSci- exhaust plumes to predict threshold temperatures and ences, Boulder. pressuresat which contrails could form. Attempts have aBay Area EnvironmentalResearch Institute, San Fran- cisco, California. been made to validate the predictedthreshold temper- 4NASA LangleyResearch Center, Hampton, Virginia. atures using ground-basedobservations of contrails and 5NationalCenter for AtmosphericResearch, Boulder, Col- nearly colocatedradiosonde measurements of tempera- orado. ture and humidity [Peters,1993; Busen and Schumann, 6AerodyneResearch, Inc., Billerica, Massachusetts. 1995]. However,the inaccuracyof radiosondehumid- ity measurementsand the separationbetween humidity Copyright 1998 by the American GeophysicalUnion. measurements and the contrails have limited the useful- Paper number 97JD02808. nessof thesestudies. Recently, Schumann et al. [1996] 0148-0227/98/ 97JD-02808$ 09.00 reported in situ measurementsof temperature and hu-

3929 3930 JENSEN ET AL.: CONTRAIL THRESHOLD TEMPERATURES midity at contrail onset times for a flight following a Atmosphere(40øN) temperatureand pressureprofiles, moderate-sized research aircraft. Their results were in Appleman showedthat contrails should typically form aggreementwith the theoreticalpredictions of threshold under ambient ice-saturated conditionsat pressuresbe- temperatures assumingliquid saturation in the plume low about 300 mbar. This analysis also indicated that is required for contrail formation. formation of contrails in dry ambient air should typi- In addition to determination of how frequently con- cally only be possiblenear 220 mbar. trails form, it is also important to distinguishbetween In situ investigationsof cirrus clouds over the past transient and persistent contrails. The vast majority of severalyears have indicated that ice nucleationcan oc- visible contrails dissipate within a few minutes. Such cur in air that is subsaturated with respect to liquid contrails have little potential for affectingthe radiation water but saturated with respect to ice. This distinc- balance or climate. Occasionally,contrails are observed tion is particularly important in the very cold upper to persist and even grow. Sometimes,the contrailslast troposphere. For example, at-65øC, liquid saturation for hours and spreadinto sheets. The persistenceand is not achieved until the saturation with respect to ice spreading of contrails will determine their impact on (RHI) reaches185%. Measurementsin waveclouds over the radiation balance. Hence, we need to determine the Rocky Mountains during SUCCESS showedclear the relationship between contrail lifetime and ambient evidence of ice crystal formation in air at relative hu- conditions. miditywith respectto liquidwater (RH) < 90% [Jensen In this study, we have used measurementsmade dur- et al., 1997]. Theseresults suggest that the mostbasic ing the SubsonicAircraft: Contrail and Cloud Effects requirement for ice crystal formation in aircraft plumes Special Study (SUCCESS) to correlateenvironmen- might be that ice saturation be achievedat sometime tal conditions with contrail formation and persistence. during the plume cooling. Hence, we should attempt Very precisein situ measurementsof temperature and to distinguishwhether liquid saturation is actually re- water vapor concentrationwere made with various in- quired for contrail formation. struments on the NASA DC-8 aircraft during SUC- To calculate the threshold temperature numerically, CESS. Contrails from the DC-8 were identified by chase we must derive an expressionfor the evolutionof the the aircraft, ground observers,DC-8 aft video, ER-2 nadir saturationwith respectto ice (Si) in the plume. We imaging, and satellite images. Below, we briefly re- begin by writing the plume water vapor mixing ratio view the standard theoretical treatment for predicting in terms of the changein temperature(see Schumann thresholdconditions for contrailformation [Appleman, [1996]for a moredetailed derivation): 1953]. Next, we describethe SUCCESStemperature, humidity, and contrail observations. Then we compare Wplm--Wamb + cpATEI•t•.o (1) the theoretical predictions and observedthreshold conditions. Finally, we discussthe implications of these where Wambis the ambient water vapor mixing ratio, results for current theories of contrail formation pro- cp is the specificheat capacityof air, AT is the differ- cesses. ence between the plume temperature and the ambient 2. Theoretical Prediction of Contrail temperature, EI•.o is the emissionindex for water vapor, and Elheat is the emissionindex for heat. Next, Threshold Temperatures we convert from mixing ratio to water vapor number The first detailed analysisof the environmentalcon- density: ditionsrequired for contrailformation was conducted by naircpEIH•.oATRw Appleman[1953], and the topic was recently reviewed in nw,plm--nw,amb q- ElneatRd (2) detailby Schumann[1996]. These analyses were based primarilyon the assumptionthat the temperatureand where hair is the ambient air number density,nw,amb water vapor mixing ratio in the plume are both con- and rlw,plmare the ambient and plume water vapor trolled by entrainmentof ambientair into the plume. number densities,and Rd and Rw are the gas constants Usingthis assumptionwe can calculatethe watervapor for dry air and water vapor. Then the ice saturation in concentrationand temperaturein the plume giventhe the plume is given by ambienttemperature and water vaporconcentration as well as the emission indices for heat and water vapor nw,ambnai,.cpEI•t•.oATR• (3) (heatand water vapor mass per gram of fuelburned). Si,plm-Tlsat,i (Wplm)q-Tlsat,i (Wplm)Z_[heatRd Appleman[1953] argued that wheneverthe available moisturewas greaterthan that requiredto reachsat- where rlsat,iis the saturationwater vapor numberden- uration in the plume, a contrail could form. He thus sity. The saturation with respect to liquid water can calculated threshold temperatures for contrail forma- be calculated in an analogousmanner. The plume sat- tion graphically.The thresholdtemperature turns out uration dependsupon the ambient air density and hu- to be a function of both the ambient relative humid- midity. As pointedout by Busenand Schumann[1995], ity andthe ambientpressure. Using the U.S. Standardnot all of the energy liberated by the engine combus- JENSEN ET AL.- CONTRAIL THRESHOLD TEMPERATURES 3931

2.0

Tomb

-58. -42. 1.5 -46.

! ! -.54. !

O.5 - •-,•"•'• • '- - .•__•__-=__-•- • -,• =,-= ß-•' "" -• _ Solid:Ice saturation ratio, Siomb = 0.7 _ Dashed:Liquid saturation ratio, Siomb = 0.7 _ Dotted:Liquid saturation ratio, Siø,. b= 1.6 0.0/ i , ...... I , ...... I 0.01 0.10 1.00 10.00 100.00 Tp,u.•.- Tomb (K) , I I I I i I I .... 5.9x10• 1.0x10'5.8x104 1.5x104 8400 4800 i500 Plume dilution

50.0I , ,10.0 I , 5.0I 2.0I 1.0I 0.4I 0I.2 Plume age (s)

i I , I ... I . I .. I I 12.00 2.40 1.20 0.48 0.24 O. 10 0.05 Distancebehind eircraft Plate 1. Plume saturationratios (ice and liquid water) are plotted versusplume temperature enhancementfor various ambient temperatures and an ambient RHI of 0.7. Axes below the plot indicatethe plumedilution, age, and distancebehind the aircraft(assuming air speedof 240 m/s) correspondingto the temperatureenhancement. The plumeage was estimatedon the basisof resultsfrom a detailedfluid dynamicsmodel of plumeevolution [Miake-Lye et al., 1993].For the ambientpressure and RHI assumedin this calculation,the plumenever reaches water saturation for ambient temperaturesabove about -48øC. The dotted curve correspondsto ambient conditionsthat are just barely supersaturatedwith respectto liquid water.

tion is converted to heat in the exhaust plume. Hence, very hot and hencethe saturation ratio is near 0. As the the heat emissionindex, Elheat, must by replaced by gas coolsdue to entrainment of ambient air, the satura- EI•,eat(1- r/) wherer/is the efficiencyof the engine.For tion vapor density decreasesrapidly, so the saturation modern commercialjet engines,the efficiencyshould be ratio increasesrapidly. Eventually, the detrainment of about 0.3. This is an important correction: the thresh- water vapor takes over, and the saturation ratio de- old temperature calculatedby assumingr/- 0.3 is about creasesagain. The peak saturation ratio increaseswith 2 K higher than that for r/= 0. decreasingambient temperature. The plume saturation is plotted versus AT for vari- It is important to note that the plume saturation ver- ous ambient temperatures in Plate 1. Parameter values susAT, calculatedwith this method, shouldbe very re- used (typical for modernjet engines)were EI•,eat '- alistic in spite of the simplicity of the model. The result 4.2 x 1011erg/g fuel, r/= 0.3, EIH20 = 1.25g H•O/g does not depend upon the detailed three-dimensional fuel. The temperature differenceAT is inverselypro- structure of the plume. The only assumption made is portional to the plume dilution N. Using a dynami- that the water vapor and temperature in the plume are cal modelof the plumeevolution [Miake-Lye, 1993], we controlled by entrainment of ambient air. At a given have also related the plume dilution to distance behind position behind the aircraft, the plume dilution may be the aircraft. Just behind the engine, the plume gas is very different at differentdistances and directionsfrom 3932 JENSEN ET AL.' CONTRAIL THRESHOLD TEMPERATURES

Theoretical Contrail Critical Temperatures I .... I / / / // 300 // // // // // 2OO // // l l // l l // l l // l l l /// l l // T•c •:=0.3 l l / TLC E=0.3 l 100 l l ß.=-,---- TLC •:=0.2 / / / / / / / / / / / / / / / / / / / / //

-65 -60 -55 -50 -45 -40 -35 CriticalTemperature (C) Figure 1. Thresholdtemperatures based on Appleman[1953] theory are plotted versus ambient pressurefor ambientRHIs of 0, 50, 100, and 140%. Contrailformation is predictedto occur wheneverthe ambienttemperature is to the left of the solidcurves (corresponding to the ambient RHI and pressure).Solid curves are basedon the assumptionthat liquidsaturation in the plume is requiredfor contrail formation. Dashedcurves correspond to the assumptionthat only ice saturationis required.Threshold temperatures are alsoshown for an engineefficiency of 0.2.

the plume axis. However,at any given point, the di- sensitivityto this parameter.Schumann [1996] used a lution should control the temperature, the water vapor slightly differentmethod to calculatethe thresholdtem- mixing ratio, and the RHI. peratures. He used the fact that under thresholdcon- Appleman[1953] defined the thresholdtemperature ditions the curve of partial pressureof water vapor in for contrail formation as the ambient temperature such the plume versusplume temperaturewill be tangent to that the maximum plume RH would just exceed1. We the saturation vapor pressureversus temperature. The shall refer to this temperature as the liquid-saturation thresholdtemperatures calculated here agreewell with thresholdtemperature (T•e). As discussedabove, an- thosereported by Schumann.Coleman [1996] devel- other logicalthreshold temperature is that suchthat oped an analytic expressionfor the thresholdtempera- saturation with respect to ice is just barely achieved ture as a function of pressureand water vapor mixing within the plume (Tie). Thesethreshold temperatures ratio. The threshold temperatures calculated here dif- are functions of the ambient pressureand humidity. fered from the analytic expressionby no more than 0.5 Given the ambient environmental conditions,the thresh- K. old temperaturescan be calculatedby findingthe high- est ambient temperaturefor which the peak plume sat- 3. SUCCESS Measurements of uration ratio is just above1.0. We havewritten a simple Environmental Conditions and Contrails FORTRAN code for this calculation. The code can be obtained from the authors on request. The environmental conditions that determine whether Figure 1 showsthe thresholdtemperatures plotted contrail formation is possible are temperature, rela- versuspressure for variousassumed relative humidities tive humidity,and total atmosphericpressure. As part with respect to ice. A curve is includedfor a lower of the meteorologicalmeasurement system (MMS) on- engineefficiency (0.2) to indicatethe relativelylarge board the NASA DC-8, temperature was measuredwith JENSEN ET AL.- CONTRAIL THRESHOLD TEMPERATURES 3933 three different Rosemount probes. Special maneuvers contrailsidentified lasted no longerthan a few minutes. were conducted on most flights to calibrate the tem- A few casesof persistentcontrails are included.Also, perature measurements and to allow correction for air- several "threshold" cases are included when the DC-8 craft speedand attitude. The uncertainty of the MMS had just begunto generatea contrail. Finally, we iden- temperature measurement has been estimated to be no tified a few cases when the environment was near the more than 0.3 K. threshold temperature, but contrails were not visible. Water vapor concentration was measured with a laser The difference between the observedtemperatures hygrometer[Collins et al., 1995]on the DC-8. The ac- and the threshold temperaturescorresponding to the curacy of the laser hygrometerdew point measurement observedpressure and RHI are plotted versustemper- is estimated to be about 0.2 K. Along with the tem- ature for the DC-8 contrail casesin Figure 2. Within perature uncertainty discussedabove, this gives an ab- the uncertainties in the measurements,visible contrails solute uncertainty in RHI of about 5-10%, depending from the DC-8 were observedonly when the ambient on the temperature. On April 24 the laser hygrom- temperature was below Tic, suggestingthat liquid sat- eter data was affected by a voltage offset, so we have uration in the plume is indeed required for contrail for- usedresults from the cryogenichygrometer on the DC-8 mation. For the thresholdcases (shown as squaresin [Heymsfieldand Miloshevich,1993]. During time peri- Figure 2) the ambient temperatureranged from 0 to ods when both water vapor instrumentswere operating 2 K below Ttc. The only caseswith ambient tempera- at temperatures above about -50øC, the results were tures slightly above Ttc were the relatively warm con- in good agreement(Anderson, B. E., private commu- trail cases.In these casesthe ambient air was very near nication, 1997). The combinationof uncertaintiesin liquid water saturation(see discussion below). If only measuredpressure (about 5-10%) and RHI lead to un- ice saturation were required for contrail formation, then certainties in the calculated threshold temperatures of contrailsshould have been observedat temperatures4- 1-2 K. Uncertaintyin engineefficiency (about 25%) also 5 K higher than indicated by the threshold casesin Ta- contributes substantially to the uncertainty in Tic. ble 1. Even if the engineefficiency were as high as 0.4, In several cases,contrails were generatedin the pres- the observedthreshold temperatures would still be more ence of ambient cirrus clouds. As discussedbelow, it is consistent with the liquid saturation threshold than the possiblethat water vapor from ice crystalsingested into ice saturation threshold. the engines may have been partly responsiblefor con- We have also identified times when Tic < T < Tic us- trail formation in some cases. To estimate the ambient ing the DC-8 in situ data and examined the T-39 video ice water content, we used the Counterflow Virtual Im- during these times. No visible contrails were apparent. pactor (CVI) measurements.This instrumentsamples It is possible that some ice crystals do nucleate under all particles with radii larger than about 3 ttm and mea- these conditions but not enough to generate a visible surestheir water content with a downstreamLyman-t• contrail. When the ambient temperature is between Tic hygrometer[Twohy et al., 1997]. and Tic, the plume is ice supersaturatedfor lessthan a Times when the DC-8 was (or was not) generating second(see Plate 1). Hence,under these conditions, ice contrails were identified several ways: on some of the crystals would have very little time to grow, and a rel- flights, the NASA T-39 trailed the DC-8 with a for- atively small number density of ice crystals would not ward video camera; an aft directed video camera was generate a visible contrail. mounted on the DC-8; on flights over the Department If the ambient air is not supersaturatedwith respect of Energy (DOE) AtmosphericRadiation TestbedSite to ice, then contrail formation shouldnot be possibleat (CART) in northern Oklahoma,ground observers pho- temperaturesabove about-45øC to-50øC, for a typi- tographedand recordedDC-8 contrails;nadir imaging cal range of pressuresat commercial aircraft cruise al- from the ER-2 (flying above the DC-8) occasionally titudes(see the 100%RHI Tic curvein Figure1). How- showedthe presenceof a DC-8 contrail. In some cases ever, severalDC-8 contrailswere observedat higher the lifetime of the DC-8 contrail could be estimatedby temperatures; as an extreme example, on April 20, observationsfrom the DC-8 (when the aircraft was cir- 1996, the DC-8 generateda contrail at an ambient tem- cling), from the T-39 when it was trailing the DC-8, or perature of-36.2øC. In all of these warm contrail cases by ground observers. the ambient air was strongly supersaturatedwith re- 4. Results spectto ice(RHI _>140%). In thewarmest cases (events number5, 8, 9, and 10 in Table 1) the ambientair was A list of contrail casesidentified is given in Table very nearly supersaturatedwith respectto liquid water. 1, along with the temperatures, pressures,and relative Even thoughthe RHIs in thesecases were very large, humidities with respectto ice measuredby the DC-8 in- the temperatureswere slightlyabove the calculatedTic strumentation. The liquid saturation contrail threshold (seeFigure 2). Patchy,diffuse cirrus were also present temperatures(Tic) correspondingto the ambient pres- when thesewarm contrailswere generated. sure and RHI are also given. For some of the cases, The occurrence of these warm contrails can be ex- estimates of the contrail lifetime are given. Most of the plained two ways. First, within the limits of mea- 3934 JENSEN ET AL.' CONTRAIL THRESHOLD TEMPERATURES

Table 1. Contrail Events From SUCCESS

• Dateb Time,UT Temp,øCC p, mbc RHI, %c RH, %c Lifetime Ticd Tice

1 4/15 a 18.4222 -50.9 277 26 16 _• 2 min -49.0 -45.3 2 4/16 17.9417 -49.0 278 135 84 _• 2 min -44.6 NA f 3 4/16 18.4833 -57.2 240 105 60 2-3 min -48.3 NA 4 4/16 21.4483 -56.0 242 114 66 •2 min -47.7 NA 5 4/20 17.0193 -36.2 367 143 99 >_15min -38.3 NA 6 4/20 19.7890 -55.3 265 125 73 5-10 min -46.3 NA 7 4/20 20.3489 -54.9 267 128 75 5-10 min -46.1 NA 8 4/24 17.2496 -39.9 303 146 99 threshold -40.3 NA 9 4/24 17.9894 -39.5 307 146 99 _•5 min -40.2 NA 10 4/24 18.0017 -40.2 303 144 97 _•5 min -41.3 NA 11 4/24 18.0494 -43.1 287 146 96 <_5min -42.2 NA 12 4/24 18.1706 -51.0 250 125 76 _<5rain -46.6 NA 13 4/29 19.4025 -53.7 180 5 3 threshold -53.6 -49.8 14 4/30 18.1714 -50.1 272 58 36 threshold -48.4 -44.0 15 5/3 18.2417 -50.3 257 55 34 threshold -49.0 -44.6 16 5/8 17.5722 -49.5 261 138 85 2-3 min -45.2 NA 17 5/12 22.9500 -52.4 - 239 160 96 3 hours -44.1 NA 18 4/15 18.3131 -49.0 277 18 11 No contrail -49.1 -45.5 19 4/29 18.4475 -44.0 375.4 83 54 No contrail -44.2 -39.1 20 4/29 19.4164 -53.3 179.9 5 3 No contrail -53.6 -49.8

aRead 4/15 as April 15. ball dates axe 1996. CAmbient. dLiquid-saturationthreshold temperature: Ambient temperature requiredsuch that saturation with respectto water will be reached in the plume. eIce-saturation threshold temperature. fNot Applicable:Ambient RHI is > 100%, so T < Tic always,by definition.

surement uncertainty, the ambient air may have been 30 K aboveambient (see Plate 1), and the ice crystals slightly supersaturated with respect to liquid water. would sublimaterapidly. In our simpleplume model, The plume saturation would then have evolved as in- this vapor sourceis equivalentto increasingthe water dicated by the magenta dotted curve in Plate 1. As vapor emissionindex. Including this vapor sourcefor the plume diluted sufficiently,water saturation would cases where ambient ice water content measurements be reached. In this case, the contrail formation would wereavailable (cases 5, 9, and 10), we foundthat the have essentiallybeen driven by the presenceof numer- plumes would have reached water saturation. In other ous ice nuclei from the exhaust as the plume. That words,in thesecases the ambientair wasvery near liq- is, the only reason the contrail would be visible is that uid water saturation,so only a slightaddition of vapor within the plume far more ice nuclei and ice crystals to the plume from the engine exhaust and entrained were present than in the ambient air. ambientice would have resultedin liquid saturationas The problem with the above explanation is that no the plume approachedthe ambienttemperature. optically thick cirrus clouds were present. If the ambi- An obviousconclusion from this analysis is that when- ent air were supersaturated with respect to liquid wa- ever contrailsform at temperaturesabove about-44øC ter, then most of the ambient aerosolsshould have been to -50øC, they will be formingin ice-supersaturatedair activated(and subsequentlyfrozen), resultingin thick and shouldbe able to grow and persist. It wouldbe in- cirrus. The ambient air was probably just below liq- terestingto examinethe statisticsof contrailheights us- uid saturation, since optically thick cirrus clouds were ing lidar observations.Contrail frequencyat tempera- not present. An alternative explanation is that the am- turesabove • -50øCshould correspond to the frequency bient diffuse cirrus present was necessaryfor the con- of saturationwith respectto ice at thesetemperatures. trail formation. Water vapor available for ice nucleation in the plume will include ambient water vapor, 5. Contrail Persistence water vapor from the fuel combustion,and sublimated ice crystals that entered the engine or were entrained The vast majorityof contrailsobserved during the early in the plume evolution. Air enteringthe engine SUCCESSexperiment lasted no longerthan a fewmin- and air entrained into the plume before the dilution utes. The contrailstypically dissipated rapidly when reaches1000 would be heated to temperatures at least the aircraftvortices became unstable and brokeup. In JENSEN ET AL' CONTRAIL THRESHOLD TEMPERATURES 3935

SUCCESScontrail events

-5 -

-lO -

o Contrail observed

[] Threshold -15 - _ X No Contrail

-60 -55 -50 -45 -40 -35 Tamb (C) Figure 2. Differencesbetween the ambient temperature and the liquid saturation threshold temperature(Ttc) are shownfor the SUCCESScontrail cases (see Table 1). The ambientrelative humidities with respect to ice are shown above each point. Within the range of uncertainty, contrails were only observedwhen the ambient temperature was at or below Ttc, such that the plume would have reachedsaturation with respectto liquid water. Severalthreshold cases are shown(squares) when the contrailshad just begunforming. The crossesindicate caseswhen contrails were definitely not visible.

all of the cases listed in Table I for which the contrails may be very important for contrail persistence. If the persistedlonger than a few minutes,the ambientair was air just below the contrail formation level is very dry, substantiallysupersaturated with respectto ice. This then the ice crystals will sublimate as soonas they grow result is reasonable since under subsaturated ambient large enoughto begin sedimenting.Relatively small ice conditions,the plume will not remain saturatedlonger crystals could remain in the humid layer, resulting in a than about 10 seconds,even without vapor depletion persistent subvisiblecontrail. Aged, subvisiblecontrails due to crystal growth (seePlate 1), and ice crystals were often observedduring the SUCCESS experiment in the contrails with sizes of a few microns or less will with ground-basedlidar (Sassen,K., personalcommu- sublimate in less than a minute after the plume is sub- nication, 1997). saturated. On May 12, 1996, the DC-8 generateda contrailoff 6. Discussion the coast of California which persistedfor over 3 hours [Minniset al., 1997].The contrail generated by theDC- We have used precise in situ measurements from the 8 racetrackflight pattern wasclearly visible in GOES- SUCCESS experiment to evaluate the threshold tem- 8 satellite imagesas the contrail advectedover Cali- peratures for visible contrail formation. The results of fornia. The relative humidities with respect to ice in- this analysis are consistent with the theoretical calcula- dicatedby the laserhygrometer ranged from 110%to tions assuming that liquid saturation must be reached 170% throughoutthe racetrackflight path. Hence,the in the plume for contrail formation. In several of the DC-8 was flying in an extensiveregion with highly su- contrail events listed in Table 1, large ambient supersat- persaturatedair and patchy,diffuse cirrus. The persis- urations with respect to ice existed with either diffuse tent contrailsobserved during SUCCESS often formed cirrus present, or no cirrus present. These regions were in regionswith patchy cirrus present. prime for contrail formation. The lack of optically thick For someof the caseslisted in Table 1 (e.g., cases2, cirrus in these regionssuggests that relatively few effec- 5, and 16), the ambientair wassubstantially supersatu- tive heterogeneousice nuclei were present, and upper rated with respectto ice, but the contrailsstill persisted tropospheric clouds may be very sensitive to introduc- no longerthan a few minutes. Closeranalysis of the tion of effective heterogeneousnuclei. in situ water vapor and temperaturemeasurements for Contrails are frequently visible at distancesas closeas these cases indicates that these contrails often formed in 25-35 m behind the aircraft engines[Busen and Schu- narrow vertical layersor small patchesof high humidity. mann, 1995]. Using an analytical model of ice crys- The vertical structure of the ambient relative humidity tal growthin the exhaustplume, [K•ircheret al., 1996] 3936 JENSEN ET AL.: CONTRAIL THRESHOLD TEMPERATURES showedthat the number density of ice crystalsnucleated References must be at least about 104 cm-3 in order for the contrail Ackerman, A. S., O. B. Toon, J.P. Taylor, D. W. John- to be visible this quickly. Simulations of aerosoland son, P. V. Hobbs, and R. J. Ferek, Effects of aerosolson ice crystal nucleationand growth in the exhaustplume cloud albedo: Evaluation of Twomey's parameterization suggestthat the most likely mechanismfor nucleation of cloud susceptibilityusing measurementsof ship tracks, of the contrail crystalsis heterogeneousfleezing of sul- J. Atmos. $ci., in press, 1997. Appleman, H., The formation of exhaust contrails by jet fate coatedsoot particles[Kiircher et al., 1995, 1996; aircraft, Bull. Am. Meteorol. $oc., 3•, 14-20, 1953. Brown et al., 1996, 1997]. The simulationssuggested Brown, R. C., R. C. Miake-Lye, M. R. Anderson, and C. E. that fleshlynucleated sulfuric acid/water aerosols could Kolb, Aerosol dynamics in near-field aircraft plumes, J. not grow large enoughto spontaneouslyfreeze in the Geophys.Res., 101, 22,939-22,953, 1996. time required. Brown, R. C., R. C. Miake-Lye, M. R. Anderson, and C. E. Kolb, Aircraft sulfur emissionsand the formation of The simulationsreported by Kiircheret al. [1995, visible contrails, Geophys.Res. Left., œ•, 385, 1997. 1996]and Brownet al. [1997]were run underambient Busen, R., and U. Schumann, Visible contrail formation conditions where saturation with respectto liquid water from fuels with different sulfur contents, Geophys.Res. was not quite reachedwithin the plume. However,the Left., 22, 1357, 1995. SUCCESS measurementsreported here along with the Coleman, R. F., A new formulation for the critical temperature for contrail formation, J. Appl. Meteorok, 35, in situ observationsreported by Schumannet al. [1996] 2270-2282, 1996. suggestthat liquid saturationin the plumeis indeedre- Collins, J. E., Jr., G. W. Sachse,L. G. Burney, and L. O. quired for visible contrail formation. Future modeling Wade, A novel external path water vapor sensor, paper studies of contrail formation should attempt to explain presentedat the 5th Annual Meeting of the Atmospheric the suddenonset of visible contrailsat Tic. As suggested Effectsof Aviation Program, NASA, Virginia Beach,VA, April 23-28, 1995. by Kiircheret al. [1996],the higherplume supersat- Heymsfield, A. J., and L. M. Miloshevich,Homogeneous ice uration may allow growth and fleezing of fleshly nu- nucleation and super cooled liquid water in orographic cleatedH2SO4/H20 droplets(without internally mixed wave clouds, J. Atmos. $ci., 50, 2335-2353, 1993. sootparticles) to play a substantialrole in the contrail Jensen, E. J., et al., Ice nucleation processesin upper tropo- formation process.Also, if contrailsdo not form until sphericwave-clouds observed during SUCCESS, Geophys. the ambienttemperatures drop belowTtc, then the soot Res. Left., in press, 1997. K'•rcher, B., T. Peter, and R. Ottmann, Contrail forma- particlesare not necessarilyvery effectiveheterogeneous tion: Homogeneousnucleation of H2SO4/H20 droplets, fleezingnuclei. Kiircheret al. [1996]estimated that an Geophys.Res. Left., 22, 1501, 1995. ice germ/soot contact angle lessthan about 60ø was K'•rcher, B., T. Peter, U. M. Biemann, and U. Schumann, required to explain contrail formation at temperatures The initial composition of jet condensationtrails, J. At- mos. S½i., 53, 3066, 1996. just above Ttc. The results from this study suggestthat Liou, K.-N., S.C. Ou, and C. Koenig, An investigationon T must be below Ttc for contrail formation, so the con- the climatic effect of contrail cirrus, Lecture Notes in En- tact anglemay be larger (correspondingto lesseffective gineering, Vol. 60, pp. 138-153, Springer-Verlag,New fleezingnuclei). The fact that the ambienttemperature York, 1990. needsto be below Ttc may also indicate that liquid wa- Miake-Lye, R., et al., Plume and wake dynamics,mixing and chemistry behind a high speed civil transport aircraft, J. ter dropletsmust be activated(and subsequentlyfreeze Alter., 30, 467-479, 1993. in order to generate a sufficientnumber of ice crystals Minnis, P., et al., Transformation of contrails into cirrus for a visible contrail. clouds during SUCCESS, Geophys.Res. Left., in press, If some Ëaction of the aircraft exhaust particles are 1997. effective ice nuclei, then some ice crystals should nucle- Peters, J. L., New techniquesfor contrail forecasting,AWS/ TR-93/001, AD-A269 686, pp. 31, Air Weather Service, ate in the exhaust plume even when only ice supersat- Scott Air Force Base, Ill., 1993. uration is reached. Attempts were made to sample ex- Schumann, U., J., On conditions for contrail formation from haust plumeswithout contrailsduring SUCCESS;how- aircraft exhausts, Meteorol. Z., 5, 4-23, 1996. ever, the plumes were very difficult to find without a Schumann, U., J., et al., In situ observationsof particles visible contrail present, so relatively few clear plumes in jet aircraft exhaust and contrails for different sulfur- were sampled. Apparently, no plumes were sampled containingfuels, J. Geophys.Res., 101, 6853, 1996. Twohy, C. H., A. J. Schanot, and W. A. Cooper, Measure- under ice-supersaturated and water-subsaturated con- ment of condensedwater content in liquid and ice clouds ditions. Hence, all we know for certain now is that usingan airborne counterflowvirtual impactor, J. Atmos. visible contrails do not form when Ttc< T < Tic. It is Ocean. Tech., 1•, 198-202, 1997. possiblethat some ice crystals form under these condi- Twomey, S., Pollution and the planetary albedo, Atmos. tions but too few for a visible contrail. In future field Environ., 8, 1251-1256, 1974. studies of exhaust plumes and contrails, it would be in- teresting to investigatethis regime to see whether ice Eric Jensen, NASA Ames Research Center, MS 245-4, crystals are indeed formed under these conditions. Moffett Field, CA 94035,(e-mail: [email protected].nasa. gov)

Acknowledgment. This researchwas supportedby NASAsSubsonic Assessment Program, directed by Howard (ReceivedApril 7, 1997; revisedAugust 25, 1997; Wesoky. acceptedSeptember 25, 1997.)