No. 8 Spring 1976 - University Corporation for Atmospheric Research

INSTRUMENTATION in cloud m i c r o p h y s i c sN A T IO N A L C E N T E R FOR ATM OSPHERI H NUM BER 8 SPRING 197

ATMOSPHERIC TECHNOLOGY NUMBER 8 -SPRING 1976INSTRUMENTATIONIN CLOUD MICROPHYSICSCONTRIBUTORSTheodore W. CannonNicholas J. CarreraHarvey E. DytchA. J. HeymsfieldPeter V. HobbsJames E. JiustoG. Garland LalaLawrence F. RadkeRobert E. RuskinW. David RustPaul Spyers-DuranFrancis M. TurnerScientific Editor: James E. Dye; Assistant Scientific Editor: Thomas G. Kyle.ATMOSPHERIC TECHNOLOGY Steering Committee: Harold W. Baynton, John C. Gi Ile, Paul Julian,Charles Knight, Donald H. Lenschow, Chester Newton, Harry B. Vaughan.ATMOSPHERIC TECHNOLOGY contains invited articles and brief, unsolicited articles that summarize thestate of the art in instrumentation and techniques or describe specific new developments for probingthe atmosphere. Contributions are semitechnical and provide timely information to a broad readershipinterested in the tools for atmospheric research. In some instances their contents will comprisesummaries of papers that have been or will be submitted for publication in a primary journal. Articlesare edited but not refereed or copyrighted, and their inclusion in ATMOSPHERIC TECHNOLOGY is not to beregarded as formal publication.National Center for Atmospheric Research, Boulder, Colorado.

INSTRUMENTATION IN CLOUD MICROPHYSICSContents1 Introduction3 Measuring the Size, Concentration, and Structural Properties of Hydrometeorsin Clouds with Impactor and Replicating Devices— Paul Spyers-Duran10 Cloud Droplet Spectrometry by Means of Light-Scattering Techniques—Harvey E. Dytch and Nicholas J. Carrera17 Particle Size Distribution Measurement: An Evaluation of the KnollenbergOptical Array Probes— A. j. Heymsfield25 Optical Techniques for Counting Ice Particles in Mixed-Phase Clouds—Francis M. Turner, Lawrence F. Radke, and Peter V. Flobbs32 Imaging Devices— Theodore W. Cannon38 Liquid Water Content Devices— Robert E. Ruskin43 Cloud Condensation Nucleus Counters— James E. Jiusto50 Instrumentation for the Detection of Ice Nuclei--- G. Garland Lata57 A Review of Recently Developed Instrumentation to Measure ElectricFields inside Clouds—W. David RustATMOSPHERIC TECHNOLOGY is published semiannually for the NCAR Atmospheric Technology Division, Clifford J. Murino, Director. Facilities of the AtmosphericTechnology Division and their heads are: Research Aviation Facility, Harry B. Vaughan; National Scientific Balloon Facility, Alfred Shipley; Computing Facility, G. S.Patterson Jr.; Field Observing Facility, Robert Serafin; Research Systems Facility, David W. Bargen.Editor: Merry Maisel; Editorial Staff: Elmer Armstrong, K. Redman; Production Editor: Marie Boyko; Art Director: William Hemphill; Designer: Howard Crosslen;Graphics Staff: John Bazil, Lee Fortier, Wilbur Garcia, Justin Kitsutaka, Barbara Mericle, Michael Shibao; Printing Staff: Clarence Cook, Carl Edwards, Arthur Gray,Donald Green, Geraldine Morris, James Puhr.The National Center for Atmospheric Research (NCAR) is operated by the nonprofit University Corporation for Atmospheric Research (UCAR) under the sponsorshipof the National Science Foundation. NCAR is an equal opportunity employer.UCAR Member Institutions: University of Alaska, University of Arizona, California Institute of Technology, University of California, Catholic University of America,University of Chicago, Colorado State University, University of Colorado, Cornell University, University of Denver, Drexel University, Florida State University, HarvardUniversity, University of Hawaii, University of Illinois at Urbana-Champaign, Iowa State University, Johns Hopkins University, University of Maryland, MassachusettsInstitute of Technology, McGill University, University of Miami, University of Michigan, University of Minnesota, University of Missouri, University of Nevada, NewMexico Institute of Mining and Technology, State University of New York at Albany, New York University, Ohio State University, University of Oklahoma, OregonState University, Pennsylvania State University, Purdue University, Rice University, Saint Louis University, Stanford University, Texas A&M University, University ofTexas, University of Toronto, Utah State University, University of Utah, University of Washington, University of Wisconsin, Woods Hole Oceanographic Institution.Published by the National Center for Atmospheric Research, Post Office Box 3000, Boulder, Colorado 80303. Subscriptions free on request.

NUMBER 8 -S P R IN G 1976IntroductionThe increase in our knowledge o f cloud m icrophysicalprocesses and m eteorology in general has o fte n been lim ite dby a lack o f instrum ents capable o f giving us the desiredobservations. C loud m icrophysics requires the m easurement o fconcentrations and sizes ranging fro m 10 1° /m 3 fo r subm icronA itk e n nuclei, through 109/m 3 fo r m icron-sized clouddroplets and about 103/m 3 fo r m illim eter-sized particles, toless than 1 /m 3 fo r centim eter-sized hailstones. In addition todeterm ining size and concentration, it is fre q u e n tly necessaryto be able to distinguish water fro m ice.I f we hope to ob tain measurements over this entire range, anum ber o f d iffe re n t instrum ents m ust be used. Moreover, theobservations m ust be made in the clo ud environm ent, whichcan sometimes be very hostile, and where extrem es inh u m id ity , tem perature, turbulence, and hail can fre q u e n tly befo u n d . Also, the measurements are usually made fro m anairborne p la tfo rm w ith the inherent problem s o f vib ratio n andelectrical noise. An in strum ent which may perform perfectlyin the la boratory may be useless on an a irc ra ft in a cloud.R e lia b ility o f operation in this en viro nm en t m ust be anim p o rta n t consideration in the design and selection o finstrum ents.Because o f the ir great beauty and variety, snow crystalswere the object o f many o f the earliest observations in cloudphysics. The firs t sketches o f snow crystals are said to havebeen drawn by Olaus Magnus, the A rchbishop o f Uppsala,about 1550. The firs t scientific record o f snow crystals waspublished by Descartes in 1635, and the invention o f them icroscope in the last h a lf o f the 17th century led to evencloser exam ination. These early observations were p rim a rilydevoted to determ ining the d iffe re n t types o f particles. Fewstudies were directed at understanding the physical processesu n til the 20th century, and in pa rticular the last 40 years.The airplane has played a m ajor role in the developm ent o fcloud physics during this century. It n o t o n ly made it possibleto make observations d ire c tly in clouds, b u t also provided them o tiva tio n fo r investigating what con d itio n s could beexpected there. Pilots soon learned th a t the accretion o f supercooledcloud droplets could lead to a bu ild up o f ice on thewings, thereby greatly reducing the perform ance o f the a ircra ftand m aking fu rth e r flig h t quite hazardous. The need todeterm ine when such conditions could be expected helpedm otiva te m any o f the early a ircra ft observations.In the beginning, measurements were made fro m opencockpits. Particles were e ith er collected or im pacted on objectssuch as rods covered w ith black velvet o r glass slides coatedw ith o il. A fte r colle ctio n they were exam ined w ith a m ic ro ­scope d ire c tly in the aircraft. Later, im provem ents in sam plingtechniques made it possible to replicate the im pacted particlesso th a t they could be exam ined back in the la boratory.A lthough much o f our understanding o f cloud m icrophysicshas been gained using these im paction o r collectio n techniques,the analysis process is tedious, tim e-consum ing, andsometimes d iffic u lt to in te rp re t. W ithin the past few years, anum ber o f new instrum ents have appeared th a t can giveautom atic, continuous measurements o f certain parameters.A ll o f these new instrum ents hold prom ise o f greatlyincreasing o u r understanding o f the grow th and developm ento f particles w ith in clouds, but, as in the past, in te rp re ta tio nand analysis o f data rem ain tedious and ambiguous processes.There is no substitute fo r careful analysis.In view o f the new developm ents in cloud physics in stru ­ments, it seems p a rtic u la rly appropriate to devote an issue o fAtmospheric Technology to this topic. Several o f the articlesin this issue (fo r exam ple, those by H. D ytch and N. Carrera;A. H eym sfield; F. T urner, L. Radke, and P. Hobbs; andT. Cannon) discuss the newer instrum ents, some o f which haveno t been described previously. O ther articles (those byP. Spyers-Duran and R. Ruskin) describe more standard buts till w idely used techniques. In all o f the articles we haveattem pted to present a balanced view o f the principles o foperation, advantages, lim ita tio n s, and operational re lia b ilityo f these instrum ents, and to give some examples o f data fro mthem . There are additional instrum ents in existence th a t havebeen used w ith varying degrees o f success, but w hich, in theop in ion o f the editors, have serious shortcom ings or are n o t ingeneral use; therefore they have n o t been included in thesearticles.In a d dition to the articles concerned w ith cloud particlemeasurements, there are tw o articles on techniques used in themeasurement o f cloud condensation nuclei (that by J. Jiusto)and ice nuclei (th a t by G. Lala). In the case o f cloud condensation nuclei, we now have fa irly reliable and proven m ethods o fobservation. B ut fo r ice nuclei the story is quite d iffe re n t. Weare s till in a very active stage o f learning and exp erim e ntatio n,but, as the article shows, some understanding is beginning toemerge and some techniques have been developed which maygive us the observations we so badly need fo r a m ore com pleteunderstanding o f the role o f ice-form ing nuclei in clouds.Last, b u t by no means least, there is an article (by W. Rust)on instrum ents cu rre n tly being used fo r the measurement o felectric fie ld in and around clouds. Research in clo ud e le c trificationhas largely been isolated fro m research in cloud m icro ­physics, in spite o f the fa c t th a t e le ctrifica tio n probably doesplay a m ajor role in the developm ent o f p re cip ita tio n in largeconvective clouds. The merging o f the fields o f cloud physics1

ATMOSPHERIC TECHNOLOGYand cloud e le c trific a tio n is one o f the fro n tie rs in cloudphysics research and measurement, and an area in w hich muchmore emphasis should be given to fu tu re in stru m e n t development and research.O ther areas in which continued developm ent is lik e ly tohave considerable im pact on cloud m icrophysics are those o fremote-sensing techniques and com puter analysis o f data usingpa tte rn-reco gn ition techniques. A ltho ug h remote-sensingtechniques are n o t ro u tin e ly used at the present tim e, they arebeginning to appear in special projects and have the potentialo f pro vid in g measurements in clouds fro m ground-basedobservations. Measurements fro m m ultiple-w avelength lidarhave now reached the p o in t th a t ice can be distinguished fromwater; qu an tita tive measurements o f size and con cen tra tionmay soon fo llo w . I t soon should be possible to use dualwavelengthradars to detect the presence and location o f hailand to determ ine liq u id water contents and rainfall ratesw ith in thunderstorm s. D ual-doppler radars are already givingmeasurements o f air m o tio n w ith in p re cip ita tin g clouds, fro mw hich p re c ip ita tio n pa rticle trajectories can be in ferred, andtrip le -d o p p le r radars now being developed w ill give even bettermeasurements, especially o f the im p o rta n t vertical fie ld o fm o tio n .W hile remote-sensing techniques can provide valuablein fo rm a tio n , the need fo r d ire c t measurements fro m a ircra ftw ill u n d o u b te d ly continue. Perhaps the largest presentlim ita tio n o f a irc ra ft measurements is the need fo r hum anjudgm ent in data reductio n. We now have instrum ents th a t cangive us images o f clo ud particles, fro m w hich concentration,size, and shape can be derived (e.g., the Cannon particlecamera and the Particle Measuring Systems’ tw o-dim ensionalo p tical spectrom eter), b u t no way o f rapid ly analyzing thedata. A uto m ated data analysis is a necessity if we are to handleand in te rp re t the enorm ous quantities o f data which theseinstrum ents are capable o f producing. Some w o rk has alreadybeen started using pattern-recognition techniques and photo-densitom etric scanning b u t much remains to be done.A lthough this issue is concerned w ith instrum ents used inmeasuring m icrophysical properties o f clouds, m easurement o fthe tem perature, pressure, h u m id ity , and m o tio n o f the air(both ho rizo ntal and vertical) in and around a cloud is alsoessential fo r understanding its developm ent and behavior.Instrum ents and techniques used in measuring these quantitieshave been discussed in previous issues of AtmosphericTechnology and therefore have n o t been included here. Theinterested reader m ay refer to Atmospheric Technology 4, 6,and 7.£ $James E. DyeJanuary 19762

NUMBER 8 -S P R IN G 1976Measuring the Size, Concentration,and Structural Properties of Hydrometeors inClouds with Impactor and Replicating DevicesPaul Spyers-DuranA irb o rn e investigations o f clouds have been carried o u t bycloud physicists fo r m ore than 30 years. D eterm ination o f thesizes, concentrations, and shapes o f hydrom eteors (liq u id orsolid cloud particles) has been o f central interest, requiring insitu observation o f the developm ent and grow th o f clouds andd istin ctio n am ong the various p re cip ita tio n mechanisms.measured pa rticle size distrib u tio n s fo r the collectio n e fficiencyo f the device. The collectio n efficiencies o f d iffe re n tco lle cto r shapes have been calculated by Langm uir andB lodgett (1946) and by Ranz and Wong (1952). In ad dition,the shape o f the probe and the sam pling s lit make the collectione fficie n cy de term in ation even more d iffic u lt.A t present, the size o f clouds and th e ir relatively short lifecycle demand the use o f an airplane as a sam pling p la tfo rm .Sam pling hydrom eteors using the direct m ethods describedhere involves carrying the probe in to the cloud and capturingthe particles on a suitably prepared and exposed surface. Thehigh speed o f the a irc ra ft introduces d iffic u ltie s and errorsduring sam pling, and no single im p actor or replicating in strument has been devised th a t is capable o f sam pling the entirerange o f particle sizes. The problem s in sam pling 109 clouddroplets ( < 5 0 ^ ) per cubic m eter are quite d iffe re n t fromthose in sam pling ten precipitation-sized particles (> 2 0 0 nm)per cubic meter.The accuracy o f the measurements obtained from a givendirect m ethod is affected by the fo llo w in g :• Collection Efficiency o f the Sampler. Particles approachingthe c o lle cto r w ill tend to fo llo w stream lines and w ill bedeflected around the c o lle c to r to a degree dependent on the irsize. M ost collectors w ill discrim inate m ore against the smallestcloud particles, and therefore it is necessary to corre ct the• Representativeness o f the Sample. The usefulness o f thesample w ill depend on the size o f the volum e o f air requiredfo r the sample relative to the size o f the cloud and the rate o fchange o f the d ro p le t po pu latio n.• Proper Exposure. Coalescence o f droplets can occur onoverexposed slides if the covering fra ctio n exceeds 0.1.• Calibration o f the Impressions. The collected dropletsmay diffuse in to sam pling media, such as oils, th a t arepermeable to water, so im m ediate photography is a must.Also, accurate calibrations are needed in w ind tunnels or w ithw h irlin g arm facilities to relate the size o f impressions to theoriginal drop size, because particles deform upon im pact w iththe collectin g m edium . The degree o f d is to rtio n is dependentupon the nature o f the surface coating and im paction speed(see Fig. 1).• Shattering. Larger particles w ill shatter in to m any sm allerparticles upon h ittin g exposed surfaces at high speeds. Thiscould make the sample very d iffic u lt to evaluate.• Icing. Unheated leading edges w ill accrete ice in supercooledclouds, d is to rtin g the sampler config u ra tio n andchanging the colle ctio n efficie ncy.AuthorPaul Spyers-Duran received a B.S. in meteorology from theUniversity o f Vienna in 1959. After graduate studies at theUniversity o f Chicago, he worked with the Cloud PhysicsLaboratory there for 13 years, developing and testing airbornecloud physics instrumentation. He joined the NCAR ResearchAviation Facility in 1974.In all direct m ethods, a suitable exposed surface records theimpressions or shapes o f the particles. The data reduction islaborious and tim e-consum ing. O ften, hum an in te rp re ta tio n isnecessary to evaluate the q u a lity and representativeness o fsamples. In spite o f these drawbacks, large am ounts o f datahave been reduced, and o u r present understanding o f cloudphysics has com e largely fro m these conceptually sim plem ethods.3

ATMOSPHERIC TECHNOLOGYSlide Impactors• Oil-Coated Slides. One o f the earliest measurements o fcloud d ro p le t sizes was made by Fuchs and P e trja n o ff (1937).A clean glass slide coated w ith a m ix tu re o f lig h t m ineral oiland petroleum je lly was used to capture cloud droplets andkeep them submerged u n til they had been photographicallyrecorded. The m ethod was im proved and used in an a ircra ft byMazur (1943), w ho saturated the m ineral oil w ith distilledwater to prevent the droplets fro m d iffu s in g in to the oil.Slides coated w ith castor oil were used in the firs t extensiveset o f measurements o f d ro p le t sizes in c u m u lifo rm clouds byW eickm an and A u fm Kampe (1953). W ith this m ethod it waspossible to c o lle c t droplets as large as 200 jum in diam eter w ithno apparent shattering i f the im pact velocity was less than100 m/s.IMPRESSION DIAMETER Ifim )Fig. 1 Relation between droplet diameter and impression diameter fo rvarious sampling media.A large am ount o f d ro p le t data was obtained by an automatedsam pler designed by Brown and W ille tt (1955). In the irsampler three slides coated w ith silicone oil moved in rapidsuccession through an airstream and were photographed undera m icroscope in a cold cabin. The results fo r mean d ro p le td is trib u tio n s in trade-w ind and sum m ertim e U.S. continentalcum u li were reported by Braham, Battan, and Byers (1957)and by Battan and Reitan (1957).Fig. 2 Drop impressions in soot layer. (Photo courtesy o f J. Dye,NCAR.)These m ethods gave us the firs t details o f the d ro p le tspectrum at d iffe re n t geographical locations. A disadvantage o fusing them is th a t the sample m ust be recorded im m ediately, aprocedure w hich is d iffic u lt in tu rb u le n t air. It is also necessaryto know the exact tim e th a t elapsed between sam plingand recording in order to apply d iffu s io n corrections.• Magnesium Oxide Method. A n o th e r w idely used m ethodinvolves coating a clean glass slide w ith a th in film o f magnesiumoxide. D roplets im pinging on the film leave roundholes which are p ro p o rtio n a l to th e ir size.This technique was developed and calibrated by May (1950)fo r drop diam eters between 10 and 240 ^m . For dropletslarger than 20 ^ m in diam eter, the ra tio o f dro plet diam eter toim pression diam eter was fo u n d to be 0.86; it remains constantw ith d ro p le t size. Data reduction requires the investigator toexam ine the slide surfaces by m icroscope w ith a strong transmitte d lig h t and record the images photographically. Thism ethod is n o t affected by drop d iffu s io n and droplets cannotcoalesce, so the samples can be preserved. One disadvantage isth a t this layer is rather fragile in textu re and can break o ffbecause o f b u ffe tin g by the airstream . A n o th e r is th a t dropsbelow 8 /im in diam eter cannot be sized w ith c e rta in ty becausethe te xtu re and grain size o f the magnesium oxide interfere.Squires and G illespie (1952) used this technique in a gun-typesampler w hich could be reloaded in about 50 s du rin g flig hts.They exposed ten rods 3 mm w ide (yie lding a high collectio n4

NUMBER 8 -S P R IN G 1976e fficie n cy) fro m a magazine every 3 s to study the dro pletspectrum variation inside cum ulus clouds.• Carbon Film. The m ost w idely used substrate is a carbonfilm , because its textu re enables it to w ithstand b u ffe tin g bythe airstream (see Fig. 2). Neiburger (1949), fo r exam ple,coated small glass slides w ith a film o f lam pblack and exposedthem through stratus clouds during flig h ts in a blim p. Hederived the drop determ ination fro m the size o f the ringswhich were obtained in the soot layer. Squires (1958a, b, c)also used this m ethod during his investigation o f continentaland m aritim e cum u li. His technique made possible one o f them ost sig nifica nt advances so far in our understanding o fdifferences between the behavior o f continental and m aritim ecum ulus clouds. Clague (1965) developed an autom atedsampler w hich cou ld expose 18 slides 3 mm wide in rapidsuccession. Warner (1969) used this device extensively tostudy the m icro stru ctu re o f cum ulus clouds over A ustralia.Fig. 3 Drop impressions in gelatin substrate. (Photo courtesy o f]. Jiusto, State University o f New York at Albany.)• Gelatin-Coated Slides. This m ethod o f recording cloudd ro p le t sizes uses a gelatin substrate, as in Fig. 3. Liddell andW otten (1 957) used glass slides coated w ith gelatin containingwater-soluble dye. Cloud and fog droplets im pinging upon theslide dissolve the gelatin, leaving a clear area w ith an intensering caused by the concentration o f dye. Jiusto (1965) firs tused the gelatin sam pling technique in airborne cloud studies.Cloud droplets im pacting on a gelatin-coated slide (w ith o u twater-soluble dye) dissolve the hygroscopic gelatin andredeposit it in a characteristic “ m oon cra te r” manner. Pena etal. (1970) reported a continuous cloud sampler th a t used a16 mm gelatin-coated film capable o f sam pling over a cloudpath length o f m ore than 30 km . Because o f its collectio nefficie ncy, this sampler can o n ly detect drops no sm aller than2.5 nm in diam eter; it is lim ite d to use in clouds w ith liq u idwater conte n t less than 0.35 g/kg.The highly sensitive gelatin m ethod can resolve droplets1 /um in diam eter. It has a m inim al problem w ith drop evaporation and coalescence, although the samples have to beprotected fro m high h u m id ity . Using a phase-contrast m ic ro ­scope during data analysis helps to enhance the images fo rsizing.Foil ImpactorsD roplets larger than 5 0 /im in diam eter are fo u n d in lowcon cen tra tion in natural clouds. The “ single sh o t” samplersfail to collect large droplets because o f the small volum e theysweep o u t du rin g a short exposure (to m eet the 0.1 coveringfa c to r requirem ents). T heir substrates are n o t suitable fo rrecording large droplets, because such droplets often shatter,giving erroneous counts as small droplets on oth e r parts o f thesample.Foil im pactors were developed by various research groups asa means to measure precipitation-sized particles during cloudpenetrations. B row n (1961) perfected a fo il sampler w hich iscapable o f recording droplets larger than 250 jum in diam eterw hile n o t recording the smaller sizes (see Fig. 4). His deviceconsists o f a lead fo il m ounted and supported on a continuouscopper-mesh belt about 5 m long, which is driven past anaperture (1.24 X 1.59 cm ) at a speed o f 1.27 cm/s. Theim pinging drop creates an im p rin t which is a fu n c tio n o f itssize and the airspeed. The smallest im p rin t th a t can berecorded is th a t o f a drop 250 /im in diam eter at an airspeed o f76 m/s.A sim ilar device using alum inum fo il 0.025 mm th ic k wasreported by Duncan (1966). The fo il is exposed over a ridgeddrum fo r 0.4 s by a shutter arrangem ent, then the exposedp o rtio n is w ound past the aperture during the fo llo w in g 6.1 s.This exposure tim e was fo u n d to be the m axim um possible inheavy rain (ap pro xim ately 250 drops per cubic m eter) w ith o u thaving the drop im p rints overlap. The ra tio o f im p rin tdiam eter to drop diam eter was found to be ~ 1 .3 fo r drops o f0.75 - 3 m m ; it increased to 1.5 fo r drops o f 5 mm at anairspeed o f 62 m/s.A sim ilar drop sampler is com m ercially available fromM eteorology Research, Inc., Altadena, C a liforn ia . Schecter andRuss (1970) perform ed a detailed calibra tion between dropsize and im p rin t diam eter on this in strum ent at speeds o f 72and 118 m/s. T hey fo u n d th a t the ratios o f im p rin t to dropdiam eter fo r drops fro m 0.5 to 5 mm in diam eter range from1.0 to 1.30. U n fo rtu n a te ly , the ir paper does n o t indicate w hatthickness or type o f fo il they used.5

ATMOSPHERIC TECHNOLOGYThe fo il im pactors provided o u r firs t drop-size d is trib u tio n sfo r precipitation-sized particles (> 2 5 0 /um in diam eter) innatural clouds. Y e t there are still problem s w ith these devices.One is th a t it is d iffic u lt to obtain a representative sample o flarge drops if they occur in low concentrations. Increasing thesam pling volum e yields an overexposure o f the sm aller drops,resulting in overlapping o f the im p rints. M oreover, du rin g icingcon d itio n s in stru m en t re lia b ility depends on ho w successfullydeicing elements were em ployed. It is n o t always possible tomake a clear d istin ctio n between liq u id and solid h y d ro ­m eteors fro m impressions obtained in a mixed-phase cloud.C a libration o f im p rin ts is n o t available fo r solid particles.Under all conditions, data handling and analysis are tediousand subjective.Replicator DevicesR eplicator devices have been w idely used in cloud physicsstudies. T hey are mechanized sam pling devices using the wellknow n Form var technique to capture and perm anentlyencapsulate cloud particles (see MacCready and T odd, 1964;Spyers-Duran and Braham, 1967; Spyers-Duran, 1972a). Theyu tiliz e a M yla r tape (usually a 16 mm polyester leader) whichis coated w ith a solutio n o f Form var plastic and ch lo ro fo rm .(F o r airborne use, c h lo ro fo rm is preferred since it is n o n fla m ­mable.) The co n tin u o u s ly m oving ribbon o f Form var isexposed in a cloud through a sam pling s lit several m illim eterswide and then carried in to a d ry in g com p artm en t where theplastic sets q u ic k ly . There, the encapsulated particles evaporate,leaving behind perm anent replicas which can be sized andcounted after suitable m agnification. Since it is possible toobtain a continuous record through a cloud, the replicas canprovide small-scale resolution o f changes in cloud m icro stru c­ture. R ecently developed devices have the cap a b ility o fview ing the replicas s h o rtly after exposure so th a t tape speedand Form var thickness can be adjusted in flig h t, as reported byChristensen, K eller, and H a lle tt (1974). A n o th e r m ajoradvantage o f this m ethod is th a t sim ultaneous recording o fcloud droplets and ice crystals is possible (see Fig. 5).Basically these devices are sim ple; however, they require agreat deal o f design com prom ise. In designs in which coatings(a)Fig. 4Impressions o f precipitation-sizedparticles in a lead fo il exposed on theUniversity o f Chicago Lodestar aircraft (a),and in aluminum fo il exposed on the SouthDakota School o f Mines and Technology T - 28aircraft (b). Smallest diameter is 250 vm.(Sample a, courtesy o f E. Brown, NCA R;sample b, courtesy o f C. Knight, NCAR.)(b)Fig. 5Replicas o f supercooled droplets (A), a snow pellet (B), and icecrystal columns (C).6

NUMBER 8 -S P R IN G 1976are applied d ire c tly , they m ust be applied in the rig htthickness and viscosity to avoid being blow n o ff; in designs fo rw hich the tape is precoated, enough tim e m ust be a llo tte d fo rthe Form var to soften before exposure. E vaporation o f thesolvent cools the film to the p o in t th a t am bient m oisture maycondense on it, causing blushing. W ith careful design, condensationcan be elim inated by adding heat to the system, b u texcessive heat can m elt ice crystals also. High airspeeds tend tomake replica tion m ore d iffic u lt: the solution can blow o ff andparticles can shatter du rin g replication. Data analysis islaborious, and hum an in te rp re ta tio n is ofte n necessary duringdata reduction to distinguish poor records fro m good ones andto determ ine areas where droplets are distorted. The largeam ounts o f data generated during each flig h t are hard tomanage.In spite o f all these drawbacks, replicating devices havebecome a very valuable tool in a num ber o f cloud physicsstudies. Cloud d ro p le t spectra have been measured in h o riz o n ­tal and vertical profiles o f d iffe re n t types o f clouds (Spyers-Duran, 1970, 1972a). A study o f the onset o f glaciation wasmade by MacCready and Takeuchi (1968); observations o f icecrystals in a cum ulus cloud seeded by silver iodide wereobtained by W einstein and Takeuchi (1970); and studies o f icec o n te n t in clouds were undertaken by Mossop, O no, andHeffernan (1967). Mossop and O no (1969) have verified cirruscrystal survival in clear air, and Braham and Spyers-Duran(1967) have docum ented seeding o f m iddle-level clouds. Thereplicating device has also made it possible to compareobserved condensation and d ro p le t grow th w ith theory(F itzgerald, 1972) and to measure the e ffe c t o f p o llu tio n onclo ud drop po pu latio n in an urban atm osphere (Fitzgerald andSpyers-Duran, 1973; Spyers-Duran, 1972b; Eagan, Hobbs, andRadke, 1974a, b).the particles in the air w ith respect to the a irc ra ft by reducingthe a irflo w past the sampler.A n ideal decelerator w o uld m eet the fo llo w in g designcriteria:• Gradual deceleration in order to achieve low turbulencedu rin g deceleration (Fragile crystals m ig ht break updurin g rapid o scilla tion.)• Small divergent angle so the ve lo city p ro file remainssym m etrical over the diam eter w ith no flo w Separation atthe walls• S u ffic ie n t decelerator length to allo w particles to bedecelerated to an im paction speed o f less than 15 m /s• K now n velo city at the sam pling p o in t• Deicing o f the leading edges and surfaces when fly in gthrough supercooled clouds.Mossop et al. (1967) and Mossop and O no (1969) designedand evaluated three d iffe re n t types o f decelerators. The bestresults were reported fro m a sam pling tube 3.7 m long, w ith afro n t d iffu ser section having an intake diam eter o f 4.1 cm andflared o u t at an angle o f 6° to an internal diam eter o f 9.9 cm.W ith this device the velocity was reduced fro m 60 to 20 m/sand hexagonal plates, capped colum ns, and fra il stellar crystalswere successfully replicated.Yam ashita (1969) reports a decelerator th a t sets up areverse a irflo w to reduce the ve lo city o f the particles beforereplication.In the hands o f a s k illfu l operator, w ho can make properadjustm ents to sample the variety o f clouds studied at d iffe r­e nt geographical locations, the Form var rep lica to r m ethod hasseveral advantages. There is no problem w ith evaporation ordrop coalescence. It is possible to obtain , fro m slo w -flyin ga irc ra ft or by the use o f decelerators, ice crystal replicas inwhich the features and h a b it o f the original crystal arerecognizable.DeceleratorsR eplicator devices w o uld be an ideal way to study icecrystal h a bit, size, and concentrations in natural clouds ifshattering o f the delicate crystals during replica tion could beavoided. Experience shows th a t on ly the sturdier crystalfo rm s — nam ely, small prisms, bullets, and h o llo wcolum ns— survive im pact and the replica tion process and th a tm ost crystals shatter in to small fragm ents. T o circu m vent thisproblem , devices have been designed to low er the velocities o fIn a detailed study, Hobbs, Farber, and Joppa (1973)reported a new approach to decelerators, one in which thedeceleration occurs upstream o f the e n try duct. T w o types o fdecelerators are discussed. One is a rectangular du ct 23 cmlong w ith a constant w idth o f 12.7 cm . The area o f the replication p o in t is five tim es th a t o f the e x it p o in t, thus the velocityproduced at the replication p o in t is o n e -fifth o f the freeairstream velo city. This device was used on a slo w -flyin gaircra ft.The second decelerator was used on a faster fly in g B - 23aircraft. In this device air deceleration is contro lle d over alonger length (91 cm ). The device has a circu lar cross sectionw ith entrance and e x it cones, and a velo city section 30 cmlong to decelerate the particles to o n e -fifth o f airstreamvelocity.Both decelerators were tested in a w ind tunnel where thevelocity p ro file was determ ined. The authors do n o t givecollectio n efficiencies o f the decelerators fo r various types o f7

ATMOSPHERIC TECHNOLOGYice crystals, so accurate values o f concentra tion cannot bedeterm ined. T h e ir paper does include descriptions o f severalsamples w hich were obtained fro m an aircra ft, in d ica tin g thatice crystals larger than 1 m m in diam eter can be collectedw ith o u t fragm e ntatio n.A n o th e r decelerator was designed and tested by Davis andVeal (1974) fo r a 10:1 reduction in v e lo c ity ; its straightsection o f 102 cm provides s u ffic ie n t length to allo w largedendrites and plates to be decelerated by at least a fa c to r o ffive. The u n it is 230 cm in length. The entrance cone w ith asemi-angle o f 4° is 98 cm in length; the straight section has adiam eter o f 20 cm ; and the e x it cone is 30 cm long. Thisdevice was extensively tested in a w ind tunnel to obtain itsve lo city p ro file and colle ctio n efficiencies. The collectio nefficiencies were fo u n d to exceed 0.9 fo r particles larger than75 nm and were less than 0.06 fo r diam eters sm aller thanabout 15 /Um. Samples obtained w ith this decelerator show thedelicate structure o f crystals preserved through replica tion.ReferencesBattan, L. J., and C. H. Reitan, 1957: D ro p le t size measurementsin convective clouds. In Artificial Stimulation o f Rain(H. W eickm ann and W. S m ith, Eds.), Pergamon Press, Inc.,E lm sford, N .Y ., 1 8 4 -1 9 1 .Braham, R. R. Jr., L. J. Battan, and H. R. Byers, 1975: A r t if i­cial nucleation o f cum ulus clouds. Met. Monogr. 2 (11),4 7 - 85.------------- , and P. A. Spyers-Duran, 1967: Survival o f cirruscrystals in clear a ir ./. Appi. Met. 6, 1053 - 1061.B row n, E. N., 1961: A continuous-recording pre cip ita tio nsampler. J. Appi. Met. 18, 815 - 818.---------- , and J. W. W ille tt, 1955: A three-slide clo ud d ro p le tsampler. Bull. Am. Met. Soc. 36, 123 - 127.Decelerators can also be used fo r d ire ct capture o f iceparticles in subfreezing silicone o il. T hey can then be broughtback to the la boratory fo r studies as reported by Schreck,T o u te n h o o fd , and K n ig h t (1974).Table 1 summarizes the practical lim its w ith in which dropsamples can be obtained by the various measurem ent techniques.The low er lim it depends on the collectio n e fficie n cy o fthe sam pling device, the upper lim it on im paction speed andon drop breakup, w hich is also a fu n c tio n o f substratethickness.Christensen, L., V. K eller, and J. H a lle tt, 1974: Ice particlecon cen tra tion in prehail convective cloud in Colorado. InP rep rin t V o l., Conference on C loud Physics, held 21 - 24O ctober, Tucson, A riz .; A m erican M eteorological Society,Boston, Mass., 191 -1 9 4 .»Clague, L. F., 1965: A n im proved device fo r obtain ing cloudd ro p le t samples. J. Appl. Met. 4, 549 - 551.Davis, C. I., and D. L. Veal, 1974: Interim Progress ReportNo. 9, D epartm ent o f A tm ospheric Research, U niversity o fW yom ing, 10 pp.OilSubstrateTable IS i/e Range o f Various Measurement TechniquesMagnesium oxideCarbonGelatinLead fo ilAlum inum fo ilFormvarDrop SizeDiarneier (jum)4 - 2008 - 2404 - 1 ,0 0 02.5 - 502 5 0 - 5,000500 -5 ,0 0 04 - 50Cloud [)rop>YesYesYesYesNoNoYesPrecipitationSliced DropsDrizzle sizeDrizzle sizeDrizzle sizeNoYesYesNoDuncan, A. D., 1966: The m easurement o f shower rainfallusing an airborne fo il im p a c to r./. Appi. Met. 5, 198 - 204.Eagan, R. C., P. V. Hobbs, and L. F. Radke, 1974a: Measurementso f cloud condensation nuclei and cloud d ro p le t sized is trib u tio n s in the v ic in ity o f forest fires. / . Appl. Met. 13,553 - 557.------------- , --------------, a n d --------------, 1974b: Particle emissionsfro m a large K ra ft paper m ill and th e ir effects on the m icro ­structure o f warm c lo u d s ./. Appi. Met. 13, 535 - 552.Fitzgerald, J. W., 1972: A Study o f the Initial Phase o f CloudDroplet Growth by Condensation. Technical N ote No. 44,Cloud Physics La bo ratory, U niversity o f Chicago, 144 pp.(Clearinghouse fo r Federal Science In fo rm a tio nPB 211322).------------- , and P. A Spyers-Duran, 1973: Changes in cloudnucleus con cen tra tion and clo ud d ro p le t size d is trib u tio nassociated w ith p o llu tio n fro m St. L o u is ./. Appi. Met. 12,511 - 516.

NUMBER 8 -S P R IN G 1976Fuchs, N., and I. P etrjanoff, 1937: M icroscopic exam inationo f fog and rain droplets. Nature 139, 1 1 1 -1 1 2 .Hobbs, P. V ., R. J. Farber, and R. G. Joppa, 1973: C o llectiono f ice particles fro m a irc ra ft using d e celerators./. Appl.Met. 12, 522 - 528.Jiusto, J. E., 1965: Cloud Particle Sampling. R eport No. 6,NSF G - 24850, D epartm ent o f M eteorology, PennsylvaniaState U niversity, 19 pp.Langm uir, J., and K. B. B lodgett, 1946: A MathematicalInvestigation o f Water Droplet Trajectories. A rm y A ir ForceTechnical R eport No. 5418, 65 pp. (N ational TechnicalIn fo rm a tio n Service PB - 27565). A vailable fromX L Lib ra ry o f Congress P hotoduplicatin g Service,W ashington, D.C.Liddell, H. F., and N. W. W otten, 1957: The detection andm easurem ent o f water droplets. Q. J. R. Met. Soc. 83,263 - 266.MacCready, P. B. Jr., and C. J. Todd, 1964: C ontinuouspa rticle sampler. J. Appl. Met. 3, 450 - 460.------------- , and D. M. Takeuchi, 1968: P recipitation in itia tio nmechanism and d ro p le t characteristics o f some convectiveclo ud c o re s ./. Appl. Met. 7, 591 - 601.May, K. R., 1950: Measurements o f airborne droplets bymagnesium oxide m e th o d ./. Sci. Instrum. 27, 128 - 130.Mazur, J., 1943: An Investigation on the Number and SizeDistribution o f Water Particles in Natural Clouds. M eteorologicalResearch Paper No. 109, M eteorological O ffice ,London.Mossop, S. C., and A. Ono, 1969: Measurements o f ice crystalcon cen tra tion in c lo u d s ./. Atmos. Sci. 26, 1 3 0 - 137.------------- , ------------- , and K. J. H effernan, 19 6 7 : Studies o f icecrystals in natural c lo u d s ./. Rech. Atmos. 3, 45 - 64.Neiburger, M ., 1949: R eflection, absorption and transmissiono f isolation stratus cloud. / . Met. 6, 9 8 - 1 0 4 .Pena, J., R. de Pena, R. L. Lavoie, and J. Lease, 1970: Ac ontinuous clo ud sampler. In P reprint V o l., Conference onC loud Physics, held 24 - 27 August, Ft. C ollins, C olo.;Am erican M eteorological S ociety, Boston, Mass., 93 - 94.Ranz, W. E., and J. B. Wong, 1952: Im paction o f dust andsmoke particles on surface and body collectors. Ind. Engng.Chem. 44, 1371.Schecter, R. M ., and R. G. Russ, 1970: The relation betweenim p rin t size and drop diam eter fro m an airborne drops a m p le r./. Appl. Met. 9, 123 - 126.Schreck, R. I., V. T o u te n h o o fd , and C. A. K nigh t, 1974: Asim ple, airborne ice particle c o lle c to r./. Appl. Met. 13,949 - 950.Spyers-Duran, P. A ., 1970: Observations o f m icro stru ctu re intw o cum u li. In P reprint V o l., Conference on C loud Physics,held 24 - 27 August, F t. Collins, C olo.; Am erican M eteorologicalSociety, Boston, Mass., 173 - 174.------------- , 1972a. Systematic Measurements o f Cloud ParticleSpectra in Middle Level Clouds. Technical N ote No. 43,C loud Physics Laboratory, U niversity o f Chicago, 54 pp.(Clearinghouse fo r Federal Science In fo rm a tio nPB 211321).------------- , 1972b: Upwind and Downwind Cloud-Base Microstructure.Technical N ote No. 45, Cloud Physics Laboratory , U niversity o f Chicago, 98 pp. (Clearinghouse fo rFederal Science In fo rm a tio n , PB 218678).------------- , and R. R. Braham Jr., 1967: An airborne continuousclo ud particle replicator. / . Appl. Met. 6, 1108 - 1113.Squires, P., 1958a: The m icro stru ctu re and colloidal s ta b ilityo f warm clouds. Part I. The relation between structure ands ta b ility . Tellus 10, 256 - 261.------------- , 1958b: The m icro stru ctu re and colloidal s ta b ility o fwarm clouds. Part II. The causes and variations in m icro ­structure. Tellus 10, 262 - 271.------------- , 1958c: The spatial variation o f liq u id w ater andd ro p le t concentration in cum u li. Tellus 10, 372 - 380.------------- , and C. A. G illespie, 1952: A clo ud d ro p le t samplerfo r use on aircra ft. Q. J. R. Met. Soc. 78, 387 - 393.W arner, J., 1969: The m icro stru ctu re o f cum ulus cloud. Part I.General features o f the d ro p le t s p e c tru m ./. Atmos. Sci. 26,1 0 4 9 -1 0 5 9 .W eickm ann, H. K., and H. J. A u fm Kampe, 1953: Physicalproperties o f cum ulus clouds. / . Met. 10, 204 -2 1 1 .W einstein, A. I., and D. M. Takeuchi, 1970: Observations o fice crystals in a cum ulus cloud seeded by vertical-fallp y ro te c h n ic s ./. Appl. Met. 9, 265 - 268.Yam ashita, A ., 1969: M ethods o f continuous pa rticle sam plingby a irc ra ft./. Met. Soc. Japan 47, 86 - 97.9

Cloud Droplet Spectrometry by Means ofLight-Scattering TechniquesATMOSPHERIC TECHNOLOGYHarvey E. Dytch and Nicholas J. Carrera, University of ChicagoThe use o f direct sam pling techniques fo r the measuremento f the concentra tion and size d is trib u tio n o f w ater droplets inclouds presents a num ber o f d iffic u ltie s . Sam pling necessarilyinvolves distu rbing the airstream and is anisokinetic, w ithattendant colle ctio n efficie n cy problem s; the technique iso fte n discontinuous and provides data o n ly through a fewsections o f a c lo u d ; and analysis is a slow, tedious, p o st-flig h tprocess. O ptical techniques o ffe r an attractive m ethod o fcircu m ventin g many o f these problem s, and, fo r the smalld ro p le t size range, the use o f the light-scattering properties o findividual droplets to c o u n t and size cloud particles has beenthe basis o f several recent designs.General Theory: Advantages and ProblemsThe pow er scattered per u n it solid angle by a homogeneous,dielectric sphere o f radius r when illum inate d by a plane-parallel beam o f m onoch rom atic lig h t o f wavelength X can becom puted as a fu n c tio n o f scattering angle by the use o f Mietheory. The scattering in te n sity is a com plex fu n c tio n o f theparticle refractive index, absorption c o e fficie n t, size param eter(a = 27T/-/X), and scattering angle, and can be com pletelyspecified o n ly fo r spherical particles. However, fo r transparentAuthorsHarvey E. Dytch is a meteorologist for the Cloud PhysicsLaboratory o f the University o f Chicago, where he earned abachelor’s degree in geophysical sciences in 1971. His presentfield o f interest, in connection with the Metropolitan MeteorologicalExperiment (M ETROMEX), is the influence o f urbanareas on warm cloud microstructure.Nicholas j. Carrera, also currently engaged in M ETRO M EXresearch, is a graduate o f Harvard University and has receivedM.S. and Ph.D. degrees in physics from the University o fIllinois, Urbana. He worked at the National Bureau ofStandards and Brookhaven National Laboratory prior tojoining the University o f Chicago Cloud Physics Laboratory in1973.spheres o f know n com p osition (e.g., water droplets), thescattered pow er integrated over a given solid angle by ac o lle c to r at a p a rticula r scattering angle can be related to thesize o f the scattering pa rticle i f the in te n sity o f the illu m i­nating radiation is know n. This is the basis o f d ro p le t sizing bymeans o f lig h t scattering.I f the scattered pow er as determ ined by Mie theory isintegrated over a p a rticula r range o f scattering angles and isexam ined as a fu n c tio n o f particle size param eter (a) over awide range o f a, one notes a region o f oscilla tion o f the Miein te n sity fu n ctio n s bounded by Rayleigh scattering (r6dependence) fo r small a and geom etrical scattering (r2 dependence)fo r large a. These oscillations are caused by the in te r­ference effects o f re fra ctio n , internal re fle ctio n , and d iffra c ­tio n . The use o f visible o r near-infrared lig h t sources[ty p ic a lly , helium -neon (He-Ne) lasers w ith X = 6328 A areem ployed fo r practical considerations] fo r the sizing o f clouddroplets involves a range o f a fo r which these effects m ay besig nificant. The a m p litu de o f these oscillations and the rangeo f a over which they occur are d ire c tly dependent upon thescattering angle and the solid angle chosen fo r lig h t colle ctio n ,m aking it increasingly d iffic u lt unam biguously to associateparticle size w ith scattered lig h t in te n sity at larger scatteringangles. M oreover, at larger scattering angles, the relativein te n sity o f the scattered radiation-decreases w ith particle size.Thus, insofar as it is com p atib le w ith the restraints imposed bythe sam pling geom etry, small scattering angles are desirable fo rkeeping the scattered signal pow er high and as nearlym o n o to n ic a fu n c tio n o f d ro p le t size as possible.The basic elem ents, then, com m on to m ost light-scatteringd ro p le t spectrom etry schemes, are:• A source o f in c id e n t radiation• A means o f d e fin in g (o p tic a lly or electronically) somep o rtio n o f this illu m in a tin g beam as a sam pling volum ethrough w hich the sampled airstream passes• C o lle cting optics and electronics to gather the radiationscattered at a given angle during d ro p le t transit throughthis sam pling volum e and to convert it in to an electronicsignal• A means o f com paring the relative pow er o f each10

NUMBER 8 -S P R IN G 1976scattering event and associating it w ith a particular size o fparticle• An associated data acquisition system.One im p o rta n t advantage o f a scattering technique lies in itsnonm echanical d e fin itio n o f a sam pling volum e. This allowsisokinetic sam pling in the free airstream outside surfaceboundary layers w ith o u t distu rbing o r m o d ify in g the cloudsample in the m easurement process.The d e fin itio n o f this sample volum e, however, also presentsone o f the m ost severe basic lim ita tio n s to scattering spectrom ­e try . In order to c la rify this, le t us define the “ static sam plingv o lu m e ” as sim ply the w o rkin g volum e under illu m in a tio nw ith in which valid scattering events can occur, definedo p tic a lly when the in stru m ent is at rest w ith respect to theairstream . When the probe is in m o tio n , the “ dynam icsam pling v o lu m e ” is then defined as the p ro d u ct o f the area o fthe static volum e orthogonal to the airstream through whichdroplets pass (the sam pling area), the relative ve lo city o f theairstream, and the response tim e o f the in stru m en t, as lim ite dby the photodetectio n or electronic processing m ethodsem ployed. In general, this dynam ic volum e is far larger thanthe static volum e, and it is the dynam ic volum e th a t is herecalled the sam pling volum e. The sam pling rate o f the in stru ­m ent is ju st the pro d u ct o f the sam pling area and the relativeairstream velocity.A basic lim ita tio n in scattering technique, then, is thein c o m p a tib ility o f m axim izing the sam pling rate in order toachieve satisfactory sam pling statistics (p a rticula rly fo r thelarger droplets) and m in im iz in g the sam pling volum e in orderto ensure th a t scattering occurs at a unique scattering angleand th a t the droplets are in d iv id u a lly sized and counted. I f ahigh sam pling rate is achieved at the cost o f a large sam plingvolum e, sig nifica nt coincidence e rro r can arise fro m m u ltip lescattering events when m ore than one d ro p le t is present in thesam pling volum e.I f the average num ber o f droplets in the sam pling volum e issmall, the p ro b a b ility o f coincidence errors may be determ inedby Poisson statistics. I f P(n) is the p ro b a b ility o f the presenceo f n droplets, N is the total concentration o f droplets per u n itvolum e, V is the sam pling volum e, and m is the average num bero f droplets in the sam pling volum e, then n = NV andThe ra tio o f m u ltip le scattering events, Pm, to valid singlescattering occurrences, Ps, is then^m _ 1 - e M (1 + y ) ^fo r n < 1, i.e., ap pro xim ately the ra tio o f do u b le t to singletevents fo r small /u. Thus, fo r exam ple, concentrations o f 100and 1,000 droplets per cubic centim eter require dynam icsam pling volum es o f less than about 2 X 10“ 4 and2 X 10~5 cm 3, respectively, to obtain o n ly 1% as m anycoincidence errors as valid measurements. So small a sam plingvolum e as 2 X 10 “ 4 cm 3 , fo r exam ple, provides very poorsam pling statistics fo r the larger, less num erous cloud droplets.W hile concentrations o f droplets sm aller than 10 /um indiam eter may be in the hundreds per cubic centim eter,concentration falls o ff ra p id ly w ith size, and droplets largerthan ab o u t 50 jum may num ber o n ly a few per cubic c e n timeter or tw o orders o f m agnitude low er in concentration.Therefore, a scattering instrum ent, if it is designed to operateover a size range encompassing the abundant sm aller dropletsin natural clouds, is necessarily restricted in spatial resolutionfo r the larger cloud particles because o f sam pling statistics.The sam pling volum e can be increased som ewhat to obtainbetter sam pling statistics, w ith resultant increase in c o in c i­dence errors, b u t this may be acceptable.T o take some figures typica l o f dro plet spectrom eterscu rre n tly in use, fo r a sam pling area o f 4 X 10 “ 3 cm 2, anairspeed o f 100 m/s, and an effective in stru m ent response tim eo f 10 ms (particles detected at a rate o f 100 kHz w ould be seenas single scattering events), the sam pling volum e is4 X 10 -4 cm 3. For concentrations o f N = 100 and N = 1,000drops per cubic centim eter, this gives 2% and 23% probablecoincidence errors, respectively. A total con cen tra tion o f1,000 drops per cubic centim eter is rather high fo r m ostclouds; perhaps a m ore typical figure w ould be 500 drops percubic centim eter, which gives a 10% probable coincidenceerror. I f greater accuracy is needed, it may be necessary to usem u ltip le instrum ents, each restricted in d ro p le t size range.W ith in the static sam pling volum e itself, a fie ld o f illu m in a ­tio n o f perfe ctly u n ifo rm in te nsity is desirable, so th a t thescattering signal collected fro m a d ro p le t o f given size isindependent o f the path through the sam pling volum e. Inpractice, however, laser lig h t sources e m it n o n u n ifo rm in te n ­sity profiles, varying fro m a Gaussian d is trib u tio n fo r thelowest mode to higher order modes fo r which the p ro b a b ilityth a t a particle transiting the beam w ill intersect a peakisophote is as high as 90%. This e ffe c t is m ost im p o rta n t aslig h t in te nsity rolls o ff at the edge o f the beam, since a dro p le tpassing through near the edge w ill scatter less light. Anadditional problem derives fro m the shorter path length nearthe edge, shortening d ro p le t transit tim es under flig h tc o n ditions; because o f electronics tim e constants, this mayresult in an effective reduction o f the peak signal am plitudeso f these shorter pulses. The com b in ation o f these effectsresults in “ edge-effect erro rs” : since droplets are n o t m echanicallyconstrained to pass through the center o f the illu m in a tin gbeam, undersizing o f droplets passing through the edgesoccurs. In the absence o f c irc u itry designed to reject edge-11

ATMOSPHERIC TECHNOLOGYscattering events, errors o f the firs t type , at least, may becorrected fo r th e o re tica lly, or through la boratory calibra tionw ith m onodisperse droplets, by tra n sfo rm in g the raw d ro p le tsize d is trib u tio n s obtained, once the pro ba b ilitie s o f in corre ctsizing are know n.A d d itio n a l errors in light-scattering measurements, p a rtic u ­la rly fo r the smallest d ro p le t sizes measured, may arise whenthe m in im um detectable scattering signal am plitude is near thenoise level o f the in stru m en t, both fro m lig h t contam in ationo f the collectin g optics by background radiation and fromslight flu c tu a tio n s in the pow er o u tp u t o f the illu m in a tin gsource. The use o f high-pow er laser lig h t sources, opticalshielding and filte rin g o f the colle ctin g optics, and carefulm o n ito rin g o f laser pow er o u tp u t can m in im ize theseproblem s. F ina lly, precise la boratory d e term in ation o f thedim ensions o f the in stru m en t sam pling area and in -flig h tdeterm in ation o f the true airspeed are necessary fo r accuratemeasurem ent o f absolute d ro p le t concentrations.For size c a lib ra tio n , a m onodisperse beam o f particles ispassed through the in stru m e n t and the response determ ined.R outine generation o f in dividu al water droplets is possible fo rdiam eters o f 20 - 30 yum and larger, w ith size variations o f on lya fe w percent. Glass o r plastic beads can be used fo r smallerdiam eters, and corre ctio n made fo r the difference in index o frefractio n. V ery nearly m onodisperse beads are com m erciallyavailable (w ith , e.g., standard deviation less than 2% o fnom inal size), b u t less accurate sizes may be acceptable in viewo f oth er lim its to the in stru m en t resolution in the sm aller sizecategories.tAs discussed above, droplets which pass near the edge o f thesam pling area w ill be in a less intense p o rtio n o f the lig h tbeam, and w ill be sized as sm aller particles. For a Gaussianlaser mode, as m any as h a lf o f the droplets may be undersized(m ost o f them by o n ly one size category). This d is trib u tio nin to smaller size categories should be determ ined du rin g thecalibra tion by measuring the response fo r know n particle sizesas particles pass through d iffe re n t parts o f the sam pling area.CalibrationThe ideal calib ra tio n fo r any m easuring in stru m en t is tocom pare it w ith a standard and see w h a t it reads. This shouldbe done over the fu ll range o f the in stru m e n t and under actualor sim ulated con d itio n s o f use. K now n in p u t and measuredresponse then determ ine a transfer fu n c tio n fo r the in strum entw hich allow s corre ction fo r all the in s tru m e n t’s errors. Such acalib ra tio n fo r the cloud d ro p le t spectrom eter w o uld involvepreparation o f sized droplets in know n concentrations, w iththese quantities to be varied over the entire measuring range o fthe instrum ent. These “ standard clo u d s” w o uld then be passedthrough the in stru m en t at airspeeds to be used du rin g fie ldmeasurements. In fact, the calib ra tio n should really be done insitu, to take in to account such things as v ib ra tio n , disturbedflo w over a irc ra ft surfaces, electrical noise, and oth er d is tu r­bances th a t may n o t be foreseen in a la boratory calibra tion. A tpresent such a thorough c a lib ra tio n is im practical orim possible. We attem p t, then, to consider w hich parts o f thisideal calibra tion are m ost im p o rta n t, and to arrive at somepractical calib ra tio n procedures.S ingle-droplet spectrom eters make o n ly tw o measurements:they size particles and they c o u n t them . The in fo rm a tio n th a twe desire, though, is size and concentra tion. T o obtainconcentra tion fro m counts, the sam pling area and airspeedneed to be know n. Airspeed is norm a lly measured by aseparate in stru m ent, which has its ow n calib ra tio n problem s.We w ill consider here the calibra tion o f sizing, cou ntin g, andsam pling area. As we shall see, these determ inations are n o t atall independent.The question o f edge effects and undersizing is closelyconnected w ith the problem o f de fining and m easuring thesample area. The effective sample area— the area w ith inw hich a d ro p le t w ill be c o rre ctly sized— differs w ith dro pletsize. For d ro p le t sizes approaching the beam w id th , thepassage o f a d ro p le t center to o far fro m the beam center lineresults in scattering by o n ly part o f the dro plet, and consequentundersizing. In a d d itio n , we m ust rem em ber th a t thefa llin g o f f o f lig h t in te n s ity near the edge o f the beam meansth a t even small droplets near the edge b u t w ith in the geom etricallim its o f the beam w ill be undersized. It is convenient totake the sample area to be the same fo r all sizes o f dropletsand to be th a t area w ith in which any particle scatters lig h tabove the noise level. Careful surveying o f this area w ith sizeddroplets can then determ ine the fra c tio n and degree o f undersizingo f a u n ifo rm con cen tra tion o f droplets. This w illn o rm ally vary w ith d ro p le t size, and can be applied as acorre ction to the indicated size spectrum to corre ct both fo rin te n sity fa llo ff and fo r partial d ro p le t scattering near theedge.Besides scattered lig h t in te nsity, the other param etermeasured by the in stru m e n t is counts per u n it o f tim e. Sincethe tim e base fo r electronic logic circu its may be con tro lledq u ite accurately, no errors w ill arise here as long as thesam pling rate is low . When the airspeed and d ro p le t concentration com bine to create overlap or near-overlap o f particlepassage through the sample area, counts w ill be lost, asdiscussed above. When this occurs, size in fo rm a tio n w ill be lostas w ell. It may be possible to determ ine w hat percentage o fcounts is lost due to coincidence, b u t if coincidence errors areappreciable, the indicated spectrum cannot be corrected fo rthe errors w ith o u t m aking some assumptions about the true12

NUMBER 8-S P R IN G 1976size spectrum . Such assumptions may be inappropriate andmisleading. Every a tte m p t should be made to operate w ith inthe lim ita tio n s imposed by the in strum ent design, andestimates o f coincidence error should be quoted as a q u a lific a ­tio n to the in te rp re ta tio n o f published data, rather thanapplied as a “ c o rre c tio n ” to obtain data which are thenclaim ed to be more accurate.Again, careful size calibra tion, sample area determ ination,and a tte n tio n to the operating lim its o f the in stru m ent areo n ly an approach to the fu ll-b lo w n calibra tion over all sizes, allconcentrations, and all airspeeds fo r which the d ro p le tspectrom eter w ill be used. In a d d itio n , errors may arise afterin sta lla tion on the a ircra ft fro m noise pickup , vib ratio n,fatigue, tu rb u le n t flo w , and so on. P ost-installation checks oncalibra tion m ig ht include sam pling passes through sim ilarclouds at d iffe re n t airspeeds to check on possible coincidenceeffects, and ground checks w ith a “ standard” or at leastreproducible dro plet spray, to check on day-to-day consistency.Size calibra tion can also be checked in the fie ld w ithsized water droplets or beads.beam enters and leaves the airstream may be a problem duringpasses through p re cip ita tio n o r heavy clouds. In practice, aslight recessing o f these optical surfaces below the protectiveenclosure seems to keep them dry in all b u t the heaviestp re c ip ita tio n . Condensation on external optics may occur infly in g fro m colder to warm er air, but is lik e ly to be o f shortduration. In supercooled o r m ixed clouds, deicing o f structuralin stru m en t parts w ill be necessary fo r prolonged flig h t. Underthese con ditions, disturbed a irflo w may occur through thesam pling area, and optical alignm ent changes may result fromincreased strain on external members which are n o t deiced.E lectrical noise is a general problem w ith electronic in stru ­m entation, and the usual a tte n tio n to shielding, voltage regulation , separation fro m noise sources, and so on w ill help reducenoise p ickup. A recent discussion o f a irc ra ft electrical noiseaffectin g atm ospheric measurements can be fo u n d in Ruskinand S cott (1974).Optical Cloud Particle SpectrometerField ProblemsThe norm al problem s o f alignm ent and s ta b ility o f opticalsystems are exacerbated in an airborne environm ent, w ith itshigh vib ratio n levels and rapid pressure and tem peraturechanges. A simple and rugged design can help m inim ize theseproblem s, b u t some adjustm ent c a p a b ility is necessary ifalignm ent is affected by vib ra tio n or airfram e flexing . Signalin te nsity may be affected by aging o f the lig h t source, voltagefluctu ation s, alignm ent changes, o r d irt on optical surfaces.This can be m onitore d and p a rtia lly compensated fo r by usinga reference signal fro m the lig h t source to com pare w ith thescattered signal. W etting o f optical com ponents where theThe optical clo ud particle spectrom eter (Ryan et al., 1972)is m anufactured by E nvironm ental Research and Technology,Inc., o f C oncord, Massachusetts. Its present config u ra tio nFig. 1 Diagram o f the cloud droplet spectrometer.(Courtesy o f H. Biau, Environmental Research andTechnology, Inc.)13

ATMOSPHERIC TECHNOLOGY(shown schem atically in Fig. 1) uses a 4 mW He-Ne gas laser(6328 A ) m ounted inside the a irc ra ft as its lig h t source. Thelaser o u tp u t is focused through a beam -defining aperture andthen reflected and re-imaged o u t o f the a irc ra ft o n to a m irro rm ounted in an e x te rio r cow l assembly. The scattering detectorused is an S - 20 p h o to m u ltip lie r, in fro n t o f w hich are locateda bandpass interference filte r th a t is 100 A wide (centered at6328 A ) and a scattering detector aperture th a t is reflectedand re-imaged outside the a irc ra ft in to the sample space. Theintersection o f this detector aperture image w ith the laserbeam image at about 45° defines a 1.3 X 10 -4 cm 3 staticsample volum e about 15 cm o u t fro m the a irc ra ft skin, thesam pling area o f which is 0.4 m m 2, norm al to the flig h t direction . Scattering signals fro m the p h o to m u ltip lie r are fed to apulse-height analyzer, operating as a peak detector, whichdirects each scattering event in to one o f 12 channels correspondingto 12 dro plet size ranges, covering diameters from4.4 to 110 jum, varying in w id th fro m 1.5 to 32.0 n m. The useo f an e x te rio r cow l w ith blackened inner walls, together w iththe interference filte r in fro n t o f the p h o to m u ltip lie r,m inim izes the lim itin g e ffe c t o f stray background radiation.The laser and detector w indow s and the e x te rio r cow l-m ounted m irro r are recessed and canted at about 10° to thea irflo w to avoid w etting.Fig. 2 Diagram o f the axiaiiy scattering spectrometer probe.(Courtesy o f R. Knollenberg, Particle Measuring Systems, inc.)G round calibra tion is accom plished by recording theresponse o f the in stru m en t to water droplets o f selected sizeproduced by applying a d ire ct curre nt potentia l to a h y p o ­derm ic needle connected to a water reservoir. The dro pletbeam is directed through the sample volum e and adjusted togive m axim um scattered signal as recorded by an oscilloscope.Sim ultaneous photographs o f the oscilloscope trace andphotom icrographs o f the sample volum e its e lf are made byslaving a stroboscope th a t illum inates the sample volum e tothe scattering pulse. A calibra tion curve is then constructedfro m these photographs, relating d ro p le t size to o u tp u tvoltage. C orrections fo r edge-effect errors are determ ined byscanning the generated m onodisperse d ro p le t beam at aconstant speed through ten equal horizontal increm ents andrecording the resultant d ro p le t cou nt. By p e rfo rm in g thisexp erim e nt fo r three d ro p le t diameters near the low , m iddle,and upper size ranges fo r a given channel, correction factorsare determ ined. Using this procedure, Ryan et al. (1972)determ ined th a t 50% o f the droplets were pro perly sized, 25%were counted in the n e x t low er channel, 8% in the ne xtchannel, and the rem aining 17% alm ost evenly distribu te d inthe next five channels, w ith no dependence o f correctionfa c to r on d ro p le t diam eter fo r the 30 - 60 /um drops used. Thesam pling area is determ ined exp erim e ntally by placing a planediffu ser in the static sam pling volum e perpendicular to theflig h t directio n and m easuring its illum inate d area; the errorassociated w ith this determ in ation is estim ated to be ±5%.E rror in determ ining the size d is trib u tio n , estim ated by theFig. 3 A xia lly scattering spectrometer probe (black mast on le ft)m ounted on University o f Chicago Cloud Physics Laboratoryairplane. Silver probe (to rig ht o f spectrometer) is an earlier,prototype scattering probe.

NUMBER 8 -S P R IN G 1976observed scatter in calibra tion measurements using m onodispersedroplets, is th o u g h t to be ±15% o f the reported value fo reach size interval.Axially Scattering Spectrometer ProbeA 1 mW He-Ne laser (6328 A), operating in a high-orderm u ltim o d e state, provides the source o f illu m in a tio n o f theaxially scattering spectrom eter probe shown schem atically inFig. 2 and m ounted on an aircraft in Fig. 3; the in stru m en t ism anufactured by Particle Measuring Systems, Inc., o f Boulder,C olorado. The laser tube, m ounted w ith the detector electronicsand associated optics in a small external a irfo il,projects a beam o f lig h t which is focused to appro xim ately200 jum in diam eter in the center o f a sam pling aperture in thea irfo il located about 51 cm fro m the a irc ra ft skin. Thescattered energy fro m droplets passing through the laser beamin the sam pling aperture is collected by a pair o f condensinglenses sandw iching a 6328 A interference filte r, w hile theunscattered laser beam is intercepted by a central stop on thefirs t collectin g lens. A 50% beam -splitter prism then dividesthe scattered lig h t in to a d ire ctly transm itte d p o rtio n , which isfocused on the scattering signal ph otodio de, and a reflectedp o rtio n w hich passes through the 90° prism face. A centralstop on this face defines an annular colle ctin g fie ld , allow inglig h t transm ission fro m d ro p le t scattering on ly when thedroplets are s u ffic ie n tly displaced fro m the o b je ct plane in thecenter o f the sam pling aperture. By com paring the am p lifie dsignals fro m the scattering and annulus detectors, a givenscattering event is rejected or accepted as valid, depending onw hether it is w ith in a 4 - 5 mm electronically determ ineddepth o f fie ld . T his depth-o f-fie ld lim ita tio n thus defines astatic sam pling volum e o f ab out 1 X 10“ 4 cm 3 in the center o fthe sam pling aperture w ith a sam pling area o f approxim ately7 X 10~3 cm 2 .The scattering signal fro m droplets w ith in the proper deptho f fie ld is then passed to a pulse-height detector whichcompares the am plitude o f the signal to a reference voltagederived by m o n ito rin g the laser o u tp u t w ith a referencep h o to d e te cto r in back o f the laser to cancel o u t changes inillu m in a tin g in te nsity. A square-weighted resistor array,measuring m axim um pulse am plitude, and a string o f voltagecom parators associate each scattering event w ith one o f 15 sizeclasses 2jum wide, covering d ro p le t diameters fro m 2 to30 jum. The in stru m en t also has a range-switching o p tio np e rm ittin g the user to choose ad ditional size ranges o f0.5 - 7.5, 1 - 15, o r 3 - 45 /um; the size range lim its are o n lypro pe rly linearized fo r the 2 - 30 and 3 - 45 /im size ranges,however. The prim e in stru m en t calibra tion, in the 2 - 30 jumrange, is perform ed in the la boratory using water droplets andglass beads w ith a theoretical corre ction fo r refractive index.A n o th e r o p tio n available is edge-effect reject c irc u itry th a tmeasures scattering pulse w idths and compares the runningmean particle transit tim e to the instantaneous transit tim e o feach droplet. O nly those scattering events w ith pulse w idthslong enough to have passed through the central 62% o f thelaser beam diam eter are considered valid. W ith this optionalc irc u itry , size errors are ±10% or ±2 /um, whichever is greater,w hile the static sam pling volum e is effe ctive ly reduced toabout 4 X 10 _s c m 3 and the sam pling area to 4 X 10~ 3 cm 2 .The in stru m ent is designed to fu n c tio n at airspeeds fro m 0.1to 125.0 m/s and has a m axim um particle rate o f 100 kHz.Particle Instrumentation by Laser Light ScatteringCustom-designed instrum ents fo r measuring water dro p le tspectra are made by E nvironm ental Systems C o rpo ra tion ,K n o xville , Tennessee. W hile those made to date have been fo rpow er plan t cooling-tow er m o n ito rin g , the dro p le t sizes are inthe natural cloud size range, and airborne operation o f suchinstrum ents appears feasible. Solid-state lasers operating in theinfrared have been used in a pulsed mode, w ith 200 ns pulsesoccurring at 300 Hz. Forw ard scattering and pulse-heightdetection are used to size individual drops. The laser, optics,and detector are com pactly m ounted in a cylin d ric a l container(typica l overall dim ensions are about 60 cm by 7.6 cm indiam eter). For one instrum ent, the static sam pling volum e was5 X 10~3 cm 3 and the size range covered was 3 to 100 jum indiam eter. O ther models have been made fo r e ith er larger orsm aller d ro p le t sizes and have ranged fro m 10~6 to 10 cm 3 instatic sam pling volum e. C a libration is done w ith a dro p le tgenerator th a t produces in dividu al m onodisperse drops rangingfro m 30 to 900 /um in diam eter, w ith standard deviations o f5%.Early Cloud Droplet SpectrometersDeveloped in the USSRP rior to the easy a va ila b ility o f small lasers, some interestingcloud spectrom eter designs were developed in the USSR. Theyused incandescent lig h t sources b u t were otherw ise quitesim ilar to the devices described above. Kazas, Konyshev, andL a ktio n o v (1965) used a forw ard-scattering in stru m ent th a tsized droplets over the range o f 20 - 150in diam eter, usingten size categories. The problem o f de fining static sam plingvolum e was treated m echanically, by using an in le t tubealigned w ith the airstream to d ire c t droplets through thesam pling area. A later in stru m en t by Konyshev and Laktionov(1966) used 90° scattering and defined the sam pling volum eby the intersection o f the lig h t source beam and the detectoracceptance beam, sim ilar to the arrangem ent in the opticalcloud particle spectrom eter described above (Ryan et al.,1972). K onyshev’s in stru m en t is described as being usable overthe range o f 25 - 150 fim in diam eter, w ith 20% coincidenceerrors and o n ly 6% o f the droplets in corre ctly sized.15

ATMOSPHERIC TECHNOLOGYIVan de H ulst, H. C., 1957: Light Scattering by Small Particles.John W iley and Sons, Inc., New Y o rk , N .Y ., 4 2 0 pp.ReferencesKazas, V. I., Y u. V . Konyshev, and A . G. L a ktio n o v, 1965:A irb o rn e device fo r the con tinuous m easurem ent o f thedim ensions and concentrations o f large-scale clo ud drops.Izw. Atmos. Oceanic Phys. 7, 710 - 712.Konyshev, Y u. V ., and A. G. L a ktio n o v, 1966: A n airbornep h o to e le ctric measurem ent device fo r clo ud droplets, izv.Atmos. Oceanic Phys. 2, 462 - 464.Ruskin, R. E., and W. D. S cott, 1974: W eather m o d ific a tio ninstrum ents and the ir use. Chapter 4 in Weather and ClimateModification (W. N. Hess, Ed.), John W iley and Sons, Inc.,New Y o rk , N .Y ., 1 3 6 - 205.Ryan, R. T ., H. H. Blau Jr., P. C. von T hiin a , and M. L. Cohen,1972: C loud m icro stru ctu re as determ ined by an opticalclo ud pa rticle spectrom eter. J. Appl. Met. 17, 149 - 156.For Further ReadingBlau, H. H. Jr., D. J. McCleese, and D. W atson, 1970: Scatteringby individual transparent spheres. Appl. Opt. 9,2 5 2 2 - 2528.------------- , M. L. Cohen, L. B. Lapson, P. von T hiina, R. T.Ryan, and D. W atson, 1970: A p ro to ty p e clo ud physicslaser nephelom eter. Appl. Opt. 9, 1 7 9 8 - 1803.H odkinson, R., 1966: Particle sizing by means o f the fo rw a rdscattering lobe. Appl. Opt. 5, 839 - 844.Mie, G ., 1908: C o n trib u tio n s to the optics o f tu rb id media,especially colloidal m etal solutions. Ann. Physik 25,377 -4 4 5 .V onnegut, B., and J. Neubauer, 1952: P roduction o fm o n o -disperse liq u id particles by electrical a to m iz a tio n. }. ColloidSci. 7 ,6 1 6 - 6 1 8 .For Further Information about EquipmentSized BeadsD ow C orning C o rporationB ioproducts D epartm entM idland, M ichigan 4 8 640Particle In fo rm a tio n Services, Inc.P.O. Box 702G rants Pass, Oregon 97526T elephone: (5 0 3 ) 4 7 6 - 9 6 7 6Droplet SpectrometersE nvironm ental Research and T echnology, Inc.696 V irg in ia RoadC oncord, Massachusetts 01742T elephone: (617) 369-8910E nvironm ental Systems Corp.P.O. Box 2525K n o x v ille , Tennessee 37901Telephone: (615) 6 3 7 -4 7 4 1Particle Measuring Systems, Inc.1855 57th C o u rtB oulder, C olorado 80301Telephone: (303) 443 - 710016

ATMOSPHERIC TECHNOLOGYthe optical system. Each second o f particle in fo rm a tio ncollected by the BMS is dum ped in proper fo rm fo r recordingon a nine-track increm ental com p uter tape recorder.SpecificationsFig. 7 Meteorology Research, Inc., Navajo aircraft w ith precipitationprobe m ounted on wing pod.In fo rm a tio n fro m each PC o f the array is acquired in thein stru m en t, and a particle is measured by accum ulating thenum ber o f flip -flo p circu its triggered. This in fo rm a tio n isacquired by a b u ffe r m em ory system (BM S). The u n it particlesize results fro m a determ ination o f the num ber o f elementsset, the size o f each array elem ent, and the m agnification o fThe systems no w being b u ilt are equipped to size in to 15equal-sized channels. There are tw o basic size ranges c u rre n tlyin use: 20 - 300 ;um fo r cloud probes, and generally200 - 4,500 jum fo r p re c ip ita tio n . In the p re cip ita tio n sizerange, some models size between 200 and 3,000 ;um, andothers between 300 and 4,500 ^m , each divided in to 15equal-sized channels. Detailed specifications o f these probestaken fro m PMS handbooks appear in Table 1.The sam pling volum es o f these probes are o f im portance tousers. K nollenberg (1970) used glass beads and opaque disks o fvarious diam eters placed on glass slides to determ ine the irshadow size and lig h t-in te n s ity d is trib u tio n in a parallel planeas a fu n c tio n o f distance. Using coherent illu m in a tio n , hefo u n d th a t the particle was in sharp focus at short distancesand all o f the decrease in lig h t in te nsity was measured w ith in adistance equal to the pa rticle diam eter. One w o uld expect thise ffe ct fro m basic optics (Stone, 1963); fo r nonspherical solidparticles, one w o uld expect th a t at short distances all o f thedecrease in lig h t in te n sity w o uld be in an area equivalent tothe cross-sectional area o f the particle. A t distances far fro mfljCLOUD PROBEPRECIPITATION PROBERange 1 Range 2PowerDimensions1 1 5 V , 50-400 Hz or 60 Hz, 60 WDeice: 70 W/28 V or 100 W /115 V heaters suppliedCylinder: 71 cm long, 16 cm in diameterOptical Extensions (2): 25 cm long, 2.5 cm indiameterSeparation between sampling arms 6.1 cm 6.1 cm 26.3 cmWeight12.6 kgNumber o f size channels 15 15 15Size range 20-300 M m 200-3,000 M m 300-4,500 M mM inimum detectable size 20 ium 200 Mm 300 M mSize resolution 20 M m 200 M m 300 M mMaximum particle rate100,000 HzCoincidence errors less than 0.1% w ith concentrations o f 10“ 3/cm 3(computed fo r 1 0 Mm size)Maximum particle velocitySampling volume (m 3/s)TAS is true airspeed. X is channel numberExamples o f sampling volume (m 3/s for1 s o f sampling at TAS o f 1 00 m/s)Channel 1Channel 5Channel 10Channel 1 5125 m/s(for 20-160 M m particles) 2.0 X 10"8 (23-X)TAS(for 160-300 M m particles) 1.22 X 10“ 8 (23-X)TAS(20 M m ) 4.4 X 10_s(100 M m ) 9.0 X 10“ 4(200 M m ) 1.59 X 10~3(300 M m ) 9.76 X 10~41.22 X 10’ 5 (23-X)TAS 7.95 X 10~6 (23-X)TAS(200 M m ) 0.02691,000 M m ) 0.022(2,000 M m ) 0.016(3,000 M m ) 0.010(300 M m ) 0.175(1,500 M m ) 0.143(3,000 M m ) 0.104(4,500 M m ) 0.06518

NUMBER 8 - SPRING 1976the particle diam eter, the decrease in in te nsity was found to bedistrib u te d over a much greater distance than the particlediam eter. For the measured shadow size to be w ith in ±10% o fthe actual particle size (w hich corresponded to a 40% reduction in lig h t in te nsity) the depth o f fie ld , DF, in nondim ensionalunits was fo u n d to be DF= ±3. The detection thresholdo f a 50% reduction in lig h t level before the circu its are trig ­gered was derived fro m this crite rio n . The depth o f fie ld indim ensional units as a fu n c tio n o f particle radius, r, andillu m in a tio n wavelength, A, is therefore D F= ±3r2/X. For aHe-Ne laser w ith A = 0.6328 jum, fo r a particle 0.2 mm indiam eter DF = 9.4 cm , and fo r a particle 1.0 mm in diam eter,DF = 948 cm . The depth o f fie ld in the sam pling volum ebetween arms o f the probes is indicated nearly to scale inFig. 2.The sam pling volum e is the p ro d u c t o f the depth o f fie ld ,the effective array w id th (which is determ ined by the w id th o fthe array m inus the diam eter o f the particle), and the true airspeed. I f the depth o f fie ld fo r a given particle diam eter isw ider than the separation between sam pling arms, then thedepth o f fie ld is equal to the arm w id th . This is true o f theprobes fo r diam eters larger than 160 p m . The sam pling volum efo r I s o f sam pling at a true air speed o f 100 m /s is indicatedin Table 1. F or the 20 - 100 jum p o rtio n o f the size range, thesam pling volum e is extrem ely small. A probe sizing in therange o f 300 - 4,5 00 /im has a larger sam pling volum e than onesizing in the range o f 200 - 3,000 jum because it has largerseparation between arms. Diagrams showing the separationdistance between arms o f the three d iffe re n t probes aredepicted nearly to scale, w ith the beam in its norm al verticalo rie n ta tio n , in Fig. 2. By way o f com parison w ith otherinstrum ents, the fo il has a sam pling volum e o f 0.14 m 3 at ana irc ra ft speed o f 100 m /s fo r 1 s, and under the same c o n d i­tions the Form var replica tor has one 1.5 X 10"- 3 m 3Mounting LocationsFor unpressurized aircraft, the PMS probes can be m ountedin the free airstream wherever m ost convenient. For pressurizedaircra ft, m o u n tin g is usually m ore d iffic u lt. PMSprobes have been m ounted on pressurized a irc ra ft through anemergency e x it door, above the aircraft, and below the wingsin pods.(a )BEAMsJ2 0 ' 3 0 0 pmU 6 0 pmIDI II II__ I40 pm/ T D rI0 0 20^- | I t -------,------1-J- \ FO CAL-*■--------- ------ t 4/lr, 0;jm „ „ V P LA N EIIIIIiIIILFig. 2 Diagram (a) shows theexternal u n it o f the cioudprobe nearly to scale as i t isoriented in space, and thedepth o f fie ld o f various-sizedspheres. Diagram (b) shows theexternal u n it o f both precipitationprobes (200 - 3,000 nmand 300 - 4,500 tim ) nearly toscale.( b )300 -4500pm3 00 pm V i200pm2 00 -3 00 0 pm200 pm300ymJ71Operating Conditions and ReliabilityM eteorology Research, Inc. (M R I), has instrum ented andoperated six PMS optical array probes on three a irc ra ft duringthe past 18 m onths. The a irc ra ft— a Navajo (shown in Fig. 1),a Cessna C ita tio n jet, and a WB 57F je t— were operatedth ro u g h o u t the U nited States; the la tte r tw o were also

ATMOSPHERIC TECHNOLOGYFig. 3Average ice crystal spectra fo r a 2 min pass measured in a deepcyclonic storm system near Medford, Oregon. Solid lines representmeasurements taken by the precipitation probe. The dashed line is theoverlap o f the cloud probe in the range from 110 to 300 nm. N j i s thetotal number concentration o f particles > 100 nm. The inset at upperle ft represents cloud probe measurements in the size range from 20 to90 nm ; these have been norm alized to 200 nm size intervals fo rcomparison with precipitation probe measurements.9.5 km49.1 °C3.76/m3CLOUDPROBE 8.9 km 7.6 km 6.9 km-41,2°C-38.1 °C34.5° C7.42 X 103/m 39.4 X 103/m1.33 X 104/m4.4 kmm; 6.9 kmCONCENTRATION (N/m3 for the 0.2 mm size class)Truncatedat 90 /jmDIMENSION (Mm)6.3 km30.6° C3.7 X 104/m5.6 km 5.4 km 4.8 km4.4 km19.3°C7.4 X 104/mDIMENSION (mm)20

NUMBER 8 -S P R IN G 1976operated in the Marshall Islands in a w ide variety o f m eteorologicalsituations. A reasonably accurate estim ate o f the rangeo f operating conditions and re lia b ility o f these PMS opticalarray probes can be obtained by lo o kin g through the recordsobtained fo r these six instrum ents.• Operating Conditions. The probes were operated over atem perature range o f +20 to -8 5 ° C . When the a irc ra ft wasferried to its sam pling location w ith the optical array probesn o t operating at tem peratures low er than —60°C, the arrayscracked. However, when the instrum ents were in operation atthese extrem ely lo w tem peratures, we had no problem s w iththe array cracking. In heavy rain o r snow on level passes orascents we had no d iffic u ltie s w ith condensation or pre cip itationsedim enting on any o f the lenses. In rapid descents o f305 - 610 m /m in (1,000 - 2,000 ft/m in ), the heaters wereapparently n o t s u ffic ie n t to keep the lenses fro m fogging, w itha consequent loss o f data.• Reliability. We have fo u n d th a t the probes are n o tm aintenance-free and often require considerable e ffo rt toensure accurate data acquisition. I f careful m aintenance is n o tperform ed, in correct spectra may be obtained. The m ostcom m on problem w ith the probes was m isalignm ent o f thelaser beam. This caused “ noise” on several channels, m ost o fwhich could be filte re d o u t through data processing. O therproblem s included failure o f com ponents on c irc u it boards,which generally caused loss o f data fro m one channel (b u tw hich could be corrected fo r in the processing); occasionally,pow er supplies and lasers burned out, w hich caused com pleteloss o f data. We had continual problem s w ith one o f the cloudprobes which were never resolved.Sample of Data Obtained with ProbesFigure 3 presents data obtained fro m the optical arrayprobes by sam pling w ith the C ita tio n a irc ra ft near M edford,Oregon, in a deep, intense storm system consisting o f all icecrystals. P recipitation and cloud probe data are indicated inthe figure. The d is trib u tio n s s h ift tow a rd the rig h t to largersizes and the concentrations increase considerably fro m 9.5 to4.4 km , indica ting crystal grow th dow nw ard. The d is trib u tio n sdecrease exp o n e n tia lly w ith increasing size. The concentrationsmeasured where the tw o probes overlap do n o t appear tohave the same values. This could have been due to resolutionproblem s caused by nonspherical particles in the firs t size classo f the p re c ip ita tio n probe (w hich is discussed later in thisarticle) and by the small sam pling volum e in the last fewchannels in the cloud probe.C loud probe data at selected levels in the size range fro m 20to 90 p.m appear in the upper le ft hand inset in Fig. 3. Thed istrib u tio n s also decrease in concentra tion exp o n e n tia lly w ithincreasing length.Accuracy of Measurements• Calibration. The three p re c ip ita tio n probes operated byM RI were n o t calibrated according to specifications at thetim e o f delivery. A zoom lens, which is m ounted inside thein stru m ent to co n tro l the m agnification o f the particle imageand therefore the calibra tion, had n o t been set co rre c tly , b u t arelatively sim ple calibra tion brought the instrum ents to specifications.When the probes were delivered to us, they werecalibrated fo r square photodiodes, b u t the photodiodessupplied were rectangular. The sizes o f the firs t fo u r classeshad to be recalibrated. The original arrays have been replacedby arrays which have square photodiodes. From our exp erienceit is desirable to calibrate the in stru m ent before placingconfidence in its accuracy.• Rain. The probes were calibrated fo r spherical particles,and therefore, clo ud droplets and raindrops should be sizedand counted m ost accurately. Tracings o f ty p ic a lly sized clouddroplets and raindrops, and th e ir original dim ensions, appearin Fig. 4, parts A and F, respectively.The cloud probe seems to c o u n t and size cloud dropletsvery accurately. Let us then consider the p re c ip ita tio n probeaccuracy in rain. A n ideal way to estim ate the accuracy o f thisprobe is to sample rain clouds o f d iffe re n t intensities, tomeasure size d is trib u tio n s, to calculate the radar re fle c tiv ityfactor, Z (w hich is related to the sum o f the sixth powers o fthe diam eters), and sim ultaneously to measure the sameparam eter by radar in nearly the same volumes. This was doneby Takeuchi, Peace, and Howard (1975), using the Navajoa irc ra ft and the N ational Severe Storm s Laboratory (NSSL)radar in Norm an, O klahom a. A com parison o f the data appearsin Fig. 5. Each p o in t is a 1 km pass average, w ith 137 to ta lpoints. The dashed line is the 45° line; all points should fall onthis line. The solid line is f i t in the fo rm PMS = A ' + B1(radar),and fits very closely to a 45° line. The accuracy o f the radar is±3 dB Z, and therefore there is excellent agreement betweenthe radar-measured and the probe-calculated radar re fle c tiv ityfa c to r in rain.There are some problem s in evaluating large raindrop sizes,because they become m ore oblate as the ir diameters increase.When the water con te n t and the radar re fle c tiv ity fa c to r arecalculated, the nonspherical nature o f large drops m ust beconsidered.• Ice and Snow Particles. Tracings o f typica l ice and snowparticles as they w o uld be seen when viewed fro m above o r asthey pass through the vertica lly oriented laser beam and th e iroriginal dim ensions appear in Fig. 4 (B-E and G -L). Thede pth-o f-fie ld calibrations o f the probes were made fo rspherical particles. Let us consider the depth o f fie ld o f typicalcloud ice particles and any inaccuracies due to th e ir nonsphericalcross sections.21

ATMOSPHERIC TECHNOLOGYCLOUD PROBE P R E C IP IT A T IO N PROBEo25 pm20 p rrX

ATMOSPHERIC TECHNOLOGY(200 - 3,000 Mm), w hich, at low altitudes, have large num berso f particles in the largest size class. Since, in general, the icecrystal mass w ill go as the second power and the radar reflectivity fa c to r as the fo u rth pow er o f length, the particles n o tcounted are very im p o rta n t to the cloud m icrophysics, watercon tent, and radar re fle c tiv ity factor. The 300 - 4,5 00 /umprobe im proves the m axim um sam pling length, b u t suffersfro m resolution problem s at the lo w end o f the spectrum . Itw o uld be advantageous to have a probe sizing to 2,000 //m in200 Mm steps, and then reduce the resolution fo r particleslarger than 2,000 ^ m to 400 /um steps. Thus, the probe w ouldsize particles fro m 200 to 4,0 00 jum, w ith high resolution,possibly w ith the same electronics.ReferencesBarnes, A. A ., L. D. Nelson, and J. M etcalf, 1974: W eatherdocum entation at K w ajalein Missile Range. In Proc.Conference on Aerospace and Aeronautical Meteorology,El Paso, T e x.; A m erican M eteorological S ociety, Boston,Mass., 66 - 69.H eym sfield, A. J., and R. G. K nollenberg, 1972: Properties o fcirrus generating cells. / . Atmos. Sci. 9, 1 3 5 8 - 1366.• Radar Interference. We attem pted correlating the radarre fle c tiv ity fa c to r calculated fro m the optical array probespectra w ith th a t measured by a radar in nearly the samesam pling volum e. The WB 57F was fly in g in clo ud , a trackingradar beam was fix e d on the aircra ft, and re fle c tiv ity wasmeasured ju st ahead o f the aircraft. There was com pletecontam in ation o f the particle spectra measurements due toradar interference. By p ro vid ing additional shielding o f theprobes and po sitio n in g the radar beam about 150 m ahead o fthe aircraft, the problem was elim inated. Such po sitio n in g o fthe beam resulted in o n ly a 2 s tim e lag between radar andairc ra ft measurements, ensuring sam pling o f the same cloudm icrostructure.ConclusionsThe K nollenberg optical array probes provide a m ethod byw hich large num bers o f clo ud particle spectra can be reducedra p id ly and measurements can be made quite accurately.However, the probes require considerable care and a tte n tio n toensure accurate data acquisition. A ccurate particle spectrameasurements are obtained fo r raindrops; less accurate resultsare obtained in ice clouds. Particle spectra in clouds containingsingle plates, aggregates o f crystals, and b u lle t rosettes can bemeasured accurately; in clouds contain ing colum nar icecrystals, the num ber concentrations and sizes are underestimated and in clouds w ith d e n d ritic crystals, the particles areundersized. The p re c ip ita tio n probe in the size range o f200 - 3,000 jum has a much higher resolution in clouds w ithsmall ice crystals than the 300 - 4,5 00 /um probe; however, the300 - 4,500 /im probe has the advantage o f a larger sizingc a p a b ility , m aking it m ore useful in clouds contain ing large icecrystals.Acknowledgments. The au thor wishes to thank L. Jahnsen,R. Davey, and D. T akeuchi fo r the ir thorough fie ld testing andoperation o f these instrum ents; J. Foreman fo r d ra ftin g thefigures; and H. W ilkinson fo r ty p in g the m anuscript.K nollenberg, R. G., 1970: The op tical array: A n alternative toscattering or e x tin c tio n fo r airborne particle size determ ination ./. Appl. Met. 9 (J), 86 -1 0 3 .------------- , 1973: C irrus-contrail cloud spectra studies w ith theSabreliner. Atmos. Technol. 7, 52 - 55.Nathan, A. M., and L. Bennet, 1966: Precipitation WaterContent Meter. Final R eport No. 1,004, New Y o rk U niversity,New Y o rk , 42 pp.Stone, J. M., 1963: Radiation and Optics. M cG raw -H ill BookC om pany, New Y o rk , 544 pp.Takeuchi, D., R. Peace, and S. H ow ard, 1975: Final Report tothe Bureau o f Reclamation. R eport No. M RI 75 FR-1294,M eteorology Research, Inc., Altadena, C alif., 407 pp.For Further ReadingW ilm o t, R. A ., C. Cisneros, and F. G uiberson, 1974: Highcloud measurements applicable to ballistic missile systemstesting. In Proc. Conference on Aerospace and AeronauticalMeteorology, El Paso, T ex.; A m erican M eteorologicalS ociety, Boston, Mass., 1 9 4 - 199.For Further Information about EquipmentM eteorology Research, Inc.464 West W ood bu ry RoadA ltadena, C a lifo rn ia 91001Particle Measuring Systems, Inc.1855 57th C o u rtBoulder, C olorado 8030124

NUMBER 8 -S P R IN G 1976Optical Techniques for Counting Ice Particlesin Mixed-Phase CloudsFrancis M. Turner, Lawrence F. Radke, and Peter V. Hobbs, University of WashingtonOne o f the firs t attem pts to distinguish ice particles fromw ater drops in the atmosphere was made m ore than 30 yearsago in the T hunderstorm Project (Byers and Braham, 1949),during which it was noted th a t ice particles and water dropsproduced d iffe re n t sounds when they im pacted on the Plexiglascanopy o f an aircraft. This sim ple m ethod has theim p o rta n t advantage o f pro vid ing real-tim e in fo rm a tio n ,although its usefulness is probably confined to particles th a thave already reached p re cip ita tio n sizes. Subsequently,a tte n tio n was directed towards the developm ent and use o fvarious particle colle ctio n and im paction devices (e.g.,Form var replication techniques and metal fo il im pactors),which are discussed in the article by Spyers-Duran in this issue.AuthorsW hile these devices can provide valuable in fo rm a tio n on thetypes o f particles in the air, determ ining the concentrations o fice particles is d iffic u lt: some o f the ice particles fragm ent onim pact when the Form var replica tor is used, and particles250 ^m in diam eter are the smallest th a t can be measuredwhen fo il im pactors are used. Moreover, these devices do n o tprovide real-tim e data on ice particle concentrations andrequire much tedious p o s tflig h t analysis to deduceconcentrations.In the past fe w years, tw o groups (the C loud Physics G roupat the U niversity o f W ashington and Mee Industries, Inc.,Rosemead, C alifornia) have w orked independently on thedevelopm ent o f optical devices fo r the detection and countingo f ice particles in clouds in real tim e fro m aircraft. In thisarticle we describe the m ost recent versions o f these tw odevices and the m ethods th a t have been used to calibratethem , and present a fe w examples o f results obtained in thefie ld .Francis M. Turner was affiliated with the University ofWashington from 1968 to 1975, receiving B.S., M.S., andPh. D. degrees in electrical engineering. He served as a teachingassistant and research associate there, and designed severalinstruments for studying air pollution and cioud microstructure.During the past few years, his interest has centered onthe development o f the University o f Washington’s optical iceparticle counter.Lawrence F. Radke, a research associate professor at theUniversity o f Washington since 1972, received all his academictraining there. He earned an M.S. in 1966 and a Ph.D. in 1968,both in atmospheric sciences. He is presently engaged inairborne studies o f cyclonic storm microstructure, industrialand natural sources o f trace gases and particles, and inadvertentweather modification.Peter V. Hobbs, founder and Director o f the Cloud PhysicsResearch Group at the University o f Washington and a professoro f atmospheric sciences since 1970, joined the university in1963, after earning B.S. and Ph.D. degrees in physics from theimperial College o f Science, University o f London. He hasserved as an associate editor o f several publications and as aNA TO visiting lecturer in Europe, has published a book on icephysics, and is currently collaborating on an introductorytextbook in atmospheric science.Description of the InstrumentsFigure 1 shows a schematic representation o f the latestversion o f the U niversity o f W ashington’s autom atic iceparticle counter (UW -IPC). This in stru m ent uses a linearlypolarized helium -neon laser to illu m in a te any particle thatpasses through the sample p o rt. The prim ary purpose o f theseries o f c o llim a tin g disks is to keep outside am bient lig h tfro m entering the detection system. A polarizing filte r isplaced at the end o f the detection system and is set fo rm axim um e x tin c tio n o f the incident, linearly polarized light.The main beam o f the laser is absorbed in the lig h t trap so th a tdirect lig h t is n o t detected by the p h o to m u ltip lie r tube. Theforw ard-scattered lig h t th a t enters the detection system islim ite d to forw ard-scattering angles o f approxim ately0.5 - 3.5°. An aperture on the p h o to m u ltip lie r-tu b e side o f thep o larizing filte r lim its the azim uthal angles o f the forw a rdscatteredlig h t to ±5° on each side o f the main planes o fscattering. The lig h t then passes through an interference filte r(0.01 /um band-pass at 0.6328 Mm wavelength) and is detectedby the p h o to m u ltip lie r tube.Figure 2 shows a schematic diagram o f the Mee IndustriesModel 120 ice crystal counter (Mee-IPC). A pro je ction lamp isused as a lig h t source. The lig h t passes through an optical

ATMOSPHERIC TECHNOLOGYLaser Interference Filtersystem so th a t linearly polarized infrared lig h t is in cid e n t onparticles passing through the sample pipe. The lig h t sensor (asolid-state device) detects the infrared lig h t th a t is scattered atan angle o f 90° (actually a cone o f lig h t around a scatteringangle o f 90°) after the lig h t passes through an optical systemcontain ing a polarizing filte r set fo r m axim um e x tin c tio n .It can be seen fro m the description o f the tw o instrum entsth a t the UW-IPC detects forw ard-scattered lig h t w h ile theMee-IPC detects lig h t scattered at an angle o f 90°. As we w illsee below , this difference becomes im p o rta n t in discussing themechanisms fo r the detection o f ice particles in the tw osystems.Mechanisms for Ice DetectionThere are three possible mechanisms fo r the detection o f iceparticles in the tw o devices described above (see T urner andRadke, 1973, fo r a m ore detailed discussion):• The ro ta tio n o f the plane o f p o la rizatio n o f the lig h tbeam, produced by the birefring en t pro p e rty o f ice, as itpasses through the ice particles• The detection o f specularly reflected lig h t fro m theexternal faces o f ice crystalsFig. 1Schematic representation o f the University o f Washington'sautom atic optical ice particle counter.• The detection o f lig h t scattered at a preferred angle bythe ice particles.O w ing to the b ire frin g e n t (o r double refracting) p ro p e rty o fice, linea rly polarized lig h t th a t is transm itted through ice w ill,generally, be rotated in such a way th a t although the lig h tMirrorWindowMirrorFig. 2 Schematic diagram o f the MeeIndustries autom atic optical ice particlecounter. The p o d is 45.7 cm (18 in.)long and 17.8 cm (7 in.) in diameter.The cylindrical opening in its center is5.1 cm (2 in.) in diameter.InfraredPolarizingFilterPhotodetectorLight TrapAmplifierHousing ‘26

NUMBER 8 -S P R IN G 1976w hich leaves the crystal w ill s till be linearly polarized, its planeo f p o la rizatio n w ill d iffe r fro m th a t o f the in cident light. Forexam ple, an ice crystal 226 /xm th ic k th a t is o p tim a lly orientedw ill rotate in cident, linearly polarized lig h t through an angle o f90°. The am o unt o f ro ta tio n depends on both the thicknessand o rie n ta tio n o f the ice crystal.I t is clear fro m the geom etry o f the UW-IPC th a t theb ire frin g e n t p ro p e rty provides a direct, firs t-o rd e r detectionmechanism fo r ice particles. However, fo r this mechanism tobe effective in the Mee-IPC, the in cid e n t lig h t w ould have tobe transm itted through the ice, reflected at an internal surface,and transm itte d back through the ice particle. Moreover, inorder fo r a signal to be detected, this lig h t w ould have to e x itat a suitable angle to enter the lig h t sensor.In the case o f the Mee-IPC, this mechanism is alm ostcerta in ly secondary in im portance to specular reflections fro mthe external faces o f ice crystals. The la tte r mechanism suffersfro m the fa c t that, since the refle ction o f lig h t fro m thespecular face o f an ice particle is a relatively discrete event, thedetection o f a signal w ill be strongly dependent on the orientation o f the ice particle. O rie n ta tio n is also im p o rta n t in thecase o f reflections fro m internal crystal faces; however, thetransm ission o f lig h t through the crystal w ill pro bably cause agreater divergence o f the lig h t than external reflections w ill.The im portance o f o rie n ta tio n w ill, o f course, ra p id ly decreasew ith increasing size and c o m p le x ity o f the crystals. Thedetection o f ice particles by reflections is n o t th o u g h t to beim p o rta n t in the UW-IPC, although it is possible fo r glancingreflections to occur.The th ird possible mechanism fo r the detection o f iceparticles is the existence o f a preferred scattering angle.H u ffm an and T hursby (1969) and H u ffm an (1970) havecom pared the lig h t scattered fro m w ater drops and ice crystals.These measurements show th a t the difference in relativescattering fu n c tio n (in term s o f the am o unt o f lig h t scatteredfro m ice crystals) is largest at a scattering angle o f about 100°.This difference is probably due to the external refle ctionmechanism discussed above.depolarized. Thus, rejection o f w ater drops is achieved byc o n fin in g the detection to angles close to the directio n o ffo rw a rd scattering. F urther im provem ent is achieved by meanso f the sectored aperture (Fig. 1) w hich lim its the lig h t enteringthe sensor to azim uthal angles close to the principal planes o fscattering. In com m on w ith the Mee-IPC, the UW-IPC w illdetect water drops o n ly if they are large enough to d iffe rappreciably fro m spheres.Calibration of the InstrumentsA num ber o f calibra tion tests have been perform ed on theUW-IPC. First, w ind tunnel tests have shown th a t there is lessthan 1% difference between the velo city o f air through thesample p o rt and the free-stream air velo city. In ad dition,extensive la boratory tests have been carried o u t to determ inethe sensitivity o f the in stru m ent to w ater drops. These testshave shown th a t water drops w ith diameters less than 600 jumare n o t detected by the in stru m en t. Above 600 iUm anincreasing percentage o f water drops is counted. F lig h t tests infa irly heavy rain have shown th a t w hile some raindrops aredetected, generally the cou ntin g rate is less than 0.1 drops perlite r, w hich is the m in im u m calibrated cou n tin g rate (theUW-IPC is calibrated to cover a countin g rate o f 0.1 to 1,000particles per lite r and has autom atic true airspeed corre ction ).The sensitivity o f the in stru m en t was m aintained at a constantlevel fo r the tests described above as well as fo r the ice-crystalcalibra tion tests and fie ld observations described below.Extensive tests o f the UW-IPC have been carried o u t in acold room . These tests were designed to determ ine thepercentage o f ice crystals o f various types and sizes th a t can bedetected by the counter. Clouds were created by sprayingsmall water droplets in to the cold room and then seeding theclouds w ith dry ice. Ice crystals grew rapid ly in the seededclouds and fell o u t o n to the flo o r o f the cold room . Hand-heldslide replicas were used to determ ine the con cen tra tion , types,and sizes o f the ice crystals. The ice crystals were drawnthrough the UW-IPC w ith an axial blow er, b u t otherw ise thecounter was operated norm ally.The degree to which w ater drops are n o t counted when theypass through the sample ports o f both instrum ents underdiscussion is o f prim e interest. The basic reason w hy waterdrops give signals o f a m uch smaller m agnitude than iceparticles o f the same size in the Mee-IPC is understandable interm s o f the external refle ction mechanism, since the sphericalsurfaces o f w ater drops w ill produce much sm aller signals thanthe specular faces o f sim ila rly sized ice particles. The reasonfo r the rejection o f water drops in the UW-IPC is m ore subtle.In general, when incident, linearly polarized lig h t is scatteredfro m a spherical water drop, it is e llip tic a lly polarized.However, the lig h t scattered in the fo rw a rd directio n (0°scattered lig h t o n ly ) and in the tw o principal planes is notFigure 3 shows the results o f m any cold-room tests w ithhexagonal plate crystals. Shown in this figure is the percentageo f crystals detected versus th e ir m axim um dim ensions. It canbe seen th a t when the m axim um dim ension o f the crystal is100 /im , about 35% o f the crystals are detected. I f the p lo t isextrapolated linearly to the p o in t where 100% o f the crystalsare detected, a m axim um dim ension o f 250 /im is obtained (itwas n o t possible to achieve this size in the cold room becausethe crystals fell o u t before they attained it). The results o fsim ilar tests w ith solid hexagonal colum nar ice crystals areshown in Fig. 4; the results were sim ilar to those fo r hexagonalplates.27

ATMOSPHERIC TECHNOLOGY• Some plates and short colum ns w ith m axim um dim ensionso f 50 - 100 Mm were detected.• The signals generated in the counter by rim ed or o th e r­wise distorted cryscals were substantially sm aller than thoseproduced by regular crystals. (This is to be expected in view o fthe assumed mode o f detection by specular and internalreflections.)• The use o f cross-polarizing filte rs resulted in slightim provem ent in the a b ility o f the counter to discrim inatebetween ice and water.Fig. 3 Percentage o f hexagonal plate ice crystals detected by theUW-IPC.• W ater droplets dow n to 100 Mm in diam eter weredetected; however, o n ly drops larger than 1 mm in diam eterproduced signals com parable to those produced by50 -1 0 0 Mm ice crystals.In a d d itio n , flig h t tests o f the Mee-IPC have shown th a t itcan be operated to exclude m ost water drops occurring in lightrain (Sax and W illis, 1974). However, heavy rain fre q u e n tlyproduced countin g rates equivalent to a fe w ice particles perlite r. (The Mee-IPC does n o t provide a direct readout o f icepa rticle con cen tra tion , b u t counts fo r fix e d tim e periods orcon tin u o u sly. The in stru m ent operator m ust convert thecounts obtained in this way to actual concentrations.) Theflig h t tests also showed th a t the smallest ice particle detectedis ab out 250 Mm in m axim um dim ension.Some Field ObservationsA V E R A G E C R Y S T A L M A X IM U M D IM E N S IO N ( ^ tm )Fig. 4 Percentage o f columnar ice crystals detected by the UW-IPC.The UW-IPC has been flo w n on m any data-gathering flig htsduring the past fe w years, in cludin g studies o f cirrus clouds,weather m o d ific a tio n experim ents, and investigations o fcyclo nic storm s in W ashington State.O ther cold-room tests showed th a t the UW-IPC detectssome frozen drops dow n to about 15 /um in diam eter. Rimedice crystals are also detected. The counter responded in thesame way to lig h tly-to -m o d e ra te ly rim ed ice crystals in thecold room as it did to unrim ed crystals.L a boratory calib ra tio n o f the Mee-IPC has been reported bySheets and O dencrantz (19 74 ). The procedures fo llo w e d weresim ilar to those described above, except th a t the critica lquestion o f cou n tin g e fficie n cy was n o t addressed. Thisproblem is made m ore d iffic u lt by the fa c t th a t the counterhas a fro n t panel sensitivity adjustm ent. The results o f Sheetsand O de ncran tz’s calib ra tio n e ffo rts can be sum m arized asfo llo w s:T o illustrate the types o f results obtained, we present somedata obtained on 2 January 1975, du rin g a flig h t in a cyclo nicstorm . The measurements were made in largely stra tifo rmclouds containing some im bedded cum ulus. L iq u id watercontents were measured w ith a cloud water m eter fromJohnson-W illiam s Products, M ountain V ie w , C a liforn ia , andcloud d ro p le t concentrations (from 3 to 45 Mm in diam eter)w ith a m od ifie d a xia lly scattering spectrom eter probe (M odelASSP -1 0 0 ) m anufactured by Particle Measuring Systems,Inc., B oulder, Colorado.Figure 5 shows three series o f measurements obtained on2 January 1975. Figure 5(a) shows measurements obtainednear Seattle at an a ltitu d e o f 2.1 km , where the tem peraturewas —3°C . N o ice particles were detected in the clo ud; theliq u id water co n te n t rem ained at about 0.1 - 0.2 g /m 3 ; and thecloud d ro p le t con cen tra tion rem ained fa irly constant w ith a28

NUMBER 8 -S P R IN G 1976Fig. 5 Measurements o f ice particle concentration (UW-IPC), cloudwater content (lohnson-W illiams meter), and cloud droplet concentration(Particle Measuring Systems ASSP - 100) obtained on 2 January1975 near Seattle (a), near O lym pia (b), and over Hoquiam,Washington (c).m axim um value o f about 500 drops per cubic centim eter.Figure 5(b) shows measurements obtained near O lym pia,W ashington, at an a ltitu d e o f 1.8 km where the tem peraturewas about —4°C. There was no measurable liq u id water andthe d ro p le t concentrations were very low . However, the iceparticle concentrations reached peak values o f ten particles perlite r. Visual observations o f particles im pacting on a black rodshowed th a t m illim eter-sized ice particles were present.Figure 5(c) shows the measurements obtained at an a ltitu d e o f2.1 km over H oquiam , W ashington. The tem perature wasin itia lly —4°C b u t decreased to —6°C du rin g the traverse. Theliq u id water co n te n t rose to a m axim um value o f 1.3 g /m 3durin g the penetration o f a convective cell, and the d ro p le tconcentration rose to about 700 drops per cubic centim eter.The con cen tra tion o f ice particles showed the opposite trend,having a m axim um value o f 100 particles per lite r in the layerclouds and decreasing to a m in im um value o f about tenparticles per lite r in the convective cell.The Mee-IPC has been used m ost extensively by theN ational H urricane and E xperim ental M eteorology La boratoryo f the N ational Oceanic and A tm ospheric A d m in is tra tio n .Shown in Fig. 6 are data fro m fo u r sequential penetrations o fa convective “ b u b b le " rising o u t o f a mass o f tropical cum ulus.The heights o f penetration o f the a ircra ft were changed successivelyin an a tte m p t to remain in the rising bubble. For thefirs t penetration these results also show th a t when the concentratio n o f ice particles increases, the liq u id water contentdecreases, and vice versa. However, in the second penetration(4 m in later) these tw o quantities were varying together in thesame m anner, and by the th ird penetration the cloud wasessentially glaciated.ConclusionsThe tw o autom atic ice particle counters described in thisarticle have the potential to provide, fo r the firs t tim e,accurate and real-tim e measurements o f the concentrations o fice particles in clouds. The instrum ents are already semioperational.However, a few cau tion ary remarks should bemade. First, neither in strum ent is capable o f absolutediscrim in atio n between ice and w ater over the entire range o fsizes at w hich these particles occur in natural clouds. As thesensitivity o f the instrum ents is increased in order to detectsm aller ice particles, the m in im um size o f detected water dropsD IST A N C E (km)29

ATMOSPHERIC TECHNOLOGYFig. 6 M icrophysical data obtained w ithin a growing tropical cumulusbubble at 4.6 km (a), 5.2 km (b), and 5.8 km (c and d). Shown are thecloud liq u id water content (Johnson-Williams meter), ice particleconcentration (Mee-IPC), hydrom eteor water mass, and concentrationo f drops >0.93 mm in diameter (fo il impactor). (Courtesy o f R. Sax;fo r a more complete discussion o f these measurements, see Sax andWillis, 1974.)(a)(b)MASS (g/m3) LIQUID WATER (g/m3) MASS (g/m3) LIQUID WATER (g/m3)1 -3 -2 -1 -SECONDS FROM BEGIN TIME(c)i---------T T T I• L I Q U I D W A T E R — — IC EC O N T E N T Aki P A R T I C L E _CI/ i f i r C O U N T• /AM10 20 30 ' t li. 50i -------1-------1-------1-------r■H Y D R O M E T E O R W A T E R M A S SD R O P .C O N C E N T R A T I O NII60r702.01.51.00.5SECONDS FROM BEGIN TIME(d)>J3mcro3)o -oo zn m>"0>33HOr-mC/5DJ3O ■oO zo33>_l____L> -I—£ I____I____L10 20 30 40 50 600.070SECONDS FROM BEGIN TIMESECONDS FROM BEGIN TIME30

NUMBER 8 -S P R IN G 1976is also lowered. It is our feeling th a t the latest design o f theUW-IPC, described in this article, is approaching the o p tim u min this respect. It can detect some ice particles down to about15 Mm in m axim um dim ension w hile rejecting all water dropsw ith diam eters smaller than 600 jum. Therefore, unless thecloud contains rather large precipitation-sized w ater drops(w hich can be readily determ ined by other means), o n ly iceparticles w ill be counted. Second, it should be noted th a t thecounters do n o t detect all o f the particles dow n to them in im u m threshold size. Instead, as the size o f the ice particlesdecreases the percentage th a t is counted decreases (e.g., theUW-IPC can detect some ice particles dow n to ab out 15 £im inm axim um dim ension, b u t the particles have to be 250 nmbefore they are all counted). T h ird , fu rth e r careful calibrationso f both instrum ents are necessary in order to establish theeffects o f crystal habits, rim ing, and aggregation on the ircou ntin g efficiencies over a wide range o f ice particle sizes.F ina lly, we recom m end tha t, in order to ensure the com p atibility o f fu tu re data collected w ith these instrum ents, they becalibrated together both in the laboratory and in the fie ld .Agreem ent should also be reached as to the levels o f sensitivityat w hich they should be operated so th a t the m in im um size o fice particles th a t can be detected, the countin g efficiencies,and the m axim um size o f w ater drops th a t are rejected areknow n.The preparation o f this paper was supported by theA tm ospheric Sciences Section o f the N ational Science Foundation (G rant G A 40806) and C ontract F 19628 - 74 - C - 0066fro m the A ir Force Systems Com m and.Sheets, R. C., and F. K. O dencrantz, 1974: Response characteristicso f tw o autom atic ice particle counters, / . Appl.Met. 13, 148 - 155.T urne r, F. M ., and L. F. Radke, 1973: The design and evaluation o f an airborne optical ice particle c o u n te r./. Appl.Met. 12, 1309 - 1318.For Further Information about EquipmentMee Ice Crystal CounterMee Industries, Inc.4939 N o rth Earle StreetRosemead, C a liforn ia 91770Cloud Water MeterJohnson-W illiam s ProductsBacharach Instrum ent Co.(D ivision o f A M B A C Industries, Inc.)2300 Leghorn StreetM ountain V iew , C a lifo rn ia 94040Axially Scattering Spectrometer ProbeParticle Measuring Systems, Inc.1855 57th C o urtBoulder, C olorado 80301ReferencesByers, H. R., and R. R. Braham Jr., 1949: The Thunderstorm.U.S. W eather Bureau, W ashington, D.C., 287 pp.H u ffm an , P., 1970: P olarization o f lig h t scattered by icec ry s ta ls ./. Atmos. Sci. 27, 1207 - 1208.------------- , and W. R. T hursby Jr., 1969: Lig h t scattering by icec ry s ta ls ./. Atmos. Sci. 26, 1073 - 1077.Sax, R. I., and P. T. W illis, 1974: N atural glaciation characteristicso f an ensemble o f F lorida cum ulus clouds. In P reprintV o l., Conference on C loud Physics, held 21 - 25 O ctober atTucson, A riz .; Am erican M eteorological Society, Boston,Mass., 152 - 157.31

ATMOSPHERIC TECHNOLOGYImaging DevicesTheodore W. CannonIn order to understand clouds, the scientist makes measurementsfro m a irc ra ft o f cloud air tem perature, w ater vapor, airm o tio n , electric fie ld , and particles. Particle attribu tesmeasured include chem ical com p osition, size, shape, concentratio n (num ber per u n it volum e), physical state (solid orliq u id ), and liq u id o r ice water conte n t or both. This articleconcerns instrum ents unique in th e ir a b ility to record particleshape, although size, concentra tion, and liq u id water conte n tmay be measured as w ell. Such instrum ents are norm a llyrestricted to measurements o f those cloud particles producedby condensation o r sub lim a tion o f atm ospheric water vapor;aerosol particles are usually to o small to be detected by them .The problem o f o b tain ing particle shape is a d iffic u lt one.The particles generally range in size over fo u r orders o fm agnitude, fro m tin y ice crystals and cloud droplets a fewm icrom eters in diam eter to ice particles several centim eters indiam eter. C oncentrations vary fro m a fe w per cubic m eter fo rthe larger p re c ip ita tio n particles to several thousand per cubiccentim eter fo r clo ud droplets in continental clouds—a rangeo f nine orders o f m agnitude. A n in stru m ent designed to recordthe sm aller particles w ill, as a rule, n o t sample a s u ffic ie n tvolum e to record the larger, rarer particles in any statistica llysig nifica nt num ber. Consequently it is necessary to have m orethan one in stru m en t to ob tain shapes fo r all particle sizes.Measurements are made fro m a irc ra ft fly in g at true airspeedsfro m about 4 0 m /s fo r research sailplanes to 200 m /s o r higherfo r je t a ircra ft, so th a t the sam pling tim e fo r each pa rticle isvery short. C o nditions o f high h u m id ity and lo w tem peraturescan produce icing; electronic and mechanical problem scom pound the d iffic u ltie s o f o b tain ing these data.AuthorTheodore W. Cannon has been involved in cloud physicsresearch at NCA R since 1966, specializing in laboratory andfield measurements o f atmospheric cloud particles and kinematiccloud modeling. He received a B.S. in electrical engineeringfrom Oregon State University in 1956; after five yearswith General Electric Company, he returned to Oregon Stateto earn a Ph. D. in nuclear physics, in 19 75, he was recipient o fa service award from the Society o f Photographic Scientistsand Engineers.A straig htforw a rd technique, used fo r a num ber o f years, isto allow the particles to make an im pression in th in fo ils o r tobe replicated in plastic (see the article by Spyers-Duran, thisissue), but the particles are subject to splashing, shattering,freezing, and sometimes m elting as they approach andencounter the fo il or plastic surfaces at a irc ra ft speeds. Thetrajectories o f the particles are influenced by the presence o fthe a irc ra ft and the colle ctin g in stru m en t in the airstream soth a t the colle ctio n efficie n cy (a fu n c tio n o f particle size andairspeed) m ust be know n in order to calculate concentrationso f the particles in the cloud.Im aging devices obtain in fo rm a tio n on particles in the irnatural state, as far as possible, w ith a m in im um o f disturbanceto the sample by the a ircra ft and by the instrum ent. Thus theproblem s o f the collectin g devices can largely be circum vented.The Need for Information on Particle ShapesIm aging devices are very useful fo r p ro vid ing in fo rm a tio nleading to an understanding o f im p o rta n t physical processes inclouds. Some o f the questions answered by using these devicesare as fo llo w s :• Are the in dividu al clo ud particles composed o f water, ice,or a m ix tu re o f the tw o (as in w et or slushy ice)? This in fo rm a ­tio n , together w ith data on clo ud tem perature and air m otions,indicates w hether the particles are grow ing by dro pletcoalescence, an all-ice process, or both.• In the case o f ice particles, w hat specific grow th processesare involved, e.g., accretion o f supercooled droplets ontoem b ryon ic ice particles w ith subsequent freezing, stickingtogether o f ice particles to fo rm aggregates, freezing o fsupercooled raindrops to fo rm ice spheres, etc.? The physicalappearance o f the ice particles provides understanding o f thesegrow th processes.• A re ice crystals in the shape o f needles, colum ns, dendriticplates, or oth er form s? The shapes o f ice crystals giveinsight in to the tem perature and h u m id ity con d itio n s underwhich they grow fro m the vapor phase.• W hat does the radar re fle c tiv ity fro m a cloud mean interm s o f sizes, concentrations, and types o f particles present?Radar re fle c tiv ity depends on the physical state (ice or water)and shape o f individual particles and, if they are asym m etrical,on th e ir aspect relative to the dire ctio n o f propagation o f the32

NUMBER 8 -S P R IN G 1976radar pulses. Studies o f hydrom eteors using airborne im agingdevices correlated w ith one or more radars observing the sameregion o f cloud or p re c ip ita tio n as the a irc ra ft w ill lead tofu rth e r understanding o f storms.• Are a u tom atic particle -co unting and sizing instrum entsgiving an accurate representation o f the cloud being observed?By giving the scientist a “ lo o k ” at the cloud, im aging devicesserve as a check on the in te g rity o f data com ing fro m oth erc o u n tin g and sizing instrum ents. Data fro m electronic andelectro-optical devices m ay be influenced by electronic noise,icing, or colle ctio n efficie n cy problem s th a t m ig h t otherw isego undetected if the data lo ok reasonable. A u to m a tic sizinginstrum ents give the size o f the projected image; the measuredsize depends on o rie n ta tio n o f the particles. This orien ta tioncan often be determ ined w ith im aging instrum ents. Theaccuracy and in te rp re ta tio n o f impressions or replicas can alsobe checked by com parison w ith images obtained using imagingdevices.We w ill describe three im aging devices cu rre n tly in use—the tw o-dim ensional optical array spectrom eter probe o fParticle Measuring Systems (PMS), B oulder, C olorado; theN C A R particle camera; and the airborne holography system o fScience A pplications, Inc., El Segundo, C alifornia.The Optical Array Spectrometer ProbeThe PMS tw o-dim ensional array spectrom eter probe, developedunder the directio n o f R obert K nollenberg (o f PMS), is aunique variation o f the optical array particle-sizing probedescribed in the article by A . H eym sfield in this issue. In thesizing probe, the particles pass through a parallel lig h t beamfro m a con tin u o u sly radiating laser and a shadow is cast on alinear array o f lig h t sensors (photodiodes) located in the beam.The size o f each particle is determ ined by electronicallycou ntin g the num ber o f sensors w ith lig h t extinguished belowa 50% in te n sity threshold by the shadow. In the tw o-dim ensional probe, the entire array o f sensors is sampled at a“ slice” rate p ro p o rtio n a l to the true airspeed. By use o f ahigh-speed, front-e nd data storage register, each sensor cantransm it up to 1,024 bits o f shadow in fo rm a tio n fo r eachparticle. A series o f “ image slices” is recorded across theshadow to develop a true tw o-dim ensional image.Figure 1 illustrates the data fo rm a t fo r a d e n d ritic icecrystal. A ll together in the basic probe there are 32 sensors inthe linear array, each im aging a p o rtio n o f the lig h t beam25 jum in diam eter. Because o f the shadow cast by the particle,the lig h t level falls to the threshold o f d e te c ta b ility (50%S H A D O W P A T T E R NFig. 1 Star-shaped dendritic ice particle passing across optical array o fPMS two-dimensional probe is imaged as a m atrix o f zeros and ones,depending on absence or presence o f lig h t (at 50% extinction) on theoptical sensors during each slice.L IN E A R A R R A YD E N D R IT IC IC E C R Y S T A LQJ 00> o §E =H

ATMOSPHERIC TECHNOLOGYe x tin c tio n ) o f the sensors at each p o in t m arked zero on thediagram ; the points m arked one denote sensors s till receivinglig h t above the threshold. A t a particle speed o f 100 m /s, aslice rate o f 4 M Hz (4 X 106 cycles per second) w o uld result inan u n disto rte d image, a slower rate in a compressed image, anda higher rate in an elongated image. The particle shown inFig. 1, being ap pro xim ately 475 /um wide, images across 19slices. The images can be viewed on a cathode ray tube displayduring flig h t and stored, along w ith the tim e between passageo f the particles, on m agnetic tape fo r later com p uter analysis.A com p uter can be program m ed to classify particles accordingto linear o r areal size, and concentrations can be determ inedby c o u n tin g the num ber o f particles fro m a know n volum e o fcloud passing through the space seen by the probe. Figure 2shows com puter-reconstructed images o f several particles incum ulus cloud. The basic tw o-dim ensional probe can be usedto image particles fro m 25 to 800 jum in diam eter, butm ag nifying lenses may be used to extend the range to diam ­eters o f 400 - 6,000 |Um. For the basic probe the sample isabout 49 2/km or 4.9 C/s at a true airspeed o f 100 m /s. Asample volum e o f 784 8/km or 78.4 C/s is obtained w ith thelargest (400 - 6,000 nm) size range. W ith the basic probe, theFig. 2 Examples o f images o f natural ice particles p lo tte d by com puterfrom two-dimensional probe data. The width o f each band o f images is800 txm, corresponding to 32 sensors, each sensing a portion o f the lig h tbeam 25 nm in diameter.sample volum e fo r particles less than 155/um in diam eterdecreases w ith decreasing particle size.Because the images and the tim e between images areobtained and stored in com p uter-co m patib le (digital) fo rm ,autom atic data analysis fo r concentration, size, and spatiald is trib u tio n is straig htforw a rd, w ith o u t necessity fo r any in te r­m ediate steps. An on-board video display allows the scientistto m o n ito r the cloud particle con te n t in real tim e and to makeon-the-spot judgm ents on the subsequent course o f the flig h t.G enerally, ice cannot be distinguished fro m w ater fo rspherical particles, since the shadow images have the sameappearance fo r both. No internal details are recorded in theimages except fo r regions o f relatively high optical transparency,and it is sometimes d iffic u lt to in te rp re t images asrepresentative o f know n particle shapes.The Particle CameraBy using a com b in a tio n o f high-speed flash lamps (about10 /is flash du ratio n) and a ro ta tin g m irro r synchronized to thetrue airspeed, it is possible to stop image m o tio n adequately tomake sharp photographs o f cloud particles fro m aircraft. Aschem atic diagram o f a clo ud particle camera developed atN C A R by the author is shown in Fig. 3. The camera ism ounted on The Explorer, the research sailplane owned by theNational Oceanic and A tm ospheric A d m in is tra tio n andoperated by N C A R .The camera is composed o f tw o basic units. The firs t is ac o c k p it package th a t contains a 35 m m film transport, a filmmagazine (n o t show n), intervalom eter, tw o xenon flash lamps,a 135 m m focal length lens, a ro ta tin g m irro r w ith drivem o to r, and electronic and electro-optical com ponents fo rcamera c o n tro l and synchronization o f the shutter, ro ta tin gm irro r, and lamps. The second u n it is an airfoil-shaped opticalhousing, located 59 cm above the canopy and supported byfo u r struts, th a t contains tw o corner reflectors and provides afla t-b la c k background fo r the photographs. The camera isfocused in a plane 34 cm above the top o f the canopy; thesample is illum inate d by crossed beams o f lig h t fro m the flashlamps and reflected back through the volum e by the cornerreflectors. This system o f illu m in a tio n provides fro n t lightin gnecessary fo r the more opaque particles and at the same tim eprovides back lig h tin g fo r transparent particles and waterdrops.W ith the present sailplane camera, photographs are taken atrates o f up to tw o fram es per second on 35 m m black-andwh ite recording film . Because the camera cannot be reloadedin flig h t, a special m agazine-film com b in a tio n was developed.The special magazine was developed by R obert W oltz34

NUMBER 8 -S P R IN G 1976Associates o f N e w port Beach, C a liforn ia , fo r attachm ent tothe existing electric N ikon camera film transport. Using anextra th in base (0.0025 in.) film , the m agazine-film com bination extends the capacity to 3,200 frames per flig h t, equivalent to 27 m in o f continuous operation at a m axim um rate o ftw o frames per second.Fig. 3 Schematic diagram o f NCA R particle camera installed inresearch sailplane (see text fo r details). From Cannon (1974).The camera images particles 8 /im in diam eter and larger.Water particles larger than 100/um are distinguished fro m iceby th e ir characteristic signatures. Because a drop acts like alens, fo rm in g images o f the tw o lig h t sources, the image o f anin-focus drop is a "d o t-p a ir” signature. O ut-of-focus dropsappear as overlapping disks or, fo r very far out-of-focus drops,as overlapping septagons (the shape o f the camera aperture)(see Fig. 4). Ice particles th a t are to o badly o u t o f focus toshow detail are imaged as irregular shapes o r as a single disk(both o f which are easily distinguished fro m the drop signature).The camera has the a b ility to give excellent detail fo rin-focus ice particle images, as shown in Fig. 5. W hile them a jo rity o f images are o u t o f focus, con cen tra tion and sizedata can be obtained fro m the out-of-focus images bymeasuring the size and optical density o f each image.Special charts, prepared fro m out-o f-fo cus photographs o fparticles taken in the la boratory are then used to determ inethe size and position o f each particle at the tim e the p h o to ­graph was taken. The m ethod can be used w ith o u t statisticalcorrections on ly when all o f the ice particles have the sameoptical transparency and re fle c tiv ity (all graupel, all clear icecrystals, etc.). When this size-density m ethod is used, thesample volum e depends on pa rticle size, ranging from2.6 ± 0.5 cm 3 per photograph fo r particles 8jum in diam eter ata camera m agnification o f one up to about 1 £ per photographfo r particles larger than about 5 m m at a camera m agnificationo f 0.5.Fig. 4 Sharply focused (left), out-of-focus (middle), and far out-of-focus (right) images o f raindrops in precipitation, photographed withthe NCAR particle camera. (From Cannon, 1974.)The images fro m the camera are o f generally high q u a lityand show internal structure. By using d iffe re n t lenses, a widerange o f particle sizes and volum es can be imaged w ith onecamera. Data reduction is tedious, requiring manual measurements o f size and optical density fo r each image.The Airborne Holography SystemThe hologram is an interference pattern recorded on p h o to ­graphic film . The pattern results fro m coherent m onoch romatic lig h t (generally fro m a laser) scattered by objects w ith inthe beam arriving at the photographic em ulsion w ith d iffe re n tphase relationships relative to unscattered (reference) lig h t inthe beam; constructive and destructive interference resultsw ith in the photosensitive em ulsion. By shining laser lig h t onthe exposed and subsequently developed hologram , it ispossible to reconstruct, in three dim ensions, the original sceneFig. 5Ice particle photographed with the NCA R particle camera.*I------1Imm

ATMOSPHERIC TECHNOLOGYas it existed when the hologram was made. The reconstructedimages come in to focus w ith in three-dim ensional space in thesame relative locations in which they existed when the h o lo ­gram was made. They can be photographed and viewed fo rsizing, cou ntin g, or detailed structural in fo rm a tio n .It is possible to make hologram s o f clo ud particles fro ma irc ra ft using a holocam era developed under the dire ctio n o fJames T rolinge r o f Science A pplications, Inc. Illu m in a tio n isprovided by a pulsed rub y laser. Hologram s o f particles w ith ina volum e o f ap pro xim ately 300 cm 3 located between theprobe tips o f the holocam era (Fig. 6) are made w ith10 -1 5 jum resolution on 70 m m film . The technique has theadvantages o f stopping particle m o tio n by the use o f a veryshort exposure tim e (10 ns; i.e., 10 “ 8 s) and allow in g in-focusreconstruction o f images o f all detectable particles w ith in thesample volum e, so there are no out-of-focus images such as areobtained w ith the pa rticle camera. T w o reconstructed iceparticle images fro m cirrus clouds are shown in Fig. 7.The holocam era has been used in cirrus clouds as high as18,300 m above sea level and at tem peratures as lo w as —62°C.One version o f this in stru m en t is being flo w n on a CessnaC ita tio n a irc ra ft and is used by the A ir Force Space andMissiles Systems O rganization fo r weather d e fin itio n .The sam pling rate is rather lo w ; o n ly tw o hologram s can bemade per m inute. Science A pp lication s is developing a Y A G(y ttriu m , alum inum , garnet) laser system which w ill increasethe sam pling rate to ten hologram s per second.Fig. 6 Holocamera designed fo r aircraft operation. L ig h t from pulsedruby laser illum inates a 300 cm 3 sample volume between the twoextending probes. Hologram is made on high-resolution film located in70 mm film transport at rig ht end o f low er tube. Camera is m ounted sothat probes extend either o u t o f the nose or out o f the side o f aircraft.(Diagram reprinted from Trolinger, 1974, p. 7, by permission o fScience Applications, inc., and the Instrum ent Society o f America.)Future Improvements and Data ReductionThe im aging devices described in this article are all relativelynew; the N C A R particle camera was firs t used in 1971, and thetw o-dim ensional probe and holocam era in 1974. W ork is underway to im prove these instrum ents, especially in the areas o fresolution o f particle detail and sample volum e. Im provem entsin resolution w ill enable scientists to distinguish ice fro m wateras well as to study details in ice particles at the ir earliest stageso f gro w th. A m ultiflash technique, higher fram e rate, andlarger size o f film w ill be used on fu tu re particle cameras toobtain m ore images over a given flig h t distance.Data reduction presents one o f the rather form id ableproblem s w ith im aging devices. D uring one flig h t m anythousands o f images may be obtained, and each one m ust beexam ined fo r size and shape, and fo r any special characteristicso f interest to the scientist. In a d d itio n , the num ber o f particlesper u n it volum e m ust often be determ ined by size class fromparticle counts. O f the three instrum ents described, the PMStw o-dim ensional probe, by using an electronic im aging techniqueth a t rejects any out-of-focus images and by storing thedata in com puter-ready fo rm , provides the greatest fle x ib ilityfo r rapid data analysis. The holocam era allows reconstructiono f all images above the size threshold o f d e te c ta b ility by use o fa reconstruction system, b u t the procedure is tedious andseveral hours o f manual analysis are generally required toanalyze each hologram . Analysis o f photographic images fromthe N C A R pa rticle camera requires a great deal o f manuale ffo rt, since both size and optical density m ust be determ inedfo r each image. There are several autom atic image-analyzingcom puters com m e rcially available, and in the fu tu re it w ill bepossible to use these fo r data red uctio n fo r both the h o lo ­camera and the particle camera. By using one o f theseanalyzers, size, optical image density, and shape classificationscan be obtained fo r all images w ith in a photograph w ith in afew seconds, b u t the details o f particle type, appearance,am ount o f aggregation, etc., m ust generally be decided byexam ination o f each image by the scientist.Fig. 7 Reconstructed holocamera images o f ice particles from cirrusclouds. (By permission o f Science Applications, Inc.)[200 Microns36

NUMBER 8 -S P R IN G 1976Table 1 presents the com parative advantages and lim ita tio n so f the instrum ents described in this article. The scientist maychoose among these instrum ents depending on need andbudget, although use o f m ore than one in stru m ent on the samea ircra ft is desirable to extend the ranges o f size and samplevolum e and fo r com parison over the ranges o f overlappingdetectable particle sizes.For Further Information about EquipmentOptical Array Spectrometer ProbeParticle Measuring Systems, Inc.1855 57th C ourtBoulder, C olorado 80301Airborne Holography SystemReferencesScience A pplications, Inc.101 C ontinental B uilding, S uite 310El Segundo, C a liforn ia 90245Cannon, T. W., 1974: A camera fo r photography o f atm o­spheric particles from aircraft. Rev. Sci. Instrum. 45,1 4 4 8 - 1455.T rolinger, J. D., 1974: A n airborne holography system fo rcloud particle analysis in weather studies. PaperNo. 74 - 627 in Advances in Instrumentation 29, Proc.International In s tru m e n ta tio n -A u to m a tio n Conference andE x h ib it, held 28 - 31 O ctober, New Y o rk , N. Y .; Instrum entSociety o f A m erica, P ittsburgh, Pa.IM AG ING DEVICE i i *PMS Two-Dimensional Optical ProbeNCAR Particle CameraScience Applications HolocameraA utom atic electronic data acquisition o f images andtime between images with no intermediate datareduction prior to computer analysisOn-board display allowing scientists to make on-thespotdecision on flig h t path based on particle contentobserved in real timeLarge sample volume fo r given length o f flig h tHigh-quality images w ith internal structureVery wide range o f particle sizes (depending onmagnification used) w ith one instrumentWater drops distinguished from ice particles bycharacteristic signaturesLarge volume in each sampleNo out-of-focus imagesHigh-quality images w ith internal structureShort exposure time obviating use o f rotating m irrorShadow image showing no internal structure exceptregions o f relatively high optical transparencyGenerally cannot distinguish ice from water fo rspherical or near-spherical particlesImages sometimes d iffic u lt to interpret asrepresentative o f known particle shapesOut-of-focus images requiring special considerationin data reductionTedious data reductionSample rate rather low in present systemTedious data reduction37

ATMOSPHERIC TECHNOLOGYLiquid Water Content DevicesRobert E. Ruskin, Naval Research LaboratoryM easurement o f am bient liq u id w ater co n te n t (LW C) fro ma irc ra ft is com plicated by several factors: d iffe re n t sizes o fdrops are present in clouds, d iffe re n t p re c ip ita tio n trajectoriesoccur around an a irc ra ft fuselage, and c o lle ctio n efficienciesd iffe r w ith d iffe rin g projections fro m the a irc ra ft and fro minternal to external instrum ents. The c o lle ctio n efficie n cy isthe fra c tio n o f drops s trik in g an o b stru ctio n as the rem ainingdrops are diverted around it by the a irflo w . Unless the dropsenter an in stru m en t w ith isokinetic flo w (no back pressure), ap o rtio n o f the drops w ill be d ive rte d .1 P ro p o rtio n a lly moresmall droplets than large drops are diverted around obstructions.For all sizes, a greater fra c tio n is collected at fasterspeeds, at low er air densities, and on sm aller obstructions.For cross sections as large as the fuselage o f a large plane,m any o f the drops are diverted as far away fro m the fuselageas a couple o f meters. A d ro p le t-countin g in stru m e n t locatedw ith in a m eter o f the fuselage may undercount o r overcountby more than a fa c to r o f ten. For this reason it is generallypreferable to m o u n t instrum ents fo r LWC determ in ation undera w ing or on a boom ahead o f the plane. C o llection efficie ncyis discussed in detail by Langm uir and B lo dgett (1949), and asim p lifie d sum m ary curve is shown in Ruskin and S cott(1974).T w o o f the earliest LWC instrum ents were V o n n e g u t’scap illary c o lle cto r (see Mason, 1971), and the W arner andNewnham (1952) paper tape in stru m en t, w hich was developed' Achieving isokinetic flow into an instrument requires that the properamount o f suction be provided (usually by the aerodynamics at thesample exit) so that during high-speed entry the droplets are neitherdiverted around the obstruction caused by increased pressure at theentry nor diverted into the entry by a suction.AuthorRobert E. Ruskin joined the Naval Research Laboratory(N RL) in 1942, after academic training in physics at KansasState Teachers College and at the University o f Missouri. He isa member o f numerous professional and honorary societies,and his awards include listing in W ho’s W ho in Science andIndustry. His present work at N RL involves research on marinefog and haze and their effects on electro-optical systems.in A ustra lia and was w idely used during the 1950s. Item ployed a porous paper tape to which there had been applieda solution th a t conducts e le c tric ity when w et b u t n o t whendry. The tape moved over tw o fix e d electrodes between whichthe electrical resistance was measured. Its princip al lim ita tio n swere th a t collectio n e fficie n cy fo r small droplets was lo w andth a t at higher flig h t speeds and large LWC the tape softenedand broke, thereby becom ing inoperative fo r the rem ainder o fa flig h t.Hot-Wire-Type LWC InstrumentsThe m ost com m on a ircra ft in stru m en t fo r measuring thesm all-droplet p o rtio n o f cloud water is the Johnson-W illiam s(J-W) ho t-w ire device m anufactured by Johnson-W illiam sProducts, M ountain V ie w , C a lifo rn ia (Neel, 1955), which usesa h o t n ickel-iron w ire w ith a know n tem perature c o e ffic ie n t o fresistance (see Fig. 1). This w ire is heated by a constant electriccurre nt; the resistance resulting fro m its tem peraturechange is measured using a bridge c irc u it. In dry air the wirem aintains a steady-state tem perature, a result o f balancebetween the electrical heat supplied and the heat removed bythe dry a irflo w . However, if droplets are present in the air,they collide w ith the wire and cause a degree o f coolingd ire c tly related to the am o unt o f liq u id water in the airstream .V ariatio ns in airspeed, w hich cause variations in heat transfer,are compensated fo r in the electrical c irc u it by manualadjustm ent o f a potentio m eter, b u t m in o r variations due toflu c tu a tio n s o f airspeed fro m turbulence or cloud a c tiv ity aren o t taken in to account. The e ffe c t o f am bient tem peraturechanges is compensated fo r electrica lly by a reference h o t wirein the in stru m ent positioned parallel to the airstream so th a t itis affected by the a irflo w b u t n o t by the droplets. This referencew ire form s a com pensating leg o f the bridge. In operation , the in stru m en t is responsive to cloud water. However,because the reference w ire gradually becomes w et in heavyp re c ip ita tio n , the in d ica tio n tends to d r ift lo w and give anegative reading after return in g to clear air. This e ffe ct isgreatest when the probe is oriented so th a t the reference wireis at the b o tto m .The J-W hot-w ire device has the a b ility to produce a realtime electrical readout o f the liq u id water contained in clouddroplets, m ainly in those sm aller than about 20 /um in radius(Levine, 1965). The reduced reading w ith large drops probablyis due to splattering o f the drops. Though the in strum entbecomes saturated if the cloud water is more than about38

NUMBER 8 -S P R IN G 1976Fig. 1The Naval Research L aboratory’s pylon-m ounted,evaporator Lyman-alpha instrum ent fo r to ta l water contentis now being manufactured by General Eastern Corp., Watertown,Massachusetts. A j-W , h o t-w ire LWC probe is m ounted on therig ht end o f the pylon.3 g /m 3 (at an airspeed o f 250 k t), it can be valuable fo r thedetection o f young, grow ing cells because o f its discrim in atio nagainst larger drops, and so it can act as a seeding c rite rio n byin dica ting the presence o f a usable am ount o f unfrozen waterin small droplets.A n o th e r in stru m en t using wires heated by a constantcurre nt has been designed by Levine (1965). His in strum entconsists o f tw o portions, called the “ cloud-w ater in s tru m e n t”and “ rain-w ater in s tru m e n t.” The cloud-w ater in stru m en t is aplastic yoke strung w ith a heated w ire o f sim ilar diam eter toth a t in the J-W in stru m ent b u t longer. The rain-w ater in stru ­m ent is a threaded ceram ic cone about 5 cm in diam eterw ound w ith a sim ilar wire. In the ory, the collectio n efficie ncycharacteristics and drop breakup problem s th a t make the J-Win strum ent insensitive to larger drops apply on ly to the cloudwater instrum ent. The cone-shaped rain-water sensor has acollectio n efficie n cy w hich favors larger raindrops. This in stru ­m ent system has had problem s o f d r ift during extended flig h tsin clouds. U n like the J-W instrum ent, it provides no autom atictem perature com pensation.A n o th e r ho t-w ire instrum ent, w hich has been in use fo rseveral years in the U nion o f Soviet S ocialist Republics, isreported to respond w ith equal sensitivity fo r all drop sizes.This in stru m ent consists o f a w ire filte r screen o f a p p ro ximately 100 cm 2 heated by high-frequency current. As w iththe J-W and Levine instrum ents, the wire tem perature (andhence the resistance) is a calibrated fu n c tio n o f LWC.A n im provem ent in hot-w ire LWC in stru m enta tion is basedon operation in a constant-tem perature mode instead o f theconstant-current mode. T w o instrum ents o f this type use fo rthe ir LWC calibra tion the electric curre nt required to m aintaina constant tem perature. One was designed by T. K yle o fN C A R fo r LWC as high as 30 g /m 3 . The other, called a“ n im b io m e te r,” was designed by M erceret and Schricker(1975) at the N ational H urricane Research Laboratory(N H R L ), where it has undergone considerable flig h t use,including through hurricanes. This in stru m ent is reported toprovide a tim e response o f a fe w m illiseconds and to consumeless curre nt per volum e sampled than either constant-currentor total-evaporation types o f hot-w ire instrum ents. Itssam pling area o f about 0.5 cm 2 consists o f about 30 cm o fw ire 0.177 m m in diam eter. The small diam eter perm its a fastresponse, b u t increases the v u ln e ra b ility to breakage by hail orgraupel.A lim ita tio n fo r m ost ho t-w ire instrum ents is the change incalibra tion because o f cooling when they are operated in ice ormixed-phase p re cip ita tio n or clouds. A n o th e r lim ita tio n inaccuracy, p a rticula rly fo r fast-response, hot-w ire instrum ents,is the coo lin g due to rapid flu c tu a tio n s in the speed o f the airpassing through the probe du rin g flig h t in tu rb u le n t air (whichis usually the case in clouds).Evaporation Lyman-Alpha InstrumentsThe problem o f varied coo lin g under mixed-phasecon ditions is overcom e by measuring o n ly the vapor densityresulting fro m total evaporation o f the water, as opposed tomeasuring the cooling e ffe c t on the h o t wires. T w o in stru ­ments based on evaporation measurem ent em ploy a type o ffast-response, vapor-density sensor in which w ater vapor causeslig h t absorption in the spectral line o f atom ic hydrogen at thewavelength o f 121.56 nm . This spectral line is very highlyabsorbed by water vapor: vapor w ith a dew p o in t o f 0°Ccauses the lig h t to be reduced to 1 /e o f its dry-air value in0.5 cm o f path length. The sensor is based on the Lym an-alphah u m id iom e te r discussed in W exler and Ruskin (1965), in thechapters by T illm a n , by Randall et al., and by Ruskin. Thetw o instrum ents th a t use this sensor are described by Ruskin(1967) and by K yle (1975). A lth o u g h the sensor (see Fig. 2)has a response tim e o f a few m illiseconds, the overall response39

ATMOSPHERIC TECHNOLOGYSCREENWATER VAPOR DENSITYMEASUREMENT BYL - a LIGHT EXTINCTIONFREE AIR -ANDLIQUID W ATER.TO SUCTIONDISCHARGEFig. 2 Schematic cross section o f the NRL totalwater content probe. A few turns o f the "Cairod"heater brazed into the entry provide deicing; mazeconsisting o f heater and screens evaporates cloudwater in 0.01 s; Lyman-alpha detector indicateschanges in resultant vapor density in a few m illiseconds.WATER AND ICEEVAPORATINGHEATERAIR FLOWFig. 3 Vertical cross section o f the optical flowmeter,illustrating the size and relative position o f components.(Courtesy o f E. Brown, NCAR.)INNERCONEWATERFig. 4 Results obtained during a comparison test comparingthe optical flow/meter and a hot-wire instrument on 2 September1970. Data shown at 46 and 51 min are from cumulusclouds with tops at 6,000 m and with rain on the exit side.4.0LIG HT BEAMPROJECTIONEXIT ORIFICEAND WIREPHOTO DIODE (C)ISOLATOR£o\3.02.0MOUNT RINGSQ.OCD 1.01—ZUJ 0.0»- 4 0—I__1— 1__I__ L~i— i— i— i— r “ 1— I--- 1----I--- 1----1----1— I--- 1— I— T" 1 — i— i— i— rHYSTERESISSYNCHRONOUSMOTOR$h-OoL>A 0trUJh-< ? 0£1.00.045 46 47I CMT IM E IN M IN U T E SFig. 5 Block diagram o f the instrument being developed by MeteorologyResearch, Inc., to measure total water content.40

NUMBER 8 -SPRING 1976time o f the instrum ent depends largely on the time required tobring the sample in and evaporate its water. Ruskin gives thistime as a few hundredths o f a second except w ith ice, when itbecomes a few tenths. Kyle lists a response o f 1 s. In bothinstruments the temperature o f the open-mesh heating elementis maintained somewhat below the boiling point o f water;therefore, the water striking the heater spreads quickly over it(with a reduced surface tension) and evaporates almost instantaneously.The Kyle design has the advantage o f an automatictemperature control so that LWC as high as 40 g/m 3 can beevaporated.These evaporator instruments provide a direct measuremento f the total amount o f water present in all three phases (ice,liquid, and vapor). Data in this form are convenient fo r studieso f cloud air entrainm ent or conservative properties. However,fo r LWC data it is necessary to subtract the vapor density o fthe air between the cloud droplets. For this purpose a therm o­electric dew-point hygrometer has been used w ith a water-separating sample entry, as illustrated by Ruskin and Scott(1974).Although the Lyman-alpha sensors in the evaporator instrumentscan be calibrated essentially continuously while flyingin clear air, still the accuracy o f LWC data is lim ited by thecombined random errors in the dew-point and Lyman-alphainstruments and by the slower (4 s) response time o f the dew-point instrument. This response time lim itation is obviated bythe addition o f a second Lyman-alpha sensor fo r the water-separated sample. Two new developments are expected toimprove the evaporator Lyman-alpha instruments: reductiono f the combined random error o f tw o sensors by time-sharinga single sensor between the tw o samples (in development atthe Naval Research Laboratory— N RL), and increasing thesample inlet area from 0.8 cm2 to 10 cm 2 (in development forthe U.S. A ir Force at Aerospace Corporation, Los Angeles,California).Sampling Volum e ConsiderationsFor measuring LWC in precipitation the problem o fsampling volume is a major one because o f the large apparentfluctuations in LWC caused by the difference in size betweenraindrops and cloud droplets. The LWC o f a 2 mm raindrop isa m illion times that o f a 20 /um cloud droplet and, correspondingly,there are a m illionth as many raindrops as cloud dropletsper u nit volume fo r the same LWC. To provide statisticalvalidity fo r rain comparable to that obtained in the same flig h tdistance from a 1 mm2 sampling area fo r cloud droplets, anarea o f 1,000 cm2 is required. For further discussion o f thesampling problem see Ruskin and Scott (1974).Centrifugal Separator TypesA move toward proper sampling area is achieved in theoptical flow m eter o f Brown (1971, 1973) at NCAR (Fig. 3).In this instrum ent the sampling area is 10 cm2 . The collectionefficiency fo r cloud droplets is smaller than fo r raindropsbecause little air passes through to carry the smaller droplets.For this reason it may measure less cloud water than waspresent. Raindrops are collected and centrifuged to the outerwall o f a cup or bowl as it spins on its horizontal axis. Thecentrifugal force drives the water through a small hole inwhich is mounted a 0.2 mm silvered wire. The wire-watersurface tension and aerodynamic drag cause the form ation o f aweb o f water. The width o f this web as measured by a lightsource and photodiode detector provides inform ation on thewater flo w rate, hence LWC. On the developmental modelsdeicing is not available. A comparison o f flig h t data from thisinstrum ent and a J-W hot-wire type is shown in Fig. 4. Theoptical flowm eter, which is more sensitive to raindrops, showslarge amounts o f rain water which the hot-wire instrum entmisses. Near the cloud edges the hot-wire instrum ent showsmore LWC, probably because the smaller droplet sizes favorthat instrument.A promising instrument, now under development at MeteorologyResearch, Inc., Altadena, California, is called the“ cyclone separator.” This instrum ent improves the usuallyinadequate sampling volume by incorporating a 100 cm2entry. As shown schematically in Fig. 5, the sample impingeson a heated (deiced) plate which slopes down into a ce ntrifugalseparator. The plate is continually covered with a liquid inwhich the water drops become imbedded and flo w into thecyclone separator where the air is expelled from the liquidsample. The liquid is then pumped to a detector cell where thefraction o f water in the carrier liquid is measured by changesin capacitance resulting from changes between the high dielectricconstant o f the water and a lower value fo r the carrierliquid. This instrum ent can be quite accurate at LWC values aslow as 0.01 g/m 3 and with drop sizes from the largest raindown to 5 or 1 0 /im in radius. In the first models a lim itationwas the instrum ent’s slow resolving time o f about 6 s plus anadditional 6 s o f constant delay time.Each o f the LWC instruments has advantages and lim itations.Some are better fo r young clouds, others fo r precipitation.Some have fast response, others good sampling volumes.Some operate in ice or mixed phase; others do not. Thereforethe researcher must consider the priorities fo r his particularLWC measurements and choose his instruments carefully.Whichever instrum ent is chosen, it w ill require some “ livingw ith ” before meaningful and reliable data can be expected,even after it has been carefully mounted in an aerodynamicallycorrect location on the aircraft.41

ATMOSPHERIC TECHNOLOGYReferencesBrown, E. N., 1971: A Prototype Optical Flowmeter for theMeasurement o f Liquid Water Content. NCAR TechnicalNote EDD - 61, NCAR, Boulder, Colo., 25 pp.------------ , 1973: A flow m eter to measure cloud liquid water.Atmos. Technol. 1, 49 - 51.Kyle, T. G., 1975: The measurement o f water content by anevaporator. J. Appl. Met. 14, 327 - 332.Langmuir, I., and K. B. Blodgett, 1949: Mathematical Investigationo f Water Droplet Trajectories. Final Report underArm y Contract W - 33 - 038 - ac - 9151, Report No. RL - 225from the General Electric Company, 46 pp.Levine, J., 1965: The Dynamics o f Cumulus Convection in theTrades, a Combined Observation and Theoretical Study.Ph.D. thesis, Dept, o f Meteorology, Woods Hole OceanographicInstitution, Woods Hole, Mass., 129 pp.Mason, B. J., 1971: The Physics o f Clouds. Clarendon Press,O xford, 675 pp.Merceret, F. J., and T. L. Schricker, 1975: A new hot-wireliquid cloud water meter. /. Appl. Met. 14, 319 - 326.Neel, C. B., 1955: A Heated-Wire Liquid-Water-Content Instrumentand Results on Initial Flight Tests in Icing Conditions.National Advisory Committee fo r Aeronautics ResearchMemo R M A54123, 33 pp.------------ , and W. D. Scott, 1974: Weather m odification instruments and their use. Chapter 4 in Weather and ClimateModification (W. N. Hess, Ed.), John Wiley and Sons, Inc.,New Y ork, 136 - 205.Warner, J., and T. D. Newnham, 1952: A new method o fmeasurement o f cloud-water content. Q. J. R. Met. Soc. 78,46.Wexler, A., and R. E. Ruskin (Eds.), 1965: Humidity andMoisture. Vol. /, Principles and Methods o f MeasuringHumidity in Gases. Reinhold Pub. Corp., New York, 704 pp.For Further In fo rm a tio n about Equipm entJ-W Hot-Wire InstrumentJohnson-Williams ProductsBacharach Instrument Co.(Division o f AM BAC Industries, Inc.)2300 Leghorn StreetMountain View, California 94040Cyclone SeparatorMeteorology Research, Inc.464 West Woodbury RoadAltadena, California 91001N R L Lyman-Alpha InstrumentGeneral Eastern Corp.36 Maple StreetWatertown, Massachusetts 02172Ruskin, R. E., 1967: Measurements o f water-ice budgetchanges at —5C in Agl-seeded tropical cum ulus./. Appl.Met. 6, 7 2 -8 1 .42

NUMBER 8 — SPRING 1976Cloud Condensation Nucleus CountersJames E. Jiusto, State U niversity o f New Y o rk at AlbanyDiffusion cloud chambers were devised almost 40 years agofo r the purpose o f maintaining high supersaturations anddetecting nuclear particles. Langsdorf (1936) developed athermal-gradient diffusion chamber in which heated alcoholvapor at 75°C diffused downward to a base plate refrigeratedto —45°C by dry ice. The supersaturations o f several hundredpercent produced in this manner were sufficient to condensevapor on ions le ft in the wake o f cosmic rays, thereby form ingvisible drops.Twenty years were to pass before the diffusion chamberprinciple was first used to detect cloud nuclei active at supersaturationso f only a few percent (Wieland, 1956). It remainedfo r Twomey (1959, 1963) to pioneer the development o f suchinstruments and to stimulate interest in the betterunderstanding o f cloud condensation nuclei (CCN).CCN are those atmospheric particles having sufficient sizeand water a ffin ity to act as centers fo r the form ation o f clouddrops at the slight supersaturations (

ATMOSPHERIC TECHNOLOGYCLOUD CHAMBERTOP AND BO TTO M WATER RESE R VO IR SS(%)N (/cm )SAM PLEINLETLIG H TBEAMP6 6 0 6 6 61.0 -0.75--1400-1280DETECTOR(C a m e ra , e tc .)0.5-- 1 0 2 00.5--10100.25--560TEMPERATURE, TFig. 2 Photograph o f condensed drops formed in a cloud chamberfrom successive air samples at indicated supersaturations from 0.25 to1.0%.Fig. 1 Sketch o f a thermal-gradient diffusion chamber for detectingcloud condensation nuclei. Temperature difference between upper andlower wet surfaces produces supersaturation, S, as illustrated in phasediagram (bottom) and in resultant droplet growth on cloud nuclei. In thediagram, e represents ambient vapor pressure at about mid-level o f thechamber, and es(T) represents the saturation vapor pressure with respectto water at that level.by an imbedded thermoelectric cooler. Temperatures o f bothplates are monitored by thermistors or thermocouples, and thetemperature difference (AT) varied in steps over an equivalent5 range o f approxim ately 0.2 - 3%. An intense light source(e.g., a 100 - 200 W mercury arc) is collimated into a ribbon o flight and the known volume is photographed at a 90° anglew ith a Polaroid system, 35 mm still camera, or 8 mm moviecamera. The droplets, as shown in Fig. 2, are then ratherlaboriously counted w ith the help o f a magnifying lens orlarge-screen projection, resulting in CCN spectra such as thoseo f Fig. 3. From results such as these, certain maritime andcontinental aerosols can be characterized.Several design principles must be considered in constructinga reliable diffusion chamber. Twomey (1967), Saxena andKassner (1970), and Squires (1972) have reviewed some o f theoptim um characteristics o f CCN counters. Table 1 summarizescertain lim itations and design criteria to be considered. Inessence, these guideline specifications are aimed at satisfyingthe follow ing criteria:• Providing adequate droplet growth time at S > 0.1 or0.2%. Chamber height, h, must be restricted because the timefo r an air sample to equilibrate to chamber conditions isproportional to / 72 . Conversely,/? < 1 cm enhances drop fa lloutand complicates illum ination optics.• Minimizing wall effects. An aspect ratio (d/h) > 5, whered is the diameter, generally assures that warmer walls willneither perturb the temperature and the desired 5 fieldstoward the center o f the chamber, nor introduce substantialconvection currents. Gagin and Terliuc (1968) furtherm inim ize such effects by using metal fo r the conducting wallsin contrast to the Lucite or glass custom arily used.• Avoiding large, transient supersaturations. Because themolecular diffusion o f water vapor is about 20% greater thanthat o f heat, the introduction o f relatively cold, moist air intothe chamber can theoretically produce spurious high S transients(Saxena, Burford, and Kassner, 1970; Fitzgerald, 1970).Intake air at about the temperature o f the top warm plate isrecommended. While experiments by Radke and Hegg (1971)do not confirm these transients in their chamber (perhapsbecause o f stronger initial eddy diffusion or complex growth44

NUMBER 8 -SPRING 1976kinetics), it nevertheless would seem advisable to follow thisconservative, easily achieved criterion. As the phase diagram inFig. 1 suggests, saturated intake air colder than Tc or warmerthan T w could well produce S transients.which enables a range o f S samples to be recorded on a singlefilm (similar to that o f Fig. 2). Light from a 200 W mercury-arc source is collimated by lenses and slits to a ribbon approximately0.2 cm deep, 0.5 cm high, and several centimeters long.• Maintaining desired supersaturations. If excessivenumbers o f nuclei compete fo r the available water vapor, theprescribed S value will not be achieved. To avoid this, Twomey(1959) and Squires (1972) recommend diluting the samplewhenever nucleus concentrations exceed about 1 X 103 nucleiper cubic centimeter at 5 = 1%.• Minimizing line losses and successive sample contamination.As indicated in the table, modest tube lengths, theavoidance o f sharp bends, and adequate chamber flushing willassure that representative ambient samples are containedw ithin the diffusion chamber.Details o f the CCN counters o f this type tested at theFt. Collins workshop can be found elsewhere (Grant, 1971).The Calspan instrum ent (Kocmond, 1971), in use since late1964, has demonstrated its reliability at several comparisonworkshops. One o f its distinctive features is the Polaroidcamera (ASA film speed o f 3,000) and oscilloscope mount,Fig. 3 Typical spectra o f concentrations o f CCN vs supersaturation.Data can often be approximated by power functions o f the formN = cS^, where c and k are constants for a given aerosol. The dashedportion o f the line for oceanside at Hilo, Hawaii, represents extrapolatedvalues.The Naval Research Laboratory u nit (Ruskin and Dinger,1971), very similar in several respects to Tw om ey’s instrument(Twomey, 1963), also employs a mercury light source and 90°photography. However, an 8 mm movie camera is used toobtain a time sequence o f frames, from which the maximumdroplet concentration is ascertained. An innovation is thesimultaneous viewing o f droplet form ation with a videocamera; immediate playback and subsequent (video recorder)stop-frame counting are provided. Like some others (see, e.g.,Bonner and Low, 1971), this system incorporates a preconditioningchamber fo r sample equilibration before injection intothe chamber. When used at altitude in an aircraft, this featureis quite valuable.A utom atic Static ChambersPhotographic recording o f droplets, while proven quitereliable, presents tedious data analysis problems. It tends toin h ib it studies o f CCN variations over fine time and spacescales. Recent attempts to overcome this lim itation haveresorted to estimating droplet concentrations by means o flight-scattering methods— either from a large volume o fscatterers or from discrete drops individually counted.The volume scattering approach is employed in a Universityo f Washington CCN counter (Radke, 1971; Radke and Hobbs,1969). Because light is scattered by drops in proportion to thesquares o f their sizes, it is essential to this method that theTable IDesign Characteristics o f Static D iffusion ChambersChamber aspect ratio (diameter to height) > 5Maximum chamber height1 to 2 cmMinimum reliable supersaturation 0.1 to 0.2%Typical maximum supersaturation 1 to 3%Maximum droplet concentrationIntake sample air temperatureIntake sample airflow, F(a) Type1 to 2 X 103/cm3Top plate TwNonturbulent(b) Chamber flushings > 4 to 5(c) Tube length, L, vs flow F(cm 3/s) > 4 ^Minimum drop size detectedr ~ 1 Mm45

ATMOSPHERIC TECHNOLOGYdrops be o f reasonably uniform and known size fo r concentrationdeterminations. Standard calculations o f droplet growth,beginning w ith a broad range o f particle sizes and chemicalcom position, show that at S > 1 %, a cloud rapidly becomesmonodisperse (Jiusto, 1967). A t lower supersaturations, dropspectrum broadening increases but may well be at acceptableaccuracy levels down to approxim ately 0.2% 5, as stated forthis unit. A xenon flash tube illuminates a large volume o f thechamber when the cloud has become reasonably monodisperse.A laser beam at a 20° forward-scattering angle helpsdetermine drop spread and the first Mie peak. A t this point,the scattering coefficient, ks, and mean size, r, o f the drops areknown, so that the cloud’s scattered light, b, measured w ith aphototube, yields droplet concentration, N, by the follow ingrelation:A nother fu lly automatic CCN counter is that o f the StateUniversity o f New Y ork (Lala and Jiusto).2 The basic instrumenthas been in use since 1968 and retains some o f thegeneral features o f the u nit at Calspan, where it was in itiallydesigned. These include the 90° Polaroid recording camera anda m odified optics system. In addition to this absolute countingmethod, a corresponding light-scattering method (apparentlysimilar to Mee’s) recently has been incorporated fo r automaticoperation. A 150 W quartzline lamp and highly sensitivephotodetector (at 45° forward scatter) provide ample signal o froughly 0.4 - 1.5 m V per drop. The instrum ent can beprogrammed to operate fu lly autom atically and unattendedw ith several options:• A ir samples every 1 min to 1 h or moreN = b/(kgnr2) (-])This instrum ent compared favorably with the other CCNcounters examined at the Ft. Collins workshop. (A commercialversion o f the Radke counter is Model 1521, manufactured byMeteorology Research, Inc., Altadena, California.)A nother light-scattering instrum ent is manufactured by MeeIndustries, Inc., o f Rosemead, California (Model 130). It iscompact (15 X 43 X 50 cm), portable (18 kg), and supposedlycapable o f unattended operation at a fixed supersaturationor semiautomatic operation over a range o f 5 < 4%. Itcan be operated from either alternating or direct currentpower sources. A thermal diffusion chamber (30 cm 3 involume) is illum inated with a collimated beam from a low-intensity (25 W) tungsten lamp. Light scattered from the cloudis received in the forward direction by a high-sensitivity photodetector.Correspondence between the droplet concentrationand total light scattered is accomplished by microscopiccalibration. The cloud drops are viewed w ith a microscope (atthe top o f the instrum ent); their concentration is determinedby means o f a ruled reticle, and the scattered light analog isadjusted to correspond. Because o f low lamp intensity (tominim ize power needs when airborne), the drops are d iffic u ltto see until they have reached a size at which they aresedimenting. Then concentration estimates are generallypossible, though less reliable. As described below, meandroplet size will vary w ith supersaturation, so that this calibrationshould be done at various S levels fo r acceptable accuracy,a feature not presently provided for. These lim itations (lowlamp power and a single overall calibration) presumably couldbe readily correctable. The nucleus count (per cubic centimeter)is displayed in digital form on the fro n t panel, alongw ith Tw and A T; all inform ation can be appropriatelyrecorded.• Chamber operation at four sequential supersaturations(e.g., 0.25, 0.5, 0.75, and 1%S) fo r every designatedsampling interval, or at one fixed 5• Digital display and recording o f A', Tw, and A TMeasurements indicate that the mean drop size at peak cloudconcentration (prior to fallout) increases from approximatelyr = 1 /im to 2.5 /urn as S increases from 0.25 to 1.0%. The sizeincrease is reflected in the character o f the scattered lightspectrum and is appropriately calibrated via the photographicmethod. Thus far, the calibrations— one fo r each operatingsupersaturation level— appear consistent and stable (seeFig. 4).A promising method fo r rapidly processing droplet imageson photographs and negatives is with an autom atic imageanalyzer. Preliminary tests with a Leitz instrument (ClassimatModel or Texture Analyzer System, E. Leitz, Inc., Rockleigh,New Jersey) showed general agreement to w ithin 15% o f themanual photographic counting method.A direct scheme fo r autom atically measuring dropletconcentrations is to pass the cloud through an optical counter(e.g., Climet, Royco, Southern Research Institute) in whicheach drop is individually detected. Using forward-scatteroptics, these instruments can size and count drops greater thanabout 0.3 /um in diameter. For static diffusion chambers, thisis not practical, but dynamic- or continuous-airflow counterscan use this principle to advantage.2G. Lala and J. Jiusto: An automatic recording cloud-diffusionchamber. Paper available from the authors at the AtmosphericSciences Research Center, State University of New York, Albany; ithas also been submitted to J. Appl. Met.46

NUMBER 8 - SPRING 1976D ynam ic-Flow ChambersContinuous-flow CCN instruments apparently had theirorigin in Russia. Laktionov’s vertical counter (1968) consistso f two coaxial, cylindrical, wet tubes through which sample airpasses. Droplets leaving the diffusion chamber are individuallycounted photoelectrically. Such chambers introduce a certaindesign com plexity but proponents claim an advantage inphysically separating the function o f droplet growth at theprescribed supersaturation from that o f droplet counting.Hudson and Squires (1973) have developed a dynamicsystem that features a laminar air-sample flo w (~1 cm3/s)between tw o streams o f particle-free sheath air along horizontalplates (h = 1.3 cm; traverse length and depth ^ 2 9 cm)and individual droplet sensing with a Royco Model 225(Royco Instruments, Inc., Menlo Park, California) opticalcounter just downwind o f the chamber. The system has theadvantages o f a high sampling rate and an ability to size drops(haze discrim ination); moreover, the authors claim that theproblems associated with static chambers are lessened with thisinstrument.Another dynamic chamber o f horizontal parallel-platedesign has been developed (Fukutaand Saxena, 1974; Fukuta,Saxena, and Gorove, 1974) which simultaneously produces airstreams o f varying supersaturations. The plates are 122 cmlong, 30.5 cm wide, and 1.1 cm apart. Heat is conductedacross the upper plate, down a conducting wall, and backthrough the bottom plate to a nonconducting wall. Varyingvertical temperature gradients are thus produced across thechamber and, combined w ith moisture surfaces (filte r paper)o f different widths, a range o f S from ~0.1 7 to 4% and d iffe r­ent droplet growth times can be achieved. A Clim etFig. 4 Scattered light signal vs CCN concentration. The sample volumewas ~ 0 .1 cm3.Model 0294 - 1 optical counter (Clim et Instruments Co.,Redlands, California) rides on a platform at the downstreame xit o f the chamber, perpendicular to the airflow ; a completescan o f droplet concentrations over the S range is accomplishedin 15 s. Careful control o f sample flo w rate versus 5must be maintained to prevent droplet fa llo u t; alternately, thesheath air surrounding the air sample entering an opticalcounter must not be allowed to heat up and evaporatedroplets.It is not yet evident that these dynamic CCN countersprovide substantially greater accuracy or reliability thansimpler static-diffusion chambers. Until comparison data areavailable, one cannot judge. Their strength appears to be theautomatic droplet-counting feature, which is gained at theexpense o f added com plexity in the control o f sample flo wand temperature fields.Lower Supersaturation ChambersMost inform ation to date on cloud nuclei has been acquiredat S > 0.1 or 0.2%. Y et it is known that many clouds orportions o f clouds have supersaturations below this threshold.Two problems arise in detecting active cloud nuclei at low 5:the growth rate o f drops and fa llo u t is much slower prior todetection; and the drops must be allowed to grow large enoughto be distinguished from haze. To help circumvent theselim itations, tall vertical chambers with continuous (dynamic)flo w are being developed.Sinnarwalla and Alofs (1973) have constructed a tube100 cm high with a plate separation o f 1 cm. The growth timeis approxim ately eight times that o f a conventional chamber.Such configurations and time scales introduce problems o fconvection and phoretic (vapor and thermal diffusion) forcesthat can deflect drops toward the colder wall surface. Theseeffects reportedly are accounted fo r in this system, however,and the resultant drops are optically counted with a phototubeand laser beam.In unique fashion, Hudson and Squires (1974) merely tilttheir horizontal dynamic counter (previously described) onend to achieve a 40 cm fall distance. Operating the counteralternately in the horizontal and vertical modes, they reportconcentration agreement to 1% across the overlapping range ofabout 0.3 to 0.7% 5.Perhaps the most ingenious and simplest approach— andthe one potentially yielding the lowest values o f S — has beenadvanced by Laktionov (1972) and reviewed in the literatureby Alofs and Podzimek (1974). Laktionov derived a simpleCCN (/c m 3)47

ATMOSPHERIC TECHNOLOGYrelationship between critical supersaturation, S*, and the sizeo f droplets at 100% relative hum idity that have condensed onhygroscopic nuclei:S* ~ O.Ob/r

NUMBER 8 -SPRING 1976------------ , 1971: Cornell Aeronautical Lab. (Calspan) thermaldiffusion chamber. In Second International Workshop onCondensation and Ice Nuclei (L. Grant, Ed.), August 1970,Colorado State University, Ft. Collins, Colo., 40 - 41.Laktionov, A., 1968: Photoelectric measurements o fcondensation cloud n u cle i./. Rech. Atmos. 3, 63 - 70.------------ , J. Burford, and J. Kassner, 1970: Operation o f athermal diffusion chamber fo r measurements on CCN./. Atmos. Sci. 27, 73 - 80.Sinnarwalla, A., and D. Alofs, 1973: A cloud nucleus counterw ith long available growth tim e ./. Appl. Met. 12,831 - 835.------------ , 1972: Isothermal method fo r the determination o fcloud condensation nuclei concentrations. Izv. Akad. Nauk.SSSR, Fiz. Atmos. Okeana 8, 672 - 677.Langsdorf, A., 1936: A continuously sensitive cloud chamber.Phys. Rev. 49, 422.Radke, L., 1971: An improved autom atic cloud condensationnucleus counter. In Second International Workshop onCondensation and Ice Nuclei (L. Grant, Ed.), August 1970,Colorado State University, Ft. Collins, Colo., 41 - 42.Squires, P., 1972: Diffusion chambers fo r the measurement o fcloud n u cle i./. Rech. Atmos. 6, 565 - 572.Twomey, S., 1959: The nuclei o f natural cloud form ation, I.Geofis. Pura Appl. 43, 227 - 242.------------ , 1963: Measurements o f natural cloud nuclei./. Rech. Atmos. 1, 101 - 105.------------ , 1967: Remarks on the photographic counting o fcloud n u cle i./. Rech. Atmos. 3, 85 - 90.------------ , and D. Hegg, 1971: Experimental observations on“ Non-steady-state supersaturations in thermal diffusionchambers.” /- Appl. Met. 10, 340 - 341.Wieland, W., 1956: Die Wasserdampfkondensation an naturlichemAerosol bei geringen Ubersattigungen. Z. Angew.Math. Phys. 7, 428 - 459.------------ , and P. Flobbs, 1969: An autom atic CCN counter./ . Appl. Met. 8, 105 - 109.Ruskin, R., and J. Dinger, 1971: Description o f N RL CCNequipment. In Second International Workshop on Condensationand Ice Nuclei (L. Grant, Ed.), August 1970,Colorado State University, Ft. Collins, Colo., 43 - 44.------------ , and W. Kocmond, 1971: Summary o f CN investigationsat the 1970 International Workshop. In Second InternationalWorkshop on Condensation and Ice Nuclei(L. Grant, Ed.), August 1970, Colorado State University,Ft. Collins, Colo., 9 2 -9 7 .Saxena, V., and J. Kassner, 1970: Thermal-diffusion chambersas cloud-nuclei counters. In Precipitation Scavenging, AECSymposium Series 22, 217 - 238.49

ATMOSPHERIC TECHNOLOGYInstrumentation for the Detection of Ice NucleiG. Garland Lala, SU N Y at AlbanyThe importance o f the ice phase in the form ation o f precipitationat high and middle latitudes has long been accepted bycloud physicists. The role o f ice nuclei in the initiation o f theice phase in clouds has been well established, but the relationshipo f the concentration o f ice nuclei to the concentration o fice crystals in clouds has remained elusive. This is owed in partto the inadequacy o f our techniques fo r measuring ice nucleusconcentrations and in part to the complicated nature o f icem ultiplication processes, whereby ice crystals increase innumber by interacting w ith crystals and droplets in the cloudenvironment. The discrepancy among various ice nucleuscounters was clearly demonstrated at the Second InternationalWorkshop on Condensation and Ice Nuclei, where the typicalspread in the measurements was 100:1, w ith extremes of10,000:1 (Bigg, 1971). The treatment given here w ill notattem pt to explain these differences, but rather w ill concentrateon the principle o f operation o f various instruments andthe relationship o f their results to ice nucleation mechanisms.Ice Nucleation MechanismsFreezing nucleation (heterogeneous initiation o f the liquid-solid phase transition) may occur either by a nucleus makingcontact w ith a drop or by a nucleus immersed in a drop. Adistinction is made between tw o types o f freezing nucleationbecause experimental laboratory evidence has shown contactnucleation to be more effective than immersion nucleation. Innature, contact nucleation can occur by the collection o f a dryparticle through any o f several scavenging mechanisms.Immersion o f a particle may occur in the same manner,depending on the nature o f the particle and the circumstanceso f collection, or through condensation on a particle. If thecondensation takes place on a mixed nucleus (partiallysoluble) that contains an ice nucleus, freezing nucleation willbe complicated by the freezing point depression due to thepresence o f a dissolved salt.Homogeneous freezing occurs through the random form a­tion o f an embryo in the liquid phase. This mechanism occursprim arily in cirrifo rm clouds and cumulus anvils, since thenucleation rate fo r homogeneous freezing becomes im portantonly at temperatures approaching —40°C. Ice nucleus detectorsare rarely, if ever, operated at temperatures this low; thushomogeneous freezing can be excluded from practicalconsideration.Four processes may be considered as possible mechanismsfo r ice phase nucleation in the atmosphere: deposition (sublimation), immersion freezing, contact freezing, and homogeneousfreezing. Deposition nucleation does not require thepresence o f water droplets fo r activation. This is usuallyconsidered as the direct form ation o f ice on a nucleus from thevapor (Fletcher, 1962). A related process, called sorptionnucleation, involves adsorption o f water molecules onto anucleus, follow ed by freezing. As referred to here, depositionnucleation w ill include both o f these processes since they areindistinguishable in most ice nucleus detectors.Ice Nucleus Counters: M inim um RequirementsMost ice nucleus counters operate by bringing an air oraerosol sample into an environment that simulates, as nearly aspossible, conditions found in supercooled clouds. Some o f theparticles in the sample act as nuclei; the resulting ice crystalsgrow to a detectable size and are counted by some method.Thus the results o f nucleation events (ice crystals) are counted,not the events themselves. The performance o f an ice nucleusdetector is very sensitive both to how well it simulates supercooledcloud conditions and to how well it is able to maintainthese conditions reproducibly. Individual ice nucleus detectorsd iffe r prim arily in the way the simulated environment isproduced and in the way the resulting crystals are counted.A u th o rG. Garland Lala, a research associate at the A tmosphericSciences Research Center o f the State University o f New York(SUNY) at Albany, received a B.S. from Massachusetts Instituteo f Technology in 1967, and M.S. and Ph.D. degrees fromS U N Y at Albany in 1970 and 1972, respectively. His researchinterests center on the study and measurement o f cloudcondensation and ice nuclei, as well as numerical modeling o fthe formation o f radiation fog.The significant variables in simulating a cloud environmentare temperature, hum idity, droplet size and concentration, andtime available fo r nucleation to occur. Instruments usingdifferent methods do not produce the same conditions, whichundoubtedly accounts fo r most o f the variations among icenucleus counters. Because o f these differences in operatingconditions, care must be exercised in extending the results toprocesses in clouds.Estimates o f the volume o f sample required can be obtainedfrom examination o f crystal concentrations measured in50

NUMBER 8 -SPRING 1976clouds and o f data from the various instruments. The extremeso f cloud crystal concentrations range from 100 crystals percubic meter to values approaching 106 crystals per cubicmeter. This sets a lower lim it on sample volume o f 10- 2 m 3and at the same time requires the counting scheme to handleup to 104 ice crystals in the sample volume. Such a highconcentration is rarely observed in ice nucleus detectorsampling from the atmosphere. A more realistic value o f themaximum count to be expected from a 0.01 m 3 sample is onthe order o f 103. The measured temperature spectrum o f icenuclei (Fig. 1) shows an increase o f a factor o f ten fo r each 4°o f supercooling. For an instrum ent designed to operate overthe range o f temperature from —15 to —30°C, this correspondsto a range o f concentration o f 103 or more. Thus, anice nucleus counter must use a minim um o f 0.01 m 3 o f sampleair and be able to provide accurate counts fo r concentrationsup to 104 crystals in this volume.To place ice nucleus counts in perspective relative to thetotal aerosol count, consider a single ice nucleus at —15°C.This particle is outnumbered by Aitken nuclei by a factor of107 to 109 and by cloud condensation nuclei by a factor o f106 to 107. Thus, in terms o f the total aerosol concentration,ice nuclei are present only in trace concentrations.errors if care is not taken to prevent frost form ation on thechamber walls. Frost crystals that break o ff and fall into thesugar solution are indistinguishable from crystals formed in thecloud.• The operation of diffusion chambers is similar to that o fm ixing chambers. In the diffusion chamber, the air sample iscooled to the prescribed temperature and then hum idified byallowing water vapor from a warm reservoir to diffuse into thesample volume, resulting in a supercooled cloud. Ice crystalconcentrations are determined by counting in an illum inatedvolume or by collection w ith a supercooled sugar solution.Temperature control w ith diffusion chambers is better thanwith m ixing chambers, but the hum idity varies from highsupersaturations near the moisture source to slight saturationfar from the moisture source.• The cloud-settling chamber is similar to the diffusionchamber, but the hum idity control is better. The chamberdesigned by Ohtake (1971) consists o f an isothermal cold boxcapped by a Plexiglas cylinder which supports a warmmoisture source (Fig. 2). The unique feature o f this chamber isthat condensation occurs at near room temperature w ithin thePlexiglas section and the resulting droplets settle into the coldSeveral types o f ice nucleus counters are used routinely todetermine ice nucleus concentrations. These instruments usedifferent methods to model the cloud environment and d iffe r­ent schemes fo r counting ice crystals. The follow ing discussionw ill present an overview o f the more common instruments andprovide a rough interpretation o f their characteristics in termso f nucleation mechanisms.Chambers: M ixing, D iffusion, Cloud-SettlingFig. 7 Typical natural ice nucleus temperature spectrum exhibiting anexponential form.• Mixing chambers are cold boxes o f large volume in whichclouds are maintained by adding moisture from a dropletgenerator or by evaporation from a warm water reservoir. Theair sample to be studied is introduced into the chamber whereit mixes w ith the supercooled cloud, resulting in the form ationo f ice crystals. Crystal concentrations are determined byvisually counting the number o f ice crystals in an illum inatedvolume or by collecting crystals in a supercooled sugar solutionby sedimentation. The sugar solution provides a mediumfo r the growth o f ice crystals to m illim eter sizes in less than aminute.Temperature and hum idity control in m ixing chambers isnot very good. In regions where the sample is introduced, thehum idity is generally less than water saturation, but highsupersaturations may occur in the region near the moisturesource. Visual counting in a small illum inated volume can alsocontribute large errors, especially at low crystal concentrations.Use o f the supercooled sugar solution can also lead toTEMPERATURE (°C)51

ATMOSPHERIC TECHNOLOGYThe Colorado State University (CSU) isothermal cloudchamber (Slusher, Katz, and Grant, 1971) is a unique combinationo f both the diffusion and cloud-settling principles.This instrum ent is a large, cylindrical chamber with a workingvolume o f 1.4 m 3 whose temperature can be regulated w ithin0.1 °C. Moisture— in the form o f droplets produced by anultrasonic nebulizer— is continuously introduced into thecenter o f the chamber and distributed throughout the volume.Aerosol samples are introduced by a syringe through a smallopening in the side o f the chamber. The number o f ice crystalsform ing in the chamber is determined by observing sedimentationonto microscope slides near the bottom o f the chamber.This detection method lim its the apparatus to use fo r highconcentrations such as those one would expect from artificialnuclei.Fig. 2 Cloud-settling chamber.Fig. 3 NCAR acoustic counter.A ll o f these chambers can activate nuclei by deposition andimmersion freezing, but the available time (on the order o f3 - 5 min) is generally too short fo r any appreciable contributionby contact freezing. The cloud-settling chamber and theCSU isothermal chamber are possible exceptions: the largedroplet concentration in the form er and the long time availablein the latter make it possible to activate a measurablefraction o f the contact nuclei. Deposition nucleation isprobably inhibited in chambers w ith high transient supersaturationsbecause on most nuclei condensation w ill occurfirst. Interpretation o f the counts from m ixing and diffusionchambers is quite complicated because o f the lack o f controlover the temperature and hum idity, but the situation is betterw ith the cloud-settling chamber, where these conditions arecontrolled and reproducible. The performance o f the cloud-settling chamber at the Second International Workshopshowed it to be one o f the more consistent instruments, givingreasonable results (Bigg, 1971).NCAR Acoustical Countersection o f the chamber, avoiding the large transient supersaturationswhich can occur when warm saturated air is mixedw ith very cold air. The mean drop size can be controlled bythe temperature o f the hum idifier, w ith a typical meandiameter o f 1 0 ^m corresponding to a supersaturation o fabout 1%. The operation o f this chamber is the same as fo r adiffusion chamber, and the crystals are detected by means o f asupercooled sugar solution.The NCAR acoustical counter is essentially a continuousmixing chamber (Fig. 3). Operation consists o f adding condensationnuclei and water vapor to the sample follow ed bycooling (by m ixing and conduction o f heat to the walls o f thecold chamber), resulting in the form ation o f a supercooledcloud. A t the bottom o f the chamber, the sample is drawnthrough an acoustical detector in which the air and particlesare accelerated and then rapidly decelerated, resulting in anaudible click fo r particles greater than 20jum in diameter. Theacoustical signal is detected by a microphone and countedelectronically. Successful operation requires that the icecrystals exceed the acoustical threshold size while the clouddroplets remain smaller and pass through the sensor undetected.This condition is insured by adding sufficient saltnuclei to keep the drop size below the threshold value.52

NUMBER 8 -SPRING 1976Frost form ation is prevented by m aintaining a constant flowo f glycol down the chamber walls, which are covered w ith aporous lining to ensure even distribution. The glycol circulationis a closed system driven by a pump which carries theglycol from a reservoir to the top o f the chamber where itflows down the walls into a collector and back to the reservoir.ThermistorsThe supersaturation produced at the point o f m ixing isquite high, w ith the result that most o f the particles presumablyform ice crystals by condensation followed by freezing(immersion freezing). The sample flo w rate lim its the residencetime in the chamber to about 2 min, which is too short fo r anyappreciable nucleation by contact nuclei that may havesurvived the initial high supersaturation.One o f the greatest problems o f this counter is the lowcounting efficiency (due to the low air velocity in thechamber), which results in a considerable loss o f crystals bysedimentation before they can reach the sensor. Correctionfactors have been determined experimentally fo r several typeso f nuclei (Langer, 1973) w ith a typical value o f nine fo r atm o­spheric nuclei. This study showed a slight dependence o f icenucleus concentration on temperature as well as on the type o fnucleant.The capability o f continuous operation is a strong asset o fthis instrument, making it very useful fo r plume tracking incloud-seeding experiments. With careful calibration the instrumentshould provide consistent results, but interpretation o fthe counts w ill always be d iffic u lt because o f the high supersaturationoccurring at the sample inlet.Fig. 4 Filter processing chamber.covered w ith ice and a smaller section is the colder filte rsupport. A nother design by Gagin and Aroyo (1969) uses ageometry similar to that o f a thermal gradient diffusionchamber, w ith the top being the ice surface and the bottomthe colder filte r support (Fig. 4). Designs employing a flo w o fice-saturated air over a colder filte r have been used by Langer(1971) and by Bigg and Stevenson (1970).F ilte r TechniqueThe filte r technique represents an entirely differentapproach to the problem o f ice nucleus concentrationmeasurement. B riefly, the technique consists o f sampling aknown volume o f air by filtering, w ith subsequent activationo f ice nuclei on the filte r and growth o f ice crystals in atemperature- and hum idity-controlled chamber. The majoradvantages o f the technique are that a prescribed volume o fair can be sampled fo r processing at a later time and theconditions o f processing can be carefully controlled.The processing chambers used most recently operate on thesame principle as a dew-point hygrometer. The filte r is held ata constant temperature (Tf) in an ice-saturated environment ata warmer temperature (77). The temperature difference(7/ — Tf) determines the hum idity over the filte r. A fte r aprocessing period (from 15 min to 1 h), the ice crystals thathave grown on the nuclei are counted visually with the aid o f alow-power microscope. A design by Stevenson (1968) uses aninsulated volume in which part o f the bottom surface isThe modes o f nucleation possible in the filte r technique areprobably lim ited to deposition and condensation on mixednuclei followed by freezing. Com petition among the growingparticles would in most cases restrict the maximum achievablehum idity to an amount less than saturation, preventing theform ation o f droplets necessary fo r other modes o f nucleation.Huffman (1973) used the filte r technique at subsaturatedconditions and found that the spectrum o f deposition nucleicould be expressed asN = C Siawhere N is the number o f nuclei, S i is the supersaturation overice, and C and a are constants. In the range o f Si from 8 to22%, the range o f a fo r natural aerosol was from three toeight; fo r silver iodide the range was about the same but variedsomewhat w ith the method o f preparation.The choice o f a chamber height is quite im portant in thedesign o f a processing chamber. Work in our laboratory andexperiments at the International Workshop on Ice Nuclei(Laramie, Wyoming, 1975) have shown that decreasing the53

ATMOSPHERIC TECHNOLOGYchamber height from 1 to 0.25 cm increases the crystal countby a factor o f about four. Analysis o f the data indicates thatcrystal counts increase exponentially with decreasing chamberheight. The reason is that reduction o f the chamber heightincreases the hum idity over the filte r by increasing the availablemoisture flu x and results in the activation o f additionalnuclei.The major deficiency o f the filte r method is the lack o fcorrelation o f nucleus concentration w ith sample volume. Thiseffect was analyzed by Mossop and Thorndike (1966) andattributed to lower humidities in areas around hygroscopicnuclei. A numerical model o f a filte r processing chamber byLala and J iusto (1972) also indicated this effect and showedthat growing ice crystals also act to reduce the averagehum idity over the filte r. The results o f the model indicate thatit may not be possible to achieve water saturation in aprocessing chamber because o f the com petition o f crystals andhygroscopic particles fo r the available water vapor. Vali andHuffm an (1973) analyzed the volume effect experimentallyand theoretically and showed that a correction could bederived on the basis o f the number o f cloud and ice nuclei inthe sample. Further w ork is needed to substantiate the validityo f this correction and show its applicability to otherprocessing chambers that use a different geometry.A chamber using a flow ing stream o f air was devised byLanger (1971) in an attem pt to increase the available watervapor and reduce the magnitude o f the volume effect. A t thepresent time the author is not aware o f any comparativestudies which show this to be true.agent is required. Stevenson (1968) and Gagin and A royo(1969) have used petroleum jelly fo r this purpose, while othershave used a variety o f thick oils. Preparation w ith the petroleumjelly consists o f placing the filte r on a smooth surface o fpetroleum jelly and heating to a temperature o f about 45°C,allowing the softened petroleum jelly to fill the filte r poresw ith o u t flooding the upper surface o f the filte r. Some practiceis required to prepare filters reproducibly. Heating the filte rand the nuclei may alter the behavior o f the ice nuclei, butthere is no experimental evidence fo r this.The choice o f filte r type is also im portant fo r the success o fthe technique. Tests at the recent workshop indicate that thestandard M illipore filters are significant sinks fo r water vapor.Higher counts were obtained from Sartorius filters o f both thestandard and hydrophobic type, w ith o u t an increase in backgroundcounts. Hydrophobic Sartorius filters gave countsabout twice those o f M illipore filters. Thus, to achieve arealistic count, it is im portant to choose a filte r which doesnot interfere w ith the processing.The filte r method is potentially one o f the best methods fo rmeasuring ice nucleus concentrations under controlled conditions.If improvements in the design o f the processingchambers can overcome the volume effect and achieve reliablehumidities over the filte r, the method may well become thestandard fo r measuring ice nuclei activated by immersionfreezing and deposition.Drop-Freezing Spectrom eterAnother aspect o f filte r processing that can lead to errors isthe filte r preparation technique. To ensure good contact o f thefilte r w ith the supporting substrate, some sort o f bondingThe instruments discussed above are all used to analyze theice nucleus content o f the air samples, whereas the drop-freezing spectrometer is used to analyze the ice nucleusTable 1General Characteristics o f Ice Nucleus CountersCOUNTER CLOUD HUMIDITYTEMPERATURECONTROLNUCLEATION MECHANISMSTIME (s)' FVeezkig0 Contact Freezing DepositionMixing chamber Yes Subsaturation to slightsupersaturationDiffusion chamber Yes Slight subsaturation tosupersaturationFair Yes No Yes 200Fair Yes No Yes 200Drop settling chamber Yes Slight supersaturation Good Yes No Yes 200NCAR acoustical counter Yes High supersaturation Good Yes No No -150Filter method No Subsaturation to possiblesmall supersaturationExcellent Yes Yes Yes 900 - 5,600Drop spectrometer No Excellent Yes No No54

NUMBER 8 -SPRING 1976content o f liquid water. The ice nucleus spectrum o f a liquidsample can be used to deduce the ice nucleus concentration inprecipitation, thus providing an indication o f the number ofnuclei that have been incorporated into precipitation by cloudprocesses. The technique can also be used to study suspensionso f soil particles in liquids to give an indication o f the numbero f potential nuclei in soils.The freezing nucleus spectrum is measured by placing anumber o f droplets on a temperature-controlled surface anddetermining the number o f droplets that freeze w ithin a giventemperature interval as the temperature is lowered. Vali andStansbury (1966) have indicated that the temperaturespectrum o f ice nuclei in the sample can be determined fromthese data. An automatic system to control the cooling, tom onitor the temperature, and to record freezing events hasbeen devised and is described in detail by Vali and K nowlton(1971).environment. The wide variation o f nucleus counts fromdifferent instruments is an indication not so much o f inaccuracyo f the instrum ent as o f the sensitivity o f ice nucleationprocesses to environmental conditions. What is required is aclearer understanding o f the influence o f environmentalconditions on each mechanism o f ice nucleation. Thisundoubtedly w ill further the understanding o f ice processes inclouds and facilitate the interpretation o f ice nucleus counts interms o f nucleation mechanisms. It is not likely that any oneice nucleus counter w ill be able to provide concentrationmeasurements appropriate to all nucleus types and cloudconditions. Rather, several instruments w ith differingoperating principles w ill be required to assess the importanceo f individual nucleation mechanisms under carefully prescribedconditions.The drop-freezing technique offers a convenient way o fstudying the temperature spectrum o f freezing nuclei in liquidsuspensions. However, because other modes o f nucleationoccur in clouds, it must be used in conjunction w ith othertechniques if complete inform ation on ice nuclei in theatmosphere is to be deduced.A m odification o f the drop-freezing system fo r the measuremento f contact nuclei has been reported by Vali (1974). Inthis system, a captive population o f droplets is held in a supercooledstate while the aerosol is electrostatically deposited onthem. The ratio o f contact nuclei can be deduced by thawingand refreezing the drops. This technique produces highercounts than other types o f ice nucleus counters, but theconcentrations are still lower than would be necessary toexplain crystal concentrations in clouds.SummaryIt is very d iffic u lt to compare ice nucleus counters except inthe most general terms because o f variations in their operatingconditions and because o f the large spread in their results.Even instruments o f the same design can show significantvariations. A summary o f the general characteristics o f the icenucleus counters described here is presented in Table 1. For amore detailed comparison o f the results o f many ice nucleuscounters, the reader is referred to the report o f the SecondInternational Workshop on Condensation and Ice Nuclei(Grant, 1971).Ice nucleus counters all share the common objective o fsimulating the cloud environment in which ice nucleationoccurs. The physical principles o f operation and instrum entconfiguration vary widely and all can, to some degree, simulatesome o f the im portant aspects o f the supercooled cloudReferencesBigg, E. K., 1971: Report on the ice nucleus workshop. InSecond International Workshop on Condensation and iceNuclei (L. Grant, Ed.), August 1970. Colorado State University,Ft. Collins, Colo., 97 - 105.------------ , and C. M. Stevenson, 1970: Comparison o f ice nucleiin different parts o f the w o rld ./. Rech. Atmos. 4, 41 - 58.Fletcher, N. H., 1962: The Physics o f Raindouds. CambridgeUniversity Press, 210 pp.Gagin, A., and M. Aroyo, 1969: A thermal diffusion chamberfo r the measurement o f ice nucleus concentrations./. Rech.Atmos. 4, 115 -122.Grant, L. (Ed.), 1971: Second International Workshop onCondensation and Ice Nuclei, August 1970. Colorado StateUniversity, Ft. Collins, Colo., 149 pp.Huffm an, P. J., 1973: Supersaturation spectra o f Agl andnatural ice n u cle i./. Appi. Met. 12, 1080- 1082.Lala, G. G., and J. E. Jiusto, 1972: Numerical estimates o fhum idity in a membrane filte r ice nucleus chamber. /. Appl.Met. 7 7 ,6 7 4 - 683.Langer, G., 1971: NCAR ice nucleus analyzer. In SecondInternational Workshop on Condensation and ice Nuclei(L. Grant, Ed.), August 1970. Colorado State University,Ft. Collins, Colo., 55 - 56.55

ATMOSPHERIC TECHNOLOGY------------ , 1973: Evaluation o f NCAR ice nucleus counter.Part I: Basic o p e ratio n./. Atmos. Sci. 12, 1000 - 1011.Mossop, S. C., and N. S. C. Thorndike, 1966: The use o fmembrane filters in measurements o f ice nucleus concentrations.I. E ffect o f sampled air vo lu m e./. Appl. Met. 5,474 - 480.Ohtake, T., 1971: Cloud-settling chamber fo r ice nuclei count.I n Second International Workshop on Condensation and IceNuclei (L. Grant, Ed.), August 1970. Colorado State University,Ft. Collins, Colo., 58 - 60.Slusher, T., U. Katz, and L. Grant, 1971: CSU isothermalcloud chamber. In Second International Workshop onCondensation and Ice Nuclei (L. Grant, Ed.), August 1970.Colorado State University, Ft. Collins, Colo., 6 0 -6 1 .Stevenson, C. M., 1968: An improved M illipore filte r techniquefo r measuring the concentration o f freezing nuclei inthe atmosphere. Q. /. R. Met. Soc. 94, 35 - 43.Vali, G., 1974: Contact ice nucleation by natural and artificialaerosols. In Preprint V ol., Conference on Cloud Physics,held 21 - 25 October at Tucson, A riz.; American MeteorologicalSociety, Boston, Mass., 34 - 37.------------ , and P. J. Fluffman, 1973: The effect o f vapor depletionon ice nucleus measurements w ith membrane filters./. Appl. Met. 7 2 ,1 0 1 8 -1 0 2 4 .------------ , and D. K now lton, 1971: An automated drop freezersystem fo r determining the freezing nucleus content ofwater. In Second International Workshop on Condensationand Ice Nuclei (L. Grant, Ed.), August 1970. Colorado StateUniversity, Ft. Collins, Colo., 75 - 79.------------ , and E. J. Stansbury, 1966: Time-dependent characteristicso f the heterogeneous nucleation o f ice. Can. /.Phys. 44, 477 - 502.56

NUMBER 8 -SPRING 1976A Review ofRecently Developed Instrumentationto Measure Electric Fields inside CloudsW. David Rust, National Oceanic and A tm ospheric A dm inistrationThe reliable measurement o f electric fields inside clouds isrecognized as fundamental to attempts to describe quantitativelythe electrical conditions that exist inside clouds,including the location, polarity, and magnitude o f chargecenters and the regions where maximum fields are most likelyto occur. Some attempts were made in the early 1900s to useballoons to measure electric fields aloft (summarized inChalmers, 1967), but most o f the early measurements wereobtained when clouds engulfed measuring sites, which werefrequently located atop mountains. It is particularly im portant,however, to measure electric fields away from theboundary effects o f the earth.The measurement o f electric fields inside clouds is d ifficu lt.The apparatus and its transporting vehicle distort the quantitybeing measured; the high hum idity and precipitation generallyencountered may cause instrumental problems, such as leakageacross high-impedance insulation. The charging o f the instrumentby precipitation or other means may produce a field thatis often d iffic u lt to separate from the external field. Moreover,charged precipitation striking and splashing from the instrumentcan cause severe noise problems, and corona dischargeA uthorW. David Rust conducts research on thunderstorm electrificationand iightning suppression for the A tmospheric Physicsand Chemistry Laboratory o f the National Oceanic andAtmospheric Administration. He graduated from SouthwesternUniversity and pursued further studies in physics atNew Mexico Institute o f Mining and Technology, earning hisPh.D. in 1973. As a recipient o f a National Research Councilpostdoctoral resident associateship for two years, he studiedthe use o f balloon- and aircraft-borne instruments for measuringelectrical parameters of clouds. During the Viking IImission to Mars, his work included the use of real-time atmosphericelectricity measurements to protect rockets from thedangers o f triggered lightning.from the instrum ent in the presence o f high electric fields maygenerate a space charge that can alter the local electric fieldenough to vitiate a measurement.This article discusses several o f the relatively new instrumentsdeveloped to measure electric fields aloft w ithin clouds,including rocket-, balloon-, and aircraft-borne devices. Om ittedfrom this discussion are instruments that use radioactiveprobes as the sensing elements. Although this technique isuseful in some instances if used with care, its reliability inmaking measurements inside highly electrified clouds has beenquestioned by several investigators. The use o f radioactiveprobes has been described in detail by Vonnegut, Moore, andMallahan (1961); Chalmers (1967); Lane-Smith (1974); andothers.Instrum ented RocketsRockets carrying electric-field measuring devices throughclouds can produce nearly instantaneous profiles o f themeasured components o f the electric field along their flig h ttrajectories. It is d iffic u lt, however, to obtain measurementshorizontally through a cloud w ith rockets launched from theground. In addition, there are only a few locations w ithin thecontinental United States where airspace restrictions perm itthe firing o f adequately large rockets. In spite o f their lim itations,rockets are useful vehicles, and tw o different techniquesusing instrumented rockets are described here.One rocket-borne electric field meter has been developed byWinn and Moore (1971) for measuring the component o f theelectric field perpendicular to the longitudinal axis o f therocket. The rocket type used is the m ilitary Mark 40, Model 1,which has a solid-fuel m otor; the body o f the rocket is 70 mmin diameter and about 1.5 m long when equipped w ith theelectric-field measuring nose. The rockets are generallylaunched from Langmuir Laboratory, which is located on amountain ridge (at an elevation o f 3.2 km) in central NewMexico; they attain altitudes o f about 7 or 8.5 km, dependingon the launch technique used.57

ATMOSPHERIC TECHNOLOGYga |ROTATION SENSEfixed w ithin the epoxy nose o f the rocket.) The electric fieldand direction data are telemetered back to a ground receivingstation by using the electric-field signal to modulate directlythe oscillator o f an FM transm itter.Since the rocket-borne sensor must rotate in an externalfield to obtain a measure o f that field, direct calibration o f thedevice by placing it in known electric fields is d iffic u lt and hasnot been done. This is not a serious disadvantage since thecalibration obtained from the characteristics o f the electroniccircu itry and the cylindrical geometry o f the rocket is straightforwardand reasonably accurate. In addition, the rotating coil,which senses the earth’s magnetic field, is used to produce,once during each revolution, a calibration pulse whosemagnitude can be compared with the signal produced by theelectric field. This technique provides an in-flight systemcalibration.Fig. J Schematic o f instrumented rocket payload for measuring theelectric field component perpendicular to the rocket body. (Winn andMoore, 1971.)The instrum ent is shown schematically in Fig. 1. Thesensing electrodes are placed on the dielectric nose portion o fthe rocket and covered with shrinkable Teflon to increase theonset threshold o f the corona. The rocket, which is generallylaunched nearly vertically, is made to spin about its axis by theuse o f canted fins. As the rocket rotates, equal and oppositecharges are induced upon the sensing plates by the perpendicularcomponent o f the field. The plates are connected to chargeamplifiers so that their output voltage is independent o f therocket spin rate. Subtraction o f the signals from each plateusing a differential input am plifier allows a determination o fthe external field normal to the rocket’s longitudinal axis.As Winn and Moore (1971) state, care must always beexercised when making measurements o f electric fields w ithinclouds and caution used in interpreting the data. They discusstwo possible sources o f error in their instrum ent: charge onthe device and corona discharge from the rocket. For the first,any charge that is distributed on the rocket body results in adirect current (dc) offset o f the sinusoidal signal that isproduced by the external field. There is thus no d iffic u lty inseparating the field produced by charge on the rocket from theactual external field.Frier (1972) has warned o f possible vitiation o f themeasurements if corona is produced at the fro n t o f the rocket.Winn and Moore (1972) argued that even in the unlikely evento f corona from the dielectric nose ahead o f the sensors, thefield thus caused could always be distinguished from theexternal field unless the corona produced a sinusoidallyvarying space charge around the sensors. The more likely placefo r corona to occur, they contend, is at the fins, w ith the onlyresult norm ally being spike-type noise on the telemetry record.This is shown on several o f their recordings and is easilydistinguishable from signals due to the external field.Winn, Schwede, and Moore (1974) report on the maximumfields they have observed inside thunderclouds over NewMexico. They com m only measured field components withmagnitudes o f about 50 kV/m perpendicular to the rocket.The maximum measurement they quote as reliable is approximately160 kV /m ; they report tw o higher values, 400 kV/mand 1 M V/m , whose reliability they question.The direction o f this perpendicular component o f the fieldrelative to the earth’s magnetic field is determined bycomparing the phase o f a signal from a rotating coil and thatfrom the field sensors. (Both the coil and the field sensors areA second type o f rocket-borne instrum ent fo r the measuremento f the vertical electric field was described by Ruhnke(1971). The electric field is determined by measuring thecorona current that flows from a sharply pointed conductormounted on the nose. This current is related to the field in a58

NUMBER 8 -SPRING 1976nonlinear fashion, but a high-ohm resistor used in series withthe pointed tip makes the current approxim ately a linearfunction o f the electric field. The current is then convertedinto pulses that are used to modulate a 403 MHz transmitter.Some test flights have been made, and the data obtained froma penetration through the edge o f one thunderstorm have beenpresented (Ruhnke, 1971).Although the device can be calibrated directly by placing itin known electric fields, there appear to be several drawbacksto the instrum ent in its present configuration. Fields less thanthose necessary to initiate corona at the tip cannot bemeasured. This is, o f course, a problem only if measurement o flow fields is desired. Currents due to any charge transfer to thepoint by particle impaction cannot be distinguished fromthose caused by the external field, and the polarity o f the fieldcannot be determined. Because o f the design o f the telemetryoscillator circuitry, the telemetering capability apparentlyworks well only in fields up to about 40 kV /m ; this problemconceivably could be overcome with appropriate circuitchanges.Dropsonde Field M illEvans (1969a) developed a rotating-differential electric-fieldmill that he suspended beneath a parachute and dropped froman aircraft flying over thunderclouds. The instrum ent consistsessentially o f tw o vertical, cylindrical input electrodes that arealternately shielded from and exposed to the external electricfield by means o f a windowed cylinder that is rotated with asmall m otor. The inputs are connected differentially to anelectrometer. The output o f the electrometer is synchronouslyrectified and then telemetered to a receiving station. Theelectronics allow measurements to be made over a range o f100 V /m to 250 kV /m . Evans noted, however, that his laboratorycalibrations indicate that corona w ill be present above50 kV /m , which could result in large errors in the measurementso f fields above this value.Evans (1969a) described the data obtained when several o fthese field meters were dropped through thunderstorms. Hestated that the fields observed, which did not exceed about40 kV /m , were substantially lower than generally had beenassumed to exist w ithin thunderstorms.Evans’ measurements were questioned by Vonnegut (1969).Since the device is not electrically symmetrical about anyhorizontal plane, Vonnegut suggested that it was susceptible toerrors due to charge on the housing. A dditionally, he said thatproblems might be caused by the release o f p oint dischargeions, particularly from a “ sym m etry” rod placed verticallybeneath the instrument. Evans (1969b) argued that the d iffe r­ential input o f the instrum ent allowed a check on instrumentcharging; he added that the instrum ent had been tested boththeoretically and experimentally in the laboratory.Balloon-Borne Instrum entsThe use o f balloons to make measurements o f electric fieldsw ithin clouds is advantageous in that the creation o f spacecharge by high-speed impaction w ith cloud and precipitationparticles is not a problem. The use o f captive balloons offersthe additional advantage o f easy retrieval o f the instrumentation.U nfortunately, even when one is using a dielectric tetherline material, such as fiber glass, there must always be somequestion about the validity o f the electric field measurements,since tether lines span potential differences in thunderstormsthat may reach several m illion volts. Accordingly, when theflyin g line becomes wet, some current must flow , and this mayalter the electric field in the vicinity o f the instrument. Freeballoons, on the other hand, do not have this lim itation. A freeballoon system w ill tend to fo llo w the potential o f theenvironment, and so the ground tether contam ination problemis inherently eliminated. Probably the most serious drawbacksto the use o f free balloons is that control o f their position orflig h t path and retrieval o f the instrum entation at the end o f aflight are often quite d ifficu lt.Numerous types o f field-measuring devices have beendeveloped fo r use with both free and tethered balloons.Several attempts to measure electric fields w ithin clouds weremade with meteorological radiosondes that were m odifiedw ith radioactive probes or corona points. The reader isreferred to Moore, Vonnegut, and Botka (1958) and Chalmers(1967) fo r details o f these instruments. During the past decadeseveral balloon-borne field m ill systems have been developedto make electric field measurements inside clouds. Two o fthese, which have actually been used to obtain data w ithinthunderstorms, are reported here.One instrument, a field meter that was designed fo r use withtethered balloons by C. B. Moore (and reported by Clark,1971), consists o f two field mills recessed w ithin a smooth,spherical aluminum housing 0.3 m in diameter. The field millsare mounted so that one faces outward horizontally and theother downward vertically. The spherical shape o f the housingand the gently rounded m ounting ports for the mills are usedto minimize point discharge problems. Clark fitte d his fieldmeter w ith an internal tape recorder fo r data recording in hisefforts to determine the maximum fields w ithinthunderstorms.The field meter has subsequently been m odified by adding a1,680 MHz transm itter (Rust and Moore, 1974). This allowsreal-time display o f the data as they are transmitted by FM-FM59

ATMOSPHERIC TECHNOLOGYexposed to and shielded from the external field as the motordrivenrotor spins, and a periodic charge is thus induced on thestator by the external field. Use o f a charge am plifierconnected directly to the stator makes the measurement o f thefield independent o f rotor spin rate.The polarity o f the electric field is determined bycomparing the phase angle between the position o f the rotorand the time-varying charge on the stator. The position o f therotor is determined w ith a reference signal generated by aphototransistor mounted between tw o segments o f the statorand looking outward at the ambient light through the vanes o fthe rotor. Phase-sensitive demodulation o f the reference andinput signals allows a determination o f both the magnitudeand polarity o f the electric field.The maximum fields that can be reliably measured aredetermined by how deeply the mills are recessed into thespherical housing. In C lark’s w ork, the field meter was calibratedto fields o f about 2 M V /m . Rust and Moore used theinstrum ent prim arily fo r the study o f initial cloud electrificationand the early stages o f thunderstorm development, andthey placed the field mills so that 25 kV/m was the maximumthat could be measured. The calibration was done by applyingknown potentials directly to the spherical housing, thusproducing radial electric fields, and by placing the field meterbetween parallel plate electrodes across which known potentialdifferences were imposed. Both calibration techniques showthe instrum ent to be linear w ithin its intended range.Fig. 2 Arrangement o f spherically housed field mills for determiningthe electric field vector aloft. (Rust and Moore, 1974.)In use, the field meter was suspended from a motor-driven,rotating bearing by a Teflon-insulated, nylon line equippedw ith rain diverters. It revolved about its vertical axis at about10 rpm. The instrum ent was attached at least 30 m below theballoon in an e ffo rt to minimize any electrical perturbationscaused by the balloon.telemetry and received at a ground station. The quarter-wavetransm itting antenna stub is approximately 39 mm long and isencased in a rounded Teflon housing; the antenna protrudesthrough the sphere at a p oint diam etrically opposite thehorizontally facing field m ill, as may be seen in Fig. 2.Although the basic design o f field mills has been reportedby Chalmers (1967) and others, a brief description o f thoseused in this balloon-borne field meter is presented here. Thesensing element o f the field m ill consists o f a stationary, fla tstator having four vanes above which is placed a groundedrotor o f the same configuration. The stator is alternatelyThe data from both mills are used to determine the chargeon the instrum ent and the horizontal and vertical componentso f the electric field. The output from the horizontal millconsists o f a sinusoidal wave from the external horizontal fieldand a dc voltage offset from any charge on the instrument.The vertical, downward-facing m ill provides a measure o f thevertical component o f the field plus the field from charge onthe instrument.Two Hall-effect semiconductors mounted at right anglesw ithin the sphere allow a determination o f the instrum ent’sorientation relative to the earth’s magnetic field. From thecombined outputs o f all sensors, the electric field vectoroutside the instrum ent can be reconstructed and determinationsmade o f any charge on the instrum ent as a function o ftime.60

NUMBER 8 -SPRING 1976Another type o f balloon-borne field meter has beendeveloped by W. P. Winn (see Winn and Byerley, 1974) fo r usebeneath small, free balloons (Fig. 3). The instrum ent isdesigned to measure the horizontal component o f the electricfield during vertical traverses o f a thunderstorm as it ascendsbelow the balloon and then descends beneath a parachute afterit has been released from the rising balloon by a pressure-activated squib. The instrument is attached with nylon monofilament line about 20 m below the balloon and the parachute.Figure 4 is a sim plified schematic o f this field meter. The twohollow copper spheres placed horizontally opposite each otherserve as the housing fo r the electronics, the transm ittingantenna, and the field sensors. In flig h t the m otion o f the airpast coated balsa wood propeller blades located 1 m above thesensors causes the instrum ent to rotate about its vertical axisat approxim ately 60 rpm. The charge induced on the rotatingspheres by the external field produces a sinusoidal outputfrom which the horizontal component o f the electric field isdetermined. Any charge that is acquired by the sensorsthemselves can also be determined since the charge produces adc voltage that results in an offset o f the sinusoidal output.The field meter can be calibrated in a parallel-plateelectrode system; it is also calibrated internally during flig h tby the cyclic application o f 0 V and 2 V pulses. Winn andByerley (1974) have reported observations w ith this instrumento f horizontal electric fields as high as 100 kV /m w ithinthunderstorms.A irc ra ft Instrum entationFig. 3 Sketch o f the field meter designed to measure the horizontalelectric field by ascending below a balloon and then descending on aparachute. (After Winn and Byerley, 1974.)The use o f aircraft to make measurements o f electric fieldsw ithin clouds and thunderstorms makes easier the control o fwhen and where the data are obtained than is possible withballoons or rockets. A ircra ft also make it simple to obtainmeasurements horizontally through clouds, w ith severaltraverses usually possible in a relatively short time comparedwith the storm ’s development. Disadvantages o f using aircraftinclude the problem o f creating local space charge due tohigh-speed impaction w ith cloud or precipitation particles;corona from relatively sharp edges such as propellers andwingtips; charge on the aircraft as a result o f impaction,corona, or engine exhaust; and possible hazards to those onboard during penetration o f thunderstorms.Fig. 4 Simplified schematic diagram o f the field meter for measuringthe horizontal components o f electric field aloft. (Winn and Byerley,1974.)If reliable measurements are to be made using aircraft, caremust be exercised to design and install a field m ill system sothat it can be calibrated and so that the effects o f charge onthe aircraft may be determined directly from the measurementsor compensated fo r by either electronic or mechanicaltechniques.

ATMOSPHERIC TECHNOLOGYThere are tw o types o f field m ill systems that are generallyused on aircraft to obtain more than one component o f theelectric field aloft: those using several single-component mills,coupled to give field components that are orthogonal in theaircraft coordinate system, and those using field mills thatdetermine tw o orthogonal components o f the field w ith asingle instrument.These are used in the phase-sensitive rectification circuits sothat the tw o components o f the external field may be determined.The outputs corresponding to the field components arealigned w ith the geometrical axes o f the aircraft by electronicallyphase-shifting the reference signals. A fte r phase-sensitiverectification, the signals are filtered to give a direct currentoutput proportional to the electric field.One system that uses four single-component field mills isthat developed by Fitzgerald and Byers (1962) and used on aC -1 3 0 aircraft. They mounted one m ill facing outward oneach w ingtip; the remaining tw o were mounted fore and aftalong the underside o f the fuselage and faced downward. Thefield mills were o f the standard induction type and used thefla t stator-rotor configuration described previously; polarity o fthe field was determined by phase-sensitive rectification o finput and reference signals.Calibration o f the system was done theoretically byapproxim ating the aircraft as an ellipsoid and then experimentallydetermining estimates o f the necessary field enhancementfactors. These factors were then used to determine theapproximate magnitudes o f the measured fields.By electronically combining the outputs from appropriatemills, Fitzgerald and Byers (1962) could determine the threecomponents o f the electric field and the field due to charge onthe aircraft. The magnitude o f the charge on the aircraft wasthen calculated from the field due to the charge and thecapacitance o f the aircraft, again approximating the aircraft asan ellipsoid.Numerous cloud penetrations were made w ith the C -1 3 0and also w ith a sim ilarly instrumented F - 1 0OF jet aircraft.Fitzgerald (1965) reports fields well above 200 kV /m w ithinactive thunderstorms, with magnitudes near 50 kV/mapparently fa irly common.Shown in Fig. 5 is a cylindrical field mill designed byKasemir (1972) that is used to measure two components o fthe electric field. By m ounting one m ill on the aircraft noseand a second m ill on the underside or top o f the fuselage, allthree components o f the external field can be measured. Atypical mounting and coordinate system are shown in Fig. 6.The most recent version o f the field m ill (Kasemir andHolitza, 1972) is shown in the sim plified diagram in Fig. 7.The external electric field induces an alternating charge on thetw o cylindrical sensor segments as they rotate. This results in acurrent flo w through the rotating capacitor to the stationaryelectronic circuitry. Two reference signals, 90° out o f phase,are produced by a signal generator coupled to the field millm otor (only one reference output is shown in the figure).The field mills themselves are calibrated in the laboratorybetween parallel plate electrodes. Calibration o f the field millsystem on the aircraft is done in tw o steps. First the verticalcomponent measured on the aircraft is calibrated by using thevalue o f the vertical fine-weather field, which is measuredusing a suitable field m ill located at the ground. The tw ohorizontal components are then calibrated by flyin g theaircraft in a series o f banks, climbs, and dives whose anglesrelative to the vertical fine-weather field are determined. Thisdirect calibration compensates fo r the distortion o f the fieldcaused by the aircraft.The effects o f charge on the aircraft are reduced to am inim um by the use o f sm oothly rounded metal bosses called“ hump rings,” which are mounted on the cylindrical housingjust below the sensing elements (see Fig. 5). The procedure isto charge the aircraft in flig h t either by changing engine powersettings (exhaust charging) or preferably by placing a probeoutside the exterior o f the aircraft and applying high voltageto it. It is o f interest to note that aircraft have a tendency tobe electrically charged. The details o f aircraft charging byengine exhaust are not fu lly understood, but the emission ofcharged engine gases results in charging until a charge-limitingprocess causes equilibrium . The charge acquired in this mannerduring cruise flig h t conditions is significantly less than thatobtained by use o f the corona probe.A pplication o f high voltage to the probe results in coronacurrent that flows into the air. As the airstream flow ing by theaircraft carries away the corona-produced ions o f one polarity,the aircraft tends to be charged to the opposite polarity.Equilibrium is reached when the charge on the aircraft causesan electric field sufficient to deflect a portion o f the ions backto the aircraft skin. This generally takes only a few seconds,and at this equilibrium point, the charge on the aircraft isquite constant. The hump ring is then moved and rotated untilthe charging procedure causes no noticeable outputs from thefield m ill even on sensitive ranges. This is a slow and tedioustask since the hump ring cannot be moved in flig h t, but thetechnique is reliable.The dynamic range o f the system is determined by means ofrange-switching circuitry located w ithin a control panel. Full-scale ranges o f approxim ately 50 V/m to 500 kV/m arecovered in 13 steps that change fu ll scale by factors o f either 2or 2.5. A ircra ft equipped and calibrated as above have6 2

NUMBER 8 -SPRING 1976measured electric fields as high as 280 kV /m inside thunderstorms,with values o f 100 kV /m frequently observed even inrelatively small thunderstorms (Holitza et al., 1974)., 1 ' I P f j j g * v %Concluding RemarksSENSORSSeveral o f the field m ill systems that have been discussedhere are currently in use and are updated and improved on acontinuing basis. There are advantages and problems in eachtype o f system, and the use o f each results in some sort ofcompromise. Many investigators in atmospheric electricitybelieve, however, that the instrumentation necessary tomeasure electric fields inside the cloud and thunderstormenvironment has been developed to a degree sufficient to allowreasonably reliable measurements. The continuing use o f theseinstruments is a major part o f the e ffo rt to describe quantitativelythe electrical growth and behavior o f clouds. In additionto electric fields, the charges carried by cloud and precipitationparticles, their size distributions, and other parameters o fFig. 5 A cylindrical field m ill mounted on the nose o f an aircraft andused to measure two components o f the electric field. (Kasemir, 1972.)Fig. 6 Cylindrical field mills (indicated byasterisks) are located along tines o f aircraftsymmetry; shown is a typical coordinatesystem used to define the three componentso f the electric field, described here in termso f potential gradients, i.e., G points towardpositive charge.i------------- ----------------------------------------------------- 163

ATMOSPHERIC TECHNOLOGYcloud physics need to be measured simultaneously near andw ithin clouds and thunderstorms if we are to determine theinterrelationships between electrical and microphysicaldevelopments.Kasemir, H. W., 1972: The cylindrical field m ill. MeteoroiogischeRundschau 25, 3 3 - 38.------------ , and F. J. Holitza, 1972: Description and InstructionManual for the Cylindrical Field Mill System. NOAATechnical Memorandum, NOAA TM ERL APCL - 14,20 pp.Acknowledgments. I thank F. J. Holitza and W. P. Winn fo rtheir helpful suggestions. This w ork has been supported by theNational Research Council (NRC) under an NRC/NationalOceanic and Atmospheric A dm inistration Resident ResearchAssociateship.Lane-Smith, D. R., 1974: Review o f instrum entation fo ratmospheric electricity. Paper presented at F ifth InternationalConference on Atmospheric E lectricity (sponsoredby International Council o f Scientific Unions and WorldMeteorological Organization), held 2 - 7 September atGarmisch-Partenkirchen, Germany; book publicationpending.ReferencesChalmers, J. A., 1967: Atmospheric Electricity, 2nd ed.Pergamon Press, O xford, 515 pp.Clark, D., 1971: Balloon-Borne Electric Field Mills for Use inThunderclouds. M.S. thesis, New Mexico Institute o f Miningand Technology, Socorro, N. Mex., 60 pp.Moore, C. B., B. Vonnegut, and A. T. Botka, 1958: Results o fan experiment to determine initial precedence o f organizedelectrification and precipitation in thunderstorms. RecentAdvances in Atmospheric Electricity (L. G. Smith, Ed.),Pergamon Press, Elmsford, N.Y., 333 - 360.Ruhnke, L. H., 1971: A Rocket Borne Instrument to MeasureElectric Fields inside Electrified Clouds. NOAA TechnicalReport, NOAA TR ERL 206 APCL - 20, 26 pp.Rust, W. D., and C. B. Moore, 1974: Electrical conditions nearthe bases o f thunderclouds over New Mexico Q. J. R. Met.Soc. 100, 450 - 468.Evans, W. H., 1969a: Electric fields and conductivity inthunderclouds./. Geophys. Res. 74, 939 - 948.------------ , 1969b: R e p ly./. Geophys. Res. 74, 7056 - 70 5 7.Fitzgerald, D. R., 1965: Measurement techniques in clouds.Problems o f Atmospheric and Space Electricity (S. C oroniti,Ed.). Elsevier Publishing Company, New Y ork, N.Y.,199 - 212.------------ , and H. R. Byers, 1962: Aircraft ElectrostaticMeasurement Instrumentation and Observations o f CloudElectrification. A ir Force Cambridge Research LaboratoriesTechnical Report, AFC R L - TR - 62 - 805, AFCRL,Bedford, Mass.Frier, G. D., 1972: Comments on “ Electric field measurementsin thunderclouds using instrumented rockets,” by W. P.Winn and C. B. M oore ./. Geophys. Res. 77, 505.Holitza, F. J., H. W. Kasemir, W. E. Cobb, and W. D. Rust,1974: A ircra ft measurements o f electric fields in and belowthunderstorms (abstract). £©S, Trans. A G U 56, 1131.Vonnegut, B., 1969: Discussion o f paper by W. H. Evans,"E lectric fields and conductivity inside thunderclouds.”/. Geophys. Res. 74, 70 53 - 70 5 5.------------ , C. B. Moore, and F. J. Mallahan, 1961: Adjustablepotential-gradient measuring apparatus fo r airplane use./. Geophys. Res. 66, 2393 - 2397.Winn, W. P., and L. G. Byerley, 1974: Measurements o felectric fields in thunderclouds using a balloon-borne instrument(abstract). £©S, Trans. A G U 56, 1131.------------ , and C. B. Moore, 1971: Electric field measurementsin thunderclouds using instrumented rockets./. Geophys.Res. 76, 5003 - 5017.------------ , a n d ------------ , 1972: R e p ly./. Geophys. Res. 77,506 - 508.------------ , G. W. Schwede, and C. B. Moore, 1974: Measurementso f electric fields in thunderclouds./. Geophys.Res. 79, 1761 - 1767.64

NUMBER 8 - SPRING 1976C ontributions from our readers are printed in this “ ShortReports” section. Further inform ation on these subjectsshould be obtained from the contributors rather than from theeditor. Short reports (maximum 750 words) may deal withnew techniques fo r atmospheric measurements and dataprocessing, evaluation o f existing instrum entation, analysis o funfilled needs fo r systems and techniques, and other topicsrelated to the technology o f atmospheric measurement. A rtshould be held to a m inim um and should be submitted incamera-ready form . We also welcome letters to the editorcommenting on materials published in AtmosphericTechnology. Our intent is to encourage responsive and criticalreading o f the publication’s contents and a sharing o f opinionsamong readers. Articles may be sent to the Managing Editor o fAtmospheric Technology, Publications O ffice, NCAR.CorrectionEquation (1) o f the Short Report by E. N. Brown, “ A ircraftStatic Pressure Errors: Their Measurement and Influence,”Atmospheric Technology No. 7, Fall 1975, p. 89, was printedincorrectly. The correct equation is7 i71Our thanks to reader P. Church, who called the error toour attention.