An Example of Supercooled Drizzle Drops Formed through a Collision-Coalescence Process

View More View Less
  • 1 Cloud Physics Research Division, Atmospheric Environment Service, Downsview, Ontario, Canada
© Get Permissions
Full access

Abstract

The microphysics associated with observations of supercooled drizzle drops, which formed through a condensation and collision-coalescence process, are reported and discussed. The growth environment was an 1100-m-thick stratiform cloud with cloud-base and cloud-top temperatures of −7.5° and −12°C, respectively. The cloud was characterized by a low droplet concentration of 21 cm−3 and a large droplet median volume diameter of 29 µm, with a concentration of interstitial aerosol particles of less than 15 cm−3 (larger than 0. 13 µm in diameter). The evolution of drizzle drops was traced downward from cloud top, with a maximum diameter of 500 µm observed at cloud base. The air mass was sufficiently clean to ensure only a small number of active cloud condensation nuclei. Consequently, small concentrations of cloud droplets led to concentrations of over 300 L−1 for droplets larger than 40 µm, which set up strong conditions for continued growth by collision-coalescence. Ice crystals in concentrations of 0.08 L−1 were measured simultaneously with the drizzle drops and were not effective in glaciating the cloud, even though the drizzle drops were estimated to have taken at least 1–2 h to form.

While the growth of precipitation-sized drops through collision-coalescence has been well documented, there are few measurements of this phenomena at temperatures less than 0°C. This study provides a well-documented example of such an event at subfreezing temperatures. The applicability of this measurement in terms of hazardous aircraft icing is discussed.

Abstract

The microphysics associated with observations of supercooled drizzle drops, which formed through a condensation and collision-coalescence process, are reported and discussed. The growth environment was an 1100-m-thick stratiform cloud with cloud-base and cloud-top temperatures of −7.5° and −12°C, respectively. The cloud was characterized by a low droplet concentration of 21 cm−3 and a large droplet median volume diameter of 29 µm, with a concentration of interstitial aerosol particles of less than 15 cm−3 (larger than 0. 13 µm in diameter). The evolution of drizzle drops was traced downward from cloud top, with a maximum diameter of 500 µm observed at cloud base. The air mass was sufficiently clean to ensure only a small number of active cloud condensation nuclei. Consequently, small concentrations of cloud droplets led to concentrations of over 300 L−1 for droplets larger than 40 µm, which set up strong conditions for continued growth by collision-coalescence. Ice crystals in concentrations of 0.08 L−1 were measured simultaneously with the drizzle drops and were not effective in glaciating the cloud, even though the drizzle drops were estimated to have taken at least 1–2 h to form.

While the growth of precipitation-sized drops through collision-coalescence has been well documented, there are few measurements of this phenomena at temperatures less than 0°C. This study provides a well-documented example of such an event at subfreezing temperatures. The applicability of this measurement in terms of hazardous aircraft icing is discussed.

2250 JOURNAL OF APPLIED METEOROLOGY VOLUMe35An Example of Supercooled Drizzle Drops Formedthrough a Collision-Coalescence Process STEWART G. COBER, J. WALTER STRAPP, AND GEORGE A. ISAACCloud Physics Research Division, Atmospheric Environment Service, Downsview, Ontario, Canada(Manuscript received 26 October 1995, in final form 21 May 1996) ABSTRACT The microphysics associated with observations of supercooled drizzle drops, which formed through a condensation and collision-coalescence process, are reported and discussed. The growth environment was an1100-m-thick stratiform cloud with cloud-base and cloud-top temperatures of -7.5- and -12-C, respectively.The cloud was characterized by a low droplet concentration of 21 cm-~ and a large droplet median volumediameter of 29/~m, with a concenlration of interstitial aerosol particles of less than 15 cm-3 (larger than 0.13/~m in diameter). The evolution of drizzle drops was traced downward from cloud top, with a maximumdiameter of 500/~m observed at cloud base. The air mass was sufficiently clean to ensure only a small numberof active cloud condensation nuclei. Consequently, small concentrations of cloud droplets led to concentrations of over 300 L-~ for droplets larger than 40 ~m, which set up strong conditions for continued growth bycollision-coalescence. Ice crystals in concentrations of 0.08 L-~ were measured simultaneously with thedrizzle drops and were not effective in glaciating the cloud, even though the drizzle drops were estimated tohave taken at least 1-2 h to form. While the growth of precipitation-sized drops through collision-coalescence has been well documented, thereare few measurements of this phenomena at temperatures less than 0-C. This study provides a well-documentedexample of such an event at subfreezing temperatures. The applicability of this measurement in terms of hazardous aircraft icing is discussed.1. Introduction The concept of drizzle growth by a combination ofcondensation and collision-coalescence has long beenrecognized in cloud environments where the temperature is warmer than 0-C (see, .e.g., Langmuir 1948;Ludlum 1951). Pruppacher and Klett (1978) gives ahistorical overview of work in this area, while Beardand Ochs (1993) provides a more current review. Thisphenomenon has also been inferred in several studiesfor clouds in which the temperature of the cloud environment was colder than 0-C. The latter observationsfall into two categories: observations of supercooleddrizzle at the ground, combined with radiosonde measurements that indicated a source cloud layer that wasentirely cooler than 0-C (Ohtake 1963; Bocchieri 1980;Huffman and Norman 1988); and in situ aircraft-basedobservations of supercooled drizzle-sized drops in convective and stratiform clouds (Byers 1952; Braham1964; Isaac and Schemenauer 1979; Sand et al. 1984;Politovich 1989; Pobanz et al. 1994). Corresponding author address: Dr. Stewart Cober, Cloud PhysicsResearch Division, ARMP, Atmospheric Environment Service, 4905Dufferin Street, Downsview, ON M3H 5T4, Canada.E-mail: cobers @aestor. am.doe.ca In each case, the growth of supercooled drizzle dropslarger than 100/zm in diameter was inferred to haveoccurred through a condensation and collision-coalescence process. Strong moisture sources and vertical lifthave been suggested to be necessary for the growth ofsuch drops (Langmuir 1948; Ludlum 1951; Pmppacherand Klett 1978; Politovich 1989), while the simultaneous inefficiency of the ice phase has been explainedby observations that the minimum cloud temperatureswere usually warmer than -10-C and that there was arelative lack of active ice nuclei at such temperatures(Ohtake 1963; Braham 1964; Bocchieri 1980; Huffman and Norman 1988; Politovich 1989). Pobanz etal. (1994) have suggested that wind shear is often associated with the formation of large drops. However,few studies have provided in-depth documentation ofthe microphysics associated with the growth and evolution of supercooled drizzle drops. Rasmussen et al.(1995) have documented a recent example in shallowupslope clouds in Colorado. On a research flight during the Second Canadian Atlantic Storms Program (Stewart 1991), an instrumented aircraft entered a region of drizzle at -7.5-C atan altitude of 3100 m. Because of an excessive buildupof ice on the aircraft, an ascent was performed to exitfrom the.icing region. The vertical profile obtained extended completely through the cloud in which the drizzle formed and provided a characterization of the evoc 1996 American Meteorological SocietyDECEMBER 1996 C O B E R E T A L. 2251lution of the drizzle drops in an environment that wasentirely colder than -7-C. The microphysics associatedwith the formation and growth of the drizzle drops arereported and discussed here. The measurements are discussed in terms of the associated hazard to aircraft andare compared to the standard icing curves used for aircraft certification.2. Instrumentation The research aircraft was a Convair-580 operated bythe National Research Council of Canada. It wasequipped by the Atmospheric Environment Service(AES) for cloud microphysics measurements as described in Cober et al. (1995). Instrumentation included the following: two Rosemount temperatureprobes and a reverse flow temperature probe, whichgenerally agreed to within I-C; a Cambridge dewpointhygrometer, which measured dewpoint to within +2-C;a Particle Measuring System (PMS) forward-scatteringspectrometer probe (FSSP) 100, which measured concentrations (_+20%) of droplet sizes between 1.5 and49/.tm in diameter; a passive cavity aerosol spectrometer probe (PCASP), which measured aerosol spectra(0.13-3 ktm); two PMS King probes (King et al.1978), which measured liquid water content (LWC)to _+0.02 g m-3 for LWC less than 0.2 g m-3; a Rosemount 871FA221B icing detector; an icing cylinderfrom which ice buildup could be measured from videotape images to _+ 1 mm; a Rosemount 858 gust probe;a PMS 2DC mono probe (25-800 ~m); a PMS 2DCgray (25-1600 tzm); and a PMS 2DP mono (2006400/~m). The three PMS 2D probes provided shapesand concentrations of hydromete0rs within their respective size ranges. Horizontal winds were also measured with a Litton 91 IRS navigational system. While the instruments collectively allow detailedcharacterizations of the microphysics of the cloud environment crossed by the aircraft, there are significantdeficiencies in the instrumentation associated with themeasurements of drops in the size range between 50and 150 ~m. The FSSP sized droplets only up to 49~m, while depth of field errors associated with thechannels from 25 to 150 t~m (lowest six channels) ofthe 2DC mono and 2DC gray probes cause significantuncertainty in the concentrations measured in thesechannels, particularly in the channels below 100 ~tm.In addition, the 2D probes may also incorrectly sizedrops (Joe and List 1987; Korolev et al. 1991 ), an effect that can be significant for drops smaller than 100/~m in diameter. Consequently, the measurements ofdrops in the first four channels of the 2DC mono and2DC gray were ignored in this analysis, so that there isa measurement gap between 50 and 125/~m. The 2DCdrop spectra larger than 125/.tm were not corrected toaccount for oversizing, as demonstrated by Korolev etal. ( 1991 ). For similar reasons, measurements of concentrations based on 2DP data do not incorporate datafrom the first two channels and, consequently, reflectonly particles greater than 500/.tm in diameter. Ice buildup on instruments cmi have a significant effect on their measurements. The majority of the instruments are heated to avoid this problem, and severalprobes were modified by AES to augment icing protection. The data from each instrument was carefullyexamined for indications of ice buildup. Icing in thedrizzle region was not enough to significantly affect theFSSP, PCASP, or 2D probes, although the Rosemount858 probe was affected, and its data was not used inthe analysis.3. Measurements and discussiona. Synoptic situation On 14 March 1992, an upper-level low pressure circulation centered over Labrador dominated the synoptic pattern near Newfoundland (see Fig. la). Thewinds at 70 kPa in the flight region were predominatelyfrom the south and southwest, which was consistentwith the circulation around the upper-level low. A 99.2kPa low pressure center (Fig. lb) was associated witha wave that had evolved from the circulation of theupper-level feature. The aircraft flew north of the redeveloping low pressure region, although at the initialflight level of 68 kPa, the aircraft was well above thefrontal surface. Winds at 68 kPa (3100 m) were fromthe southwest (210-) at 13 m s-1, while winds at 59.5kPa (4200 m) were from the south ( 170-) at 42 m swhich represented a large vertical wind shear. In thelocation where the aircraft encountered the drizzle, thevertical velocity was between 4 and 7 cm s -~ based oncalculations from the Canadian Regional Finite Element Model (Benoit et al. 1989). This is consistentwith the trough in Fig. lb, which indicates a region oflarge-scale ascent. The aircraft location coincides wellwith the region of lifting above the frontal surface.b. Microphysics measurements During level flight heading east from St. John's (Fig.1 ), the aircraft encountered drizzle drops at an altitudeof 3150 m (-7.5-C) at 1329 UTC 14 March 1992. Anascent was started at 1331:30 UTC, reaching 4230 m(- 11-C) at 1337 UTC. Cloud base was entered at 1330UTC, just before starting the ascent, and cloud top wasexited at 1340 UTC. Between 1337 and 1340 UTC, theaircraft flew level, below cloud top. The cloud-topheight was observed from the aircraft to decrease fromwest to east. Consequently, while the cloud-top temperature measured from the aircraft at 1340 UTC was-ll-C, the cloud-top temperature to the west, abovethe region where the drizzle was observed, was probably colder than - 11-C by 1--2-C. The aircraft profilewas roughly perpendicular to the winds, so that themeasurements should be unbiased with respect to thecloud and drizzle drop trajectories. The horizontal ex2252 JOURNAL OF APPLIED METEOROLOGY VOLUME357O 6O 5O Longitude (West) 7O 6O 50 Longitude (West) Fro. ]. Analysis of 1200 UTC 14 March I992 for (a) 70-kPa geopotential height and (b) mean sea level surface pressure. Geopotentialheight contours are labeled in tens of meters, and mean sea levelpressures are in kilopascals. The aircraft location between 1320 and1340 UTC is marked AC.tent of the drizzle region crossed by the aircraft between 1330 and 1337 UTC was approximately 40 km.To link microphysical measurements at different altitudes, the subsequent analysis assumes that the horizontal cloud environment was relatively constantacross the 40 km. The microphysics measurements andhydrometeor evolution discussed below support this assumption. Time history graphs of LWC from the King probeand altitude are shown in Fig. 2. Images of hydrometeors from the 2DC mono and 2DP are shown in Figs.3 and 4, respectively, for selected times in Fig. 2(points A-K). Figures 3 and 4 include zero area images, shattered drops, and out of focus images, all ofwhich are rejected when a quantitative analysis is performed on the 2D data. Prior to 1327 UTC the aircraftflew through a region of glaciated cloud, characterizedby no LWC, 2DP ice crystal concentrations ( >0.5 mmin diameter) of 0.2 L-~, and PCASP aerosol concentrations (>0.13 ~m) of 40-60 cm-3. The ice crystalimages were mainly 1-4-mm dendrites, and no dropswere discernible in the 2DC or 2DP images. Between1327 and 1329 UTC, the PCASP concentrationsdropped to 5 cm-3, while the crystal nature remainedrelatively unchanged (points A and B), although the2DP concentration decreased to 0.08 L-~. Between1329 and 1330 UTC, the 2DC images showed an evolution to circular particles (points C and D), and by1330 UTC, there were only a few small irregular images on the 2DP, while the 2DC was showing mainlydrizzle drops with maximum diameters of 500/zm. Hydrometeors of less than 500-/~m diameter wouldshadow too few photo diodes to distinguish drops fromcrystals on the 2DP because of the 200-/~m resolutionof the probe. The large number of blank images in Figs.4d-i is an indication of very small particles (less thanroughly 200 /~m), which trip the electronics but passout of the sample volume before they can be imaged.This, with the simultaneous lack of large numbers oflarger images, is consistent with the existence of drizzledrops. It is evident that by 1330, the majority of the2DC images were circular with diameters of less than0.5 mm. The transition from crystals to drops was observed across a 10-km region. The aircraft continuedwith level flight until 1331:30 UTC (points E and F),when an ascent was started to exit from the drizzleregion. A region with supercooled cloud droplets was~, 0.20~ 0.15~0.10 0.05 0.00 13:25 == ! K !....................... ~ ........................... i ................... t ........................... , .............. ,~._-.,.,,~ ==...:.,,~L,,=,,:,,,:=,,i=,,.=_ i ~ ..-"i :r ',........................... .......... ..... ............... ........... ................................ .................... ~ ...... ...... L ~ ', ~.,~ .... ~3:27 ~3:~ ~3:3~ ~3:~ ~:3~ 1~:~7 ~:~ Time (UTC)FIG. 2. Time history of King LWC (solid line) and altitude (dashed line) between 1325 and 1340 UTC.Points A-K correspond to the times of 2DC and 2DP images shown in Figs. 3 and 4, respectively.DECEMBER 1996 COBER ET AL. 2253A. 2D-C mono TIME: 13:27:27:375B. 2D-C mono TIME: 13:28:12:275C. 2D-C mono TIME: 13:29:10:375D. 2D-C mono T~E: 13:29:37:575E, 2D-C mono T~E: 13:31:03:950F. 2D-C mono ~E: 13:31:34:~O. 2D-C mono T~E: 13:31:58:~H. 2D-C mono T~E: 13:32:58:~L 2D-C mo.o T~E:J. 2D-C mono T~E: 13:35:01:~K. 2D-C mono T~E: fr~~~(~lilllr!llll[l~llllrllll~l~l(lltll[lil~rll~tJ~llFFl.-lilLrlll~l~ltfl FIG. 3. The 2DC mono images at selected times between 1327 ~d 1336 UTC. The veaica] lines separating images represent a scale leng~of 800 ~m. Images without any recogDizable p~ticle ~e llkely caused by the probe triggering on a droplet that ]s roughly ode photo diode(25 ~m) in diameter and consequently too small to resolve as an image. Some i~egu]~ images ~e caused by hydrometeors ~at ~e out offocus.entered at 1330 UTC as indicated by an increase in theFSSP droplet concentrations and King probe LWC(Fig. 2). Figures 3f-j show a decrease in the 2DC dropsizes as the aircraft ascended through the cloud, and by1336 (Fig. 3k) the drops were too small to resolve theirshapes with the 2DC. In the tipper regions of the cloud(points J and K), the 2DP measured concentrations of0.08 L -~ for ice particles with diameters larger than 0.5mm. No 2D images are shown after 1337 UTC becausethe images are similar to those in frames k of Figs. 3and 4. The temperature profile measured by the aircraft between 3150 and 4230 m around 1335 UTC is shown inFig. 5, along with a more complete vertical profiletaken at 1520 UTC, about 300 km north of the areawhere the drizzle was measured. No layer warmer than0-C was observed in either profile, implying that thedrizzle drops did not form through a melting process.Vertical profiles through the drizzle.region of LWC,FSSP droplet concentration, FSSP droplet mean volume diameter, and wind speed and direction are shownin Fig. 6. Figure 6a shows that the aircraft was in liquidcloud throughout the ascent, with an LWC between0.05 and 0.2 g m-3. The FSSP droplet concentrationvaried between 10 and 20 cm-3 below 3800 m, andincreased with altitude to 40 cm-3 at 4000 m. The meanvolume diameter of cloud droplets was 22/~m and wasrelatively constant with altitude, although this is basedon FSSP measurements that do not include contributions from drops larger than 49/~m. An increase in themean volume diameter to values greater than 25/~m at3800 m is well correlated with a corresponding decrease in the droplet concentration. High mean volumediameters at 3800 and 4200 m may indicate that theseareas are favorable source regions for the initiation ofdrizzle. The PCASP aerosol spectra (0.13-3/~m) are shownin Fig. 7. The average spectrum in glaciated cloud priorto the drizzle region (curve A) had a concentrati6n of38 cm-3, with a mean volume diameter of 1.0/~m. In2254 JOURNAL OF APPLIED METEOROLOGY VOLUME35A. 2D-P mono TIME: 13:27:26:050B. 2D-P mono TIME: 13:28:13:525C. 2D-P mono TIME: 13:29:10:550D. 2D-P mono TIM~: 13:2-:3'7:550E. 2D-Pmono TIME: 13:31:03:550F. 2D-P mono TIME: 13:31:33:550G. 2D-Pmono TIME: 13:31:57:550H. 2D-P mono TIME: 13:32:57:550I. 2D-P mono TIME: 13:33:58:050J. 2D-P mono TIME: 13:35:04:450K. 2D-P mono TIME: 13:35:58:850 Fro. 4. The 2DP images at the same selected times as in Fig. 3. The vertical spacing lines represent a scale length of 6.4 mm.the liquid cloud region (curve B), the concentrationwas 11 cm-3 with a mean volume diameter of 1.9/_tm.The reduction in the aerosol concentration is consistentwith the aerosol particles acting as cloud condensationnuclei (CCN), although some may have been scavenged by the drizzle drops. The change in the PCASPconcentration from 38 to 11 cm-3 is roughly consistentwith the average cloud drop concentration of :21 cm-3and supports the contention that the aerosols larger than0.13/~m were serving as CCN. Aerosol concentrationsof this magnitude in this size range are extremely lowand are indicative of clean background concentrations(Leaitch and Isaac 1991). A water sample was collected with a heated cloud water collector immediatelyafter leaving the drizzle region. It is believed that thesample resulted from drizzle drops that froze upon impaction with the collector and subsequently melted intothe collection bottle. The sample was analyzed for major inorganic ions. Sulphate, chloride, sodium, and potassium were below detection level [detection levelswere, respectively, 3.4, 0.8, 1.0, and 1.5 microequivalents per liter (~E L- ~ ) ], while nitrate and ammoniumwere present at about twice detection levels (2 ~EL-~). These concentrations are very low in comparisonto similar measurements in Other cloud studies (Leaitchet al. 1992) and support the idea that the air mass wasquite clean. The FSSP spectra are shown in Fig. 8 for 6-min averages taken before the aircraft entered the drizzle region (curve A) and for the ascent profile (curve B).Curve A likely shows the FSSP response to ice crystals(Gardiner and Hallett 1985) as there are nearly equalparticle concentrations for 14/~m and larger. The incloud droplet distribution (curve B) had an averageconcentration of 21 cm-3, mean volume diameter of 22/~m, and median volume diameter of 29/~m. The concentration of droplets larger than 40/.tm averaged 300L-~ throughout the cloud. The evolution of the FSSP and 2DC spectra aregiven in Fig. 9 for 90-180-s averages, which correspond to specific altitude intervals. The FSSP and 2DCspectra are interpolated between the last channel of theFSSP and the fifth' channel of the 2DC. The 2DC concentrations in the 25-, 50-, 75-, and 100-bm channelsare not shown because of the large errors associatedwith both sizing and concentrations in these channels.DECEMBER 1996 COBER ET AL. 2255 ~ ~ . ....'r..o :~;..~....u.~..c ...................... i .................. iiii.~.;'. ................. --T,, 'ia:~ u*c ...,.dr ................. 100 .... ~ .... I .... i .... I .... -25 -20 -15 -10 -5 0 Temperature (C) FIG. 5. Vertical temperature profile of the ascent through the drizzleregion at 1335 UTC (solid curve). The altitude spanned 3150-4230m, which is labeled for comparison to Fig. 6. Vertical profiles oftemperature (dashed curve) and dewpoint (dotted curve) measured at1520 UTC are shown for comparison. The latter measurements weremade with the aircraft approximately 2 h after the profile through thedrizzle region and approximately 300 km further north.The analysis of 2DC data was performed with a centerin technique similar to that described by Heymsfieldand Parrish (1978), with acceptance of particles withbox-area ratios greater than 0.65 and box-axis ratiosbetween 1.0 and 1.5. The rejection criteria discussedabove are effective in screening out noncircular images, out of focus particles, and zero area images,thereby leaving presumably only well-imaged drizzledrops for numerical analysis. A careful inspection ofall 2D records during the period spanning frames (e) (k) (Figs. 3 and 4) showed two distinct populations,small circular drizzle drops and larger noncircular icecrystals. The existence of drizzle drops below cloudbase was verified by the icing signature on the icingrod and the Rosemount icing detector, and it is reasonable to assume that these are from the same populationas those observed in-cloud. It is unlikely that significantnumbers of the small in-cloud circular images are icecrystals, since it is improbable that small crystals withnear-circular faces would coexist with drizzle drops toa maximum diameter of 500 pm and then reappear atmuch larger diameters with a dendritic structure. In thiswater-saturated environment, ice crystals would growquickly to millimeter sizes. Thus using the restrictiveacceptance criteria summarized above should eliminateall of the larger noncircular ice crystals and leave onlythe small circular images that are interpreted to be drizzle drops. Curve A (Fig. 9) represents drizzle measurementstaken at cloud base (altitude 3150 m), while curve Drepresents measurements near cloud top (4160-4220m). Curves B and C represent averages through thelower (3150-3615 m) and upper (3615-4160 m)halves of the cloud, respectively. The FSSP concentrations were highest in the upper half of the cloud, wherethe 2DC measurements showed a relative absence ofdrizzle drops. Conversely, the drizzle spectra clearlyevolve to larger sizes from cloud top to cloud base.These observations are consistent with growth by collision-coalescence. The cloud-top region was the Wind Direction (degrees)0 O0 180 270 ! ' ~ . i a ! ilbll ~~c ~ Id ; I[ ~ ~.Z ~ ....... ~ ...... ': ....... !'i ........ 'i 'i3600 i ~ ~ ...... ~ '* ......... ]:! i i, If ....:. :: ii i~ ,,,, -. - .... '. i ......... 4.;0"0.~"0.'~0"0.~"0.~" 0 ' ~0 '~' ~'~' ~o .... ~ .... ~ .... ~'"~o ~ ' ~" ~ ' Liquid W~er ~e~ (O m~ ~n~mr~on (~1 ~ean Volume DI~ ~) ~nd 8~ (m ~'~} F~G. 6. Vertical profiles for the ascent between 1331:30 and 1337:00 UTC: (a) ~ng probe LWC, (b) FSSPconcentration, (c) FSSP mean volume diameter, and (d) wind speed (WS) and direction (WD). The entire ascent wasin cloud. The temperature profile for this region is shown in Fig. 5.2256 JOURNAL OF APPLIED METEOROLOGY VOLUME3510' ............................ ......... i....L..i...',..i......i..'i .................................. - .Co 107 .............~ lo* ~ ~ ~ A lo* , ~ i i iliil i 0.1 1 Partiole Diameter ~) ~[~. 7. ~-minut~-~v~m~d PCA~P sp~c~ ~o~ (~) th~ n~on o~~]~ci~t~ c]ou~ ~m~t~]~ pdo~ to th~ ~z~]~ ~ion (~ 3~-]3~8~T~) ~nd (b) th~ ~fic~) pmfi~ ~ou~h ~ ~zz]~ ~ cloud ~on(]~]-]~7 ~TC).source region for some of the drizzle drops, whic:h grewby collision-coalescence once their terminal velocityexceeded the upward vertical velocity. The temperaturein this region was -11-C. Although ice particles wereobserved with the 2DP throughout the cloud region(concentrations of 0.08 L-~ for particles larger than 0.5mm), they were ineffective in glaciating the cloud anddrizzle region. Photographs of the icing cylinder indicated that 6.0+ 1.5 mm of ice built up between 1330 and 1337 UTC.This implies an average LWC of 0.12 + 0.03 g m-3,assuming an ice density of 0.8 g cm-3. The King probe(short wire version) average LWC for the same periodwas 0.09 + 0.02. The difference may be associatedwith the large errors in measuring the ice thickness anduncertainty in the ice density. In addition, the Kingprobes are known to underestimate the LWC incorporated in drops greater than 50 ktm (Biter et al. 1987).Therefore, considering the uncertainties noted above,the measurements of the icing rod and the King probeare consistent. Because of the enhanced aircraft icing hazard associated with supercooled drizzle drops, as compared tothat from supercooled cloud droplets (see section 4),it is useful to determine the fraction of the LWC incorporated in drops larger than 50 ktm in diameter. Using the 2DC and interpolated LWC as described above,the total LWCs for curves A-D in Fig. 9 we. re 0.06+__ 0.02, 0.14 2 0.03, 0.12 +_ 0.02, and 0.08 +__ 0.02g m-3, respectively, of which 62%, 53%, 17%, and 6%of the LWC was associated with drops larger than 50Fm. Figure 10 gives the fraction of mass incorporatedin drops smaller than a specified size and demonstratesthe evolution of the LWC from cloud droplets to drizzledrops as the drizzle moved lower in the cloud. At cloudbase (curve A), roughly 56% of the mass was incorporated in drops larger than 100 /~m. Conversely, atcloud top (curve D), over 95% of the mass was indroplets smaller than 50 tzm. The uncertainties in concentrations and sizing for the 2DC mono, and in theinterpolation of the droplet spectrum between 50 and125 ktm, cause the LWC estimates for drops greaterthan 50 Fm to be accurate within a factor of 2. Regardless, these errors do not affect the trends observedin the evolution of the drop spectra from cloud top tocloud base as shown in Fig. 10.c. Simple growth calculations Johnson (1980) has modeled the evolution of dropletspectra from an initial aerosol spectra. This model wasapplied to the conditions of the case study reportedhere. It predicted that a droplet distribution with a concentration of 24 cm-3, mean volume diameter of 25/~m, and liquid water content of 0.2 g m-3 could beobtained within 200 m from cloud base, assuming acloud-base temperature of - 8-C, adiabatic lifting of 0.1m s-t, and a maritime aerosol distribution with a concentration of aerosols greater than 0.13 /~m equal to11 cm-3. These values are quite similar to those measured in this case. Unfortunately, the vertical velocityin the drizzle region could not be measured from theaircraft, although the vertical velocity of the air mass~lo*1040 10 ~0 30 40 50Droplet Diameter ~m) Fro. 8. Six-minute-averaged FSSP spectra corresponding to thesame intervals as in Fig. 7. The first channel of the FSSP is biasedbecause of noise problems.DECEMBER 1996 10?COBER10~lOo10-~i ---A 3150 m100 1000Drop Diameter ~m) Fro. 9. FSSP and 2DC spectra averaged over 90-180-s periods,which correspond to specific altitude intervals: (A) level flight at3150 m between 1330:00 and 1331:30 UTC, (B) profile from 3150to 3615 m between 1331:30 and 1333:00 UTC, (C) profile from3615 to 4160 m between 1333:00 and 1336:00 UTC, and (D) levelflight between 4160 and 4220 m between 1336:00 and 1337:30UTC. Measurements from the last FSSP channel and from each2DC channel are also shown, although data from the first four 2DCchannels have been ignored. The linear interpolation between theFSSP and fifth channel of the 2DC is included in the curves, although it does not represent actual data.was estimated to be between 5 and 7 cm s-~ (section3a). It is likely that local updrafts had larger verticalvelocities, although the stratiform nature and wide horizontal extent of the clouds implied low vertical velocities overall. The temperature profile in the drizzle region followed a saturated adiabatic lapse rate, althoughthe liquid water content was not adiabatic throughoutthe cloud. Drizzle drops growing from collision-coalescence would tend to reduce and redistribute the liquid water content significantly. In the model simulation, the droplet spectrum took 30 min to reach a heightof 200 m above cloud base, at which point the supersaturation S predicted by the model was relatively constant at 0.15%. Further growth to diameters of 40/.tmat S = 0.15% took an additional 30 min, assuming onlycondensational growth. Using the collection efficiencies of Beard and Ochs(1984), a 100-3tm drop was modeled falling fromcloud top (4200 m) to cloud base (3100 m), through acloud with a median volume diameter of 29/.tm and aLWC of 0.1 g m-~. The drop took 45 min to fall, duringwhich time it grew to 170 /.tin in diameter. Such anincrease is consistent with the observed change in the2DC mean volume diameter (drops greater than 100/.tm only), which varied from 112/.tm at cloud top toET AL. 2257260/.tm at cloud base. The calculated cha. nge is smaller,but does not take into account the observed increase inconcentration and sizes of drizzle drops lower in thecloud, updraft velocities larger than 0.1 m s-~, or stochastic coalescence processes. Oversizing by the 2DCprobe also accounts for some of the difference. The evolution of the cloud droplet spectrum to droplets larger than 40/~m has been discussed in detail inPruppacher and Klett (1978). They indicated that theproduction of droplets of 40-50 ~m in diameter in concentrations of several per liter was an important step inthe stochastic coalescence process for the evolution oflarger drops, which could then grow by geometricsweep out. The average FSSP droplet concentration between 1331 and 1337 UTC (Fig. 8, curve B) for droplets between 40 and 49 bm was 300 L-~, which is twoorders of magnitude larger than the necessary concentrations given by Pruppacher and Klett (1978). No attempts have been made to estimate the processeswhereby droplets between 40 and 100 ktm are formed.As outlined by Beard and Ochs (1993), this remains apoorly understood phenomenon and is not addressed inthis work. The lack of measurements in the 50-125/~m range prevents investigation of the link between thecloud droplets and the evolution of the drizzle drops,and represents an unavoidable limitation of this work.Avoiding estimating the time required for the stochasticcoalescence of drops between 40 and 100/.tm is equivalent to assuming that this process took place simultaneously with the other processes. Pobanz et al. (1994) have suggested that the development of large drops is linked to regions of wind ~) 0.8 i ~ ; ~ 0.7 i i- .>~ u. 0.6 i ................... ,~ ......... /'"'f"B' t""'-"' ," .......................................... ............. 0.3 0 0.2 0.1 0.0 ............ Drop Diamotor era) F~G. 10. Cumulative mass cu~es for FSSP ~d 2DC spec~a forthe same altitude and time inte~als shown in Fig. 9. The LWC valuesfor cu~es A, B, C, and D ~e 0.06, 0.14, 0.12, and 0.08 g mrespectively.2258 1.00JOURNAL OF APPLI0.10 0.01 1 10 100 1000 Median Volume Diameter ~m) FIG. 11. Median volume diameter versus LWC for the same FSSP/2DC spectra plotted in Fig. 10. The median volume diameter for eachspectrum corresponds to the cumulative LWC fraction of 0.5 in Fig.10. Solid curves represent the icing curves of Newton (1978) forlight, moderate, and severe icing. Dashed curves represent the maximum continuous icing curves from FAR 25-C for 0- and -10-C.Points A-D correspond to the same altitude/time intervals as in Figs.9 and 10. The Newton curves represent the maximum potential accumulation of ice (gcm-2 h-I) on a 7.6-cm cylinder, at an airspeedof 100 m s-t, at -10-C, and at 70.0 kPa.shear, with the shear causing inhomogeneous mixingand turbulence, leading to the formation of large drops.In this case, there was a strong wind shear of 0.05 s-'- in the lowest 300 m of the cloud, with a correspondingRichardson number of 0.10. In the upper half of thecloud there was a 200-m layer with a Richardson number of 0.35 and shear of 0.02 s-'. Both regions areconsistent with the discussion of Pobanz et al. (1994),who inferred that regions with shear greater than0.02 s-] and Richardson numbers smaller than 1 werehighly correlated with regions in which large dropsformed. Conversely, there was no significant shear region near cloud top as hypothesised by Pobanz et al.(1994), and 100-/~m drops were also measured within100 m of cloud top where the shear was 0.01 s -~ andthe Richardson number was 3. Hence, the mechanismof Pobanz et al. (1994) was not necessarily active inthis case. While it is not the intent of this work to fully model the growth of the measured drizzle spectrum, these cal culations have been performed to show that the mea surements support the concept that the drizzle grew by a condensation and collision-coalescence process, and to provide an estimate of the timescale for the forma tion of the drizzle drops. Using the simple time esti mates given above, and neglecting the stochastic co alescence processes for drops between 40 and 100/~m,ED METEOROLOGYVOLUME 35the minimum times necessary for the evolution of thecloud droplet spectra by condensation and drizzle spectra by collision coalescence was roughly 1-2 h. Thisis not unreasonable given the low liquid water contentof the cloud. Low aerosol concentrations produced acloud drop distribution with a low concentration andlarge median volume diameter, with a significant tailof drops larger than 40/~m, which initiated the collision-coalescence process. These observations imply that the cloud remainedsupercooled for a period of 1-2 h without being significantly glaciated. Ice particle concentrations of 0.08L-~ (greater than 0.5 mm in diameter) were measuredthroughout the cloud and drizzle region. At - 12-C, theexpected mean concentration of active ice nuclei isroughly 0.01 L-~, within one order of magnitude(Pruppacher and Klett 1978), which is consistent withthe measured concentration of ice particles measuredwith the 2DP. Assuming saturation with respect to water, the supersaturation with respect to ice is 12% at-12-C. Dendritic plates of 2-mm diameter would growin approximately 1 h under these conditions, which isconsistent with the images seen on the 2DP (Figs. 4gk) throughout the cloud. The lack of small crystals implies the absence of an effective ice multiplicationmechanism in the cloud region where the drizzle wasmeasured. The temperature of the drizzle region ( -7.5-to - 11-C) was too cold for the ice multiplication mechanism described by Hallett and Mossop (1974) andMossop (1976).4. Application Supercooled drizzle has long been recognized as presenting a significant hazard to aircraft (Lewis 1951).Icing causes a decrease in lift and increase in drag, anda rapid accumulation of ice could obscure visibilityfrom the cockpit. These problems are most dangerousduring takeoff or landing, when visibility is critical, andwhen the aircraft is closest to its stall speed. Drizzlesized drops have a significantly greater collision efficiency with aircraft surfaces than cloud droplets, andtherefore cause a faster buildup of ice for equivalentliquid water contents. In-flight observations of aircrafticing caused by supercooled drizzle have been reportedby Sand et al. (1984) and Politovich (1989). Theynoted that the drizzle drops can flow back over andunder the wings and behind deicing boots and suggested that such icing caused significantly more degradation of aircraft performance than icing caused bycloud droplets, for equivalent liquid water contents. Figure 11 shows the liquid water contents versus themedian volume diameters for the four curves in Fig. 9.The 2DC and FSSP data were interpolated together toderive the median volume diameters. The icing safetythresholds of the U.S. Federal Aviation Administrationregulation 25 appendix C (FAR 25-C) and Newton(1978) are overlaid for comparison. For both sets ofDECEMBER 1996 COBER ET AL. 2259curves, the larger the LWC and median volume diameter, the heavier the icing intensity. The FAR 25-Ccurves physically represent 0.1% exceedance probabilities for flight in continuous icing over 32.6 km and areused as an estimate of the maximum possible icing.While the FAR 25-C curves do not extend beyond 40ktm, it is evident that the flight lower in the cloud(points A and B) significantly exceeded the FAR 25C thresholds. Conversely, the icing was not even classified as moderate under the Newton scheme, whichcontradicts the assessment of the FAR 25-C curves.During 119 h of in-flight measurements in winterstorms over the North Atlantic Ocean (Cober et al.1995), there were only three occasions in which icingwas subjectively assessed as severe and the flight pathwas modified to exit from the icing region. Two ofthese (one of which is the case described here) occurred in the presence of supercooled drizzle, whichformed through a collision-coalescence process. Inboth cases, the visual assessment by the pilots was thatthe icing was an unusual and potentially unsafe event,and they took immediate action to direct the aircraft outof the icing region. Their observations and reactionsqualitatively support the FAR 25-C icing curves. Freezing drizzle is frequently observed on the Canadian east coast in the winter (McKay and Thompson1969). A climatology of freezing precipitation (Strappet al. 1996) found that St. John's, Newfoundland, hasthe highest frequency of freezing precipitation in Canada, receiving 103 h of freezing drizzle and 51 h offreezing rain per year. The maximum frequency occurred in March when freezing precipitation was measured on the ground an average of 40 h per month, or5.5% of the total hours. Strapp et al. (1996) have usedradiosonde data and surface observations to show thatapproximately two-thirds of the freezing drizzle observed at the surface in St. John's forms through a collision-coalescence process, while one-third formsthrough ice particles melting and subsequently supercooling. Observations of freezing drizzle at St. John' s,formed without a melting layer, were highly correlatedwith onshore winds. Strapp et al. (1996) inferred thatsuch clean maritime air masses were suitable for theformation of supercooled drizzle through a condensation and collision-coalescence process. Ohtake (1963)also noted that freezing drizzle events with no warmlayer aloft were correlated with winds from the ocean.The example discussed in this study, although not fromlow-level clouds during onshore winds, shared manyof the same characteristics favorable for drizzle formation by condensation and coalescence. The FAR 25-C curves represent the maximum icingconditions expected during 99.9% of icing encounters.The curves are based on in-flight icing cylinder measurements taken during the 1940s and do not distinguish median volume diameters greater than 40 /~m.Certainly, cloud regions with supercooled drizzle canhave median volume diameters larger than 40/~m, asillustrated by the case described here. The high frequency of freezing drizzle on the east coast suggeststhat aircraft will be routinely taking off and landing infreezing drizzle conditions. These points suggest thatthe hazard to aircraft from supercooled drizzle is notadequately addressed in the safety thresholds of FAR25-C. The hazard to aircraft is compounded by the inability of current forecast models to forecast supercooled drizzle in the absence of an associated meltinglayer.5. Conclusions Drizzle drops to 500 tzm in diameter formed througha condensation and collision-coalescence process attemperatures between -11- and -8-C. The air masswas characterized by low aerosol concentrations(0.13-3 ~m) of 40 cm-3, while the cloud growth environment was characterized by an average dropletconcentration of 21 cm-3, interstitial aerosol concentrations of less than 15 cm-', and an average dropletmedian volume diameter of 29/~m. These observationsare consistent with a clean air mass, with small numbersof droplets competing for available moisture. Consequently, the concentrations of cloud droplets with diameters larger than 40 t~m exceeded 300 L-l, whichwas sufficient for the initiation of growth by collisioncoalescence. The evolution of the droplet spectra in thesizes smaller than 50/~m and greater than 125 tzm wastraced vertically through the cloud, and 500-/~m dropswere measured approximately 1000 m below the sourceregion at cloud top. The ice particle concentration(greater than 0.5 mm in diameter) was 0.08 L-~throughout the drizzle formation region. Simple growthmodels were used to estimate that the ice crystals anddrizzle drops measured at cloud base had been growingfor at least 1-2 h, and that no efficient ice multiplication mechanism was operating. Comparisons of thedata with the aircraft icing curves of FAR 25-C havedemonstrated a limitation with the accepted safetycurves, which needs to be addressed. This study presents a well-documented example of drizzle-sized supercooled drops, which formed through a condensation-collision-coalescence process. Measurements ofthe hydrometeor spectra between 50 and 125 ktm werenot made, and represent a limitation of this work. Application of the microphysics data toward a better understanding of the stochastic coalescence process fordrops in the 40-100-/~m range is a subject for furtherresearch. Acknowledgments. This work was funded by the Canadian National Search and Rescue Secretariat. Funding was also provided by Boeing Commercial AirplaneGroup, the Institute for Aerospace Research of the National Research Council of Canada, and the Atmospheric Environment Service (AES). The techniciansand programmers of AES are acknowledged for their2260 JOURNAL OF APPLIED METEOROLOGY VOLUME35support in keeping the instruments operational and calibrated, and for assistance provided in analyzing the2D data. The NRC pilots are acknowledged for theirsafe and efficient operations of the Convair-580 aircraft. Thanks to Mike Patnoe for sharing the duties offlight director on this flight. Thanks also to AndreTremblay for providing the vertical velocity data f~omthe Regional Finite Element Model, and to Cathy Banicfor providing the cloud water chemistry analysis. REFERENCESBeard, K. V., and H. T. Ochs, 1984: Collection and coalescence ef ficiencies for accretion. J. Geophys. Res., 89(D), 7165-7169.--, and , 1993: Warm-rain initiation: An overview of microphysical mechanisms. J. Appl. Meteor., 32, 608-625.Benoit, R., J. Cote, and J. Mailhot, 1989: Inclusion of a TKE bound ary layer parameterization in the Canadian regional finite-ele ment model. Mort. Wea. Rev., 117, 1726-1750.Biter, C. J., J. E. Dye, D. Huffman, and W. D. King, 1987: The drop size response of the CSIRO liquid water probe. J. Atmos. Oce anic Technol., 4, 359-367.Bocchieri, J. R., 1980: The objective use of upper air soundings to specify precipitation type. Mon. Wea. Rev., 108, 596-603.Braham, R. R., 1964: What is the role of ice in summer rain showers? J. Atmos. Sci., 21, 640-645.Byers, H. R., 1952: An example of flight through a thunderstorm updraft. Bull. Amer. Meteor. Soc., 33, 211-212.Cober, S. G., G. A. Isaac, and J. W. Strapp, 1995: Aircraft icing measurements in East Coast winter storms. J. Appk Meteor., 34, 88-100.Gardiner, B. A., and J. Halletr, 1985: Degradation of in-cloud forward scattering spectrometer probe measurements in the presence of ice particles. J. Atmos. Oceanic Technol., 2, 171-180.Hallett, J., and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249, 26-28.Heymsfield, A. J., and J. L. Parrish, 1978: A computational technique for increasing the effective sampling volume of the PIVlS two dimensional particle size spectrometer. J. AppL Meteor., 17, 1566-1572.Huffman, G. ~., and G. A. Norman, 1988: The supercooled w~mn rain process and the specification of freezing precipitation. Mon. Wea. Rev., 116, 2172-2182.Isaac, G. A., and R. S. Schemenauer, 1979: Large particles in super cooled regions of northern Canadian cumulus clouds. J. Appl. Meteor., 18, 1056-1065.Joe, P., and R. List, 1987: Testing and performance of two-dimen sional optical array spectrometers with greyscale. J. Atmos. Oce anic Technol., 4, 139-150.Johnson, D. B., 1980: The influence of cloud-base temperature andpressure on droplet concentration. J. Atmos. Sci., 37, 20792085.King, W. D., D. A] Parkin, and R. J. Handsworth, 1978: A hot wire liquid water device having fully calculable response character istics. J. Appl. Meteor., 17, 1809-1813.Korolev, A. V., S. V. Kuznetsov, Y. E. Makarov, and V. S. Novikov, 1991: Evaluation of measurements of particle size and sample area from optical array probes. J. Atmos. Oceanic Technol., 8, 514-522.Langmuir, I, 1948: The production of rain by a chain reaction in cumulus clouds at temperatures above freezing. J. Meteor., 5, 175-192.Leaitch, W. R., and G. A. Isaac, 1991: Tropospheric aerosol size distributions from 1982 to 1988 over eastern North America. Atmos. Environ., 25, 601-619. - --., J. W. Strapp, C. M. Banic, and H. A. Wiebe, 1992: The relationship between cloud droplet number concentrations and anthropogenic pollution: Observations and climatic implica tions. J. Geophys. Res., 97(D), 2463-2474.Lewis, W., 1951: Meteorological aspects of aircraft icing. Compendium of Meteorology, T. F. Malone, Ed., Amer. Meteor. Soc., 1197 1203.Ludlum, F. H., 1951: The production of showers by the coalescence of cloud droplets. Quart. J. Roy. Meteor. Soc., 77, 402-417.McKay, G. A., and H. A. Thompson, 1969: Estimating the hazard of ice accretion in Canada from climatological data..J. Appl. Me teor., 8, 927-935.Mossop, S. C., 1976: Production of secondary ice particles during the growth of graupel by riming. Quart. J. Roy. Meteor. Soc., 102, 45-57.Newton, D. W, 1978: An integrated approach to the problem of aimraft icing. J. Aircraft, 15, 374-380.Ohtake, T., 1963: Hemispheric investigation of warm rain by radio sonde data. J. Appl. Meteor., 2, 594-607.Pobanz, B. M., J. D. Marwitz, and M. K. Politovich, 1994: Conditions associated with large-drop regions. J. Appl. Meteor., 33, 1366-1372.Politovich, M. K., 1989: Aircraft icing caused by large supercooled droplets. J. Appl. Meteor., 28, 856-868.Pruppacher, H. R., and J. D. Klett, 1978: Microphysics of Clouds and Precipitation. Reidel, 714 pp.Rasmussen, R. M., B. C. Bernstein, M. Murakami, G. Stossmeister, and J. Reisner, 1995: The 1990 Valentine' s Day arctic outbreak. Part I: Mesoscale and microscale structure and evolution of a Colorado Front Range shallow upslope cloud. J. Appl. Meteor., 34, 1481-1511.Sand, W. R., W. A. Cooper, M. K. Politovich, and D. L. Veal, 1984: Icing conditions encountered by a research aircraft. J. Climate Appl. Meteor., 23, 1427-1440.Stewart, R. E., 1991: Canadian Atlantic Storms Program: Progress and plans of the meteorological component. Bull. Amer. Meteor. Soc., 72, 364-371.Strapp, J. W., R. A. Stuart, and G. A. Isaac, 1996: A Canadian cli matology of freezing precipitation, and a detailed study using data from St. John's Newfoundland. Proc. FAA International Conf on Aircraft lnfiight Icing, Springfield, VA, Federal Avi ation Administration, in press.

Save