1 JUNE 1986 GLENN L. GORDO'N AND JOHN D. MARWITZ 1087Hydrometeor Evolution in Rainbands over the California Valley GLENN L. GORDON AND JOHN D. MARWITZDepartment OfAtmospheric Science, University of Wyoming, Luramie, WY 82071 (Manuscript received 8 July 1985, in final form 10 December 1985)ABSTRACTHydrometeor distributions were measured in two rainbands that passed over the California Valley. Theground radar was used to vector the University of Wyoming's instrumented King Air aircraft to the top of therainband at which time an onboard computer algorithm was used to make multiple penetrations of an ensembleof ice particles that were assumed to descend at I or 2 m s" while drifting downwind. Distribution parameterswere calculated for each penetration and the changes in these parameters were used to infer the different modesof growth that the particles undergo. It was found that there were five distinct regions of particle growth.Nucleation and depositional growth were the dominant procases near the top of the rainband. The hydrometeorsthen fell through a region of aggregation from the -20°C level to the -10°C level. Ice crystal multiplicationwas the dominant process from the -10°C level to the -4°C level. From -4 to O'C, aggregation was again thedominant process. Below the 0°C level, melting and collisional coalescence were the main processes.Bulk ice densities were calculated by comparing radar reflectivities and reflectivities calculated from thehydrometeor distributions. Bulk ice densities were very low, reaching only 0.1 g cm-' just above the melting layer.1. IntroductionOf the various types of winter precipitation systemsthat pass through northern California, rainbands makeup about 19% of the total precipitation events (Heggliet al., 1983). Much of the recent work on rainbandshas come from the CYCLES (Cyclonic ExtratropicalStorms) project in the western half of Washington state,which identified several types of rainbands associatedwith extratropical cyclones (Houze et al., 1976a; Hobbs,1978). A, seeder-feeder type .of precipitation mechanism was deduced for some of these Pacific Northwestrainbands (Hobbs et al., 1980; Herzegh and, Hobbs,1980). The ice particles grow first by deposition. in theseeder region and then fall into the feeder region wheregrowth is by deposition, riming and/or aggregation(Houze et al., 1976b; Matejka et al., 1980). These resultswere numerically simulated by Rutledge and Hobbs(1 983) with reasonable agreement.Passarelli (1 978) developed an analytical model forthe growth of snowflakes which indicated that an equilibrium snowflake spectrum results due to a balancingeffect of deposition and aggregation growth. These results were tested by Lo and Passsirelli (1 982) by collecting microphysical aircraft data in mature extra. tropical cyclones. They conclude that while particlesgrow initially by deposition and then aggregation, it isparticle breakup that causes the particle spectrum tocease evolving.In this paper we show the evolution of precipitationparticles in northern California rainbands from theinitial nucleation of small ice particles to the melting0 1986 American Meteorological Societyof precipitation size particles. As will be seen, the iceparticles go through a very systematic evolutionaryprocess.2. Data collection and flight proceduresCloud physics measurements were collected by theUniversity of Wyoming instrumented King Air aircraft(data system described by Cooper, 1978). Three ParticleMeasuring Systems, Inc. (PMS) probes were used forthe determination of cloud and precipitation particlespectra; a Forward Scattering Spectrometer Probe(FSSP) for particles in the size range 3-45 pm, an OAP2D-C (2D-C) for particles in the size range 25-800 pmand an OAP-2D-P (2D-P) for particles in the size range200-6400 pm. Table 1 shows the specifications of theprobes. The size range measured by the 2D-C and 2DP is either in the lateral dimension (perpendicular tothe flight path), or, if the time-slice data is used, thelongitudinal dimension (along the flight path). In thisstudy we use the longitudinal dimension, allowing thesize range to thereby be increased to much larger sizes(values in parentheses in Table 1). The FSSP and 2DP were located on the left wing tip, while the 2D-C waslocated on the right wing tip approximately 17 m away.Both 2D probes were vertically oriented.Data from the FSSP were divided into 15 intervalsof 3 pm each. Concentrations were corrected for deadtime and coincidence errors.The 2D probe data were divided into 20 size classes.The size classes for the 2D-C probe varied from 50 to1000 pm in width and for the 2D-P from 200 to 20001088 JOURNAL OF THE ATMOSPHERIC SCIENCES VOL. 43, No. 11TABLE 1. Specifications of PMS probes.Probe FSSP- 100 OAP-2D-C OAP-2D-PNomenclature FSSP 2D-C 2D-PSize range (pm) 3-45 25-800 200-6400Resolution (pm) 3 25 200Sample volume(liters per 100 m of flight) -0.05 -5 - 170( 10000) ( 10000)Nm in width (see Table 2). Hydrometeor spectra resemble a power law with high concentrations of smallparticles and low concentrations of large particles. Thesize classes were chosen to give a more uniform concentration in each size class and also to give bettersampling statistics at the larger size classes.During 2D data processing, artifacts were rejectedaccording to the criteria in Table 3. A11 other imageswere accepted as real particles regardless of where thecenter of the particle might lie. The sample volume`was adjusted for individual images by increasing theeffective array width by the maximum longitudinal dimension. This accounts for the particle's center beingoutside the sampling area by as much as one-half thelongitudinal dimension of each particle being sampled.No depth of field correction was applied, probably resulting in an undercounting of particles in the smallestfew size classes.Since the statistical sampling error is -N-0.5, it wasrequired that there be at least ten particles per size classforplotting. This affects only the largest few size classes.These particles are relatively rare and are sampled lessfrequently. The sampling error for large particles,therefore, was limited to a maximum of about 30%(Gordon and Manvitz, 1984).The computer onboard the King Air was proTABLE 2. PMS size classes (pm).Class FSSP , 2D-C 2D-P1 0-3 <50 200-4002- ' 3-6 50-100 400-6003 6-9 100- 150 600-8004 9-12 150-200 800- 10005 12-15 200-250 1000- 12506 15-18 250-300 1250-15007 18-2 I 300-400 1500- 17508 2 1-24 400-500 1750-20009 24-27 500-600 2000-225010 27-30 600-800 2250-250011 30-33 800-1000 2500-275012 33-36 1000-1250 2750-300013 36-39 1250-1500 3000-350014 39-42 1500-2000 3500-400015 42-45 2000-2500 4000-450016 2500-3000 4500-500017 3000-4000 5000-600018 4000-5000 6000-800019 5000-6000 8000-1000020 >6000 >loo00TABLE 3.2D-C and 2D-P artifact rejection criteria.1. Streakers (usually, water shedding across the upstream edge ofthe probe and across the aperture) are rejected if they are six timesas long (longitudinal dimension) as they are wide (lateral dimension).Images are also rejected as streakers (i) if they are three times as longas wide, (ii) if they are less than 150 pm wide, and (iii) if they do notoverlay the edge of the scan region (Cooper, 1978).2. Cases in which the probe recording circuitry are triggered butno elements are recorded as shadowed are rejected. These often correspond to true objects near the minimum size detectable by theprobe (Cooper, 1978).3. Images resulting from particles splashing off the edges of theaperture are eliminated by calculating the average distance betweenparticles based on concentration and rejecting those particles thatpass through the beam at times less than the time corresponding tothe average distance.4. Images having more than 2 gaps are rejected.grammed such that at the discretion of the flight crewan air parcel or "pointer" could be selected and subsequently penetrated. This was done by constantly integrating the true air speed and heading of the aircraftand displaying, to the pilot, the heading and distanceback to the pointer. Since the pointer and aircraft areboth drifting with the environmental ~nds, no priorknowledge of the environmental winds was necessary.Assuming a constant fallspeed for the ice crystals, thealgorithm also displayed the height,of the aircraft relative to the ice crystals.Figure 1 shows a map of the SCPP study area. Location, movement and intensity of rainbands were determined with the Bureau of Reclamation's SkywaterSWR-75 5-cm radar located at Sheridan (SHR). Theradar was used to vector the King Air to the top of therainband at which time a "pointer" was set and theexperiment began. The pointer was allowed to descendat 1 or 2 m s" while drifting downwind with the ideathat the aircraft would be sampling the same ensembleof particles each time it passed through the pointer.20 hm500FIG. 1. Map of the SCPP project area. Height contours (meters) are shown as dashed lines.I JUNE 1986 GLENN L. GORDON AND JOHN D. MARWITZ 1089iI5 FEE 82K/4 FLlGUr TRACK1518-1622LTIME-60 -40 -20 0 20 40DISTANCE FROM SHERIDAN RADAR (km)FIG. 2. King Air flight track for 15 February 1982. Solid line shows the movement of the position reference.The rate of descent was selected to approximate thefallspeed of ice crystals and yet have the aircraft descendbelow the 0°C level before topography precluded sucha descent. Bands were intercepted over the valley, westof the foothills in order to minimize orographic effects.The flight track for 15 February 1982 is shown inFig. 2. The aircraft was flown in an elongated figureeight pattern oriented parallel to the orientation of therainband with each leg of the figure-eight being 15 to25 km in length. The data were averaged over each legof the figure-eight (2-4 min). The flight legs are numbered consecutively from 1 to 18. A similar flight procedure was used by Lo and Passarelli (1982) in cyclonicstorms, except that they did a spiral descent at a constant bank angle.3. Case study of 15 February 1982a. Synoptic situationFigure 3 shows the National Weather Service surfaceand 500 mb map for 0400 LST 15 February 1982. Adecaying frontal system was just moving onto the Pacific Coast. At the time of descent through the band,the study area was in the warm sector of this decayingsystem. The climbout sounding from McClellan AirForce Base is shown in Fig. 4. The atmosphere wassaturated with respect to ice below 550 mb. Above 550mb the atmosphere was not saturated. The de traceindicates that the atmosphere was stable up to about950 mb and neutrally buoyant up to 450 mb. Becauseof the neutral stability, the barrier winds were onlyabout 12 m s-' below 1 km (Waight, 1984). The windsat crest level (3 km) were about 25 m s" 'normal tothe barrier and they veered about 30" above crest level.Figure 5 shows a PPI of reflectivity at 2.5" elevationangle from the Sheridan radar at 152 1 LST. The bandat this time is located just west of the radar with themost vigorous portion located southwest of the radar.The semicircular large high reflectivity region fromnorth through southeast centered at a range of 50 to60 km is orographically induced.The movement of the band based on the radar andthe flight track (Fig. 2) was eastward at about 20 m s-I.The track of the position reference is also the trajectoryof an ice crystal falling through the band at about I ms-I. Notice that as the ice crystal falls it moves in anortherly direction within the band indicating that thewinds in the band also have a southerly component.b. Hydrometeor dataFigure 6 shows leg-averaged FSSP total concentration, FSSP concentration for drops larger than 24 pmand FSSP liquid water content plotted as a functionof temperature and altitude. The numbers correspondto consecutive legs with 1 being at the top of the rainband. There appears to be very little liquid water abovethe -10°C level. Small quantities of liquid water arepresent at warmer temperatures with values remainingless than 0.1 g m-3 above the 0°C level. The concen546552-200558564II \I \ IFIG. 3. Surface and 500 mb map for 0400 LST 15 February 1982.1090 JOURNAL OF THE ATMOSPHERIC SCIENCES VOL. 43, No. 11I441 - 1507LnEaw 700ap 800 900 1000v)v)290 300 310 320 0 IO 20 30 rns"270'FIG. 4. Takeoff sounding from McClellan Air Force Base from the King Air from 144 1 to 1507LST 15 February 1982. The temperature and dew point are plotted on a Skew T-log p diagramon the left, the potential temperature (e) and equivalent potential temperature (e,) are plotted inthe center and the wind hodograph is plotted on the right.tration of droplets with diameters greater than 24 pmwas -0.5 cm-3 near the -5°C level. Figure 6 has beendivided into five regions (a-e). The significance of theseregions will be discussed later.from the 2D-C for each of the five regions of Fig. 6. Inthe first four legs where T < -22°C (Fig. 7a) crystalsare small. Small aggregates occasionally can be seen.Figure 7b shows representative images throughout region b in the temperature range from - 1 1 " to -22°C(legs 4-8). Dendrites and dendritic aggregates can beseen throughout region b. Images from region c wherethe temperature vanes from -3" to -1 1°C (legs 8-14)are shown in Fig. 7c. A large number of columnar typecrystals are present along with some aggregat,es. Thetemperature range of region d varied from 0" to -3°CFigure 7 shows representative hydrometeor images ,I I II10-20dB1IRAINBAND . E3 20-30482I .30dBZ I-100 II I I I-100 . -50 0 50 tooDISTANCE (km)FIG. 5. PPI of reflectivity from the Skywater radar (located at Sheridan) at an elevation angle of 2.5" at 1521 UT. The first contour is10 dBZ and the contour interval is 10 dBZ.[legs 14-17 (Fig: 7d)l and contained mostly large aggregates with relatively few pristine crystals. Imagesfrom the final region (e) had a temperature range from0" to +3"C (legs 17-1 8) and showed mostly water dropswith some aggregates still present (Fig. 7e).Figure 8 shows representative size spectra for levelsa-e of the rainband. Initially, the spectrum is very steepand narrow (T = -25°C). The spectra evolve in a systematic manner as the temperature gets warmer. At T= -13.1 "C the spectrum is a little broader. By T= -56°C the spectrum seems fully developed with abroad range of sizes and an excess of particles withsizes less than 0.1 cm. At T = -0.1 "C. large (>0.2 cm)particles are still increasing in number while the smaller(<O. 1 cm) sizes are decreasing. There is a decrease inparticle concentration by T = + 1.2"C. Presumably thisis the beginning of melting and collapse of ice particles.An exponential distribution function of the formN(D) = NO exp( -AD) (1)was fitted to the data for each leg using a least-squarestechnique where N(D) is the number density, D thediameter, No the intercept parameter and X the slopeparameter. The values of X and No along with the meantemperature and height for each leg are shown in Table4. Slope and intercept values were determined only forthat portion of the spectrum that behaved in' an exponential manner (D > 0.05 cm). Byrequiring at least.ten particles in any given bin size, we found that legswith low concentrations of particles greater than 0.05cm could not have distribution parameters calculatedfor them.The log of the slope is plotted against the log of theintercept for each leg in Fig. 9. An increase in No anda slight increase in X occurred between legs 3 and 4(region a). According to Lo and Passarelli (1982) depositional growth will cause particles of all sizes to growat the same rate (with respect to diameter) which willcause X to remain, constant and No to increase due tomore numerous smaller particles growing to largersizes. A slight increase in X in addition to an increase1 JUNE 1986 GLENN L. GORDON AND JOHN D. MARWITZ 109 1FSSP 15 FEB 82-3087Ay -20Y 653 5x4 -10% e 42+o 3WYa2kWI0 IO 20 30 0 20 0. I 0.2DROPLET CONCENTRATION DROPLETS LWC(crn-9 1 24pm (g/m3)(cm-'1FIG. 6. FSSP droplet concentration, concentration of drops larger than 24 pm and FSSP liquidwater content plotted as a function of temperature and altitude. Numbered points represent flightleg averages. Lettered regions correspond to different hydrometeor growth regimes.in No indicates that there was an input of small particlesdue to nucleation, secondary production, and/or smallparticles growing to detectable sizes. If the input ofsmall particles was only due to growth to detectablesizes, X would remain constant. The fact that X increases slightly indicates that new particles were produced at a faster rate than depositional growth.Further evidence of small particle input can be seenin Fig. 10 which shows ice crystal concentration as afunction of average temperature. The concentrationshows a slight but steady increase from leg 1 to leg 4.Figure 10 shows a drop in total concentration at leg5. This is puzzling since, in addition, the 2D imageswere smaller and no aggregates were apparent (theymay have been too small to recognize). One explanation for this might be that due to a slight horizontal15 FEB 822D-C IMAGES mFrnvariation in the flight track the aircraft may not havesampled the same ensemble of descending ice crystals.The flight track with respect to the ground and withrespect to the descending position reference suggestedthat the experiment was properly flown, however. Sincethe number of particles with D > 0.05 cm sampledwas not greater than ten per bin, no X or NO was calculated for leg 5.Both X and NO show a decrease from leg 4 (-22OC)to leg 8 (- 11 "C) in Fig. 9 (region b in Fig. 6). Thisindicates a region of rapid aggregation. There continuedto be a slight increase in concentration (Fig. 10) whiledescending through region b (ignoring leg 5). The increase in concentration in region b indicates an inputT- 5I. I ' u!fY!L" "m--mw DIAMETER (cm)FIG. 7.2D-C images collected in regions a-e of the rainband.FIG. 8. Examples of 2D-P size distributions sampled in regions a-e of the rainband.1092 JOURNAL OF THE ATMOSPHERIC SCIENCES VOL. 43, No. 11TABLE 4. Mean values of temperature (T), altitude (Z), slope (X) and intercept (No) for the 2D-C and 2D-P for each leg for the 15 February 1982 case.2D-C 2D-P12345-6 7 8 9101112131415161718-30.5-27.4-24.5-21.8-18.9- 16.0-13.1-10.7-9.2-8.1-6.8-5.6-4.3-3.3-2.0-1.1-0.11.277677356698 165636 1295672516747764529430140893863363634173 197300127562546107.0106.5163.580.0100.373.055.524.837.330.345.532.225.522.818.419.316.525.31.2912.06869.2701.0621.0271.8641.0010.0580.27 10.1560.7900.4190.2380.1840.0900.0880.0380.05376.6107.681.047.214.118.717.618.117.317.016.613.514.011.215.60.0430.3370.3080.1480.0100.0190.0220.0290.0430.0530.0590.0300.0320.0120.009of smaller particles possibly by the same processes discussed above. The 2D images show increased aggregatespresent with sizes increasing through leg 8 (Fig. 7b).The concentration of dendritic crystals also increasedfrom leg 4 (-22°C) to leg 7 (-13°C).Leg 9 showed some peculiarities, again suggestingthat the aircraft may have been slightly out of position.The 2D images showed no large aggregates and numerous small particles. At leg 10, however, the largeaggregates were again evident.Legs 8 (-11°C) through 14 (-3°C) (region c) arecharacterized by No increasing rapidly and X remainingrelatively constant. This indicates a major input ofsmall particles but with aggregation still occurring. The' O'Ol-----715 FEB 82-2.5 """""""""'0.5 1.0 1.5 2.0 2.5 LOG (A), cm-IFIG.' 9. Louo vs logX for the various flight legs using 2D-P data.presence of liquid water with some drops larger than24 pm (Fig. 6) and the temperature of these legs suggestsa rime-splintering mechanism for secondary ice crystalproduction (Hallett and Mossop, 1974; Mossop, 1976).Even though the concentration of drops larger than 24pm is rather low for a rime-splintering mechanism tooccur there is evidence that drops smaller than 13 pmcan produce protuberances upon freezing that may beimportant in the rime-splintering process (Griggs andChoularton, 1983). If aggregation was not occurring,then X would increase. This assumes that depositionalgrowth with respect to diameter is slow at large sizes.Figure 10 shows the total ice crystal concentration increasing to a maximum at leg 12 (-6°C) and then decreasing through leg 14. The 2D images show manycolumnar type crystals throughout this region with thenumber and size of aggregates increasing (Fig. 7).Both No and X decrease from leg 14 (-3°C) throughleg 17 (0°C) in Fig. 9 (region d). This suggests moreaggregational growth with no significant input ofsmaller particles. Total concentration is also decreasingfrom its maximum at leg 12 (-6°C) (Fig. IO). The 2Dimages show fewer and fewer pristine crystals withlarger and more. numerous aggregates (Fig. 7d). Giantaggregates and a few water drops were observed inleg 17.By leg 18 (1 "C) (region e) the large aggregates aremelting, collapsing and decreasing in concentration dueto increased fallspeed. This causes X to increase andNo to decrease due to the smaller crystals and aggregatesmelting and the resulting decrease in concentration.Some drops may be small enough to not be detectable.The 2D images show fewer aggregates and more largewater drops (Fig. 7e).1 JUNE 1986GLENN L. GORDON AND JOHN D. MARWITZ15 FEB 82-30e- NUCLEATIONANDI DEPOSITION 7-20 F 17bI 2LOG (A).cm"-2 -I LOG (No), c m-4hGGREGATION ANDDEPOSITIONE SlCP ANDAGGREGATION AGGREGATION! MELTING,i ANDCOLLAPSE0 5 IOCONCENTRATION(L-91093FIG. 10.2D-P values of lo@, logNo and concentration plotted as a function of temperature and altitude. Regions a-e correspond to different hydrometeor growth regimes.It can be seen in Fig. 10 that, in general, X increaseswith decreasing temperature above the 0°C level. Theseresults agree with those of Houze et al. (1979) and Loand Passarelli (1982). Houze et al. (1979) show thatNo also increases with decreasing temperature abovethe 0°C level. As can be seen in Fig. 10 such was notthe case for these data; No increased at the colder temperatures as more smaller particles were being producedor detected. Then No decreased with increasing temperature through the aggregation region (legs 4 through8). An increase in No is noted from legs 8 through 14.As mentioned before, an increase in smaller particleswas observed in this region due to secondary ice crystalproduction. No then shows a decrease through the 0°Clevel as aggregation and melting become the dominantprocesses. These results are similar to those of Lo andPassarelli ( 1982).4. Case study of 13 March 1984a. Synoptic situationA National Weather Service surface and 500 mbmap for 0400 LST 13 March 1984 is shown in Fig. 1 1.A mature frontal system was approaching the WestCoast. At the time of the descent through the band(0900 UT) a cold front was just west of the study area.An ascent sounding from McClellan Air Force Basecan be seen in Fig. 12. The atmosphere was relativelydry at the surface with moisture increasing up to about700 mb and then decreasing above 700 mb. Stable layers were present from the surface to about 900 mb andfrom 630 to 580 mb. The Oe trace indicates that theatmosphere was slightly potentially unstable from 800to 630 mb (+5 to -1OOC) but otherwise stable. Thehodograph shows a barrier parallel jet of about 22 mI. I~ ~~ ~ ~FIG. I 1. As in Fig. 3 but for 0400 LST 13 March 1984.1094JOURNAL OF THE ATMOSPHERIC SCIENCES VoL. 43, No. 1113 MAR 84 0833-0855LI~'I-I-I'IIic)Ew'5 6002 700 8009001000v):$//A1/ -10//290 300 310 320 0 IO 20ms-' 30270.FIG. 12. As in Fig. 4 but from 0833 to 0855 UT 13 March 1984.s" at approximately 0.5 km and a barrier normal windof about 24 m s-l above 1 km. Above the crest thewinds veered about 20 degrees.Figure 13 shows a PPI of reflectivity taken at 0937LST at an elevation angle of 3.0" from the CP-4 Doppler radar located 9 km due south of Sheridan. The bandat this time has just passed the radar and is moving at' '' about 15 m s-I to the east. There are regions of heavier precipitation enbedded within the band probably due to the presence of weak convection as indicated on the 8, trace.b. Hydrometeor dataThe flight track for this descent is shown in Fig. 14.The particle trajectory in this case has a much greatersoutherly component reflecting the much slower general band movement towards the east than on 15 February 1982.Figure 15 shows leg-averaged FSSP concentration,FSSP concentration for droplets larger than 24 pm andRAINBANO1001I IIFSSP liquid water content plotted as a function of temperature and altitude. In this case there is measurableliquid water present up to about the - 15°C level withthe maximum liquid water values about O,.l g m-3.Droplet concentrations are much higher than in theprevious case, perhaps a result of the more unstablenature of the atmosphere. Droplets with diameterslarger than 24 pm are present at the -5°C level butagah only in relatively small concentrations (0.5 ~m-~).Figures 16 and 17 show 2D-C images and 2D-P sizespectra, respectively, at average temperaturesof -27.0,-16.8, -4.9, -1.2 and +2.3"C (legs 2, 6, 12, 14 and16 from' Fig. 14). Looking at Figs. 16b2 and 1 7bz(-27.O"C) one can see that there are already particleslarger than 1.5 mm present in significant numbers.Even in leg 1 there were particles larger than 1 mm.This indicates that the aircraft began the descent wellbelow the top of the band causing region a, the initialprimary nucleation and particle growth regions to bemissed. Figures 1 6b2, 1 6b6 , 17b2 and 1 7b6 indicate thatthe descent began during the aggregation stage; however, the particle type evolution and particle spectraevolution are the same as in the 15 February 1982 case(Figs. 7 and 8).500-50-100 I I It I I-100 -50 0 50 100DISTANCE (kmlFIG. 13. As in Fig. 5 but for the CP-4 Doppler radar (located 9km due south of Sheridan) at an elevation angle of 3.0" at 0937 UT.-20-30-40- 40 - 20 0 20 40DISTANCE FROM SHERIOAN RADAR (km)FIG. 14. As in Fig. 2 but for 13 March 1984.1 JUNE 1986 GLENN L. GORDON AND JOHN D. MARWITZ 10950eWa=lca awnw ElFSSP 13 MAR 84-30-zob8TOTAL CONCENTRATION0 20 40 60 80 100 ' 120 1400.1 0.3 0.5DROPLET CONCENTRATION DROPLETS LWC(cm-9 2 24pm (ern-') (elm3)FIG. 15. As in Fig. 6 but for 13 March 1984.The log of X is plotted against log& for the 2D-P inFig. 18. In general, both X and NO decrease from leg 1through leg 14 indicating aggregation is occurringthroughout the band. The lack of change in X and Nobetween legs 10 through 12 is due to secondary icecrystal production as will be seen later.Figure 19 shows logX and the logNo plotted as afunction of temperature and altitude for the 2D-C and2D-P (values of X and No along with average temperature and altitude for each leg can be seen in Table 5).The figure has been divided into four regions (b-e)corresponding to the same hydrometeor growth processes as in the 15 February 1982 case (as mentionedbefore, region a-the primary nucleation and deposition region-was not sampled). Region b is characterized by X and No decreasing with increasing temperature. As previously discussed, this is due to aggregation.Region c shows X and No remaining relatively constantfor the 2D-C and 2D-P.Figure 20 contains the 2D-C and 2D-P concentrationas a function of temperature. It can be seen that the2D-C concentration increases from around 40 L" at-12°C to over 100 L" at -5°C. The same trend canbe seen in the 2D-P concentration. This is probablydue to a rime-splintering secondary ice crystal production mechanism as in the 15 February 1982 case.Evidence that these excess ice crystals were producedin this region of the cloud can be seen in Fig. 16c showing 2D-C images of columnar type crystals. Furtherevidence is shown in Fig. 2 1, a microphotograph of an13 MAR 84FIG., 16. As in Fig. 7 but for regions b-e of the rainband. Thesubscripts 2 and 6 indicate samples taken from legs 2 and 6 respectivelyin region b.%,rJ 3 0. I 0.2 0.3 0.4DIAMETER (cm)FIG. 17. As in Fig. 8 but for regions b-e of the rainband. Thesubscripts 2 and 6 indicate samples taken from legs 2 and 6 respectivelyin region b.1096 JOURNAL OF THE ATMOSPHERIC SCIENCES VOL. 43, No. 110~0~""1""'""1""~13 MAR 84sa -1.5-2.0-2.50.5 1.0 1.5 2.0 2.5 LOG (X), cm"FIG. 18. As in Fig. 9 but for 13 March 1984.oil-hexane slide collected on the King Air at a temperature of -5"C, indicating that the predominantcrystal habits are sheaths and needles (Nlc and Nla,from Magono and Lee, 1966). Notice also that thereare some rimed particles. Since the newly formed crystals are small, the 2D-P cannot see them, which accounts for the differences between the two probes.Region d on Fig. 19 shows X and NO again decreasingrapidly, indicating that aggregation is once again thedominant process. This can also be seen in region don Fig. 20, which shows a rapid decrease in particleconcentration with increasing temperature.In region e, X is increasing with increasing temperature while NO increases for the 2D-C and decreasesfor 2D-P. As in the 15 February 1982 case, the iceparticles are melting and collapsing. As the particlesmelt and become even smaller, some of them becometoo small for the 2D-P to detect; hence, the differencein the change of No between the two probes.c. Bulk ice density calculationsBulk ice densities were estimated by comparing reflectivities calculated from hydrometeor spectra andreflectivities from the CP-4 Doppler radar. The technique consists of first establishing the coordinates (withrespect to the radar) of the end points of a line alongthe flight track of a given leg of the figure-eight descent.A radar cross section is then constructed, from a volume scan, along this line that is 10 km in width. Theradar cross section for leg 9 is.shown in Fig. 22. Thatportion of the flight track that was within the box isalso shown. The reflectivity is averaged along the flighttrack. This value is compared to the reflectivity calculated from the hydrometeor spectra measured by the2D-C and 2D-P. The bulk.ice density necessary to makethe 2D reflectivities agree with that from the radar isassumed to be the correct value.The bulk ice density is plotted against temperaturein Fig. 23. Since the radar reflectivities above the - 1 1 "Clevel were low (< 10 dBZ) no bulk ice densities werecalculated above that level. The curve was drawn freehand and reflects what seems to make physical sensebelow the melting layer. The important point to noticeis that the bulk ice density is very low above the meltinglayer reaching only about 0.1 g cm-3 at the - 1 "C level.13 MAR 84LOG (No),~rn-~FIG. 19. As in Fig. 10 but for 13 March 1984. Lettered regions correspond to different hydrometeor growth regimes.1 JUNE 1986 GLENN L. GORDON AND JOHN D. MARWITZ 1097TABLE 5. Mean values of temperature (T), altitude (Z), slope (X) and intercept (No) for the 2D-C and 2D-P for each leg for the 13 March 1984 case.2DC 2D-PT Z x NO x ("C)Leg (m) (cm") (cm-3 (cm") (~m-~)NO12345678910111213141516-28.7-21.0-24.3-21.4-19.0- 16.8-14.5-12.2-10.5-8.5-6.6-4.9-2.8-1.20.72.369136694634 1597 156255307495645874248389435573225285525 162157191956.551.147.642.538.437.128.827.724.422.925.525.921.218.318.131.02.6901.8531.7451.3370.9880.9310.7400.6930.4780.3970.4460.4760.2270.1380.0690.08252.446.339.134.433.529.919.319.016.816.418.117.215.313.112.716.50.5750.4170.3330.2790.3060.2580.1440.1470.1 170.1 140.1180.1 100.0660.0380.0 190.006These low values probably reflect the lack of particlemass increase due to riming even though some rimingwas occurring.Once the bulk ice density is known as a function oftemperature, the ice water content and precipitationrate can be plotted as a function of temperature. Theice water content is calculated using Eq. (2),IWC = E piz[~3n(~)l6 (2)where pi is the bulk ice density, D the particle diameterand n(D) the concentration of particles with diameterD. The precipitation rate is calculated using Eq. (3),?rPR = - P~Z[D~~(D)V~(D)] (3)where u,(D) is the terminal velocity of particles withdiameter D. The ice water content (Fig. 24) is approximately 0.2 g m-3 above the -7°C level and increasesto around 0.6 g rnp3 at the melting layer. It decreaseddramatically below the melting layer as the concentra6-30t aW3laaaa5W2I- *\ 72I0 20 40 60 80 100 '2D-P CONCENTRATION 2D-C CONCENTRATION(1-1) ( 1-11FIG. 20.2D-C and 2D-P concentrations plotted as a function of temperature and altitude. Lettered regions correspond to different hydrometeor growth regimes.W3nc5a1098FIG. 2 1. Microphotograph of an oil-hexane slide collected in the King Air at 094438 LST at a temperature of -5°C.tion of larger particles decreases due to their greaterfall velocity. Note that when all particles are meltedthe bulk ice density is. 1.0 g m-3 and the "ice watercontent" is actually the liquid water content. The pre' , cipitation rate (Fig. 25), indicated by the 2D-C, increases from about 0.8 mm h-' at -10°C to a maximum of about 2.3 mm h" below the melting layer.The decrease in precipitation rate below the melting13 MAR 84. - . . . . - .I I I I /I I I\ I I0 8 16 24 32 40 48 56 64 12 80 DISTANCE ALONG LINE (km)FIG. 22. Cross section of reflectivity from the CP-4 Doppler radarcalculated from a PPI volume scan during the time interval 0935 to0942 UT. The cross section is along a line with end-point coordinatesof (1541) and (20, -37) (km with respect to the radar). The averagingbox width is 10 km.VOL. 43, No. 1 Ilayer shown by the 2D-P is due to relatively large iceparticles melting into smaller size water drops thatcannot be detected as efficiently.4. ConclusionsA figure-eight descent flight track was used to document the hydrometeor evolution for two rainbandsthat passed over northern California. The change inthe hydrometeor spectra with changing altitude wasused to infer hydrometeor evolution from the initialnucleation of ice particles to the melting of large aggregates into raindrops. It was found that the ice particles go through five distinct regions of growth. Thefirst region (a) at the top of the rainband is dominatedI3 MAR 84X 20-P ISx 016 3IIIIII1III 0.2 0.4 0.6 0.8 1.0 BULK ICE DENSITY (g/crn3)FIG. 23. Bulk ice density plotted as a function of temperature. Flight leg numbers are indicated for the 2D-C.I JUNE 1986 GLENN L. GORDON AND JOHN D. MARWITZ 1099by the processes of nucleation and deposition. The iceparticles then go through a region of aggregation (b).The depth of the first two regions are different betweenthe two rainbands presented. This is probably a result.of the differences in overall depth of the two rainbandsand the differences in the stability of the atmosphere.This aggregation region seems to end around the- 10°C level for both rainbands. The next region, c, isdominated by continued aggregation and secondaryice crystal production by rime splintering. These processes occur down to around the -4°C level. Lo andPassarelli (1982) conclude that particle breakup due tocrystal-crystal collisions in the temperature region between 0" and - 10°C will limit the production of largeaggregates and cause the particle spectrum to reach anequilibrium distribution. Our data show that the slopesof the exponential distributions remain relatively constant once they reach a value of about 10 to 20 cm".This agrees with Lo and Passarelli. However, the intercepts continue to evolve below the - 10°C level. Initially the intercepts increase when the temperature increases from -10" to -4°C due to an increase in thenumber of small particles by an apparent rime-splintering secondary ice crystal production mechanism asmentioned before. Collisional breakup has been ruledout since we did not observe large water drops or graupel particles (Hobbs and Farber, 1972) or a large concentration of rimed crystals (Vardiman, 1978), thoughtto be necessary for secondary ice crystals to be producedby collisions. Below the -4°C level (in region d) theintercept begins to decrease again as the smaller particles cease to be produced and aggregation is againdominant. The data presented by Lo and Passarelli(1982) can also be interpreted in a similar manner,especially their second spiral, showing the same effectalthough not as strong as in our data. This is due mainlyto the fact that by using a one-dimensional precipitationprobe they were undersampling many of the smallerparticles (Gordon and Marwitz, 1984). The fact that13 MAR 84-150 20-Cw3aaaI- 0w351 ' I I I I I0.0 0.1 0.2 0.3 0.4 0*5 0.6 0.7 ICE WATER CONTENT (g/m3)FIG. 24. As in Fig. 23 but for ice water content.-'5413 MAR 842 -10W3aaaaI-0t -5wwX 160 I50.0 0.4 0.8 1.2 1.6 2.0 2.4 PRECIPITATION RATE (mmlhr)FIG. 25. As in Fig. 23 but for precipitation rate.the slope of the distribution reaches a limiting valueof between 10 and 20 cm" is probably due to a balancebetween the small particles produced by rime-splintering and the large particles produced by aggregation.The only effect on the distribution .by these processesis to change the value of the intercept which is due tochanges in particle concentration. The final stage (T> OOC) was dominated by melting, and perhaps collisional breakup (e). Aggregation seems to be occurringthroughout most of the cold (T < 0°C) part of therainbands.Bulk ice densities were calculated by comparing radar reflectivities with reflectivities calculated from thehydrometeor spectra. Bulk ice densities increased withincreasing temperature but reached a value of only 0.1g cm-3 just above the melting layer. This reflects-thelack of mass growth of the ice particles due to riming.The results of these two case studies are similar tothe weak warm sector rainbands studied by Matejkaet al. (1 980). There was no evidence of a seeder-feedermechanism for precipitation enhancement in either ofthe rainbands studied. Also, aggregation, and not riming, appears to be the dominant process for producingprecipitation size particles. Some riming must certainlybe occurring; otherwise, the apparent secondary icecrystal production by rime-splintering observed nearthe -5°C level could not have occurred.Acknowledgment. This research was funded by theDivision of Atmospheric Water Resources Management, Bureau of Reclamation, Department of the Interior, Contract 2-07-8 1-VO256.REFERENCESCooper, W. A., 1978: Cloud physics investigations by the University of Wyoming in HIPLEX 1977. Dept. Atmos. Sci., University of Wyoming, 320 pp.Gordon, G. L., and J. D. Marwitz, 1984: An airborne comparison of three PMS probes. J. Atmos. Oceanic Technol., 1, 22-27.1100 JOURNAL OF THE ATMOSPHERIC SCIENCES VOL. 43, No. 11Gdggs, D. J., and T. W. Choularton, 1983: Freezing modes of riming droplets with application to ice splinter production. Quart. J. Roy. Meteor. SOC., 109, 243-253.Hallett, J., and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249, 26-28.Heggli, M. F., L. Vardiman, R. E. Stewart and A. Huggins, 1983:Supercooled liquid water and ice crystal distributions withinSierra Nevada winter storms. J. Climate Appl. Meteor., 22, 18751886.Herzegh, P. H., and P. V. Hobbs, 1980 The mesoscale and microscale structure and organization of clouds and precipitation in mid latitude cyclones. Part 11: Warm frontal clouds. J. Atmos. Sci.,Hobbs, P. V., 1978: Organization and structure of clouds and precipitation on the mesoscale and microscale in cyclonic storms. Rev. Geophys. Space Phys., 16,741-755.J. Rech. Atmos., 6,245-258.Jr., 1980: The mesoscale and microscale structure and organization of clouds and precipitation in midlatitude cyclones. I: Acase study of a cold front. J. Atmos. Sci., 37, 568-596.Houze, R. A., Jr., P. V. Hobbs, K. R. Biswas and W. M. Davis, 1976a: Mesoscale rainbands in extratropical cyclones. Mon. Wea.37,597-61 I.- , and R. J. Farber, 1972: Fragmentation of ice particles in clouds.- , T. J. Matejka, P. H. Herzegh, J. D. Locatelli and R. A. Houze,Rev.. 104,868-878.- , J. D. Locatelli and P. V. Hobbs, 1976b Dynamics and cloudmicrophysics of the rainbands in an occluded frontal system. J.Atmos. Sci.. 33, 1921-1936.-, P. V. Hobbs, P. H. Herzegh and D. B. Parsons, 1979: Size distributions of precipitation particles in frontal clouds. J. Atmos. Sci., 36, 156-162.Lo, K. K., and R. E. Passarelli, Jr., 1982: The growth of snow in winter storms: an airborne observational study. J. Atmos. Sci.,Magono, C., and C. Lee, 1966: Meteorological classification of natural snow crystals. J. Fac. Sci., Hokkaido University, Ser. VII, 2,Matejka, T. J., R. A. Houze, Jr. and P. V. Hobbs, 1980 Microphysics and dynamics of clouds associated with mesoscale rainbands in extratropical cyclones. Quart. J. Roy. Meteor. SOC., 106,29-56.Mossop, S. C., 1976: Production of secondary ice particles during the growth of graupel by riming. Quart. J. Roy. Meteor. Soc., 102,45-57.Passarelli, R. E., Jr., 1978: An approximate analytical model of the vapor deposition and aggregation growth of snowflakes. J. Atmos. Sci., 35, 118-124.Rutledge, S. A., and P. V. Hobbs, 1983: The mesoscale and microscale structure and organization of clouds and precipitation in mid latitude cyclones. VI1 A model for the "seeder-feeder" process in warm frontal rainbands. J. Atmos. Sci., 40, 1 185-1206.Vardiman, L., 1978: The generation of secondary ice particles in clouds by crystal-crystal collision. J. Atmos. Sci., 35,2168-2180.Waight, K. T. 111, 1984: A numerical study of the Sierra Nevada barrier jet. Dept. of Atmos. Sci. Rep. No. 148, University ofWyoming, Laramie, 119 pp.39,697-706.321-335.