Raindrop sorting induced by vertical drafts in ... - Wiley Online Library

7 downloads 3666 Views 495KB Size Report
Jul 15, 2001 - Rosenstiel School of Marine and Atmospheric Science, Division of Meteorology and ... Hurricane Research Division, NOAA/AOML, Miami FL.
GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 14, PAGES 2787-2790, JULY 15,2001

Raindrop

Sorting Induced by Vertical Drafts

in Convective

Clouds

Pavlos Kollias, B. A. Albrecht Rosenstiel School of Marine and Atmospheric Science, Division of Meteorology and Physical Oceanography, University of Miami

F. D. Marks, Jr. Hurricane ResearchDivision, NOAA/AOML,

Miami FL

94-GHz Doppler radar (• - 3.2 mm) and capitalizingon

droplets will stay aloft until they grow further, or are removed by horizontal advection and fall out of the updraft. Sporadic evidenceof the above mentioned processesare pro-

the resonant nature of the backscattering cross-sectionas

vided by aircraft observations[Carboneand Nelson, 1973;

a function of the raindrop size (Mie scattering),the verti-

Rauber et al., 1991; Szumowski et al., 1998; Atlas and U1-

cal air motionsto an accuracyof 0.1 ms-x andthe shapeof

brich,2000] . These aircraft observationsindicateda corre-

Abstract. Evidence of raindrop sorting induced by a convective updraft is presented. Using a vertically pointing

lation between the location of large raindrops and updrafts. Recirculation from the sides of the updraft and prolonged suspensionof the raindrops at the top of the updraft were suggestedas primary mechanismsfor the growth of large raindrops. However, aircraft measurements are confined (D _< 1.7 mm) that have terminal velocitieslessthan the to the flight level, limiting information on the vertical coupdraftvelocities(6-7 ms-1) anda clearabsence of drops herency of the observed precipitation field. In this paper, observations of the interaction of ver_> 3 mm are observed. Towards the updraft periphery, a gradual increase in the raindrop sizes is documented where tical drafts and raindrops retrieved from a 3-mm wavelarge raindrops(D _>3 mm) are observed.The observations length Doppler radar are presented. The observations exdemonstrate the importance of updrafts in distributing the tend within and to the boundaries of an updraft observed [Kolliaset al., 1999]at the baseof a shallowconvectivecloud raindrops in space. with cloud top below 4 km. The high temporal and spatial the raindrop size distribution are retrieved from the Doppler spectra. The interaction of vertical drafts and raindrops is documented for the first time by high resolution radar data. The updraft structure clearly causeshorizontal and vertical sorting of the raindrops. In the updraft core, small raindrops

resolution

Introduction

of the

cloud

radar

recovered

the

fine structure

In convective precipitation, updraft velocities often are greater than the full spectrum of terminal fall velocities as-

of the low-level updraft. In this paper, the retrieved DSDs within the updraft are presented. Horizontal and vertical drop sorting effects induced by the updraft are well documented in a unique data set that adds an additional dimen-

sociatedwith the drop size distribution (DSD). As a re-

sion (vertical) to that availablefrom aircraft observations.

sult, raindrops are carried upward and ejected at higher al-

titudes, influencingthe evolutionof convectiveclouds(particle "fountain"[Yuter and Houze,1995]). Despitethe importance of the interaction between raindrops and vertical drafts, there are several unresolved or poorly observed aspects of this interaction. For example, what size of raindrops exists at altitudes were the updraft magnitude overcomes the fall velocity of raindrops? Do large drops coexist with small raindrops, or doesthe kinematic field sort the particles horizontally and vertically? To date, remote sensorsand aircraft penetrations have been unable to observesuch features due to their relative coarseresolution, ambiguities related to the instruments, and difficulty in observingthe narrow updrafts within large precipitating areas. Srivastava and At-

Background Cloud Doppler radars are primarily designed for the

study of non-precipitatingclouds[Lhermitte,1987]. Due to their short wavelength(X = 3.2 mm), mm-wavelength radarshaveexcellentsensitivityto small clouddroplets(for Rayleigh scattering, the backscattering cross section orbis

proportional to 1//k4),but suffersevereattenuationwhen large raindrops are present due to Mie scattering. Thus, the use of millimeter wavelength radar for precipitation studies was considered to be inadequate as measured by the traditional means that Doppler radars have been used for pre-

las [1969],developeda simplified"top-hat" updraft model cipitation radars studies (e.g. PPI and RHI scanningfor that captured the main features of the interaction between reflectivity measurements). Lhermitte, [1988]proposeda new techniquefor making the dynamical and microphysical fields within an updraft. Their model indicates that in the presenceof updrafts, small

Copyright2001 by the AmericanGeophysical Union. Papernumber2001GL013131. 0094-8276/01/2001GL013131505.00

2787

accurate measurements of vertical air motion and raindrop size distribution using a 94-GHz Doppler radar. At 94 GHz, the scattering from all raindrops must be treated as Mie scattering. The backscattering versus diameter curve exhibits a quasi-periodic form with an exponential damping of the oscillation. The oscillating nature of the backscattering curve is caused by the superposition of the multipole terms

2788

KOLLIAS

ET AL.- RAINDROP

SORTING

INDUCED

BY PRECIPITATION

DRAFTS

described in the Mie solution. The hydrostatic model pro-

posedby [Green,1974]is usedin this analysisto describethe

I

equilibrium shape of falling raindrops. The raindrops are approximated as oblate spheroidssincelarge raindrops deviate from sphericity. The oscillations due to Mie scattering are

2

3

100

presentin the Dopplerspectrum[Lhermitte,1988;Firda et al., 1999;Kollias et al., 1999]and the shift of the first min-

10-•

imum compared with the terminal velocity of the droplet associated

with

this minimum

can be used to estimate

10-2

the

minal velocity - diameter relationship. The accuracy of the technique depends on the accuracy of the relation between terminal fallspeedand drop diameter. Once the air motion is retrieved the DSD can be retrieved from the Doppler spec-

trum usingthe methoddescribedin [Kollias et al., 1999]. The narrow beamwidth (0.24ø), the short dwell time, and

theexcellent velocityresolution (3.12cms-•) oftheDoppler spectra further enhance the precision of the retrieved DSD. It will be demonstrated through the data presented, that the raw Doppler spectra can be used for qualitative analysis

0.1

0

air motionwithinq- 0.1 ms-• usingLhermitte's[1988]ter-

0.2

0.3

0.4

0.5

Diameter (cm)

1 ooo

• 8oo

::: _ool :::i,

.?

•_.600 400

-2 0 2 4 6 8 10 -4-2

0 2 4 6 -4-2 0 2 4 6 8

of theDSDfieldin timeandspace.

1000

000

Observations

8oo

In this section, the distribution of raindrops within a lowlevel updraft is presented. The casewas analyzed in Kollias

600

et al., [1999].Theshallow convective cloud(top< 4 km)was

400

sampled by the vertically pointing 94-GHz Doppler radar with a temporal resolution of 6 s and a vertical resolution of 60 m. The data were collected on December 11, 1998, at Key Biscayne, Miami Florida. Fig. 1 showsthe retrieved air

1000 ,

•N,// 800 1• 600

800

'•.i' L 400 400[ -20 2 4 6 81012 -2 0 2 4 6 81012

-202468

Velocity (ms -1)

Velocity (ms -1)

Velocity (ms -1)

Figure 2. a) Normalizedbackscatteringcross-section as a func-

velocityfield during a 4 minute period, from 200 m (lowest tion of the diameter for oblate spheroids.b) Contoursof vertical radar gate) to 1.4 km. The updraft velocity field exhibits profilesof Doppler spectra(spectrograms)wherethe contoursare an inverted bell shape.

Due to the Mie oscillations(Fig. 2a), the Doppler spectrum may containone (D < 1.7 mm), two (1.7 mm < D _ 3 mm) dependingon the raindrop sizes within the radar sampling volume. Fig. 2a, shows the normalized backscattering cross section at wavelength A = 3.2 mm as a function of the diameter for oblate spheroids

db above the noise baseline. In the top panel, movement from left to right indicates increasing time as the displayed data approach the updraft core. In the bottom panel, movement from left to right indicates increasing time as the displayed data leave the updraft core and approach the updraft periphery. The dashed lines shows the retrieved vertical air motion for each profile.The Doppler spectra velocity is positive for downward motions. The

vertical air motion is positivefor upward motion, and the numbers 1, 2 and 3 correspond to the Mie peaks shown in Fig. 2a

Air Velocity 8

usingthe T-

matrix [Mishchenkoand Travis, 1994] ap-

proach. In the top panels of Fig. 2b, contours of Doppler

spectrapowerwith height(spectrogram)as the updraft core shown in Fig. I approachesthe radar are shown. The Mie backscattering oscillations that modulate the Doppler spec--?------.--.---.--.-.---..--•.....-..••..••,•,••;• • •i•?:' /-•,-•,,•............• trum provide a natural fingerprinting of the raindrop spectrum that is used to analyze the microphysical field and its evolution across the updraft interior. At the leading edge

980

of the updraft (top left panel, Fig 2b), the spectrograms

1

2

3

4

rn$-1

Time (min)

Figure

1. Vertical air motion field calculatedfrom the Doppler

indicate the presenceof raindrops contributing to the first and second Mie peak, corresponding to diameters _< 3 mm. Gradually, as observations are made closer to the core, the signal from the second Mie peak decreases,indicating the gradual disappearance of raindrops with sizes > 1.7 mm in

diameter. At the updraft core (top right corner, Fig. 2b) nearly all raindrops have D _< 1.7 mm. In addition to the transition occurring in the horizontal dimension, sorting ef-

m respectively. The updraft was located at the base of a cloud

fects are apparent in the vertical dimension(top right and bottom left). At higheraltitudes,the first Mie peak narrows

turret with a cloud top height of 3.5 km

indicating that the maximum size of raindrop decreasesbe-

spectra data.

The temporal and vertical resolution is 6 sec 60

KOLLIAS ET AL.' RAINDROP

SORTING INDUCED

low 1.? min. These observations are unique and clearly show a lack of drops with falling velocitieslarger than the updraft

BY PRECIPITATION

DRAFTS

2789

0

strength(5-6 ms-z ) at higherlevelswithinthe updraft.In the bottom panels of Fig. 2b, we follow the time evolution of the spectrograms as we approach the trailing edge of the updraft core. Now the reverse processoccurs as indicated

by large raindrops (gradual appearanceof the secondand third Mie peak signal) of ( D _>3 mm) appearingin the

-0.5

Doppler spectrum with increasing time. Furthermore, the large raindrop signature appears initially at the low levels near the updraft base and are gradually detected at higher altitudes, a behavior consistent with a bell-shaped updraft. The data are also consistent with the trajectories of large raindrops at the base of a cylindrical updraft as described

o2 min

t -- 2.2 min

,b= 1.8 min

t = 1.i min

by [$rivastavaand Atlas, 1969]. The systematic variability detected by the raw Doppler spectra shown in Fig. 2 is verified by the retrieved DSDs. Fig 3 showsa time-height mapping of the median volume diameter Do, retrieved using the Doppler spectra during the updraft overpass. Nine profiles of Do with height, spaced every 30 sec were used to construct Fig. 3. The data show clearly the interaction between the updraft and the raindrops. There is an apparent correlation between the upward

o14 Diameter (cm)

Figure 4. Four retrieved drop size distributions sampled at 900 m altitude at distinct times in Fig 1. (cross at 0.2 minute, diamonds at 1.1 minute, circle at 1.8 minute and square at time 2.2 minute) as we move toward the updraft core.

air motion and low Do values. The Do varies between 1.4-

tion of large drops near the updraft core. The DSDs shown in Fig. 4 are not scaled, since strong attenuation of the signal does not allow accurate measurements of power. These distributions are representative of their location relative to the updraft axis. The depletion of large raindrops is gradual and it is related to the trajectories of the larger raindrops as they escape from the updraft. The appearance of large verified by the raw Doppler spectra (bottom right spectro- drops at the updraft periphery indicates the important role gram, Fig. 2) where only signalfrom very large raindrops of the updraft in the generation of large raindrops and in (third Mie peak) is observedat low levelsand graduallythe distributing the raindrops over large areas. Atlas and U1other two Mie peaks appear at higher altitude. In the leadbrich, [2000]usingairbornemeasurements in tropicalwarm ing edge of the updraft this feature is not observedsince the rain, observedsimilar narrow DSD's in an updraft and broad raindrops fall into a layer of precipitation causedby the leadDSD's with larger raindrops in the downdraft. The warm ing part of the precipitating cloud. The retrievals verify the rain character of the shallow convective cloud should also qualitative description derived from the raw Doppler specbe considered in our interpretation. Vertical shear of the tra and clearly showhorizontal and vertical sorting. Actual horizontal wind might explain the horizontal sorting of the 1.5 mm at the updraft core to 3.0 mm at the trailing edge of the updraft. In addition, there is an asymmetry in the Do field between the leading and trailing edge of the updraft. The trailing edge of the updraft signals the end of the precipitation and the large Do are caused by the different fall velocities of raindrops of different sizes. This observation is

retrievedDSDs (Fig. 4) alsosupportthe systematicdeple- raindropssincethe updraft is convergingat all levels(0.3-1.4 km) of the observations.Anotherpossibilityfor this sorting is the ejection of the large raindrops at higher levels where the updraft is diverging. 1220

Conclusions For the first time we have strong evidence for the interaction of raindrops with the storm updraft. The cloud top was below 4 km, and only warm rain processesare relevant for raindrop growth. The data show horizontal and vertical size sorting of the raindrops in the environment of a convective updraft. The physical image that emerges from the observations shows large raindrops subsiding at the boundaries of a bell-shaped updraft and small raindrops in the interior of the updraft. In the updraft core, the maximum size of raindrops observedis determined by the updraft magnitude

980

740 500

(Dmax __< D(Vupdraft)). The updraft acts as a sieve,where

260

1

2

3

4

Time (min)

Figure 3. Time-height mapping of the medium volume diameter (Do) within the updraft core. Nine vertical profilesof retrieved Do spaced every 30 s were used to construct the image.

the diameter of the perforations is replaced by the updraft magnitude. Raindrops escape only if they reach sizes with fall velocities that overcome the updraft velocities or they reach the diverging updraft top. Since only warm rain processesare at work, there are no particles falling from above into the updraft to disturb the picture. As a result, horizon-

2790

KOLLIAS

ET

AL.:

RAINDROP

SORTING

tal raindrop sorting will be induced by horizontal advection and the different altitudes where raindrops of different sizes are released. However, more observations of this type are needed to solidify the conclusions. In addition, the observationsclearly demonstrate the potential of 94-GHz Doppler radars for decomposingthe complex precipitation field and retrieving the vertical air motion and the DSDs. The Mie backscattering oscillationsprovide a very robust transfer function between the Doppler spectra and the raindrop sizes. As a result, even a graphical inspection of the raw Doppler spectra provide a qualitative description of the kinematic and microphysical fields without the use of complicated retrieval techniques. Acknowledgments. Grant

ATM9730119

and DOE

This work was supportedby NSF Grant

DEFG0297ER62337.

References Atlas, D., and C. W. Ulbrich: An observationally based conceptual model of warm oceanic convective rain in the tropics. J. Appl. Meteor., 39, 2165-2180, 2000. Aydin, K., and Y-M, Lure: Millimeter wave scattering and propagation in rain: A computational study at 94 and 140 GHz for oblate spheroidal and spherical raindrops. IEEE Trans. Geosci. Remote $ens., 29, 593-601, 1991. Carbone, R.E., and L.D. Nelson, 1978: The evolution of raindrop spectra in warm-based convective storms as observed and numerically modeled. J. Atmos. $ci., 35, 2302-2314. Firda J. M., S. M. Sekelsky and R. E. Mcintosh:Application of dual-frequency millimeter wave Doppler spectra for the retrieval of drop size distributions and vertical air motion in rain, J. Atmos. Oceanic Tech., 16, 216-236, 1999. Green, A.W. : An approximation for the shapes of large raindrops. J. Appl. Meteor., 1J, 1578-1583, 1974. Gunn R., and G. D. Kinzer, The terminal velocity of fall for water drops in stagnant air J. Meteor., 6, 243-248, 1949.

INDUCED

BY PRECIPITATION

DRAFTS

Kollias, P., R. Lhermitte and B.A. Albrecht: Vertical air motion and raindrop size distributions in convective systems using a 94 GHz radar. Geophys.Res. Letters, 26, 3109-3112, 1999. Lhermitte, R. M.: A 94-GHz Doppler radar for cloud observations. J. Atmos. Oceanic Techol., J, 36-48, 1987. Lhermitte R., Observations of rain at vertical incidence with a 94 GHz Doppler radar: an insight of Mie scattering, Geophys. Res. Lett., 15, 1125-1128, 1988. Mishchenko, M.I., and L.D. Travis, 1994: T-matrix computations of light scattering by large spheroidal particles. Opt. Commun., 109, 16-21. Rauber, R.M., K.V. Beard, and B.M. Andrews: A mechanism for giant raindrop formation in warm, shallow convective clouds. J. Atmos. Sci., J8, 1791-1797, 1991. Srivastava, R.C., and D. Atlas: Growth, motion and concentration of precipitation particles in convective storms. J. Atmos. $ci., 26, 535-544, 1969. Szumowski, M. J., R. B. Rauber, H. T. Ochs III and K. V. Beard: The microphysical structure and evolution of Hawaiian rainband

cloud.

Part

II:

Aircraft

measurements

within

rainbands

containing high reflectivity cores. J. Atmos. $ci., 55, 208-226, 1998.

Yuter, S.E., and R.A. Houze Jr.: Three-dimensional kinematical and microphysical evolution of florida cumulonimbus. Part III: Vertical mass transport, mass divergence, and synthesis. Mon. Wea. Rev., 123, 1964-1982, 1995.

P. Kollias and B. A. Albrecht, Rosenstiel Sch. of Marine and Atmos. Sci. Division of Meteorology and Phys. Oceanography, University of Miami, 4600 Rickenbacker Cswy, Miami FL 33149-

1031. (e-mail: [email protected]) F. D. Marks, Jr. Hurricane ResearchDivision, NOAA/AOML, Miami

FL

(ReceivedMarch 6, 2001; revisedApril 16, 2001; acceptedApril 30, 2001.)