mri-nhm(5km) - 名古屋大学a 1km-resolution mri-nhm (1km-nhm). the 1km-nhm w as one-w a y nested...

6
25 30 35 40 45 50 55(dBZe) LATITUDE 35 36 139 140 141 LONGITUDE Z=3.0 km KANTO PLAIN 400 400 400 400 400 400 400 400 400 400 400 400 800 800 800 1200 1200 1600 134E 135E 136E 137E 138E 139E 140E 141E 142E 143E 133E 144E 33N 34N 35N 36N 37N 38N 39N 40N 41N MRI-NHM(5km) MRI-NHM(1km) Kanto Plain TOKYO 500 500 500 500 500 500 500 500 500 500 500 500 500 1000 1000 1000 1000 1000 1000 1000 1000 1500 1500 1500 1500 1500 2000 2000 2500 RSM(20km) MRI-NHM(5km) 120E 130E 140E 150E 100E 110E 160E 20N 30N 40N 50N CHINA JAPAN Kanto Plain TAIWAN HONSHU

Upload: others

Post on 17-Mar-2020

29 views

Category:

Documents


0 download

TRANSCRIPT

Maintenance Mechanism of a 24 May 2000 Supercell StormDeveloping in Moist Environment over the Kanto Plain, Japan

�Shingo Shimizu1), Hiroshi Uyeda1), Qoosaku Moteki1), Takeshi Maesaka2),

Yoshimasa Takaya3), Kenji Akaeda3), Teruyuki Kato3), Masanori Yoshizaki3)

1)Hydrospheric Atmospheric Research Center, Nagoya University, 2)Graduate School of Science,

Hokkaido University, 3)Meteorological Research Institute, JMA

1. IntroductionA supercell storm developing in a moist environ-

ment was investigated in order to understand itsdevelopment and maintenance mechanism. A su-percell storm in a moist environment was observedby two C-band Doppler radars on 24 May 2000over the Kanto Plain, Japan. This storm was in-vestigated by using dual Doppler analyses and nu-merical simulationswith the MRI-NHM (Meteoro-logical Research Institute Non-Hydrostatic Model(Saito et al., 2001)). The purpose of this paper isto reveal the development and maintenance mech-anism of a supercell storm developing in a moistenvironment compared with that in a dry environ-ment as in the Great Plains.

25 30 35 40 45 50 55(dBZe)

LA

TIT

UD

E

35

36

139 140 141LONGITUDE

Z=3.0 km

KANTO PLAIN

Fig. 1. The locations of two Doppler radars(Narita and Haneda airports), and the radar re- ectivity �elds at a height of 3 km from 1106 JSTto 1300 JST on 24 May 2000. The gray scaleshows re ectivity at 1200 JST. The contour (over30 dBZe every 5 dBZe) shows re ectivity at 1106JST and 1300 JST. The observation area of theDoppler radars (120 km in radius) is shown in abox at the upper-right corner.

* Corresponding author address: Shingo Shimizu,Hydrospheric Atmospheric Research Center, Nagoya Uni-versity, Nagoya 464-8601, JapanEmail: [email protected]

400

400

400

400

400

400400

400

400

400

400

400

400

800

800

800

8001200

1200 200

1600

134E 135E 136E 137E 138E 139E 140E 141E 142E 143E

133E

144E

33N

34N

35N

36N

37N

38N

39N

40N

41N

MRI-NHM(5km)

MRI-NHM(1km)

Kanto Plain

TOKYO

500

500

500

500

500

500

500

500

500

500

500

500

500

1000

1000

1000

1000

1000

1000

1000

1000

1500

15001500

15001500

2000

2000

2500

RSM(20km)

MRI-NHM(5km)

120E 130E 140E 150E

100E

110E

160E

20N

30N

40N

50N

CHINAJAPAN

Kanto Plain

TAIWAN

HONSHU

Fig. 2. The domains for the numerical simu-lations. The 5km-NHM (dashed square in Fig. 2a)was one-way nested within the output of the RegionalSpectrum Model (on the top). The 1km-NHM (solidsquare in Fig. 2b) was one-way nested within theoutputs of the 5km-NHM (on the bottom).

2. Data and model descriptionThe locations of the radar sites (Narita and

Haneda airports) and the upper-air sounding atTateno are shown in Fig. 1. Two C-band Dopplerradars, covering a 120 km in radius over the KantoPlain (Fig. 1,upper right), recorded sets of vol-ume scan data regarding re ectivity and Dopplervelocity every six or seven minutes. Upper-airsoundings of wind, temperature, and humidity at900 JST were obtained from the Tateno Aerolog-ical Observatory of Japan Meteorological Agency(JMA). Surface temperature, wind speed and di-rection, and precipitation amounts were availableevery 10 minutes from surface stations of theAMeDAS (Automated Meteorological Data Ac-quisition System of JMA) at Koga and Sakura.The numerical simulations were conducted with

a 1km-resolution MRI-NHM (1km-NHM). The1km-NHM was one-way nested within the outputsof a 5km-NHM (Fig .2). The 5km-NHM was alsoone-way nested within the outputs of a 20 km res-olution Regional Spectrum Model.

3. Results of dual Doppler analysesA storm which developed in a pressure trough

was observed by two Doppler radars from 1106JST to 1400 JST on 24 May 2000 over the KantoPlain. The radar re ectivity �elds from 1106 JSTto 1300 JST, derived from Narita airport radar,are shown in Fig. 1.A remarkable hook echo, which is the main fea-

ture of a typical supercell storm developing overthe Great Plains (e.g., Browning and Foote, 1976), was observed southwest of the storm from 1144JST to 1212 JST (Fig. 3). A square in Fig. 3 in-dicates an analysis area (20km� 20km) for a ver-tical pro�le in Fig. 4. This analysis area chasedthe center of the updraft core at a height of 3 kmabove sea level (ASL) from 1106 JST to 1400 JSTin order to investigate the time series of the max-imum re ectivity pro�le in the analysis area.The time series of the maximumre ectivity pro-

�le in the traced square from 1106 JST to 1400JST is shown in Fig. 4. A strong echo over 60dBZe was observed at about 5 km ASL just be-fore 1200 JST. This echo descended to the groundfrom 1200 JST to 1300 JST. This remarkable de-velopment in re ectivity at 1200 JST was accom-panied by a strong updraft (over 8 m s�1) andstrong vertical vorticity (over 1:0� 10�2s�1) at aheight of 5 km ASL (not shown).Radar re ectivity at 1200 JST overlaid with

storm-relative wind vectors and updraft velocity(by contour) at a height of 3 km ASL is shownin Fig. 5. There was an updraft core (over 8 m

s�1) in the center of the hook echo. The dashedarrow near the hook indicates cyclonic ow withstrong vertical vorticity over 3:0 � 10�3s�1. Thevorticity center colocated with the updraft core atan altitude of 3 km to 5 km ASL.A vertical cross section along the A-A' line (Fig.

35 40 45 50 55 60(dBZe)

10

20

30

40

-50 -40 -30 -20DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

1206 JST 24 MAY 2000(c)

3.0 km

10

20

30

40

-50 -40 -30 -20DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

1200 JST 24 MAY 2000(b)

3.0 km

TATENO

(d)

10

20

30

40

-50 -40 -30 -20DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

1212 JST 24 MAY 2000

3.0 km

TATENO

30

40

50

60

-70 -60 -50 -40DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

3.0 km

1144 JST 24 MAY 2000(a)

Fig. 3. Hook-shaped re ectivity at a height of 3 kmASL from 1144 JST to 1212 JST. A square (20�20km) indicates an analysis area for a vertical pro�le inFig. 4.

TIME (JST)

HE

IGH

T(k

m)

51

51

54 54

54

57

57

57

57

60

60

60

60

60

63

63

63

63

63

30

Fig. 4. The time-height cross-section of the max-imum re ectivity in the analysis area in Fig. 3 from1106 JST to 1400 JST.

5) through the center of the updraft core is shownin Fig. 6. A strong echo, over 45 dBZe, was over-hanging ahead of the storm. In Fig. 6, the openarrow shows the location of a strong updraft core.In ow from the southeast turned into an updraftat the location of the open arrow.

These results of dual Doppler analyses revealthat this supercell storm had characteristics simi-lar to those of a typical supercell storm: an over-hanging structure and an updraft core colocatingwith the vorticity center.

3.0 km

30 35 40 45 50 55 60(dBZe)

10

20

30

40

50

60

-50 -40 -30 -20 -10 0DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

12m/s1200 JST 24 MAY 2000

stormm

otion

A

A’TATENO

Fig. 5. Radar re ectivity over 30 dBZe over-laid with storm-relative wind vectors and updraft ve-locity (solid contour every 2 m s�1, starting from2 m s�1 ) at a height of 3 km ASL. There was astrong updraft core in the center of a remarkable hookecho. The dashed arrow near the hook echo indi-cates cyclonic ow with strong vertical vorticity over3:0 � 10�3s�1. The open arrow indicates storm mo-tion south-eastward with 50 km h�1. A vertical crosssection along the A-A' line is shown in Fig. 6.

4. Results of 1km-NHMWe conducted numerical simulations with the

1km-NHM in order to clarify precipitation parti-cle in the strong overhanging echo region (over 45dBZe) and to investigate the ow structure aroundthe storm where Doppler radar could not detect anecho. Features of the storm observed by Dopplerradar were reproduced well by the 1km-NHM, asshown in Fig. 7. However, the simulated stormshifted 30 km north and developed one hour laterthan the observed storm. The mixing ratio of rain(Qr) overlaid with storm-relative wind at a heightof 3 km ASL at 1200 JST is shown in Fig. 7.The distribution of Qr over 2 g kg�1 was similarto the shape of the hook echo observed by radar,as shown in Fig. 5. The dashed arrow indicatescyclonic ow with strong vorticity existing at thesouthwest of the hook, as shown in Fig. 5.

A vertical cross section of Qr and Qg along theline A-A' through the center of vorticity is shownin Fig. 8. The contour indicates a mixing ratioof graupel (Qg), the gray scale indicates Qr. Theopen arrow shows the location of the updraft core.Corresponding to the overhanging echo structureobserved by radar, two cores of Qg over 8 g kg�1

are shown at both the front and rear of the updraft

0123456789

10

10 20 30 406(m/s)

30(m/s)

HE

IGH

T(k

m)

A’A

30 35 40 45 50 55 60(dBZe)

Fig. 6. A vertical cross section along the A-A' linein Fig. 5. A strong echo over 45 dBZe was overhang-ing ahead of the storm. The open arrow indicates thelocation of an updraft core.

core. The particles in the two cores of Qg seem tobe hail because the two Qg cores correspond tothe strong echo (over 55 dBZe).With regard to ow structure, in ow from the

southeast of the storm existed less than approxi-mately 2 km ASL ahead of the storm. As depictedin Fig. 8, the in ow turned into an updraft at thelocation of the open arrow. Behind the storm, thewesterly wind turned to a downdraft below 8 kmASL, as shown in Fig. 9. Corresponding to therear Qg core in Fig. 8, a downdraft core over -4 ms�1 induced by loading existed at a height of 6 kmASL in Fig. 9. Another downdraft core over -2 m

s�1 induced by evaporation cooling existed below3 km ASL in Fig. 9. This downdraft below 3 kmASL produced a surface out ow in the rainshaftcorresponding to Qr over 1 g kg�1, as shown inFig. 8. Vertical shear between the surface and analtitude of 3 km ASL (0ÆC level) was strong over6:8�10�3 s�1 as represented by the solid circle inFig. 7.These characteristics of the storm are very simi-

lar to those of a typical supercell storm. However,a considerable di�erence between this storm andthe typical supercell storm is that the downdraftbelow 3 km ASL was relatively weak (less than3 m s�1). The cause of the weak downdraft isdiscussed in the next section.5. DiscussionIn a previous study (Klemp, 1987), it was found

that two factors act to extend the longevity ofa supercell storm in a dry and strong shear en-vironment. First, a strong storm-relative low-level in ow prevents the gust front from movingout of the storm (Wilhelmson and Klemp, 1978b;Thorpe and Miller, 1978; Weisman and Klemp,1982). Second, lifting pressure gradients inducean updraft growth almost directly above the sur-face gust front (Schlesinger, 1980; Rotunno andKlemp, 1982).A �rst necessary condition for the development

of a steady storm such as a supercell storm is that

0 1 2 3

20.0m/s1300 JST 24 MAY 2000

DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

0-10-20-30-40

30

40

50

60

70

stormm

otion

z*=2.88

t = 3hr

A

A’

0 10km

Fig. 7. The simulated supercell storm at 1200JST 24 May 2000. The mixing ratio of rain (grayscale) overlaid with storm-relative wind vectors. Thedashed arrow indicates strong vertical vorticity, asshown in Fig. 5. A vertical cross section along the A-A' line is shown in Fig. 8 and Fig. 9. Vertical shearbetween the surface and an altitude of 3 km ASL wasover 6:8 � 10�3s�1 as represented by the solid circlein Fig. 7.

the storm-relative in ow be strong enough to keepthe out ow from propagating away from the up-draft. If the in ow is weak, the gust front maymove ahead of the storm, and the supply of warmair to the updraft may be cut o� (Weisman andKlemp, 1982). This theory, developed during pre-vious studies is consistently supported in a dryenvironment. Because the downdraft in a dry en-vironment is often very strong due to evaporationcooling, the strength of in ow is the most impor-tant factor in a steady storm. However, in a moistenvironment, the downdraft is not always strong.Therefore, the balance of the in ow and out owstrength is the most important factor.

In the present case, the upper-air sounding atTateno a few hours before the development of thestorm under discussion is shown in Fig. 10. A drylayer existed from 650 hPa to 500 hPa. A meltinglayer existed at 680 hPa (about 3 km ASL). Be-low the melting layer, there was a relatively moistlayer (relative humidity over 70%). Because ofthis moist layer, the production of negative buoy-ancy by evaporation cooling was limited. Thisexplains why the temperature had fallen only 3degrees when the core of the storm passed overthe AMeDAS surface station at Koga (Fig. 11).Because negative buoyancy by evaporation cool-

2

4

6

8

10

12

2

4

6

8

10

12

Qg (g/kg)

10 20 30(km)0

r (t = 3h 1200 JST 24 MAY 2000)

A

HE

IGH

T(k

m)

00

0 0

1

1

2

2

33

456

7

A’3.0m/s

30.0m/sQr

( g/kg )0 1 2 3

Fig. 8. A vertical cross section along the A-A' linein Fig. 7 of Qg and Qr. The gray scale indicates Qr,and the contour (every 1 g kg�1) indicates Qg. Theopen arrow indicates the location of an updraft core.

10

12

-2

-2

0

0

0

0

5

5 1015

20

2

4

6

8

2

4

6

8

10

12

W(m/s)

10 20 30(km)0

t = 3hr (1200 JST 24 MAY 2000)

A

HE

IGH

T(k

m)

A’3.0m/s

30.0m/s0 2 4 6 8

Qg( g/kg )

Fig. 9. A vertical cross section along the A-A' linein Fig. 7 of Qg and vertical wind velocity. The grayscale indicates Qg, and the contour indicates verticalwind velocity. The contour interval of an updraft is5 m s�1 (solid contours), and the dashed contoursrepresent a downdraft every -2 m s�1.

ing was small, the downdraft below the meltinglayer was relatively weak. The surface out ow pro-duced by the weak downdraft balanced the storm-relative in ow at the location indicated by theopen arrow in Fig. 9.Regarding a second necessary condition for the

development, Weisman and Klemp (1982) re-vealed that mid-level vertical vorticity develops ina strong shear environment. Rotunno and Klemp(1982) found that mid-level vertical vorticity isderived from the horizontal vorticity of the en-vironmental shear and that the vertical vorticityinduces lifting pressure gradients, which lower thepressure and thereby induce updraft growth.In the present case, the vertical shear was strong

(6:8�10�3 s�1) and the vertical vorticity was over1:0�10�2s�1 at a height of 5 km ASL. Therefore,the lifting pressure gradients maintained a strongupdraft.Before 1230 JST, the weak out ow balanced the

storm-relative in ow with the horizontal vortic-ity of the environmental shear above the surface

300

400

500

600

700

800

900

1000

-40 -30 -20 -10 0 10 20 30

0.4 0.6 0.8 1 2 3 4 5 10 15 20 25 30

240

250

260

270

280

290

300

310

320

330

340

350

360

250

260

270

280

290

300

OC

hPa 09 JST 24 May TATENO

Fig. 10. The upper-air sounding at Tateno at 900JST. Pro�les of temperature (thick solid line), dew-point temperature (thin solid line), and wind speedand direction (on the right).

gust front. The in ow turned into an updraftwith strong vertical vorticity of over 1:0�10�2s�1.This strong vertical vorticity induced lifting pres-sure gradients and maintained the updraft. There-fore before 1230 JST, the storm had characteris-tics similar to those of a typical supercell storm(except for weak downdraft).After 1230 JST, the storm could not maintain

a steady updraft. Negative buoyancy above themelting layer gradually got stronger as the mass ofhail and graupel increased. The downdraft abovethe melting layer brought dry air between 650 hPaand 500 hPa to the relatively moist layer below 3km ASL. As the layer below 3 km ASL was drier,evaporation cooling worked more e�ectively. Thetemperature had fallen 9 degrees when the coreof the storm passed over the AMeDAS station atSakura, as shown in Fig. 11. Therefore, the out- ow was stronger because the environment wasdrier below the melting layer, and the gust frontspread (not shown), and the steady updraft couldnot be maintained. At 1306 JST, a new cell ap-peared to the south of the storm, as shown in Fig.11.In a previous numerical study (Weisman and

Klemp, 1982), convective available potential en-ergy (CAPE) and vertical shear as environmentalconditions controlled the strength of the down-draft. Greater instability, as measured by CAPE,was found to increase storm downdraft strength.

1.0 km

-100

-50

0

50

100

DISTANCE EAST OF NARITA (km)

DIS

TA

NC

EN

OR

TH

OF

NA

RIT

A(k

m)

1200 JST 24 MAY 2000

-100 -50 0 50 100

HANEDA

NARITA

1118

1130

1144

1200

1212

1224

1242

1306

KOGA

1100 1130 1200 1230 (JST)

(mm)

16

12

8

4

0

28

20

24

16

12

( C)

1200 1230 1230 1330 (JST)

(mm)

16

12

8

4

0

28

20

24

16

12

( C)

SAKURA

Fig. 11. Re ectivity over 45 dBZe at a heightof 1 km ASL from 1118 JST to 1306 JST and tracesof surface temperature (solid line), wind speed anddirection, and precipitation amounts at Sakura andKoga (solid square).

This is because a stronger updraft can suspendlarger rain amounts that create greater evapora-tion cooling and precipitation loading, resultingin a stronger downdraft (Weisman and Klemp,1982; Weisman and Klemp, 1984 ). In addi-tion, weaker unidirectional wind shear was foundto produce stronger downdraft (Wilhelmson andKlemp, 1978a). However, studies that have in-vestigated the a�ect of mid-tropospheric drynesswere very few (Gilmore and Wicker, 1998). TheeÆciency of evaporation cooling is controlled bythe dryness below the melting layer.

In the present case, when the layer below themelting layer was moist, before 1230 JST, evap-oration cooling did not work e�ectively, and theout ow was so weak that it could balance thestorm-relative in ow. When the layer below themelting layer was drier, after 1230 JST, evapo-ration cooling worked e�ectively and the out owwas so strengthened that it propagated away fromthe updraft. Thus, the strength of the out owis controlled by eÆciency of evaporation cooling.Therefore, in investigating the balance of the in- ow and out ow strength, the pro�le of relativehumidity is an important environmental factor toconsider.

6. ConclusionsA conceptual model of the 24 May 2000 su-

percell storm observed over the Kanto Plain isshown in Fig. 9a. The structure of this storm(Fig. 9a) was similar to that of a typical supercellstorm (Fig. 9b), with an overhang structure, an

updraft core with vertical vorticity near the alti-tude of the melting layer, and strong vertical shearin the lower layer. However, the downdraft be-hind the storm and the out ow were very weak inthe supercell storm of 24 May 2000 because of themoist environment below the melting layer before1230 JST. This weak out ow balanced the storm-relative in ow with the horizontal vorticity of theenvironmental vertical shear. The in ow turnedinto an updraft with mid-level vertical vorticity.Lifting pressure gradients induced by strong mid-level vertical vorticity maintained the updraft coreahead of the gust front.A supercell storm on 24 May 2000 had similar-

ities to that in a dry environment over the GreatPlains. Even in a moist environment, the balanceof the out ow and the in ow was an importantfactor for the development and maintenance of thesupercell storm. Additional research is needed toinvestigate the strength of out ow and the in u-ence of the mid-level humidity in a moist environ-ment.

Z

WEAK OUTFLOW INFLOW

IN MOIST ENVIRONMENT

STRONGVERTICAL SHEAR

0 LEVEL

TTd

WEAKDOWNDRAFT

45 dBZe

MOIST AIR

GRAUPEL AND HAILCORE

(a)

Z

STRONGDOWNDRAFT

STRONG OUTFLOW INFLOW

IN DRY ENVIRONMENT

0 LEVEL

TTd

STRONGVERTICAL SHEAR

(b)

DRY AIR

Fig. 12. Conceptual models of (a) the 24 May 2000supercell storm developing in a moist environmentand (b) a typical supercell storm developing in a dryenvironment over the Great Plains.

Acknowledgments. The authors would like toexpress their special thanks to Haneda and Naritaairports for providing radar data. The numeri-cal simulations were performed on the Hitachi SR-8000 at the Hokkaido University Computing Cen-ter.

REFERENCES

Browning, K. A. and G. B. Foote, 1976: Air owand hail growth in supercell storms and someimplications for hail suppression. Quart. J. R.Soc., 102, 499{533.

Gilmore, M. S. and L. J. Wicker, 1998: The in u-ence of midtropospheric dryness on supercellmorphology and evolution. Mon. Wea. Rev.,126, 943{958.

Klemp, J. B., 1987: Dynamic of tornadic thun-derstorms. Annu. Rev. Fluid. Mech., 19, 369{402.

Rotunno, R. and J. B. Klemp, 1982: The in u-ence of the shear-induced pressure gradient onthunderstorm motion.Mon. Wea. Rev., 110,136{151.

Saito, K., T. Kato, H. Eito and C. Muroi, 2001:Documentation of meteorological research in-stitute numerical prediction division uni�ednonhydrostatic model. Technical report, Me-terological reserch institute,Japan.

Schlesinger, R. E., 1980: A three-dimensional nu-merical model of an isolated deep thunder-storm .Part II: Dynamics of updraft splittingand mesovortex couplet evolution. J. Atoms.Sci., 37, 395{420.

Thorpe, A. J. and M. J. Miller, 1978: Numeri-cal simulations showing the role of the down-draught in cumulonimbus motion and split-ting. Quart. J. R. Met. Soc., 104, 873{893.

Weisman, M. L. and J. B. Klemp, 1982: The de-pendence of numerically simulated convectivestorms on vertical wind shear and buoyancy.Mon. Wea. Rev., 110, 504{520.

Weisman,M. L. and J. B. Klemp, 1984: The struc-ture and classi�cation of numerically simu-lated convective storms in directionally vary-ing wind shears. Mon. Wea. Rev., 112, 2479{2499.

Wilhelmson, R. B. and J. B. Klemp, 1978a: A nu-merical study of storm splitting that leads tolong-lived storms. J. Atmos. Sci., 35, 1974{1986.

Wilhelmson, R. B. and J. B. Klemp, 1978b:A three-dimensional numerical simulation ofsplitting that leads to long-lived storms. J.Atmos. Sci., 35, 1037{1063.