Deep moist atmospheric convection in a sub‐kilometer global simulation

Yoshiaki Miyamoto, Yoshiyuki Kajikawa, Ryuji Yoshida, Tsuyoshi Yamaura, Hisashi Yashiro, Hirofumi Tomita

(RIKEN AICS)

I. BackgroundII. Experimental SettingsIII. Methodology for 

detection of convectionIV. ResultsV. Conclusion

戦略分野３メソ課題研究会 March 7th, 2014 @Kobe

I. BACKGROUND

Byers and Braham (1949)

※deep moist convection := ”Convection”

Convection

Convection• Element of cloudy disturbances• Transport heat and moisture

Horizontal scale (x) 〜100 km

hard to explicitly solve in global models（x〜101 – 102 km）

←cumulus parameterization

2000〜Model development ＋enhancement of computer power => x 〜 100 km→clouds are explicitly solved in global models Still coarser or comparable to obs.

Regional model (Weismann et al., 1997) : Change around Δx <= 4 km

~100 km

~100 kmHeat & Moisture

Byers and Braham (1949)

Objective:

Reveal the dependence of the simulated convection on resolution in global model by describing the global statistical characteristics.

Experimental design

model NICAM (Tomita and Satoh 2004, Satoh et al. 2008)

Initial state 3‐day integrated results of 1‐step coarser resolution

SST NCEP analysis + nudging (Reynolds weekly SST)

land Model adjusted produced by 5 year run

Cloud physics NSW6 (Tomita 2008)

Boundary layer turbulence MYNN (Nakanishi and Niino 2004, Noda et al. 2008)

Surface flux Louis (1979)

Long and short‐wave radiation MSTRNX (Sekiguchi and Nakajima 2008)

Cumulus parameterization ‐‐

integration period（12 h）Δx

※72 h integration before producing initial fields

Computational Cost

• Nodes used: 20480 (~160000 cores)

• Wall-clock time: 53 h

• Sustained performance: 7~8 %

• Storage: 200 TB

integration period（12 h）Δx

※72 h integration before producing initial fields

Δ14.0 km

Δ3.5 km

Δ0.87 km

Δ7.0 km

Δ1.7 km

Snap shot of OLR12‐h integrations

II. METHOD OF DETECTING CONVECTION1. Detect “convective grids” by ISCCP table

2. Determine “convective core” grids

(Rossow and Schiffer 1999)

Step 1/2: Detect convective grids by ISCCP table

Δx = 14 km

Step 2/2: determine convective core grids

a) ISCCP convective grids (    )

d) Convective grids (    ) := where vertically aved w is larger than those at surrounding 8 grids

b) Find grids (    ) at which all the surrounding 8 grids  satisfy the ISCCP condition

c) Estimate horizontal gradient of vertical velocity averaged vertically in the troposphere

example（Δx=3.5 km）

w（troposphere mean）CI = 0.1 m s‐1w = 0.1 m s‐1

ISCCP convective gridConvection core grid

Δx = 3.5 km

Δx = 14 km

Δx = 7 km Δx = 1.7 km

w（troposphere mean）CI = 0.1 m s‐1w = 0.1 m s‐1

ISCCP convective gridConvection core grid

III. RESULTS Δ0.87 km

Composited structure of convection (GL13)

Convection core grid ※transform the coordinate into the cylindrical around the core gridmean of all the detected convectionsymmetric around the x axis

Composite of convection (vertical velocity)

Δx ≧ 3.5 km:– Convection is represented at 1

grid– Little dependence on resolution

Δx ≦ 1.7 km:– Convection is represented at

multiple grids– Intensify w/ resolution

Δ14.0 km Δ7.0 km Δ3.5 km Δ1.7 km Δ0.87 km

※transform the coordinate into the cylindrical around the core gridmean of all the detected convectionsymmetric around the x axisX axis is normalized by resolution

0 4 5 100

0.05

0.1

0.15

0.2

0.25

number of gird

frequ

ency

/tota

l num

ber

(b) distance between convection

14.07.03.51.70.8

0.8 1.7 3.5 7 14103

104

105

(a) number of convection

(km)

num

ber

Number and distance of convection

Δx ≧ 3.5 km:

– number: increase by factor of 4

Δx ≦ 1.7 km:

– number: decrease in increasing rate

– distance: 4 grids

– distance: 5 grids

Summary

Global simulation with a sub-kilometer resolution

Finding Convection features (structure, number, distance)

change between Δ3.5 km  Δ1.7 km- Δx should be 2.0〜3.0 km to resolve convection in global models

1.7 km3.5 km

Thank you very much for your attention!

Miyamoto, Y., Y. Kajikawa, R. Yoshida, T. Yamaura, H. Yashiro and H. Tomita, 2013: Deep moist atmospheric convection in a sub-kilometer global simulation, Geophysical Research Letters, 40, 4922-4927.

Special thanks to Drs. H. Miura, S. Iga, S. Nishizawa, M. Satoh, our colleagues, and two anonymous reviewers for fruitful discussions. The authors are grateful to  researchers and technical experts at RIKEN and FUJITSU for their kind help. The simulations were performed using the K computer at the RIKEN Advanced Institute for Computer Science. 

SUPPLEMENT

What is the general characteristics of convection?

• Resolution dependence Weismann et al. (1997): dependence of squall line

(Klemp and Wilhelmson (1979) cloud model)- Characteristics changes Δx less than and

equal to 4 km

Jorgensen&LeMone (1989)

コア

の直

径(km)

存在頻度 (積算)

1

15

Is there any threshold of resolution in realistic conditions?

Weismann et al. (1997)

w’’ w’u’ qr

http://callofduty.wikia.com/wiki/File:Cumulonimbus.jpg

Jorgensen and LeMone (1989): 50% of convection (core) has horizontal scale less than 1 km

• Isolated convection

– Element of amospheric cloudy disturbances

– Transport of heat/moisture

What is the general characteristics of convection?

Model（NICAM, Tomita & Satoh 2004, Satoh et al. 2008）

Global cloud-system resolving model

Icosahedral grid

nohydrostatic DC

explicit cloud expression:

21

Miura et al. (2007)

解析範囲：130—190E, ‐15—15N

0 1 2 3 4 5

0.8 1.7

3.5 7

14 30

a r e a ( 1 0 - 6 k m 2 )

Delta x (km)

area icp cutoff (GL08-12 t=201208250600)

0

0.02

0.04

0.06

0.08 0.1

0.8 1.7

3.5 7

14 30

a r e a ( 1 0 k m )

Delta x (km)

area con cutoff (GL08-12 t=201208250600)

0 10 20 30 40 50

0.8 1.7

3.5 7

14 30

a r e a ( 1 0 - 6 k m 2 )

Delta x (km)

area icp cutoff (GL08-12 t=201208250600)

面積

0 2 4 6 8 10

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g m - 2 s - 1 )

Delta x (km)

areal ave. (all) of z-aved mass flux (GL08-12 t=2012082506

allaveallave

解析範囲：130—190E, ‐15—15N

0 50

100

150

200

250

300

350

400

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g m - 2 s - 1 )

Delta x (km)

areal ave. (con) of z-aved mass flux (GL08-12 t=201208250

conaveconave

0 50

100

150

200

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g m - 2 s - 1 )

Delta x (km)

areal ave. (icp) of z-aved mass flux (GL08-12 t=2012082506

icpaveicpave

面積平均質量フラックス

0 1 2 3 4 5

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 3 m m h - 1 )

Delta x (km)

areal ave. (icp) of rain flux (GL08-13 t=201208250600)

icpaveicpave

0 2 4 6 8 10

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 3 m m h - 1 )

Delta x (km)

areal ave. (icp) of rain flux (GL08-13 t=201208250600)

conaveconave

0

0.05 0.1

0.15 0.2

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 m m h )

Delta x (km)

areal ave. (all) of rain flux (GL08-13 t=201208250600)

allaveallave

解析範囲：130—190E, ‐15—15N

面積平均降水量

100

120

140

160

180

200

220

240

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g s - 1 )

Delta x (km)

area-integrated (all) z-aved mass flux (GL08-13 t=201208250

alltotalltot

0 5 10 15 20

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g s - 1 )

Delta x (km)

rea-integrated (con) z-aved mass flux (GL08-13 t=20120825

contotcontot

100

120

140

160

180

200

220

240

0.8 1.7

3.5 7

14 30

m a s s f l u x ( x 1 0 3 k g s - 1 )

Delta x (km)

area-integrated (icp) z-aved mass flux (GL08-13 t=20120825

icptoticptot

解析範囲：130—190E, ‐15—15N

面積積分質量フラックス

0 1 2 3 4 5

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 3 m m h - 1 m 2 )

Delta x (km)

area-integrated (icp) of rain flux (GL08-13 t=20120825060

icptoticptot

0

0.05 0.1

0.15 0.2

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 m m h m )

Delta x (km)

area-integrated (con) of rain flux (GL08-13 t=20120825060

contotcontot

解析範囲：130—190E, ‐15—15N

0 1 2 3 4 5

0.8 1.7

3.5 7

14 30

p r e c i p i t a t i o n ( x 1 0 3 m m h - 1 m 2 )

Delta x (km)

area-integrated (all) of rain flux (GL08-13 t=20120825060

alltotalltot

面積積分降水量

200

210

220

230

240

250

0.8 1.7

3.5 7

14 30

O L R ( W m - 2 )

Delta x (km)

areal ave. (all) OLR (GL08-13 t=201208250600)

allall

100

110

120

130

140

150

0.8 1.7

3.5 7

14 30

O L R ( W m - 2 )

Delta x (km)

areal ave. (con) OLR (GL08-13 t=201208250600)

concon

100

110

120

130

140

150

0.8 1.7

3.5 7

14 30

O L R ( W m - 2 )

Delta x (km)

areal ave. (icp) OLR (GL08-13 t=201208250600)

icpicp

解析範囲：130—190E, ‐15—15N

面積平均OLR

東西風速・降水量・海面更正気圧の緯度分布

• 各解像度間に大きな差無し

• 解析値・観測値との顕著な差無し

東西風速 降水量

海面更正気圧

Skamarock (2004)

Effective resolution （~ 6-7Δx）：

それより小さい空間スケールの現象が、モデルで計算される運動エネルギースペクトルが-5/3則から外れる解像度

• 実現象で対流の存在する間隔 < 6-7Δx

– モデルでは現象と同様の間隔を再現できない

→Effective resolution以上で、且つ、実現象に最も近いスケール（=6-7Δx）に最頻値が出現

• 実現象で対流の存在する間隔 > 6-7Δx

– モデルで対流間の距離を解像可能