Azusa Takeishi

SENSITIVITY OF WRF MICROPHYSICS TO

AEROSOL CONCENTRATION

Eff ect of aerosols is one of the most uncertain factors in climate

Direct eff ect/ Indirect eff ects/ Semi-direct eff ect

Aerosol – Deep convectionEven though the areal coverage of deep convection is

small, its vertical motion is a part of local-/regional-/large-scale motion

It is important to understand the eff ect of aerosols on deep convective clouds and assess its climatic eff ects

1. INTRODUCTION

1/15

1. INTRODUCTIONHOW DOES DEEP CONVECTION CHANGE WITH AEROSOL

LOADING?

Rosenfeld et al. 2008

Cloud-aerosol

interaction

2/15

Weather Research and Forecasting (WRF) model Idealized simulation of quarter-circlular shear

supercellResolution: 2km (hori.), 500m (vert.)Maximum heating (3K) in the lower troposphere at the

center of the domain (200km*200km) -> trigger convection

3 hours of simulation with open boundary conditions

2. METHODS

-10 -5 0 5 10 15 20 25 30

-4

-3

-2

-1

0

1

2

3

4

750m

1250m

1750m

250m

4250m3250m 5250m6250m7250m

u [m/s]

v [m

/s]

3/15

Left: Skew-T log-P diagram of initial profile

Right: Hodograph of initial wind profile

Microphysics schemes: 2-moment schemes were chosen

2. METHODS

Morrison scheme (Morrison et al, 2005)

- 6 categories of hydrometeors (water vapor, cloud, rain, ice, snow, graupel/hail)- Fixed number concentration of

cloud droplets (default = 250 ) ↓

multiply the number of cloud droplets by 0.2, 0.5, 1 (control),

2, 3, 4, 5, and 6

Milbrandt-Yau scheme (Milbrandt and Yau, 2009)- 7 categories of hydrometeors (water vapor, cloud, rain, ice, snow, graupel, hail)- Calculate the number of CCN

using w, P, and T↓

multiply the number of CCN by 0.2, 0.5, 1 (control), 2, 3, 4, 5,

and 6Simulating 2 pristine cases + 1 control run + 5 polluted cases - How does the precipitation change? Why?

4/15

Accumulated precipitation slightly decreases with increased aerosols

But this decrease is quite small: ~5%

Moreover, this decreasing trend disappears when the horizontal resolution is set to 1km, when the heating is changed to 2K or 4K, and with graupel

3. RESULTS [MORRISON SCHEME]

0:000:10

0:200:30

0:400:50

1:001:10

1:201:30

1:401:50

2:002:10

2:202:30

2:402:50

3:000

0.5

1

1.5

2

2.5

3

3.5

4

50

125

250(control)

500

750

1000

1250

1500

Time

Dom

ain-

aver

aged

acc

umul

ated

pre

cipita

tion

[mm

]

5/15

Downdraft shows this decrease in precipitation intensity

Updraft does notNo interaction

between downdraft and updraft

3. RESULTS [MORRISON SCHEME]

0:000:10

0:200:30

0:400:50

1:001:10

1:201:30

1:401:50

2:002:10

2:202:30

2:402:50

3:000

5

10

15

20

25

30

35

40

45

50

55

60

-30

-27.5

-25

-22.5

-20

-17.5

-15

-12.5

-10

-7.5

-5

-2.5

0

50125250(con-trol)50075010001250150050125250(con-trol)

Time

Max

imum

ver

tical

vel

ocity

[m/s

]

Minim

um vertical velocity [m

/s]

No robust aerosol effect found in the simulations with Morrison scheme

6/15

Drastic decrease in accumulated precipitation

Almost 80% decrease in precipitation, when we compare the most pristine case and the most polluted case

Robust feature, even when the horizontal resolution or the perturbation is modified (2K or 4K)

3. RESULTS [MILBRANDT SCHEME]

0:000:10

0:200:30

0:400:50

1:001:10

1:201:30

1:401:50

2:002:10

2:202:30

2:402:50

3:000

0.5

1

1.5

2

2.5

3

3.5

4

0.2*control

0.5*control

control

2*control

3*control

4*control

5*control

6*control

Time

Dom

ain-

aver

aged

acc

umul

ated

pre

cipita

tion

[mm

]

7/15

Both downdraft and updraft show the invigoration in the storm system

Almost symmetric change implies the interaction between downdrafts and updrafts

3. RESULTS [MILBRANDT SCHEME]

0:000:10

0:200:30

0:400:50

1:001:10

1:201:30

1:401:50

2:002:10

2:202:30

2:402:50

3:000

5

10

15

20

25

30

35

40

45

50

55

60

-30

-27.5

-25

-22.5

-20

-17.5

-15

-12.5

-10

-7.5

-5

-2.5

0

0.2*control0.5*controlcontrol2*control

Time

Max

imum

ver

tical

vel

ocity

[m/s

]

Minim

um vertical velocity [m

/s]

Robust aerosol effect found in the simulations with Milbrandt-Yau schemeBut why?

8/15

WHICH HYDROMETEOR CONTRIBUTES?

Decrease in precipitation

RAIN

CLOUD

ICE SNOW GRAUPEL HAIL

Stop melting of (i) Hail (ii) Hail and graupel(iii) Hail, graupel, and snow

and see how frozen precipitation changes.

Melt

9/15

FROZEN PRECIPITATION

-> Decrease in graupel is the major reason for the decrease in total precipitation

0.2*control

control 3*control 6*control0

0.5

1

1.5

2

2.5

Control runs

liquidfrozen

CCN concentration

Accum

ula

ted p

recip

itati

on

0.2*control

control 3*control 6*control0

0.5

1

1.5

2

2.5

(i) No Melting of hail

liquidfrozen

CCN concentration

Accum

ula

ted p

recip

itati

on

0.2*control

control 3*control 6*control0

0.5

1

1.5

2

2.5

(ii) No melting of graupel & hail

liquidfrozen

CCN concentrationAccum

ula

ted p

recip

itati

on

0.2*control

control 3*control 6*control0

0.5

1

1.5

2

2.5

(iii) No melting of snow, graupel, & hail

liquidfrozen

CCN concentrationAccum

ula

ted p

recip

itati

on

10/15

GRAUPEL

11/15

CCN=0.2*control CCN=0.5*control CCN=control CCN=2*control

CCN=3*control CCN=4*control CCN=5*control CCN=6*control

[g/kg]

Domain-average graupel mixing ratio

WHY DOES GRAUPEL DECREASE?

Graupel forms when snow collects cloud droplets (riming) Only when riming rate > deposition rate Riming rate of snow is dependent on (i) Total number of snow (ii) Total number of cloud droplets(iii) Size spectrum of snow(iv) Size spectrum of cloud droplets

In pristine cases, cloud droplets are larger, soriming rate > deposition rate

-> production of graupel -> fast/effi cient precipitation

In polluted cases, cloud droplets are small, sodeposition rate > riming rate

-> production of snow -> slow/ineffi cient precipitation

12/15

SNOW

13/15

CCN=0.2*control CCN=0.5*control CCN=control CCN=2*control

CCN=3*control CCN=4*control CCN=5*control CCN=6*control

[g/kg]

Domain-average snow mixing ratio

Interaction between aerosols and deep convection is still not well understood

Simulations with Morrison scheme showed a low sensitivity to the aerosol concentration (4.9% reduction in precipitation) and this change is not robust

Simulations with Milbrandt scheme showed a high sensitivity to the aerosol concentration (79% reduction in precipitation) and this decrease is robust

Realistic simulations are necessary: types of aerosols, aerosol concentration, ideally 7 hydrometeors, inclusion of all processes, etc.

4. CONCLUSION

14/15

Thompson schemeChange humidity profi leAdd radiationConstrain simulations with observational dataUse of WRF-Chem

5. FUTURE WORK

15/15