Description of the MésoNH-AROME microphysical scheme and ... · Outlook • Introduction • Common assumptions made in microphysical schemes • The current scheme developed and

Description ofDescription of the MésoNH-AROME the MésoNH-AROME microphysical scheme and its microphysical scheme and its

evaluation by remote sensing toolsevaluation by remote sensing tools

Jean-Pierre PINTYJean-Pierre PINTY, Evelyne RICHARD,, Evelyne RICHARD,

Jean-Pierre CHABOUREAU and Frank LASCAUX Jean-Pierre CHABOUREAU and Frank LASCAUX (LA/OMP),(LA/OMP),

Yann Yann SEITYSEITY (Météo-France) (Météo-France)

11stst AROME training course, Poiana-Brasov, Romania, 21-25 November 2005 AROME training course, Poiana-Brasov, Romania, 21-25 November 2005

OutlookOutlook• Introduction• Common assumptions made in microphysical schemes• The current scheme developed and used in MésoNH

• Warm microphysics• Mixed-phase microphysics

• Examples and tools to evaluate the scheme• Conclusion and perspective

The explicit simulation of the Water Cycle is a major issue of many mesoscale studies and model

applications. Microphysical schemes are the key parametrisation to

follow the evolution of the condensed phases of water at high resolution in the atmosphere.

1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005

Explicit cloud modelingExplicit cloud modelingThere are many cloud types to simulate ! Fogs, Extended cloud sheets, Cirrus, Cumulus clouds

Wide span of particle size: ~ 4 decades (µm cm) of particle habit: many ice particle types to consider

Microphysical fields describe discontinuous and sparse objects. Clouds have sharp boundaries ( microphysical fields are ≥0) which do not fit a regular grid system ( cloud fraction)

Many interactions involving clouds and precipitation:Dynamics, Radiation, Surface, Aerosols, Chemistry, Electricity What is the level of complexity really needed for a

microphysical scheme to cover all these applications ? 1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005

Choice of the microphysical variablesChoice of the microphysical variables

• State variables: Mixing ratios (mass of water / mass of dry air)• Assumptions about number concentrations: parameterization or

Aerosol Physics• Number of cloud and precipitation variables

2 water variables for warm clouds: Cloud water (droplets), Rain water (drops)

2, 3, 4, 5 ice variables for cold clouds: Cloud ice (pristine crystals), Snow (large crystals), Aggregates (assemblage of crystals), Graupel (rimed crystals), Hail (large heavily rimed crystals)

General case Mixed-phase microphysics with liquid & ice phases1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005

Common features of many microphysical Common features of many microphysical schemesschemes

• Limited number of water species (~ 6): 1 vap. + 2 liq. + 3 ice

• Size distribution:Mathematical (parametric) distribution law [0 < D < ∞]

• Mass-size and Fall speed-size relationships:Power law analytical integration

• Uncertainties about the representation of some processes and about the value of some bulk coefficients:

collision-sticking efficiencies of collection kernels autoconversion processes (onset of precipitating particles) adjustment to saturation (pure ice-phase clouds)


Description of the microphysical scheme Description of the microphysical scheme of MésoNHof MésoNH

• Size distribution (n(D)): Generalized Gamma law

N is the total concentration g(D) is the normalized distribution lawλ is deduced from the mixing ratio (α, ν) are free shape parameters (Marshall-Palmer law: α=ν=1)

• Very useful p-moment formula

dD)D-(λexpDλν)Γ

αg(D)dDn(D)dD )(1

NN

λ

1

Γ(ν)

)αpΓ(νNdDn(D)DM(p)

p0

p


Particle size distributions in Particle size distributions in microphysical schemesmicrophysical schemes


Unnormalized Distribution Laws

Exponential decay : rain, snow, graupel, hail

Modal distribution : droplets, cloud ice

• Mass-Size relationship: m=aDb

• Fall speed-Size relationship: v=cDd . (ρ00/ρa)0.4

Category

Parameters

c

d

a

b

0.640.660.271.000.82

2071245.18008423.2e7speed

mass3

524

Cloud water

3.02.81.92.53

47019.60.020.82524

HailGraupelSnowflakeAggregate

Cloud

iceRain water

Microphysical characteristics (1)Microphysical characteristics (1)


Foote-DuToit correction

The a, b, c and d coefficients (MKS units) are adjusted from ground or in situ measurements

Fall speed of the hydrometeorsFall speed of the hydrometeors


Terminal velocity at P=700 hPa

rain

hail

graupel

snow

The total concentration of precipitating particles: rain, snow, graupel, is given by N=Cλx instead of fixed N0 value (N0=N.λ) as it is

often the case in classical Marshall-Palmer schemes

Microphysical characteristics (2)Microphysical characteristics (2)


N0

D

N(D)

decrease

dDD)-(λexpNn(D)dD

:Palmer-Marshall

)(0

Fixed value

1 23

Lin scale

Log

sca

le

N(D)

N0

D

decrease

dDD)-(λexp λn(D)dD

:lExponentia-- Gamma

)( N

1 2 3

Parameterized

Lin scale

Log

sca

le

0°C

Autoconversion

0°C

RimingAggregation

Collection Collection

Deposition

Freezing

Nucleation

Melting

gsrSedimentation Evaporation

Mixed-phase cloudsMixed-phase clouds


h

Additional part to include hail explicitly

Microphysical Scheme diagramMicrophysical Scheme diagram


Vapor r_vDroplets r_cRaindrops r_rPristine ice r_iSnow/Agg. r_sGraupel r_gHail r_h

Keys

Vapor r_vDroplets r_cRaindrops r_rPristine ice r_iSnow/Agg. r_sGraupel r_gHail r_h

Keys

Microphysical SchemesMicrophysical Schemes

1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005c

• Warm clouds: no ice phase « Kessler » scheme (1969)

• To simulate the microphysical processes inside pure warm clouds (stratus decks, shallow cumuli)

• To simulate the warm processes occuring in the low levels of deep convective clouds

• Mixed-phase clouds: additional ice phase with water-ice and ice-ice interactions full « Méso-NH » scheme (... 1998)

• To simulate ice clouds (cirrus)• To simulate heavily precipitating deep convective clouds

Aerosols

Cloud dropletsRaindrops

Activation

Autoconversion

Accretion

SedimentationEvaporation

Condensation

Water cycle in warm cloudsWater cycle in warm clouds

NOT CONSID

ERED

in the p

resen

t sch

eme

1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005c

Warm microphysics: AutoconversionWarm microphysics: Autoconversion


Formation of rain: some cloud droplets become raindrops collection growth of a few big droplets role of the turbulence, width of the droplet size distribution subject of very active research to include Nc, Dc, σc, εturb

colliding droplets precipitating drop

… a very crude but efficient parametrization

mkgc

s

with

ccMax

crit

critc

AUT

r

AUT

33

13

aa

.105.0rρa

10K

rρarρa,0.0Kt

)r(ρ

t

)r(ρ

(Time scale)-1 Threshold

Warm microphysics: AccretionWarm microphysics: Accretion

Growth of raindrops: raindrops collect droplets during their fall


1E

)2d(MEr4

Dd)D(nt

)D(

t

)r(ρ

t

)r(ρ

)DK(ρr

Dd)D(nD)D,DK(ρ6t

D

acc

acc

0

aa

a0

3cw

with

rm

m

rcrrrACC

c

ACC

r

ACC

rc

cccrcr

ACC

… the parametrization is based upon the collection kernel

Collection kernel: geometrical sweep-out

EDvD4

πD,DKDDbut

EEDvDvDD4

πD,DK

accr2rrcrc

stickcollrc2r

2crc

cDrd (ρ00/ρa)0.4

Warm microphysics: EvaporationWarm microphysics: Evaporation

Decay of raindrops: raindrops evaporate below cloud


v

Dk

EVA

v(D)DRe0.22,FwithReF1f

)0Shere()/rr(rS

P)(T,(T)e

TR

T(T)R

LP)(T,A

with

P)(T,A(D)fDS4tm(D)

wv,vsvsvwv,

vsw

v2

va

2v

w

wwv,

0

a Dd)D(nt

)D(m

t

)r(ρrrr

r

EVA

r

EVA

… a rather accurate parametrization

Evaporation rate: function of the undersaturationSv,w and of a ventilation coefficient

RH<100%or

Sv,w<0

Df

Drop Reynolds number

Warm microphysics: SedimentationWarm microphysics: Sedimentation

Fallout of raindrops: vertical flux of raindrops relative to the air


Sedimentation rate from sedimentation flux

0

dba00

0.4

0

Dd)D(nD )ρρ(caSed_flux

Dd)D(n)D(m)D(vSed_flux

rrrrrr

rr

rrrrr

Sed_fluxt

)r(ρa

zr

SED

… no true size sorting effect (size shift toward large Dr)

Automatic time-splitting for numerical stability




WA

RM

mic

roph

ysic

s

COLD microphysics

Ice nucleation processesIce nucleation processes(a very simplified treatment !)(a very simplified treatment !)


Homogeneous nucleation: Spontaneous freezing of the cloud droplets when -35°C<T<-44°CandRainGraupel, T<-35°C

tVT)()VdtT)(exp(1

JJP HOM

tt

tHOM

Basics: Freezing probability P of a droplet of volume V

Integration over the droplet size distribution

M(3)

M(6)rρ

6,

rρ

t

)r(ρa

aacHOM

ci

HON

TJt

Min

Heterogeneous nucleation: Based on ice crystals growingon ice nuclei (IN)

… the treatment is more complex when Ni is a prognostic variable

.................................SSexp

Texprrrr

C5T.......,.........

C5TC2,...

22202

11

101

2

1

iNUNU

vsivswvsivNUNU

NU

NUIN

NN

NN

N

NN

NNMaxt

ttiIN

NUi

HEN

,0m

t

)r(ρ 0a

Basics: Meyers et al. (1992) as in Ferrier (1994)

Bergeron-Findeisen effectBergeron-Findeisen effectGrowth of pristine ice crystal atthe expense of cloud droplets

… maximum effect at T=-12°C


)()( TeTe swsi

droplet

_rateDepositionn_rateEvaporatio

ice

water vapor

Autoconversion of pristine ice crystalsAutoconversion of pristine ice crystals

Small pristine ice crystals do not grow by collection !

… pristine crystals (D<150µm) poorly rime or grow by agregation


Hexagonal plates: (1) D=160 µm

Stellar crystals: (1) D=200 µm

Hexagonal columns: (1) L=70 µm

Computed dropcollection efficienciesafter Wang & Ji (1992)

Autoconversion of pristine ice crystalsAutoconversion of pristine ice crystals

Formation of snow: Some pristine ice crystals grow by deposition to get bigger than 100 ~ 200 µm


+ vapor

Ice crystal Snow particles

… modified threshold to enhance cirrus sheet precipitation

mkgi

s

with

iiMax

crit

criti

AUT

s

AUT

33

1t

3

aa

.1002.0rρa

)TT(025.0exp10K

rρarρa,0.0Kt

)r(ρ

t

)r(ρ

(Time scale)-1

Threshold

Dry growth of snow: snowflakes collect ice crystals when falling


T-T0.05exp25.0E

)2d(MEr4

Dd)D(nt

)D(

t

)r(ρ

t

)r(ρ

)DK(ρr

Dd)D(nDa)D,DK(tD

tagg

agg

0

aa

a0

ibi

i

with

m

m

sissss

AGG

i

AGG

s

AGG

si

iiisis

ACC

… a very crude but efficient parametrization

Collection kernel: geometrical sweep-out

EDvD4

πD,DKDDbut

EEDvDvDD4

πD,DK

aggs2ssisi

stickcollsi2s

2isi

cDsd (ρ00/ρa)0.4

Aggregation of pristine ice Aggregation of pristine ice snow snow

Growth by depositionGrowth by deposition

Snow and graupel grow (inside clouds) or decay (outside clouds):


νD

CC

C

v

SUBDEP

v(D)DRe;Sc,ReScwhere

ffff

D

with

P)(T,A(D)fDS4tm(D)

2131

2210

1

iiv,/

0

,,,,

/

,a

/Dd)D(n

t

)D(m

t

)r(ρgsgsgs

gs

SUBDEP

gs

SUBDEP

… a fully analytical parametrization

A generalization of the raindrop evaporation

Schmidt & Reynolds numbers

Snow Graupel

water vapor

Cap

acita

nce

Ventil

ation

Generalities about collection processesGeneralities about collection processes

Collection processes: based oncontinuous collection kernels(geometrical swept-out concept)

E)D(v)D(vDD4

π)D,DK( xyyyxxyx

2

yx )(


t

)r(ρ

t

)r(ρ

t

)r(ρ

Dd)D(nDd)D(n)D(m)D,DK(t

)r(ρ

Dd)D(nDd)D(n)D(m)D,DK(t

)r(ρ

aaa

0

yx

0

a

xxx

0

yyyyyyx

0

xa

][

][

y

COLL

x

COLL

z

COLL

yyyxxxxxy

COLL

COLL

Collection+Conversion of precipitating particles: tabulated integrals

+

Snow Rain Graupel

… but the treatment is even more complicated in the case of partial conversion (function of particle size)

Snow riming processSnow riming process

Conversion threshold: Ds > Dslim = 7 mm as in Farley et al. (1989)

Splitted snow size distribution


or

1E and EDvD4

πDKD,DKwith

D)dD(Dn)dD(D)n(D)mD,K(D

t

rρ

D)dD(Dn)dD(D)n(D)mD,K(D

t

rρ

rimrims2sssc

ss0 cccccscga

0 ss0 cccccscsa

lim

s

lim

s

s

RIM

sRIM

… a continuous conversion of snow by riming

+

Snow Droplet Snow Snow Droplet Graupel

+

Conversion when Dr > Drlim : based on the density of a mixture of a

snowflake and a raindrop and leading to

Snow collection processSnow collection process


or

Drvr~Dsvs as integrals normalized latedwith tabu

)dD(DnD

)dD(D)n(D)mD,K(Dt

rρ

)dD(DnD

)dD(D)n(D)mD,K(Dt

rρ

0 ssrrrrrsrga

0 ss0 rrrrrsrsa

lim

r

lim

r

s

COL

sCOL

… large collected raindrops convert snow into graupel

+

Snow Rain Snow Snow Rain Graupel

+

ρρρsgsr

0.5 DfD slimr

Raindrop freezing: falling raindrops capture pristine ice crystals to form graupel particles

Raindrop contact freezingRaindrop contact freezing


… collection and simultaneous conversion

1E and EDvD4

πDKD,DKwith

t

rρ

t

rρ

t

rρ

)dD(Dn)dD(D)n(D)mD,K(Dt

rρ

)dD(Dn)dD(D)n(D)mD,K(Dt

rρ

cfrcfrr2rrri

iaraga

0 rr0 iiiiiriia

0 ii0 rrrrrrira

CFRCFRCFR

rCFR

iCFR

Graupel growth: other effectsGraupel growth: other effects

takenberategrowthminimumThe must


Dry growth ( Graupel)(Sum of collection rates)

Graupel with Tsurf < 0°C

Wet growth ( Hail) (Heat balance equation)

liquid film

Shedded drops

Graupel with Tsurf = 0°C

Wet growth and water shedding of the Wet growth and water shedding of the graupels graupels initiation of hailstones initiation of hailstones


Heat balance equation: Musil (1970), Nelson (1989)

sircg t

m

t

m

t

m

Mass budget:

Wet growth rate of the graupel

vtvs

v

vvtaggtsi,

sirctm e)(Te

TR

P)(T,LT)(T)(Tfπ4T)(Tc

t

m

t

m)(TL

DkC

Heat due tocollected

water

Heat due tocollected

ice

dissipated Heatby the graupel

… water shedding rate is computed raindrops

A simple A simple wayway to initiate hail ? to initiate hail ? (…this (…this rrhh extension is still under test) extension is still under test)


A realistic « graupel hail » conversion rate can be parameterized from the theoretical

Dry growth rate of the graupel Wet growth rate of the graupel

WetDry

Dry*

t

rt

r g

hg

h

Graupel tendencybefore conversion

Weighting factorwith Dry/Wet rates

… to be improved to avoid a continuous conversion into hail

Hail rate

Melting of the ice particles (T>0°C)Melting of the ice particles (T>0°C)

Pristine ice crystals: instantaneously melted into cloud droplets


Graupel particles raindrops: heat budget equation taking into account the fall and the collection capability of the particles

Snow/aggregates graupel raindrops: heat budget equation andconversion into graupel

T=0°C

Pristine crystals

mixed-phase~ graupel

Graupel Snow/Aggregate

Sedimentation of ice particlesSedimentation of ice particles

Same as for the raindrops, the pristine ice crystalsare weakly precipitating (McFarqhar & Heymsfield)


0

,,,,,,,,,, Dd)D(n)D(m)D(vSed_flux gsigsigsigsigsi

Sed_fluxt

)r(ρ ,,a

zgsi

SED

Automatic time-splitting for numerical stability




WA

RM

mic

roph

ysic

s CND/EVA DEP/SUB

COLD microphysics

Implicit adjustment to saturationImplicit adjustment to saturation

Saturation mixing ratio: using a barycentric formula ‘no’ supersaturation rr

(T)rr(T)rrr *

i*c

isatv,

*i

wsatv,

*ciw

satv,

Cond/Evap+Dep/Subl rates: Variational adjustment

rr

rrΔr

rr

rrΔr3

(T)rrr2

0F(T)T1

r(T)rrr

r(T)Lrr

r(T)L)TT(CF(T)

*i

*c

*i

vi*i

*c

*c

vc

iwsatv,

*vv

*v

iwsatv,*

i*c

*i

vi*i

*c

*c

vw*

ph

andGet

Compute

assuchFind

and andvapor

crystal droplet crystal droplet


Temporal steppingTemporal stepping


• Sedimentation processes• Warm processes• Slow mixed-phase processes (nucleation, auto-

conversion, aggregation, deposition, etc.)• Fast mixed-phase processes (riming, collections,

melting, etc.)

• Saturation ajustment is the last integrated process

Microphysical Scheme of MésoNHMicrophysical Scheme of MésoNH

• Warm processes (Kessler scheme)• Light and Heavy riming rates of the snowflakes by

the cloud droplets and by the rain drops and the conversions into graupel particles

• Wet/Dry growth modes of the graupels• Melted particles are considered as graupels• Possible extension to simulate a « hail » phase

• Sedimentation (1st order upstream scheme)• Processes are integrated one-by-one after carefully

checking the availability of the sinking categories• On-line budgets

Summary


• Onset of precipitating drops and precipitating ice: Simulation of extended cloud sheets of moderate lifetime

(Sc, Ci) ?

• Physics of the ice phase: Collection efficiencies ? Supercooling of water ? Ice nucleation ?

Which ice crystal characteristics to choose ?

Present uncertainties …Present uncertainties …

snow graupel

needle

column

plate bullet

dendriterosette

(Pruppacher and Klett, 1997)

10-100 m

1-10 mm


W (m/s)

rc (kg/kg)

U (m/s)

rr (kg/kg)

WarmWarm 2D Orographic Cloud 2D Orographic Cloud

•1st AROME training course, Poiana-Brasov, Romania, 21-25 November 2005

Mixed-phaseMixed-phase 2D Orographic Cloud 2D Orographic CloudW (m/s)

rc + ri (kg/kg)

U (m/s)

rr + rs + rg (kg/kg)


2D Tropical Squall Line2D Tropical Squall Line

Cloud water

Rain water

Snow

Graupel

Cloud iceU

W

Dynamics:Flat terrain+R/S


How does the flow over complex terrain modify the growth mechanisms of precipitation particles?

3D Orographic precipitation (MAP)3D Orographic precipitation (MAP)

32km : 150x1508km : 145x1452km : 150x150over 51 levels

8 km

2 km

Monte Lema

S Pol

Ronsard

ECMWF32 km

3 Dopplerradars ( )


(Keil et Cardinali, 2003)

IOP8 (F<1)

IOP2a(F>1)

(F: Froude number)



IOP2a: 09/17/9918 UTCUnstable Flow over

IOP8: 10/20/9912 UTCStableBlocked flow

SnowGraupel

Hail

Cloud Rain

IceIOP2a

IOP2a ( Strong convection)- Deep system- Large amount of hail and graupel

Mean vertical distribution of the hydrometeors

IOP8 ( Stratiform event)- Shallow system- Large amount of snow

IOP8

Snow

Explained by analysing the dominant microphysical processes



MesoNH with grid_nesting (V. Ducrocq)

Arpège-> MésoNH 10km -> MésoNH 2.5km

IC : Mesoscale surface data reanalysis

BC : Aladin 3h Forecasts

12-22 TU Nîmes radar cumulated rainfall

> 300 mm

« Gard » flash flood (8 Sept. 2002)« Gard » flash flood (8 Sept. 2002)

= 260 mm


Arome 60s

304 mm

MésoNH 4s12-22 TU Nîmes radar cumulated rainfall

> 300 mm

274 mm

Single model 2.5Km

IC : Mesoscale surface data reanalysis

BC : Aladin 3h Forecasts

« Gard » flash flood (8 Sept. 2002)« Gard » flash flood (8 Sept. 2002)


MésoNH verifying toolboxMésoNH verifying toolbox

• Simulation of radar observations (Ze, VDop, ZDR, …):based on the Rayleigh diffusion approximation Z ~ NDe

6

• Simulation of satellite radiances and BTs (VIS, IR, MW): based on highly accurate radiative transfer scheme coupled to Mie/T-matrix diffusion codes or using a fast algorithm such as RTTOV


Radar reflectivities Radar reflectivities (MAP IOP2a 2 km)(MAP IOP2a 2 km)

Radar obs. (T+11h)MesoNH (hail) MesoNH (hail)

2000m 2000m

6000m 6000m


Satellite analysis of storms over GermanySatellite analysis of storms over Germany SEVIRI 10.8 µm BT (MSG) AMSUB 183+/- 1 GHz (NOAA15)

MesoNH simulation (Δx=10km) with RTTOV results after 17 hours

Deep cirrus clouds Unresolved convectionConvective event of interestfully resolved at Δx=2.5km

SEVIRI 10.8-12 µm BT (MSG)


Satellite analysis of cirrus producedSatellite analysis of cirrus producedby deep convection over Brazilby deep convection over Brazil

SEVIRI 8.7-10.8 µm BTD (MSG) MesoNH simulation (Δx=30 km) with RTTOV results after 21 hours


ConclusionConclusion

• The simulation of cloud cover and precipitation fields seems promising at high resolution even without a sophisticated initialization procedure.

• The same scheme has also potential applications in chemistry and atmospheric electricity (not shown here).

• Cloud schemes still need to be improved and tuned for the

purpose of operational applications (this should be a new concern of the cloud modeling community!)

• There is a great interest in using routine radar & satellite data to assess the quality of cloud schemes.


PerspectivesPerspectives

• Fractional cloud cover and subgrid scale precipitation. • Interactions between convection schemes (implicit

clouds) and microphysical schemes (explicit clouds) to simulate the lifetime of tropical anvil outflow.

• Scale dependence of microphysical schemes. • Computation of various ground radar and satellite

products for a multispectral and active/passive evaluation of the microphysical scheme.


Documents

Description of the MésoNH-AROME microphysical scheme and ... · Outlook • Introduction • Common assumptions made in microphysical schemes • The current scheme developed and