+ All Categories
Home > Documents > Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Date post: 14-Jan-2016
Category:
Upload: ryo
View: 35 times
Download: 0 times
Share this document with a friend
Description:
Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment Cloud-Radiation interaction Large scale subsidence Vertical Resolution Colin G Jones. SMHI Norrköping S601 76 Sweden Email: [email protected]. Stratocumulus Clouds are an important - PowerPoint PPT Presentation
Popular Tags:
30
Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment Cloud-Radiation interaction Large scale subsidence Vertical Resolution Colin G Jones. SMHI Norrköping S601 76 Sweden Email: [email protected]
Transcript
Page 1: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Simulating Stratocumulus cloudssensitivity to representation of:

DrizzleCloud top entrainment

Cloud-Radiation interactionLarge scale subsidence

Vertical Resolution

Colin G Jones.SMHI

Norrköping S601 76Sweden

Email: [email protected]

Page 2: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Stratocumulus Clouds are an importantComponent of the Climate System (esp. Sub-tropical oceanic Stratocumulus)

Also an important forecast parameterSc impact strongly on:

Downwelling Solar RadiationDownwelling Long wave Radiation (esp winter)Often (but not always) produce Drizzle

Major controls on the above are:

Cloud AmountCloud water content (LWP)Droplet size distribution & effective radiusCloud top entrainment.

Page 3: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Persistent Sc Clouds aften have a strong DiurnalCycle (LWP minimum~in mid-afternoon). Failureto capture this leads to large errrors in surface solarradiation flux

Page 4: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Stratocumulus Clouds are ubiquitous & complex

Page 5: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Solar radiation heats cloud & surface

Longwave radiation at cloud top/baseinduces turbulence

Cloud top entrainment of warm/dry airdilutes cloud water

Large scale subsidence counters cloudThickening and warms & dries cloud top.

Turbulent transport of heat & moisturesource term for cloud

Drizzle depletes cloud water

Page 6: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

EUROCS: Diurnal Cycle of StratocumulusBased of July 17-18th 1987 FIRE caseOver San Nicolas Island off California.

Single Column & Large Eddy SimulationIntercomparsion.

LES models assumed case was non-precipitatingThis may NOT be a valid assumption e.g. DYCOMS2 (Stevens etal 2003) observed frequent drizzle rates ~0.5-1mm/day fromCalifornian Sc clouds.

LES may 1. Underestimate LWP 2. Get right LWP due to excess entrainment 3. N.B. Observed LWP will be for non-precipitating clouds. Total LWP will be higher and increased LWP within precipitating clouds WILL increase cloud albedo.

Page 7: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

LWP sensitive to drizzlei.e. Auotconversion & subsequent collection

of cloud droplets to rain dropsRasch & Kristjansson J.Climate 1999

2,1

1.001.0

55.0

(,max

333

13

4

n

a

effc

critprpptlocakeffcC

rrHNqCPWAUTn

aut

lcritllaut

Ql is cloud liquid water contentN is assumed droplet density of cloud dropletsN=400 cm-3 (over land in PBL)…polluted airN=(40-150) (over sea & land above PBL)..clean aireffc=collection efficiency of cloud dropletspptloc=local incloud precip rate(mm/day)critpr=1.0 mm/day

Khairoutdinov & Kogan (MWR 2000)Multiple regression to explicit collection models:

79.1 47.2 1350

baC

NqCPWAUT bal

Page 8: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

2

1

n

kRH

kRHkRHRHCLD

n

crit

crit

49.0 100 25.0

1exp1

0

,0

p

qRH

qRHXUCLD

sat

ilp

Slingo QJRMS 1987

Xu & Randall MWR 1996

Cloud Fraction Parameterisation

Both schemes only activate if RH(k)>RH(k)crit

This is the link of cloud onset to subgrid scalerelative humidity variablity (probably resolutiondependent)

Page 9: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

K-1

K

Pin(k-1)

Microphsyics (ITER=1)

Pin(k-1) + Microphysics (ITER=1)

Microphsyics (ITER=2)

Pin(k-1) + Microphysics (ITER=1) + (ITER2)

K+1

K-1

K

Pin(k-1)=0

Autoconversion

Autoconversion + Collection

Parameterisation developed for coarse vertical resolution GCMsand includes a subgrid scale vertical parameterisation allowinginitial rain water production in a vertical layer (K) to influence further production of rain water in the same vertical layer (K) i.e. two iterations of the condensate to precipitation termis performed per vertical layer. All newly formed precipitation in 1st iteration is seen as an input to layer K in the 2nd iteration.

As vertical resolution increases the subgrid scale vertical schemeis not needed and the effect should be reduced.

.1,

2

1000112z

MAXiterkPkPiterkin

P

Vertical resolution sensitivty in cloud microphysics

Page 10: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Small changes in drizzle rates can greatly effectLWP which lies in sensitive range for cloud albedo changes. Solar flux at surface increased by ~50Wm-2

Page 11: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

LWP sensitivity to drizzle reduced at higher verticalResolution due to increased cloud top mixing andDilution of cloud water. Precip becomes noisy withIncreased vertical resolution.

Page 12: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Sensitivity of LWP and Solar flux to parameterisedCloud fraction and assumed number droplet concentrationcapnc=(150, 40) spread in Khairoutdinov & Kogan 2000)HIR40L with no parameterisation of cloud top entrainment

Page 13: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Statistical Cloud schemes and linkage to turbulenceschemes using moist conserved variables

Moist TKE schemes naturally lend themselvesto use of l and qt (conserved in non-precipitatingmoist adiabatic mixing). These schemes need a cloudfraction to determine the incloud buoyancy flux

Statistical cloud schemes naturally lend themselvesto such an approach. Cloud fraction & water contentcan be diagnosed directly from l and qt

Assume l and qt have an some distirbution aboutthe mean grid box value (Gaussian and/or skewed)Diagnose cloud fraction from normalised saturationDeficit (Q1) and s (variance of l and qt about mean)

Chaboureau & Bechtold (2002) derived cloud fractionAnd ql from CRM simulations as a function of Q1 & s

2

20 086.066.0

0

55.1arctan36.05.0

11

12

111

112.1

1

1

QQq

QQQeq

Qeq

QSTATCLD

s

l

s

l

Q

s

l

Page 14: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

slpm

sl

lltts

s

lst

aqbC

Lqa

TbTqbaqa

TqqaQ

1

2

1

2222

1

21

Q1 is normalised saturation deficit s is varianceof (aql-bTl). These terms link cloud & cloud waterterms to model grid box variables & varianceof these variables derived through turbulence scheme

21

2

221

2

2 2

z

hCb

z

q

z

hCba

z

qal l

pmtl

pmt

tkes

Ltke is a diagnosed length scale from the moist turbulence scheme (cloud buoyancy included). Hence subgrid scale aspectof cloud & cloud water evolution linked directly to turbulence.Promising framework for high resolution models.**No arbitrary thresholds for cloud onset

s parameterised using turbulence length scale

Page 15: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Single column model overestimates LWP compared toLES. But they (perhaps correctly) precipitate. TotalLWP may therefore be more correct compared to totalobserved LWP (non-precipitating LWP from modelsMaximises at ~200 g/m2)

Page 16: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

strong turbulententrainment at cloud top

zgKw mhmh

,,

Turbulence scheme represents subgrid mixing by:

b

bE

gl

eEaaw

ww

m

vv

e

e

e

1

1

0

21

23

Replace at cloud top with parameterised entrainment velocity

Cloud Top Entrainment

E is evaporative enhancement factormb average buoyancy of cloudy & clear mixtures

Page 17: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Parameterised entrainment increases mixing of dry, warmair into cloud. Dilutes cloud and reduces LWP anddrizzle rates.(ie drizzle sensitive to entrainment at cloud top)

Page 18: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Increased vertical resolution leads to more mixing atcloud top directly from turbulence scheme. Analagous impact to parameterised entrainment on LWP & drizzle.need for parameterised entrainment a function ofvertical resolution. Increased mixing also reduced LWP by day and greatly impacts surface solar flux.

Page 19: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Reduced incloud absorption of solar radiation allowslarger LWP by day and greatly improves surface solarflux. Solar flux VERY sensitive to LWP.

Page 20: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Source: Martin Koehler ECMWF

Page 21: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Large scale subsidence depresses cloud thickness andLWP and drizzle rates. Strong impact on surface solarradiation flux

Page 22: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

HIR150 level model (~25m in PBL) with moist turbulence & statistical cloud scheme and 60 second timestep. Results plotted every timestep. Small hops occur when cloud thickensby a model level.

Page 23: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

HIR150 + 60s timestep. Respective cloud fraction isused in the moist turbulence scheme. In XUCLD &RHCLD cloud water prognosed by condensationroutine. Timestep plot, noise related to cloud growthby a model level.

Page 24: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Timestep level nosie not visible in 3 hour means.but will affect model stability.

Page 25: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

XUCLD & RHCLD run at 40L with 60s timestepReduced evidence of noise. Microphysics is very sensitive to vertical resolution. Moist turbulence approach exhibits greatly reduced sensitivity

Page 26: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

The overall representation of Stratocumulus cloudis sensitive to model vertical resolution. This isprobably as important as horizontal resolution.

As vertical resolution increases, cloud top mixingsimulated by the turbulence scheme increases.This reduces cloud water and as a consequencedrizzle rates decrease and surface SW flux increases

Stratocumulus LWP lies in the critical range~50-300 g/m2 where cloud albedo changes dramatically for a small LWP change. Hence simulated surface solar fluxes are very sensitiveto LWP. This will be true to a lesser extent winterfor longwave radiation via cloud emissivity

LWP is very sensitive to the representation of drizzle.This sensitivity decreases as cloud top entrainmentincreases (i.e. As vertical resolution increases)

Daytime LWP is sensitive to the fraction of solarf luxthat is aborbed in the cloud layer. This both evaporatesthe cloud and decouples the cloud from the surface.

Moist Turbulence coupled to a statistical cloud schemethat diagnoses cloud fraction & liquid/ice water seems a promising and numerically stable manner to simulate stratocumulus clouds at high (vertical)resolution.

Conclusions

Page 27: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Diurnal Cycle of Cloud Water for FIRE Stratocumulus Case

Page 28: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Diurnal Cycle of TKE and cloud fraction (isolines)For FIRE Stratocumulus case

Page 29: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Diurnal Cycle of relative humidity forFIRE Stratocumulus case

Page 30: Simulating Stratocumulus clouds sensitivity to representation of: Drizzle Cloud top entrainment

Diurnal Cycle of cloud fraction for FIRE Stratocumulus case


Recommended