06.05.2015 Views

What LES tell us about the parameterisation of  and  - Convection

What LES tell us about the parameterisation of  and  - Convection

What LES tell us about the parameterisation of  and  - Convection

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

COST ES0905<br />

<strong>What</strong> <strong>LES</strong> <strong>tell</strong> <strong>us</strong> <strong>about</strong> <strong>the</strong><br />

<strong>parameterisation</strong> <strong>of</strong> <strong>and</strong> <br />

An attempt to stimulate convergence in<br />

<strong>parameterisation</strong> approaches<br />

Wim de Rooy et al.<br />

Entrainment <strong>and</strong> detrainment in cumul<strong>us</strong> convection: an overview.<br />

Wim C. de Rooy, Peter Bechtold, Kristina Fröhlich, Cathy Hohenegger, Harm Jonker, Dmitrii<br />

Mironov, A. Pier Siebesma, Joao Teixeira, <strong>and</strong> Jun-Ichi Yano<br />

(Pending minor revisions QJRMS)


COST ES0905<br />

Parameterisations <strong>of</strong> detrainment <strong>and</strong> entrainment are<br />

extremely important in NWP <strong>and</strong> Climate models.<br />

(see e.g. Murphy et al. 2004).<br />

They should have a sound physical bases:<br />

–Observations (e.g. Jonas 1990, Rodts et al. 2003)<br />

-<strong>LES</strong><br />

Direct translation from observation to <strong>parameterisation</strong> is<br />

cumbersome<br />

Most important tool to evaluate <strong>and</strong> diagnose <strong>and</strong>


Why are <strong>LES</strong> so important?<br />

<strong>LES</strong> domain NWP gridbox<br />

NWP- or Climate model:<br />

)<br />

t<br />

<br />

(<br />

' '<br />

w<br />

z<br />

F<br />

<br />

w ' '<br />

= total turbulent transport<br />

(sub grid for NWP models)<br />

'<br />

w ww<br />

'<br />

<br />

w'<br />

'<br />

<strong>LES</strong><br />

domain


= cloudy <strong>LES</strong> gridbox<br />

Convective transport<br />

c<br />

w'<br />

'<br />

a w'<br />

'<br />

(1<br />

a<br />

) w'<br />

'<br />

a<br />

(1 a<br />

)( w<br />

c<br />

c<br />

e<br />

c<br />

c<br />

<br />

1<br />

A<br />

c<br />

<br />

c cloudy<br />

area<br />

c<br />

c<br />

dxdy<br />

w)(<br />

<br />

)<br />

Define cloud with a conditional<br />

sampling, e.g.: 0 w<br />

0<br />

aw<br />

c<br />

v v<br />

(Siebesma & Cuijpers 95)<br />

q l<br />

c<br />

<br />

e)<br />

<br />

M<br />

( c<br />

<br />

)<br />

<br />

c( e<br />

Convective transport<br />

e<br />

c<br />

e<br />

Investigate bulk mass flux<br />

approach in <strong>LES</strong>


<strong>LES</strong> provide <strong>the</strong> experimental environment to investigate<br />

your (bulk) mass flux parametersation including all <strong>the</strong><br />

relevant processes like <strong>and</strong> <br />

Determine:<br />

Maw<br />

c<br />

z<br />

<br />

<br />

M<br />

( <br />

M<br />

z<br />

c c<br />

)<br />

c e<br />

describes <strong>the</strong><br />

updraft dilution<br />

<strong>and</strong> describe <strong>the</strong><br />

mass flux pr<strong>of</strong>ile<br />

•How well do your <strong>and</strong> compare to <strong>LES</strong>?<br />

•How do <strong>the</strong>y vary in <strong>LES</strong> <strong>and</strong> with what?<br />

•Can you capture <strong>the</strong> most important dependencies?<br />

<strong>LES</strong> should be leading in <strong>parameterisation</strong> development!<br />

However, in reality………….


Direct comparison <strong>of</strong> (dilution) with <strong>LES</strong>: Kain Fritsch<br />

BOMEX case: Kain Fritsch with optimal 0<br />

0.004<br />

<br />

KF<br />

<br />

<br />

2<br />

0 c<br />

=0.02* c<br />

2<br />

0.003<br />

0.002<br />

0.001<br />

<br />

<br />

KF<br />

2<br />

0 c<br />

0.000<br />

0.000 0.001 0.002 0.003 0.004<br />

+2h<br />

+3h<br />

+4h<br />

+5h<br />

+6h<br />

+7h<br />

+8h<br />

+9h<br />

+10h<br />

+11h<br />

+12h<br />

+13h<br />

+14h<br />

+15h<br />

<strong>LES</strong><br />

+1h<br />

=1/(z-z i<br />

+500)<br />

0.004<br />

BOMEX case: =(z-z i +500) -1<br />

+1h<br />

+2h<br />

+3h<br />

+4h<br />

0.003<br />

+5h<br />

+6h<br />

+7h<br />

+8h<br />

+9h<br />

+10h<br />

0.002<br />

+11h<br />

+12h<br />

+13h<br />

+14h<br />

+15h<br />

0.001<br />

0.000<br />

0.000 0.001 0.002 0.003 0.004<br />

<strong>LES</strong><br />

Kain Fritsch not very suitable for describing <strong>the</strong> dilution (cloud top!)


COST ES0905<br />

Dependence <strong>of</strong> on RH <strong>of</strong> <strong>the</strong> environment?<br />

Two operational<br />

<strong>parameterisation</strong>s:<br />

<br />

<br />

KF<br />

2<br />

0<br />

c<br />

0(1.3<br />

RH)f<br />

Bechtold<br />

scale<br />

Increasing with RH e !<br />

Decreasing with RH e !<br />

Setup RH experiment: Derbyshire et al. 2004<br />

Bechtold et al. (2008)<br />

Dependency RH e<br />

Not yet established


There is ano<strong>the</strong>r very important feature<br />

that should be represented correctly:<br />

The mass flux pr<strong>of</strong>ile<br />

<strong>and</strong> both influence <strong>the</strong> mass flux pr<strong>of</strong>ile<br />

However<br />

•Variations in <strong>the</strong> mass flux pr<strong>of</strong>ile are strongly<br />

dominated by !<br />

•Empirical arguments<br />

•Theoretical arguments<br />

De Rooy & Siebesma 2008, 2010


Empirical arguments<br />

Much larger variation in from case to case, hour to hour, different cloud sizes<br />

Jonker et al., 2006 Hohenegger, 2011


Empirical arguments<br />

Optimal fixed <strong>and</strong> <strong>LES</strong><br />

Optimal fixed <strong>and</strong> <strong>LES</strong><br />

ARM<br />

ARM<br />

De Rooy & Siebesma, 2008


Cloud layer<br />

Theoretical arguments<br />

Based on general in-cloud budget eq. for q t <strong>and</strong> making a distinction<br />

between small <strong>and</strong> large-scale lateral mixing (de Rooy & Siebesma 2010)<br />

Cloud ensemble<br />

(divergent condition)<br />

<br />

<br />

turb<br />

<br />

dyn<br />

turb<br />

<br />

<br />

dyn<br />

~<br />

1<br />

M<br />

1<br />

M<br />

M<br />

z<br />

<br />

<br />

M<br />

~<br />

z<br />

1<br />

~<br />

z<br />

Included in:<br />

de Rooy & Siebesma 08<br />

Neggers et al. 09


<strong>What</strong> are <strong>the</strong> consequences?<br />

As far as variations in <strong>the</strong> (non-dim) mass-flux pr<strong>of</strong>ile concerns:<br />

Use a (simple) fixed function for but a flexible <br />

=<br />

Parameterise <strong>the</strong> mass-flux pr<strong>of</strong>ile<br />

So no explicit <strong>and</strong> for M pr<strong>of</strong>ile!<br />

Unconventional:<br />

Separate entrainment (dilution) <strong>and</strong> mass flux pr<strong>of</strong>ile!<br />

De Rooy & Siebesma 2008 (RACMO, Harmonie) <strong>and</strong> Neggers et al. 2009


Parameterising <strong>the</strong> mass flux pr<strong>of</strong>ile<br />

<strong>LES</strong><br />

large c<br />

High RH e (friendly environment)<br />

Very buoyant updraft(s)<br />

Slow decrease M with height<br />

small c<br />

Low RH e (hostile environment)<br />

Marginally buoyant updraft(s)<br />

Fast decrease M with height<br />

Model <strong>parameterisation</strong><br />

Parameterisation mass flux pr<strong>of</strong>ile (or ) is dependent<br />

on properties <strong>of</strong> <strong>the</strong> environment <strong>and</strong> <strong>the</strong> updraft itself!<br />

De Rooy & Siebesma, 2008


0 (1.3 RH)<br />

Bechtold f scale<br />

Traditional scheme<br />

<br />

Bechtold et al. 2008<br />

COST ES0905<br />

Setup RH experiment: Derbyshire et al. 2004


Concl<strong>us</strong>ions/Recommendations<br />

• <strong>LES</strong> are very important:<br />

– Fundamental studies <strong>and</strong> (Romps, Dawe & A<strong>us</strong>tin)<br />

– Investigate <strong>and</strong> in bulk (model) framework<br />

• Parameterisations <strong>of</strong> , should be based on <strong>LES</strong> (i.o. tuned in<br />

model) describing <strong>the</strong> right processes.<br />

• Detrainment is responsible for variations in <strong>the</strong> mass flux pr<strong>of</strong>ile<br />

(due to variations in cloudlayer depth, environment <strong>and</strong> updraft).<br />

Separate <strong>parameterisation</strong> <strong>of</strong> mass flux <strong>and</strong> dilution ()<br />

– Theoretical <strong>and</strong> empirical arguments<br />

– More flexibility<br />

– Rob<strong>us</strong>tness<br />

• Keep parameterizations as simple as possible.<br />

J<strong>us</strong>t enough complexity to include <strong>the</strong> most relevant processes


A popular traditional scheme:<br />

Kain Fritsch (buoyancy sorting)<br />

(1-)θ vu<br />

()θ ve<br />

vu<br />

c<br />

1<br />

v<br />

positive buoyant<br />

mixtures<br />

negative buoyant<br />

mixtures<br />

ve<br />

Courtesy: Stephan de Roode<br />

0<br />

fraction environmental air ()

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!