Past, ongoing and foreseen

choppedspleenΜηχανική

21 Φεβ 2014 (πριν από 3 χρόνια και 5 μήνες)

70 εμφανίσεις

Microphysics’ evolution,

Past, ongoing and foreseen

-----

Part b
: present status and
foreseen evolutions (species’
handling, processes, geometry,
layer’s split)


R. Bro
žková, J.
-
F. Geleyn

A1WD, Ljubljana, Slovenia, 13
-
15/06/2012

Set up of first sensitivity study

ALARO Microphysics is used in a unified
stratiform

and
convective case (the 3MT scheme), in grey zone:


Resolution 4.7 km; 87 levels;


10 days’ suites with the assimilation cycling, for highly
active convective period of 21 to 30 June 2009.

Experiments (with respect to ALARO reference, going
progressively to ARPEGE choices):


Switch off pseudo
-
graupel
;


ARPEGE type of processes;


Lagrangian

sedimentation and fixed speed of
precipitations.

Cloud liquid water


average
over space/runs at +48h

Alaro

reference

No pseudo
graupel
;
slower fall, more water
in feedback

ARPEGE Processes

Lag. sed. & fixed fall
speed: faster fall higher
up, less
cloud

water in
feedback

Cloud
ice

water


average over
space/runs at +48h

Alaro

reference

No pseudo
graupel

ARPEGE Processes

Lag. sed. & fixed fall
speed

Rain
-
drops amount


average
over space/runs at +48h

Alaro

reference

No pseudo
graupel

ARPEGE Processes:
diff due to evaporation
.

Lag. sed. & fixed fall
speed


compensating
the evaporation effect.

Snow
-
flakes

amount


average
over space/runs at +48h

Alaro

reference

No pseudo
graupel



impact noticeable

ARPEGE Processes
-

little change for snow

Lag. sed. & fixed fall
speed: higher speed
diminishes snow

Precipitation structure
-

night

ref

A bit spread

no
graupel

Less rain, mainly evap.

ARP processes

More rain

SLAG, fix speed

Rain
-
drops amount



average
over space/runs at +18h

Alaro

reference

No pseudo
graupel

ARPEGE Processes

Lag. sed. & fixed fall
speed:

in the afternoon rain is
considerably enhanced

Precipitation structure
-

day

ref

similar

no
graupel

ARP processes

A bit less rain

SLAG, fix speed

More rain everywhere

Seen by budgets

Difference of
budgets due to
sedimentation,
which influences all
other processes

Seen by scores

(
1)

Humidity

scores

at

250

hPa

against

radiosoundings
:

ALARO

a

bit

better

tuned

for

convective

regime

Wind

at

10
m
:

ARPEGE

tuning

is

better
.

Seen by scores (2)

Change

in

sedimentation

is

clearly

seen

in

bias

of

precipitation

in

the

afternoon
.

Higher

bias

in

cloudiness

show

that

ARPEGE

processes

lead

to

moister

atmosphere
.


Once

the

sedimentation

is

switched

to

lagrangian

with

fixed

speed,

there

is

a

compensation
.

Species at 9km and 4.7km with 3MT

Suite_1

at
4.7km

and
Suite_2

at
9km
, average over the same domain
and time period for the afternoon (18h).

Differences are within those of changing microphysical processes and/or
sedimentation.

Influence of geometry

Convective

part

of

1
h

precipitation

sum,

afternoon
.

Left
:

maximum
-
random

overlap
;

Right



pure

random

overlap
.

In

the

picture

to

the

right,

too

much

seeded

surfaces,

thus

too

much

evaporation

and

hence

too

small

amounts

of

precipitation
.


Influence of “convective
condensate protection”

Convective part of 1h precipitation sum, afternoon.

Left: protection applied; Right


no protection, less pronounced maxima.

Some (partial) conclusions


Usefulness of the modularity => we may
detect what would be otherwise unnoticed
due to compensations and feedbacks;


Parameterization of sedimentation


bigger
effect than changing details in processes?


Multi
-
scale performance of the schemes


uneasy but not impossible;


Grey zone parameterization of deep moist
convection


feasible; geometry and
prognostic aspects are key ingredients.

Foreseen evolutions


Given the difficulties linked with the diagnostic graupel
(and in order to prepare other evolutions), we would like
to use
prognostic graupel

(at unchanged
phenomenology). Work is ongoing (Joris Van den Bergh).


After this step, we could program a functional equivalent
to the
ICE3 microphysics package of AROME under
ACACON, ACCOLL and ACEVMEL
. Technically the
work is without big hurdles. But to make the (long time
-
step oriented) algorithms compatible with the ALARO and
ARPEGE ones will require some special attention. Hence
the ‘perpetual postponement’ of the task for the past 3
years …


At the level of the current geometry one should have the
third option (like for the diagnostic radiative clouds of
Christoph Wittman) of an
intermediate between random
overlap and maximum
-
random overlap
. The work is
now at the validation stage.


The ‘core issue’ will likely be the one of the
layer’s split
.



For
both last items
, let us start with the existing:

Geometry of clouds and rain

(1/4)

Geometry of clouds and rain (2/4)

Random overlap of parts separated by clear air, maximum
overlap of adjacent parts (schematic view)

Intuitive solution:
2 inputs

and
2 outputs

for the
‘transmission/creation’ in
the considered level (the
cloud ‘homogeneises’ the
precipitation’s output) ??

This is now the correct
solution with
4 inputs

and
3 outputs

(the cloud still
homogeneises).
But why
is there input in the time
-
step non
-
seeded parts ?

Because there was a
cloud there in a previous
time
-
step and that the
precipitations it generated
did not finish falling (if not
evaporating)

Geometry of clouds and rain (3/4)

Two options are currently

coded:


-

Maximum overlap of clouds

(more realistic)


reference
;

-
Random overlap of clouds


exp 1

The impact (here shown for

evaporation of falling species)

is not negligible.

The problem cannot be treated

as linear.

Geometry of clouds and rain (4/4)

Far more complex than for radiation, but still tractable !

Maximum
-
random case (

=1), C & C
*

local & above cloudiness



(clear part),


(cloudy part) seeded proportions,

*

above again

C
C
C
C
C
C
C
C
C
C
C
C
))
,
min(
(
)
,
min(
1
))
,
max(
1
(
)
)
,
(max(
*
*
*
*
*
*












Random case (

=0)

)
1
(
*
*
*
C
C







)
1
(
]
)
1
)(
,
)[min(
1
(
)]
(
)[
1
(
1
))
,
max(
1
(
)
)
,
(max(
*
*
*
*
*
*
*
*
*
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C



























0 <


㰠<

‘Treble cloud’, a more realistic view
of the «

h+v

» geometry (1/2)

The continuous spectrum
of cloud ‘intensity’ is
replaced by two synthetic
parts, instead by the
current mean value.

There are now more
transition possibilities
from layer to layer
(4=>9). Overlap
hypotheses are even
more crucial.

After
Shonk

et al.,

‘Treble cloud’, a more realistic view
of the «

h+v

» geometry (2/2)

Ultimate sophistication, the
decorrelation depth depends
on the latitude (continuous
tropical clouds are more
‘vertical’ than frontal ones).

Here also a ‘decorrelation
factor’ exists for the ‘non
-
random’ part, parameterised
in function of the layer depth
(we could already have that).

Beware that what is named here


楳i


楮i䅌A剏R湯瑡瑩潮s⡡湤


楳i瑨敮t
獯浥瑨m湧n敬獥s

Conclusions


Key factors in microphysics, judged from the
impact on results:


Geometry of clouds and seeded parts;


Sedimentation


fall speed;


Evaporation/melting (processes)

=> Phase changes (latent heat) and moistening


dominating factors?



Autoconversion

rate (feeding precipitation)


Goal:


get correct results for good reasons


physical realism:
go for better descriptions of cloud geometry
etc
;


Attention on compensating mechanisms, over
-
tunings
and/or arbitrary tricks (penicillin
vs

acylpyrin
)
.