LES%20modeling%20of%20precipitation%20in%20Boundary%20Layer%20Clouds%20and%20parameterisation%20for%20General%20Circulation%20Model

1
LES modeling of precipitation in Boundary Layer
Clouds and parameterisation for General
Circulation Model
CNRM/GMEI/MNPCA
Olivier Geoffroy
Jean-Louis Brenguier, Frédéric Burnet, Irina
Sandu, Odile Thouron
2
The problem of modeling precipitation formation
in GCM

• Variables in GCM mean values over a large area
  in GCM.

A parameterisation of the precipitation flux
averaged over an ensemble of cells is more
relevant for the GCM resolution scale
3
Super bulk parameterisation
Pawlowska Brenguier, 2003 At the scale of an
ensemble of cloud cells quasi stationnary
state Is it feasible to express the mean
precipitation flux at cloud base Rbase as a
function of macrophysical variables that
characterise the cloud layer as a whole ?
In GCMs, H, LWP and N can be predicted at the
scale of the cloud system
4
Objectives Methodology
Objectives - To establish the relationship
between Rbase, LWP and Nact, and empirically
determine the coefficients.
Methodology 3D LES simulations of BLSC fields
with various LWP, Nact and corresponding Rbase
values
Suppose power law relationship
Regression analysis
a ? a ? ß ?
5
Outline

• Presentation of the LES microphysical scheme
• Particular focus on cloud droplet sedimentation
  parameterisation
• Validation of the microphysical scheme
• Simulation of 2 cases of ACE-2 campaign and GCSS
  Boundary layer working group intercomparaison
  exercise
• Come back to the problematic
• Results of the parameterisation of precipitation
  in BLSC

6
LES microphysical scheme

• Modified version of the Khairoutdinov Kogan
  (2000) LES bulk microphysical scheme (available
  in next version of MESONH).

• Specificities
• 2 moments
• low precipitating clouds local qc lt 1,1 g kg-1
• - coefficients tuned using an explicit
  microphysical model as data source -gt using
  realistic distributions.
• valid only for CRM.

microphysical Processes and variables
Evaporation KK (2000)
Air qv (kg/kg) ? (K)
Autoconversion KK (2000)
Cloud qc (kg/kg) Nc (m-3)
Drizzle qr (kg/kg) Nr (m-3)
Condensation Evaporation Langlois (1973)
Accretion KK (2000)
Air Aerosols C (m-3), k, µ, ß ( constant
parameters) W (m s-1) ? (K)
Na (m-3)
Activation Cohard and al (1998)
Sedimentation of cloud droplets Stokes law
generalized gamma law
Sedimentation of drizzle drops KK (2000)
7
Parameterisation of cloud droplets sedimentation
8
Results for gamma law, a3, ?2
E(d5) ()
E(d2) ()
d2
d5
100
Color number of spectra in each pixel in of
nb_max
50
0
E(deff) ()
E(deff) ()
deff
deff
only spectra at cloud top

• - Generalized gamma law best results for a3,
  ?2
• Lognormal law, similar results with sg1,2-1,3
• DYCOMS-II results
• (Van Zanten personnal communication).

9
Results for lognormal law, sg1.5
E(d5) ()
E(d2) ()
d2
d5
100
Color number of spectra in each pixel in of
nb_max
50
0
E(deff) ()
E(deff) ()
only spectra at cloud top
deff
deff
Lognormal law, with sg1.5, overestimate
sedimentation flux of cloud droplets.
10
Scheme validation
11
GCSS intercomparison exercise Case coordinator
A. Ackermann (2005)

• Case studied DYCOMS-II RF02 experiment (Stevens
  et al., 2003)
• Domain 6.4 km 6.4 km 1.5 km
• horizontal resolution 50 m,
• vertical resolution 5 m near the surface and
  the initial inversion at 795 m.
• fixed cloud droplet concentration Nc 55 cm-3
• 2 simulations
• - 1 without cloud droplet sedimentation.
• - 1 with cloud droplet sedimentation
  lognormale law with sg 1.5
• 2 Microphysical schemes tested - KK00 scheme,
  
• 
  - MESONH 2 moment scheme
• 
  Berry and Reinhardt scheme (1974).

4 simulations KK00, no sed / sed
BR74, no sed / sed
12
Results, LWP, precipitation flux
Median value of the ensemble of models
observation
LWP (g m-2) f(t)
KK00, sed
BR74, sed
Central half of the simulation ensemble
KK00, no sed

• LWP too low

Ensemble range
BR74, no sed
6H
3H
Rsurface (mm d-1) f(t)

• KK00 underestimation of precipitation flux
• by a factor 10 at surface
• - BR74 good agreement at surface

0.35 mm d-1
6H
3H
Rbase (mm d-1) f(t)

• KK00 underestimation of precipitation flux by
  only a factor 2 at cloud base
• BR74 underestimation at cloud base by a factor
  2, Rsurface Rbase ? no evaporation

NO DATA
1.29 mm d-1
13
Results, What about microphysics ?
Ndrizzle (l-1)
dvdrizzle (µm)
hsurf (m)
hsurf (m)
CT
CT
BR74
BR74
KK00
KK00
CB
CB
Averaged profils of Ndrizzle, dvdrizzle in each
30 m layer after 3 hours of simulation and
averaged value of measured Ndrizzle, dmeandrizzle
(resolution 12 km) at cloud base and at cloud
top (Van Zanten personnal communication)

• - KK00 scheme reproduce with good agreement
  microphysical variables at cloud top and cloud
  base
• BR74 scheme too few and too large drops.

14
Simulation of 2 ACE-II cases
15
Simulation of 2 ACE-II cases
Objective comparison of mean profiles of qr ,
Nr , dvr for 1 polluted and 1 marine case.

• Domain 10 km 10 km,
• resolution horizontaly 100 m, verticaly
  10 m in/above the cloud
• initialisation corresponding profile of
  thermodynamical variables.

Comparison of macrophysical variables
26 june, pristine case 26 june, pristine case 9 July, polluted case 9 July, polluted case
Macrophysical variables H (m) Nact (cm-3) H (m) Nact (cm-3)
measurements 202 51 167 256
Simulations KK00, BR74 190 48-49 170 193
Macrophysical variables for measurements
(Pawlowska and Brenguier, 2003) and simulations
after 2H20
16
Results 26 june (pristine)
KK00 / measurements
hbase
BR74 / measurements
hbase
Vertical profile of qr (g kg-1)
Vertical profile of Nr (g kg-1)
Vertical profile of dvr (g kg-1)
Mean values in each 30 m layers
17
Results 9 july (polluted)
KK00 / measurements
BR74 / measurements
BR74 values lt 10-2 l-1
BR74 values lt 10-5 g kg-1
Pristine case KK00 represents with good
agreement precipitating variables Polluted case
KK00 underestimate precipitation. BR74
underestimate precipitation by making too large
drops but with very low concentration
Vertical profile of qr (g kg-1)
Vertical profile of Nr (g kg-1) Mean values in
each 30 m layers
Vertical profile of dvr (g kg-1)
18
Results, super bulk parameterisation
19
Results, super bulk parameterisation

• Initial profiles profiles (or modified
  profiles) of ACE-2 (26 june), EUROCS, DYCOMS-RF02
  ? differents values of LWP 20 g m-2 lt LWP lt 130
  g m-2
• different values of Nact 40 cm-3 lt Nact lt 260
  cm-3
• Domain 10 km 10 km.
• horizontal resolution 100 m,
• vertical resolution 10 m near surface,
  in and above cloud

Rbase (kg m-2 s-1)
( 1,7 mm d-1)
20
Summary

• Cloud droplet sedimentation
• Best fit with a 3 , ? 2 for generalized gamma
  law,
• sg 1,2 for lognormal law.
• - Validation of the microphysical scheme
• GCSS intercomparison exercise
• The KK00 scheme shows a good agreement with
  observations for microphysical variables
• Underestimation of the precipitation flux with
  respect to observations.
• LWP too low ?
• Simulation of 2 ACE-2 case
• Good agreement with observations for
  microphysical variables for KK00
• Parameterisation of the precipitation flux for
  GCM
• Corroborates experimental results Rbase is a
  function of LWP and Nact

21
(No Transcript)
22
RF02 0800 m
KK00, sed
KK00, no sed
BR74, sed
BR74, no sed
23
RF02 gt 450 m
KK00, sed
KK00, no sed
BR74, sed
BR74, no sed
24
Profils ACE-2
9july
26 june
25
Results, What about microphysics ?
Observations
Simulations
Ndrizzle (l-1)
hsurf (m)
Nc (cm-3), Ndrizzle (l-1)
CT leg
CB leg
CT
BR74
KK00
CB
Øvdrizzle (µm)
hsurf (m)
Øgc, Øgdrizzle (µm)
CT leg
CB leg
CT
BR74
KK00
CB
Variations of mean values of N and geometrical
diameter for cloud and for drizzle, along 1 cloud
top leg,, 1 cloud base leg. Mean values over 12
km. (Van Zanten personnal communication).
Averaged profils of Ndrizzle, Øvdrizzle in each
30 m layer after 3 hours of simulation.
26
9 juillet
27
26 juin
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
sigma
dv
hsurface
34
(No Transcript)
35
Paramétrisations  bulk 
Modèle bulk
On prédit les moments de la distribution qui
représentent des propriétés densemble (bulk) de
la distribution. ex M0Ni , M3qi
Modèle explicite ou bin On prédit la distribution
elle même. 200 classes.
Modèle bulk moins de variables
36
Parametrisations bulk valides dans les GCM?
collection
accrétion
autoconversion

• Processus microphysiques (10 m, 1 s) dépendent
  non linéairement des variables locales (qc, qr,
  Nc, Nr ).
• Distribution temporelle et spatiale des variables
  non uniforme.
• le modèle doit résoudre explicitement les
  variables locales pour que paramétrisations bulk
  soient valides.
• utiliser paramétrisations bulk dans les GCM ( 50
  km, 10 min) peut être remis en question.

37
simulations

• On veut plusieurs champs avec différentes valeurs
  de ltLWPgt, ltNgt,, ltRgt.
• 7 simulation MESONH avec différentes valeurs de
  Na 25, 50, 75, 100, 200, 400, 800 cm-3.
• Fichier initial champ de donnée à 12H de la
  simulation de cycle diurne dIrina et al. sans
  schéma de précipitation.
• 24H de simulation pour chaque simulation -gt LWP
  varie (cycle diurne du nuage).
• Domaine 2,5 km 2,5 km 1220 m
• Resolution horizontale 50 mailles,
• verticale 122
  niveaux.
• Pas de temps 1 s.
• Schéma microphysique schéma modifié du schéma
  Khairoutdinov-Kogan (2000)

Fig. Profil moyen du rapport de mélange en eau
nuageuse qc en fonction du temps
Début des simulations avec schéma microphysique
38
Schéma KK modifié

• KK schéma microphysique bulk pour les
  stratocumulus. Les coefficients ont été ajustés
  avec un modèle de microphysique explicite (bin).
• Intérêt
• Nact, Nc en variables pronostiques (on veut
  différentes valeurs de N).
• schéma développé spécialement pour les
  stratocumulus (particularité pluie très faible)

39

• 7 simulations de 24 H.
• 1 sortie toutes les heures.

724 168 champs avec des valeurs différentes de
H, ltLWPgt, N, ltRgt
40
Profil moyen du rapport de mélange en eau de
pluie en fonction du temps
NCCN 25 cm-3
NCCN 50 cm-3
NCCN 400 cm-3
NCCN 100 cm-3
41
Calcul de H, LWP, N, R

• mailles nuageuses mailles ou qc gt 0,025 g kg-1
• cumulus
  sous le nuage sont rejetés.
• Calcul de H
• Définition de la base?
• Calcul de LWP
• Calcul de N
• qc gt 0,9 qadiab
• 0,4H lt h lt0,6 H
• Nr lt 0,1 cm-3
• Calcul de R
• R lt qr (Vqr-w) gt, R lt qr Vqr gt
• Sur fraction nuageuse, à la base.

42
Comparaison avec les données DYCOMS-II, ACE-2

• ACE-2
• Mesures in-situ
• -gt vitesse des ascendances w pas prise en compte
  dans le calcul du flux.
• Flux calculé sur la fraction nuageuse (dans le
  nuage)
• DYCOMS-II
• Mesures radar
• -gt mesure du moment 6 de la distribution
• -gt vitesse de chute réel. (vitesse ascendances w
  vitesse terminal des gouttes Vqr)
• Flux calculé au niveau de la base du nuage.

43
R f(H3/N)
44
R f(LWP/N)
Observation dun hystérésis Déclenchement de la
pluie avec un temps de retard. -gt Il faut prendre
en compte la tendance des variables détat?
45
Conclusion

• On retrouve bien les résultats expérimentaux
• dépendance de R en fonction des variables H ou
  LWP, N
• Hystérésis de en prononcé lorsque NCCN
  augmente (lorsque R augmente).
• gt rajouter une variable pronostique
  supplémentaire (qr) ? utiliser la tendance de LWP
  dLWP/dt ?
• Expliquer cette dépendance en isolant une seule
  cellule et en regardant comment varient qc, qr

46
The problem of modeling precipitation formation
in GCM
Presently in GCM parameterisation schemes of
precipitation directly transposed from CRM bulk
parameterization. Example
Problem Inhomogeneity of microphysical
variables. Formation of precipitation non
linear process local value have to be
explicitely resolved
LES domain
3D view of LWC 0.1 g kg-1 isocontour, from the
side and above.
47
Why studying precipitation in BLSC (Boundary
Layer Stratocumulus Clouds ) ?
Parameterization of drizzle formation and
precipitation in BLSC is a key step in numerical
modeling of the aerosol impact on climate
48
(No Transcript)