Confronting LES and SCM simulations of Marine Boundary Layer Clouds in a 3D GCM Framework

Transcription

1 Confronting LES and SCM simulations of Marine Boundary Layer Clouds in a 3D GCM Framework Alexandre Catarino, Frédérique Cheruy, Frédéric Hourdin Laboratoire de Météorologie Dynamique June 9 th, 29

2 Outline 1 Introduction 2 3 4

3 Outline Introduction Motivation Objective LMDZ 1 Introduction Motivation Objective LMDZ 2 3 4

4 Motivation Objective LMDZ Assessment of Physical Parameterizations of GCM Two main approaches have been developed to improve and validate parameterizations: Single-column model (SCM) vs explicit simulations (CRM or LES): Idealised cases or inspired on observations; Simplified forcing and/or boundary conditions. GCM vs observations; Complex cases; Realistic forcing and/or boundary conditions; Climatological setup (statistical approach); Meteorological setup (deterministic approach);

5 New Assessment approach Motivation Objective LMDZ Assessment of physical parametrizations by comparing the result of LES and SCM, initialized and forced by a GCM, with simulations performed by the same GCM.

6 Motivation Objective LMDZ New Assessment approach Assessment of physical parametrizations by comparing the result of LES and SCM, initialized and forced by a GCM, with simulations performed by the same GCM. LES: resolved turbulence; SCM: GCM physics decoupled from dynamics;

7 Motivation Objective LMDZ New Assessment approach Assessment of physical parametrizations by comparing the result of LES and SCM, initialized and forced by a GCM, with simulations performed by the same GCM. LES: resolved turbulence; SCM: GCM physics decoupled from dynamics; GCM nudged and zoomed LMDZ. IPCC AR5: new physics package; To evaluate its boundary layer parametrizations (Rio & Hourdin, 28)

8 Motivation Objective LMDZ Nudging technique on stretched grid Center of the zoom (red point): -149 o W, 17 o N. Smallest grid area: (8 km) 2. «φ t nudge = φa φ, τ φ = θ, u or v, τ = relaxation constant (1 to 6 hours). LMDZ Grid for GPCI studies Black points: cross-section

9 Outline Introduction LMDZ/CALIPSO LMDZ/MODIS 1 Introduction 2 LMDZ/CALIPSO LMDZ/MODIS 3 4

10 LMDZ/CALIPSO LMDZ/MODIS High clouds 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 3 N 3 N 16 W 14 W 12 W 1 W N High Clouds Cover (JJA 28). The ITCZ high clouds (Cb tops) represented by new parametrization (upper left) are more frequent than those observed by CALIPSO (lower left), while those by the old parametrization (upper right) are less frequent. CALIPSO simulator data cortesy of Dimitra Konsta

11 LMDZ/CALIPSO LMDZ/MODIS Low clouds 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 3 N 3 N 16 W 14 W 12 W 1 W N Low Clouds Cover (JJA 28). The new parametrization (upper left) represents low clouds that the old one (upper right) do not. Those clouds are present in the observations done by CALIPSO (lower left)..1

12 LMDZ/CALIPSO LMDZ/MODIS GPCI cross-section 4 km vertical domain height (km) height (km) latitude (degrees North) latitude (degrees North) 4 height (km) latitude (degrees North) GCM June Cloud Cover (Cross-section). First 4 km: While the GCM with any parametrization represents well the decrease amount of clouds expected in the stratocumulus to cumulus transition, it fails the cloud base and top height (and the boundary layer height). CALIPSO has observed those features.

13 LMDZ/CALIPSO LMDZ/MODIS MODIS Low clouds Better representation of shallow cumulus (amount); Thin ShCu underrepresentated; Stratocumulus not consistent with MODIS; 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W N 3 N 3 N ShCu (thin).1 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 1 3 N 3 N 3 N ShCu (thick).1 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W 16 W 14 W 12 W 1 W N 3 N 3 N StCu.2 NEW.1 AR4 MODIS

14 Outline Introduction Method Case study 1 Introduction 2 3 Method Case study 4

15 Method Case study New Assessment approach Assessment of physical parameterizations by comparing the result of LES and SCM, initialized and forced by a GCM, with simulations performed by the same GCM. LES: resolved turbulence; SCM: GCM physics decoupled from dynamics; GCM nudged and zoomed LMDZ. IPCC AR5: new physics package; To evaluate its boundary layer parameterizations (Rio & Hourdin, 28)

16 Making a case... Introduction Method Case study From a particular grid at a given time, (potential) temperature, humidity and wind profiles are chosen to be the SCM/LES initial state; Surface boundary conditions and large-scale tendencies (and vertical wind) are set to meet those calculated by the GCM for the same time integration; Above 1 km, the daily mean was used and tendencies were set to zero.

17 LMDZ GCM Introduction Method Case study Dynamics Dynamical tendencies Physics T(x,y,z) q1(x,y,z),... T(z) q1(z) T(z) q1(z)... Tendencies due to : - radiative transfer - condensation - subgrid dynamics -... Physical fields φ t = ( ) φ + t DYN ( ) φ t PHY

18 Methods Introduction Method Case study A consistency test is performed (Method 1) where SCM is forced exactly like the GCM; Method 2: forcings estimated as for LES/SCM case studies; We also test the option of prescribed or calculated radiation. «φ t LS! v t LS «θ t LS «qt t LS Method 1 (SCM only) «φ t Method 2 DYN = f( v g v ) w LS v z, = H θ w LS θ z + R = H q t w LS q t z

19 Initial state Introduction Method Case study q t (g kg 1 ) u v height (m) 2 15 height (m) θ (K) l Horizontal wind Mixed layer up to 5 m, from its top until the 1 km (PBL height) a cloud layer. Baroclinic situation of surface wind of SE, turning to S with height.

20 GCM vs SCM θ Introduction Method Case study height (m) height (m) θ l : Evolution simulated by the SCM stays close to that by the GCM, which provided the forcing and surface boundary condition (T sfc = K) time (hours) time (hours) x x 1 4 height (m) height (m) δ t θ phy : Slight differences of sum of ensemble of parameterization essentially in its distribution, due to non-exact initial conditions time (hours) time (hours)

21 GCM vs SCM Cloud cover Method Case study height (m) 2 15 height (m) time (hours) time (hours) Cloud base become different after the first hour, while cloud top is different in that period. Cloud amount is quite the same.

22 GCM vs LES θ Introduction Method Case study height (m) height (m) θ l : PBL height drops in the LES in a first moment (LES spin-up) and increases until the end of simulation time (hours) time (hours) x x 1 4 height (m) height (m) δ t θ phy : The LES presents minima related to the entrainment of warmer air from the free atmosphere, which the GCM does not present (at least without the same intensity) time (hours) time (hours)

23 GCM vs LES Cloud cover Method Case study height (m) 2 15 height (m) time (hours) time (hours) Cloud layer from LES coherent with the results shown before. LES produces less cloud amount

24 Outline 1 Introduction 2 3 4

25 Cloud simulator: New parameterisation increases cloud amount consistently with CALIPSO and MODIS observations; Cloud top too low; Thinner clouds underrepresented.

26 Cloud simulator: New parameterisation increases cloud amount consistently with CALIPSO and MODIS observations; Cloud top too low; Thinner clouds underrepresented. SCM and LES Consistent results for SCM and 3D GCM approach; Difference between LES and 3D GCM: LES consistent with satellite data (e.g. higher cloud tops) LES as a tool to investigate possible parameterisation deficiencies.

27 Cloud simulator: New parameterisation increases cloud amount consistently with CALIPSO and MODIS observations; Cloud top too low; Thinner clouds underrepresented. SCM and LES Consistent results for SCM and 3D GCM approach; Difference between LES and 3D GCM: LES consistent with satellite data (e.g. higher cloud tops) LES as a tool to investigate possible parameterisation deficiencies. Work to be done... Sensitivity to the large-scale forcing and initial conditions. Deduced from LMDZ analysis (model nudged also inside zone of interest) Other regions (e.g. BOMEX) Sensitivity to the processes involved (case selection)

28 Case selection Tendency due to PBL parameterization more important than due deep convection and wake; Neglectable rain; Prevalent subsidence; 16 W 14 W 12 W 3 N 2 N 1 N # of days

29 Thank you