A process oriented descrip-on of oceanic clouds derived from A- train observa-ons, for climate model evalua-on

(1)

A process oriented descrip-on of oceanic clouds

derived from A-‐train observa-ons,

for climate model evalua-on

D. Konsta, H. Chepfer, JL Dufresne, G. Cesana, S. Bony

LMD/IPSL

** all the model / observa-on comparisons presented in this talk use COSP**

Konsta D. et al : A process oriented descrip-on of tropical oceanic clouds for climate model evalua-on, based on a sta-s-cal analysis of day-me A-‐train high spa-al resolu-on observa-ons, submiLed to Climate Dynamics

Konsta D. et al: What do we learn about clouds in climate models from A-‐train observa-ons, to be submiLed to Climate Dynamics

(2)

SW albedo

Cloud Frac-on

Cloud Reﬂectance

a drop for Op-cal Depth SW albedo depends on

the Cloud Reﬂectance and the Cloud Frac-on

(3)

Tropics, in regimes

SW albedo

Cloud Frac-on

Cloud Reﬂectance

a drop for Cloud Op-cal Depth

(4)

Tropics, in regimes

Ver-cal Distribu-on

of the Cloud Frac-on

Obs: Calipso-‐GOCCP LMDZ-‐NP LMDZ5 PRE SU RE PRE SU RE PRE SU RE W500 (hPa/day) W500 (hPa/day) W500 (hPa/day)

(5)

Monthly/ Sesonal mean model-‐obs comparisons:

• Bias in model SW albedo at high la-tudes and in subsidence tropical

regions

… signiﬁcant evolu-on in the obs in subsidence tropical regions

• Correct SW albedo results of error compensa-ons between cloud

cover and cloud op-cal depth

• The cloud ver-cal structure is poorly reproduced by the model…

too much high thick clouds

lack of mid and Low level clouds

• LMDZ5-‐NP leads to signiﬁcant improvement compared to LMDZ5,

but s-ll far from observa-ons

(6)

Monthly/ Sesonal mean model-‐obs comparisons:

• Bias in model SW albedo at high la-tudes and in subsidence tropical

regions

… signiﬁcant evolu-on in the obs in subsidence tropical regions

• Correct SW albedo results of error compensa-ons between cloud

cover and cloud op-cal depth

• The cloud ver-cal structure is poorly reproduced by the model…

too much high thick clouds

lack of mid and Low level clouds

• LMDZ5-‐NP leads to signiﬁcant improvement compared to LMDZ5,

but s-ll far from observa-ons

Could the « A-‐train » learn us more ….about clouds in

climate models ?

(7)

This sensi-vity is usually es-mated over inter-‐annual -me scale (i.e. Webb et al., Bony et al. 2004) => can not be directly linked to parameteriza-on

R: Cloud property

T : Surface Temperature

E : Environment

For climate variability studies,

we need to characterize how cloud proper-es (R) vary with environment variables (E) around the cloud

(8)

is usually es-mated over inter-‐annual -me scale (i.e. Webb et al., Bony et al. 2004) => can not be directly linked to parameteriza-on

is es-mated over inter-‐annual -me scale (i.e. Klein and Hartmann 1993), but also at nearly instantaneous -me scale (Zhang et al 2010)

R: Cloud property T : Temperature

E : Variable of Environment

R: SW albedo

CF : Cloud Frac-on

CR : Cloud Reﬂectance (drop for Cloud Op-cal Depth)

(9)

is usually es-mated over inter-‐annual -me scale (i.e. Webb et al., Bony et al. 2004) => can not be directly linked to parameteriza-on

is es-mated over inter-‐annual -me scale (i.e. Klein and Hartmann 1993), but also at nearly instantaneous -me scale (Zhang et al 2010)

R: Cloud property T : Temperature

E : Variable of Environment

R: SW albedo

CF : Cloud Frac-on

CR : Cloud Reﬂectance (drop for Cloud Op-cal Depth)

(10)

0.2 0.6 0.4 CL O U DY RE FL EC TA N CE 0 0.2 0.4 0.6 0.8 1 0.8 1 0 0.2 0.6 0.4 0.8 1 0 0 0.2 0.4 0.6 0.8 CLOUD FRACTION 0.2 0.6 0.4 0.8 1 0 0 0.2 0.4 0.6 0.8 1 CLOUD FRACTION OBS MOD BUT A B C 1 LMDZ5-‐NP + SIM OBSERVATIONS MONTHLY LMDZ5 + SIM

CL O U DY RE FL EC TA N CE CLOUD FRACTION OBSERVATIONS INSTANTANEOUS MONTHLY _MONTHLY INSTANTANEOUS _{INSTANTANEOUS}

(11)

11 0 0.2 0.4 0.6 0.8 800 600 200 400 PRE SS U RE ( hP a) CLOUDY REFLECTANCE 0.1 0 0.3 0.2 0 3. 8.1 4 16.5 40.5 CLOUD FRACTION

CLOUD OPTICAL DEPTH

-‐  op-cally thin clouds :

low level, low CF -‐  op-cally thick clouds :

mul-layer clouds

-‐  very thick clouds :

abundance of high clouds

0.4 0 0.2 0.4 0.6 0.8 CLOUDY REFLECTANCE 4% 8% 12% 0 PDF 300 500 700 900

(12)

0.4 0.2 0.6 0.8 0.2 0.4 0.6 0.8 CLOUDY REFLECTANCE OBSERVATIONS PRE SS U RE ( hP a) 0 0 0.4 0.1 0.3 0.2 LMDZ5-‐NP + SIM LMDZ5 + SIM 0 0.2 0.4 0.6 0.8 CLOUDY REFLECTANCE CLOUDY REFLECTANCE Models biases :

-‐ more op-cally thin high clouds -‐ few op-cally thick high clouds

-‐ no mid level clouds (or not well simulated) -‐ overes-ma-on of op-cally thick low clouds

CLOUD

FRACTION CLOUD _FRACTION CLOUD _FRACTION

0 800 600 200 400 300 500 700 900

(13)

13

→ Cloud op-cal depth increases with cloud top al-tude

P_TOP (hPa) τ_cloud<0.8 0.8<τ_cloud<3 τ_cloud>3 700 800 900 700 800 900 700 800 900 CLOUDY REFLECTANCE 0.2 0.1 0.3

CLOUD OPTICAL DEPTH 3.4 1.5 5. 5 850 800 700 750 900 950 0 CL O U D T O P PRE SS U RE (h Pa) 0

(14)

14

CLOUDY REFLECTANCE CLOUD OPTICAL DEPTH

CL O U D T O P PRE SS U RE (h Pa) OBS LMDZ5 LMDZ5-‐NP

(15)

15

OBS LMDZ5 LMDZ5

NP

CCCMA

 Assess the rela-on between

δCF and δτ_cloud at the cloud scale δCF ↗

δτ_cloud↗ δτcloud>>

typically

shallowcumulus typically _{stratocumulus}

δCF ↗

(16)

Conclusion

•  A-‐train obs give access to cloud pictures at « instantaneous » -me scale: it shows how the cloud proper-es (here CF and CR) vary together under a same change of environment around the cloud

•  These pictures are sta-s-cally signiﬁcant (several years of data)

•  These pictures are powerful to evaluate cloud descrip-on in climate model, … a step towards a more straightorward evalua-on of cloud descrip-on at the process scale in the model

•  The models evaluated here are not able to reproduce instantaneous cloud pictures:

ex; in model cloud op-cal depth decreases when the cloud extends horizontally (CF increases), whereas in the obs the cloud op-cal depth increases with the cloud horizontal extent

(17)

« News »:

current development in the Lidar simulator / COSP

– the cloud phase

Cloud Phase 3D for March 2008

CALIPSO-GOCCP water clouds

LATITUDE Very preliminary

(18)

Observa-ons for model evalua-on: ARM Ground * CALIPSO-‐GOCCP sat CERES Sat CLOUDNET Ground * CLOUDSAT Sat ISCCP Sat MISR Sat MODIS Sat *

MULTI-‐SENSORS Analysis Sat * MULTI-‐SENSORS Sat *

PARASOL Sat References***

« News »: recent updates of CFMIP-‐OBS database

Contributors : S. Bony (IPSL/LMD), H. Chepfer (IPSL/LMD), M. Chiriaco (IPSL/LATMOS), J-‐L. Dufresne (IPSL/LMD), N. Loeb (NASA/LarC), S. Klein (LLNL), R. Marchand (Univ. SeaLle), D. Tanré (LOA), M. Webb (UKMO), D. Winker (NASA/LarC), R. Pincus (University of Colorado), Shaocheng Xie (LLNL), Y. Zhang (LLNL)

hLp://climserv.ipsl.polytechnique.fr/fr/cfmip-‐observa-ons.html

[email protected] [email protected]

(19)

(20)

20 La-tude LO N G IT U D E Reﬂectance MODIS 250m LATITUDE 0.8° 0.85° 0.9° 0.95° 1° 1.05° HORIZONTAL DISTRIBUTION 189.65° 198.7° 189.75° 189.8° Need of: a) Instantaneous and colocated obs

b) The highest possible resolu-on

→ the full A-‐Train

capabili-es are required

0.0 3 0.0 1 0.0 7 0.0 5 0.0 9 0.1 1 CALIPSO trace CALIPSO-‐GOCCP VERTICAL DISTRIBUTION 2 4 6 8 AL TIT U D E (k m ) clouds LATITUDE 0.8° 0.85° 0.9° 0.95° 1° 1.05° 0 1 0

(21)

21

1-‐CF

All Sky Reﬂ=0.04

Cloudy Reﬂ=0.07

Clear Sky Reﬂ=0.02

=0.4 PD F PD F 0 0.02 0.04 0.06 0.08 0.1 REFLECTANCE

From case study to sta-s-cal study

conserving the

informa-on contained in high resolu-on obs

or model grid box CALIPSO trace 1 CF 1-‐CF 1° 1° PD F CDF

(22)

High clouds

HighCloud Frac-on SC AT TE RIN G RA TIO

HighCloud Frac-on HighCloud Frac-on

A process oriented descrip-on of oceanic clouds derived from A- train observa-ons, for climate model evalua-on