Atmospheric Hydrological Cycle in the Tropics in Twentieth Century Coupled Climate Simulations

Atmospheric Hydrological Cycle in the Tropics in Twentieth Century Coupled Climate Simulations

Hailan Wang and William Lau

Laboratory for Atmospheres, NASA/GSFC

30th Climate Diagnostics and Prediction WorkshopOctober 26, 2005

Climate Model Evaluation Project (CMEP)

Motivation

• Identify and understand long-term change of tropical hydrological cycle in 20th Century climate simulations by Coupled GCMs

• Precipitation• Clouds

• Provide input for IPCC AR4 in 2007

Coupled GCMs

• State-of-the-art

• Fully coupled– Time signature differs

• Driven by time-varying external climate forcings– No agreed-upon forcing functions – The diversity in external forcing in the CGCMs is

regarded as a measure of forcing uncertainties

http://www.giss.nasa.gov/research/modeling/

NASA GISS_E

Coupled GCMs (Cont’d)

– 16 CGCMs analyzed

– 1 run of each CGCM used• Monthly mean fields

• 1900-1999

– Linear trend (actual linear change over time period concerned)

• Models: 1950-1999

• Observations– 1979-1999 for GPCP precip

– 1984-1999 for ISCCP clouds

Model No.

Model Acronym Modeling group Country Atmosphere Resolution Ocean Resolution at Equator

1 CGCM3.1(T47) CCCMA Canada T47L31 1.851.85L29

2 CNRM-CM3 CNRM France T63L45 20.5L31

3 CSIRO-Mk3.0 CSIRO Australia T63L18 1.8750.84L31

4 GFDL-CM2.0 GFDL USA 2.52L24 11/3L50

5 GFDL-CM2.1 GFDL USA 2.52L24 11/3L50

6 GISS-EH NASA/GISS USA 54L20 22L16

7 GISS-ER NASA/GISS USA 54L20 54L13

8 FGOALS-g1.0 LASG/IAP China T42L26 11L33

9 INM-CM3.0 INM Russia 54L21 2.52L33

10 IPSL-CM4 IPSL France 3.752.5L19 21L31

11 MIROC3.2(hires) CCSR/NIES/FRCGC Japan T106L56 0.280.1875L47

12 MIROC3.2(medres) CCSR/NIES/FRCGC Japan T42L20 1.40.5L43

13 ECHAM5 MPI Germany T63L31 1.51.5L40

14 CCSM3 NCAR USA T85L26 1.1250.27L40

15 UKMO-HadCM3 UKMO UK 3.752.5L19 1.251.25L20

16 UKMO-HadGEM1 UKMO UK 1.8751.25L38 11/3L40

USA: 5; France: 2; Japan: 2; UK: 2; Australia: 1; Canada: 1; China:1; Germany: 1; Russia: 1

HadCRU (#17); NCEP CAMS (#18); NOAA extended SST (#19)

Linear Change of Surface TempAnnual Mean 1950-1999

GPCP (#17)

Linear Change of PrecipAnnual Mean 1979-1999

Linear Trend of Surface Temp (1950-99) and Precip (1979-99)

16 AR4 Model EnsMean Obs

Rain Rate

Light: <1mm/day

Medium: 2-8mm/day

Heavy: >9mm/day

Distribution of GPCP Rain as a function of Rain RateAnnual Mean 1979-1999; Tropical Ocean

Clim

Trend

GFDL CM2.0

NASA GISS ER

MIROC3.2 hires

NCAR CCSM3

UKMO HadCM3

GPCP

Trend_Model*4

ISCCP (#17)

Linear Change of Total Cloud Cover

Models (1950-1999) vs ISCCP/4 (1984-1999)

[1000mb-10mb]

[30S-30N]

[0-360E; 30S-30N]

Clim and Linear Trend of 3-D Cloud in GFDL CM2.0

ClimLinear Trend

Upward motion enhances

OLR reduces

Chen et al (2002); Wielicki et al (2002)

Linear Change over 1950-1999

500mb

OLR at TOA

Cool ClimateTropical Ocean

TropopauseOLR

Surface Evaporation

Tropopause

OLR

More cold and bright high cloud at tropopause and lower stratosphere

Less mid-to-low cloud

Less high cloud

Less OLR

Enhanced heavy rain

Greatly reducedmoderate rain

Increased light rain

Intensified deep convection

Strengthened updraft

Ocean surface

Warmer Tropical OceanWarm ClimateOcean surface

Enhanced Surface Evaporation

Conclusions• CGCMs are reasonably consistent in depicting aspects of long term

changes in the 20th Century climate and the tropical hydrological cycle:

– Surface warming over tropical ocean and land– Increasing/decreasing precipitation over tropical ocean/land– Increasing heavy and light rain, but decreasing moderate rain– A reduction in total cloud cover in the tropics.

• CGCMs significantly underestimate the magnitudes of the observations, by a factor of at least 4.

– This likely leads to gross errors in model simulation of tropical radiative fluxes

• Difference between CGCMs and observations– Time scale– CGCMs: certain processes may be missing– Obs: e.g. problems in retrieving high level thin clouds

Future

• Improvement in representing physical processes associated with clouds and their interaction with radiation in the CGCMs

• Observational data– Quality– Long term