1 2 chapter 8: climate models and their evaluation 3 4 ... · ipsl-cm4 miroc3.2(hires)...

First-Order Draft Chapter 8 IPCC WG1 Fourth Assessment Report

Do Not Cite or Quote 8-113 Total pages: 42

1 Chapter 8: Climate Models and Their Evaluation 2

3 Coordinating Lead Authors: David Randall, Richard Wood 4 5 Lead Authors: Sandrine Bony, Robert Colman, Thierry Fichefet, John Fyfe, Vladimir Kattsov, Andrew 6 Pitman, Jagadish Shukla, Jayaraman Srinivasan, Ron Stouffer, Akimasa Sumi, Karl Taylor 7 8 Contributing Authors: K. AchutaRao (PCMDI), R. Allan (Univ. Reading), A. Berger, H. Blatter, C. 9 Bonfils (LLNL), A. Boone, C. Bretherton (Univ. Seattle), T. Broccoli, V. Brovkin, W. Cai, M. Claussen, P. 10 Dirmeyer (COLA), C. Doutriaux (PCMDI), H. Drange (BCCR), J.-L. Dufresne (LMD), S. Emori, A. Frei, P. 11 Gent, P. Gleckler (PCMDI), H. Goosse, R. Graham, J. Gregory (CGAM), R. Gudgel (GFDL), A. Hall, S. 12 Hallegatte (METEO-FRANCE), H. Hasumi, A. Henderson-Sellers, H. Hendon, K. Hodges (Univ. Reading), 13 M. Holland (NCAR), B. Holtslag (Wageningen Univ.), E. Hunke, P. Huybrechts, W. Ingram (Oxford), F. 14 Joos, B. Kirtmann, S. Klein (PCMDI), R. Koster (NASA), P. Kushner, J. Lanzante, M. Latif (Univ. Kiel), G. 15 Lau, A.H. Monahan, J. Murphy (Hadley Centre), T. Osborn, T. Pavlova, V. Petoukhov, T. Phillips 16 (PCMDI), S. Power, S. Rahmstorf, S. Raper, H. Renssen, D. Rind, M. Roberts, A. Rosati (GFDL), C. Schär 17 (ETH), J. Scinnoca, A. Schmittner (OSU), D. Seidov, D. Smith, B. Soden (Univ. Miami), W. Stern (GFDL), 18 K.Sudo, T.Takemura, G. Tselioudis (NASA GISS), M. Webb (Hadley Centre), M. Wild (ETH). 19 20 Review Editors: Elisa Manzini, Taroh Matsuno, Bryant McAvaney 21 22 Date of Draft: 12 August 2005 23 24 Notes: This is the TSU compiled version 25 26

27



Figures 1 2 3

4 5 Figure 8.2.1. Long--term mean Sea surface temperature (°C) and ocean current velocities at 100-m 6 depth(vectors, unit: m s–1) around the Kuroshio and the Kuroshio Extension, simulated by MIROC-hi (“HI-7 ocn. HI-atm”) and MIROC-mid (“MID-ocn. MID-atm”) in the control experiment forced by pre-industrial 8 conditions (control-run; averaged for 100 years). Also shown is the result from a model with the atmospheric 9 resolution of MIROC-hi but the ocan resolution of MIROC-mid (“MID-ocn HI-atm”). More structure in the 10 temperature and velocity fields is captured on scales of of a few degree lat/long with the high resolution 11 ocean, even though the mid-resolution ocean could be argued to (marginally) resolve these scales. 12 13

14



1 2

3 4 Figure 8.2.2. The land-atmosphere coupling strength diagnostic for boreal summer (the difference, 5 dimensionless, describing the impact of soil moisture on precipitation), averaged across the 12 models 6 participating in GLACE. (Insets) Areally averaged coupling strengths for the 12 individual models over the 7 outlined, representative hotspot regions. No signal appears in southern South America or at the southern tip 8 of Africa. 9 10

11



(a) 1

-5

-4

-3

-2

-1

0

1

2

3

4

5

-120 -60 0 60 120

-60

-30

0

30

60

216224

256256

256262

268

268

276

276

276

284

284

292

292

300300 300

300

300

300

300

2 (b) 3

0

1

2

3

4

5

6

7

8

-120 -60 0 60 120

-60

-30

0

30

60

4 5 Figure 8.3.1. Observed climatological annual-mean sea surface temperature (SST) and, over land, surface 6 air temperature (labeled contours in panel a) and the multi-model mean error in these temperatures, 7 simulated minus observed (color-shaded contours in panel a); also root-mean-square model error in this 8 temperature, based on all available IPCC model simulations (panel b). The observations are from the CRU 9 merged SST and surface air temperature dataset for the period 1961–1990 (Jones, 1999), and the model 10 results are from years 1980–1999 of the CMIP 20th Century simulations. Temperature units are degrees 11 kelvin (K). 12 13

14



1 2

-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

-120 -60 0 60 120

-60

-30

0

30

60

22

2

2

2 2 2

2

2

4

4 44

4

6

6

6

8

8

10

1010

12

1212

1212

14

1416

3 4 Figure 8.3.2. Observed standard deviation (labeled contours) of sea surface temperature (SST) and, over 5 land, surface air temperature, computed over the climatological monthly mean annual cycle, and the multi-6 model mean error in these temperatures, simulated minus observed (color-shaded contours). The 7 observations are from the CRU merged SST and surface air temperature dataset for the period 1961–1990 8 (Jones, 1999), and the model results are from years 1980–1999 of the CMIP 20th Century simulations. 9 Temperature units are degrees kelvin (K). 10 11

12



1 ' 2

90S6030EQ306090N0

4

8

12

16D

iurn

al T

empe

ratu

re R

ange

(K

)

BCC-CM1BCCR-BCM2.0BCC-CM1BCCR-BCM2.0CCSM3

BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)

CNRM-CM3


CNRM-CM3CSIRO-Mk3.0


CNRM-CM3CSIRO-Mk3.0ECHAM5/MPI-OMECHO-G

FGOALS-g1.0BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


FGOALS-g1.0

GFDL-CM2.0BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOM



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EH



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ER



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0IPSL-CM4



FGOALS-g1.0

GFDL-CM2.0GFDL-CM2.1GISS-AOMGISS-EHGISS-ERINM-CM3.0IPSL-CM4MIROC3.2(hires)



FGOALS-g1.0


MIROC3.2(medres)BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean Model



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (CRU)

3 4 Figure 8.3.3. Diurnal range of surface air temperature, averaged zonally over land areas and averaged 5 annually. The observations are from the CRU surface air temperature dataset for the period 1961–1990 (New 6 et al., 1999), and the model results are from years 1980–1999 of the CMIP 20th Century simulations. Results 7 are not shown where observations are sparse (e.g., Antarctica). 8 9

10



1 2

-7.5

-6

-4.5

-3

-1.5

0

1.5

3

4.5

6

7.5

EQ 30 60 90S90N 60 30

1000

200

10

300

700

500

100 200210

210220 220

220220

240250

260

270

280

290

3 4 Figure 8.3.4. Observed climatological annual-mean air temperature (K), averaged zonally (labeled 5 contours), and the multi-model mean error in this field, simulated minus observed (color-filled contours). 6 The observational estimate is from the 40-year European Reanalysis (ERA40, Uppala et al., 2005) based on 7 observations over the period 1980–1999. The model results are from the same period of the CMIP 20th 8 Century simulations. 9 10

11



1 a) 2

90S6030EQ306090N25

50

75

100

125

150O

utgo

ing

Cle

ar-S

ky S

W (

W/m

^2)

3 4 b) 5

90S6030EQ306090N25

50

75

100

125

150

Out

goin

g S

W R

adia

tion

(W/m

^2)


BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)



CNRM-CM3


CNRM-CM3CSIRO-Mk3.0


CNRM-CM3CSIRO-Mk3.0ECHAM5/MPI-OM







FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (ERBE)

6 7 Figure 8.3.5. Annual-mean, zonally-averaged shortwave radiation scattered and reflected to space under 8 clear-sky conditions (panel a) and under "all-sky" conditions (both clear and cloudy, panel b). The 9 observational estimates are from radiometers flown on satellites during the period 1985–1989 (ERBE, 10 Barkstrom et al., 1989). The model results are from years 1980–1999 of the CMIP 20th Century simulations. 11 12

13



1 2

90S6030EQ306090N0

10

20

30

40O

utgo

ing

SW

Rad

iatio

n (W

/m^2

)





CNRM-CM3


CNRM-CM3CSIRO-Mk3.0









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)


3 4 Figure 8.3.6. Root-mean-square (RMS) model error, as a function of latitude, in simulation of outgoing 5 shortwave radiation scattered and reflected to .space. The RMS error is calculated over all longitudes and 6 over all months. The mean model result is computed by first calculating the multi-model monthly mean 7 fields, and then calculating the RMS error (i.e., it is not the mean of the individual model results). The 8 observational estimates are from radiometers carried by satellites during the period 1985–1989 (ERBE, 9 Barkstrom et al., 1989). The model results are from years 1980–1999 of the CMIP 20th Century simulations. 10 11

12



1 2 (a) 3

90S6030EQ306090N100

150

200

250

300O

utgo

ing

LW R

adia

tion

(W/m

^2)

4 5 (b) 6

90S6030EQ306090N0

5

10

15

20

25

30

35

Out

goin

g LW

Rad

iatio

n (W

/m^2

)





CNRM-CM3


CNRM-CM3CSIRO-Mk3.0









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)


7 8 Figure 8.3.7. Annual-mean, zonally-averaged outgoing longwave radiation at the top of the atmosphere 9 (panel a), and root-mean-square (RMS) model error, as a function of latitude (panel b). The RMS error is 10 calculated over all longitudes and over all months. The mean model result is computed by first calculating 11 the multi-model monthly mean fields, and then calculating the RMS error (i.e., it is not the mean of the 12 individual model results). The observational estimates are from radiometers flown on satellites during the 13 period 1985–1989 (ERBE, Barkstrom et al., 1989), and the model results are from years 1980–1999 of the 14 CMIP 20th Century simulations. 15 16

17



1 2

90S7560453015Eq153045607590N

Latitude

-6

-4

-2

0

2

4

6PW

UKMO-HadGEM1UKMO-HadCM3MRI-CGCM2.3.2ECHAM5/MPI-OMECHO-GMIROC3.2(medres)

INM-CM3.0FGOALS-g1.0

GISS-AOMGFDL-CM2.1CNRM-CM3CGCM3.1(T63)

CGCM3.1(T47)ERBE

3 4 Figure 8.3.8. Total energy transport by the oceans and the atmosphere, averaged zonally and over time, as 5 implied by the net flux of radiation at the top of the atmosphere. The observational estimates are from 6 radiometers flown on satellites during the period 1985–1989 (ERBE, Barkstrom et al., 1989). The model 7 results are from years 1980–1999 of climate of the CMIP 20th Century simulations. 8 9

10



1 2

30S30N90N Eq60N 60S90S0

2

4

6

8T

otal

Pre

cipi

tatio

n R

ate

(mm

/day

)

BCC-CM1BCC-CM1BCCR-BCM2.0BCC-CM1BCCR-BCM2.0CCSM3




CNRM-CM3


CNRM-CM3CSIRO-Mk3.0









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (Xie-Arkin)

3 4 Figure 8.3.9. Time-mean of the zonally-averaged precipitation rate. The observational estimates are from 5 Xie and Arkin (1997) for the period 1979–1993, and the model results are from years 1980–1999 of the 6 CMIP 20th Century simulations. 7 8

9



1 2

(a) 3

0

1

2

3

4

5

6

7

8

9

10

11

60 120 180 240 300

-60

-30

0

30

60

4 5

(b) 6

0

1

2

3

4

5

6

7

8

9

10

11

60 120 180 240 300

-60

-30

0

30

60

7 8 9 Figure 8.3.10. Annual-mean precipitation rate (mm/day), observed (panel a) and model simulated (panel b). 10 The observational estimates are from Xie and Arkin (1997) for the period 1979–1993, and the model results 11 are from years 19801999 of the CMIP 20th Century simulations. 12 13

14



1 2

30S30N90N Eq60N 60S90S0

4

8

12

16T

otal

Pre

cipi

tatio

n R

ate

(mm

/day

)





CNRM-CM3


CNRM-CM3CSIRO-Mk3.0









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (Xie-Arkin)

3 4 Figure 8.3.11. Observed and model simulated precipitation rate in the eastern Pacific averaged over a sector 5 from 120°W to 100°W and averaged over the months of March, April, and May. The observational estimates 6 are from Xie and Arkin (1997) for the period 1979–1993, and the model results are from years 1980–1999 of 7 the CMIP 20th Century simulations. 8 9

10



1 2

-50

-40

-30

-20

-10

0

10

20

30

40

50

EQ 30 60 90S90N 60 30

1000

200

10

300

700

500

100

0.05

0.2

0.4

0.8

1.6

3.2

6.4

12.8

3 4 Figure 8.3.12. Observed specific humidity (g/kg), averaged zonally and annually (labeled contours), and the 5 multi-model mean fractional error in this field, simulated minus observed, divided by observed (color-filled 6 contours). The observational estimate is from the 40-year European Reanalysis (ERA40, Uppala et al., 2005) 7 based on observations over the period 1980–1999. The model results are from the same period of the CMIP 8 20th Century simulations. 9 10

11



1 2

90S6030EQ306090N-80

-60

-40

-20

0

20

40

60

80N

et S

urfa

ce H

eat F

lux

(W/m

^2)




CNRM-CM3









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (COADS)

3 4 Figure 8.3.13. Annual mean, zonally averaged, total surface heat flux into the oceans. The observational 5 estimates are from da Silva (1994), and are based on COADS observations over the period 1945–1989. The 6 model results are from years 1980–1999 of the CMIP 20th Century simulations. 7 8

9



1 2

90S7560453015Eq153045607590N

Latitude

-2

-1

0

1

2

3

PW

UKMO-HadGEM1UKMO-HadCM3MRI-CGCM2.3.2ECHAM5/MPI-OMECHO-GMIROC3.2(medres)INM-CM3.0FGOALS-g1.0

GISS-AOMGFDL-CM2.1CNRM-CM3CGCM3.1(T63)CGCM3.1(T47)ERA (Trenberth)NCEP (Trenberth)

Implied Global Ocean Heat TransportTwentieth Century Simulations: 1980-1999

3 4 Figure 8.3.14. Annual mean, zonally averaged implied oceanic heat transport. The observational estimates 5 are from NCEP and ERA40 reanalyses.The model results are from years 1980–1999 of the CMIP 20th 6 Century simulations. 7 8

9



1 2

90S6030EQ306090N-4

-2

0

2

4

6

8

10W

ater

Flu

x In

to O

cean

(m

m/d

ay)



CNRM-CM3


CNRM-CM3CSIRO-Mk3.0



ECHO-GFGOALS-g1.0



ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0


IPSL-CM4MIROC3.2(hires)MIROC3.2(medres)



ECHO-GFGOALS-g1.0



MRI-CGCM2.3.2



ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0



MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean Model 3

4 Figure 8.3.15. Annual mean net rate of fresh water added to the ocean from: 1) the atmosphere (i.e., 5 precipitation minus evaporation), 2) runoff at continental margins, and 3) any net flux due to imbalances in 6 the freezing and melting of sea ice. Land areas are ignored in computing the zonal means. Model results are 7 from years 1980–1999 of the CMIP 20th Century simulations. 8 9

10



1 2

90S7560453015Eq153045607590N

Latitude

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

10**

9 kg

/s

UKMO-HadGEM1UKMO-HadCM3MRI-CGCM2.3.2MIROC3.2(medres)

INM-CM3.0FGOALS-g1.0

GISS-ERCSIRO-Mk3.0CNRM-CM3CGCM3.1(T63)

Implied Ocean Freshwater TransportTwentieth Century Simulations: 1980-1999

3 4 Figure 8.3.16. Annual mean, zonally averaged implied fresh water transport by the world's oceans. The 5 model results are from years 1980–1999 of the CMIP 20th Century simulations. 6 7

8



1 2

90S6030EQ306090N-0.1

0

0.1

0.2

Zon

al W

ind

Str

ess

at S

urfa

ce (

N/m

^2)

Zonal Wind Stress at Surface: Zonal Mean, Global O

BCC-CM1BCCR-BCM2.0CCSM3


CNRM-CM3


CNRM-CM3CSIRO-Mk3.0









FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean ModelObs (ERA40)

3 4 Figure 8.3.17. Surface zonal wind stress, annually and zonally averaged over the oceans. The 5 observationally-based estimates are from the 40-year European Reanalysis (ERA40, Uppala et al., 2005) for 6 the period 1960–2000.The model results are from years 1980–1999 of climate of the CMIP 20th Century 7 simulations. 8 9

10



1 2

90S6030EQ306090N-4

-3

-2

-1

0

1

2

3

4S

ea S

urfa

ce T

empe

ratu

re (

K)

Difference from HADISST



CNRM-CM3


CNRM-CM3CSIRO-Mk3.0





ECHO-GBCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


ECHO-GFGOALS-g1.0



ECHO-GFGOALS-g1.0

GFDL-CM2.0



ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0




ECHO-GFGOALS-g1.0


IPSL-CM4BCC-CM1BCCR-BCM2.0CCSM3CGCM3.1(T47)CGCM3.1(T63)


ECHO-GFGOALS-g1.0


IPSL-CM4MIROC3.2(hires)



ECHO-GFGOALS-g1.0





ECHO-GFGOALS-g1.0



MRI-CGCM2.3.2



ECHO-GFGOALS-g1.0



MRI-CGCM2.3.2PCM



ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0






ECHO-GFGOALS-g1.0



MRI-CGCM2.3.2PCMUKMO-HadCM3UKMO-HadGEM1Mean Model 3

4 Figure 8.3.18. Zonally averaged, time mean sea surface temperature (SST) error, simulated minus observed. 5 The observations are from the HadISST data sets for the period 1980–1999 (Rayner et al. 2003), and the 6 model results are from years 1980–1999 of climate of the CMIP 20th Century simulations. 7 8

9



1 (a) 2

-5

-4

-3

-2

-1

0

1

2

3

4

5

60 120 180 240 300

-60

-30

0

30

60

276

276276276 276

276

279 282

285

285

288 288

288

288288

291 291 291291 291

291291

294

294

294

294

294294294

297 297

297

297 297

300 300

300300300

300 300300

3 4

(b) 5

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

60 120 180 240 300

-60

-30

0

30

60

6 7 Figure 8.3.19. Observed sea surface temperature (labeled contours, panel a), multi-model mean SST error, 8 simulated minus observed (color-filled contours, panel a), and the root-mean-squared SST error, computed 9 over all models (panel b). Regions with sea ice have been masked because SST is unavailable from most 10 models in these regions. The observations are from the HadISST SST data set for the period 1961–1990 11 (Rayner et al. 2003), and model results are from years 1980–1999 of the CMIP 20th Century simulations. 12 Temperature units are Kelvin (K). 13 14

15



1 2

90S6030EQ306090N-4

-3

-2

-1

0

1

2

3

4S

urfa

ce S

alin

ity (

ppt)

Difference from Levitus





CNRM-CM3


CNRM-CM3CSIRO-Mk3.0





FGOALS-g1.0



FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0




FGOALS-g1.0


MIROC3.2(medres)



FGOALS-g1.0


MIROC3.2(medres)

MRI-CGCM2.3.2PCM



FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)




FGOALS-g1.0


MIROC3.2(medres)


3 4 Figure 8.3.20. Annual mean, zonally averaged, sea surface salinity error (PSU???), simulated minus 5 observed. The observations are from the 2004 World Ocean Atlas (WOA-2004) compiled by Levitus et al. 6 (2005), and model results are from years 1950–1999 of the CMIP 20th Century simulations. 7 8

9



1 2

(a) 3

-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

60 120 180 240 300

-60

-30

0

30

60

30303131 31 31

32

32

32 32

32

33

33

33

33

34

3434 34

34 343434

343434

34

34

3434

343434

353535

35

35 35

3535 35

35

3535

35

35

36

3636

3636

36

37

3739

4 5

(b) 6

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

60 120 180 240 300

-60

-30

0

30

60

7 8 Figure 8.3.21. Observed surface salinity (psu, labeled contours, panel a), multi-model mean surface salinity 9 error, simulated minus observed (color-filled contours, panel a), and the root-mean-squared surface salinity 10 error, computed over all models (panel b). The observations are from the 2004 World Ocean Atlas (WOA-11 2004) compiled by Levitus et al. (2005), and model results are from years 1950–1999 of climate of the 12 CMIP 20th Century simulations. 13 14

15



1 (a) 2

-5

-4

-3

-2

-1

0

1

2

3

4

5

0

200

400

600

800

-5

-4

-3

-2

-1

0

1

2

3

4

5

EQ 30 60 90S90N 60 30

1000

2000

5000

3000

4000

272.5272.5

275277.5

277.5

277.5

280

282.5

285290

272.5

272.5

272.5

275

277.5

3 4

(b) 5

0

0.4

0.8

1.2

1.6

2

2.4

2.8

3.2

3.6

0

200

400

600

800

0

0.4

0.8

1.2

1.6

2

2.4

2.8

3.2

3.6

EQ 30 60 90S90N 60 30

1000

2000

5000

3000

4000

6 7 Figure 8.3.22. Observed climatological annual-mean potential temperature, zonally averaged over all ocean 8 basins (panel a, labeled contours), multi-model mean error in this field, simulated minus observed (color-9 filled contours, panel a), and the root-mean-squared error in this field, computed over all models (panel b). 10 The observations are from the 2004 World Ocean Atlas (WOA-2004) compiled by Levitus et al. (2005), and 11 model results are from years 1950–1999 of the CMIP 20th Century simulations. Temperature units are 12 Kelvin (K). 13

14



1 2

(a) 3

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

0

200

400

600

800

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

EQ 30 60 90S90N 60 30

1000

2000

5000

3000

4000

3434

34.5

34.534.5

34.75

34.75

35

3535.25 35.25

34.5

34.75

34.75

34.75 34.75

34.7535

4 5

(b) 6

0

0.25

0.5

0.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

0

200

400

600

800

0

0.25

0.5

0.75

1

1.25

1.5

1.75

2

2.25

2.5

2.75

3

EQ 30 60 90S90N 60 30

1000

2000

5000

3000

4000

7 8 Figure 8.3.23. Observed climatological annual-mean salinity, zonally averaged over all ocean basins but 9 excluding isolated inland seas and lakes (labeled contours, panel a), multi-model mean error in this field, 10 simulated minus observed (color-filled contours, panel a), and the root-mean-squared error in this field, 11 computed over all models (panel b). The observations are from the 2004 World Ocean Atlas (WOA-2004) 12 compiled by Levitus et al. (2005), and model results are from years 1950–1999 of the CMIP 20th Century 13 simulations. Salinity units are psu. 14 15

16



1 2

-25

-20

-15

-10

-5

0

5

10

15

20

25

0

200

400

600

800

-25

-20

-15

-10

-5

0

5

10

15

20

25

EQ 30 60 90S90N 60 30

1000

2000

5000

3000

4000

3 4 Figure 8.3.24. Zonally averaged merdional streamfunction computed across all basins (Sv). Positive values 5 (brown colors) imply subsidence on the northern side, southward flow below, and ascending water to the 6 south. The flow is in the opposite direction around negative values (blue colors). 7 8

9



1

2 3

4 March September 5 6 Figure 8.3.25. Baseline climate (1980–1999) sea-ice distribution in the Northern (upper panels) and 7 Southern (lower panels) Hemispheres simulated by fourteen of the AOGCMs listed in Table 8.3.1 for March 8 (left) and September (right), adapted from Arzel et al. (2005). For each 2.5° × 2.5° longitude-latitude grid 9 cell, the figure indicates the number of models that have at least 15% of the area covered by sea ice. The 10 observed 15%-concentration boundaries (red line) are based on HadISST (Rayner et al., 2003). 11 12

13



1 2

3 Figure 8.3.26. Baseline climate (1980–1999) terrestrial snow cover distribution in February in the Northern 4 Hemisphere simulated by eight of the AOGCMs listed in Table 8.3.1. For each 2.5° × 2.5° longitude-latitude 5 grid cell, the figure indicates the number of models that have at least 2.5 cm depth of snow cover. The 6 observed 20% area coverage boundaries (red line) are based on observational data available from 7 http://climate.rutgers.edu/snowcover (Robinson and Frei, 2000; Robinson et al., 1993) and averaged over the 8 same time period. 9 10

11



1 2

3 4 Figure 8.3.27. Decadal scale variability (DSV) of observed and modeled North American snow covered area 5 (SCA) based on time series from 11 AOGCMs. DSV is defined as the range (maximum-minimum) of values 6 in the detrended nine-year running mean time series of January NA-SCA for the years 1919–1993. Model 7 number zero shows observed values: B=Brown (2000), F=Frei et al. (1999) are historical reconstructions 8 based on station observations. For each model, the large symbol is the DSV for the ensemble mean. For 9 models with >1 ensemble member, individual ensemble members are shown using smaller symbols. See text 10 for further explanation. Adapted from Frei and Gong (2005). 11 12

13



1 2

3 4 Figure 8.3.28. Global annual mean solar radiation budgets at the surface, in the atmosphere and at the TOA 5 in 20 GCMs participating in AMIPII. 6 7

8



1 2

All Seasons Climatologyglobal statistics

3 4 Figure 8.3.29. Changes in statistics characterizing AMIP model performance, based on the composite multi-5 model median fields (see text for further description). The fields analyzed were: 500 hPa geopotential height 6 (Z500), 200 hPa zonal and meridional wind (U200 and V200), zonal and meridional components of surface wind 7 stress over the oceans (τu and τv), mean sea level pressure over the oceans, (PSL), precipitation (P), cloud 8 fraction (CLT), outgoing longwave radiation (OLR), 200 hPa temperature (T200), 860 hPa specific humidity 9 (Q850), surface air temperature over land, and surface sensible and latent heat flux (SH and LH). Simulated 10 fields were compared to ERA-15 (Gibson et al., 1997), with the following exceptions: precipitation was 11 compared to CPC (Xie and Arkin, 1997), cloud fraction was compared to ISCCP (Shiffer and Rossow, 12 1985), OLR was compared to ERBE (Barkstrom et al., 1989), TAS was compared to CRU (Jones, 1999), 13 and SH, LH, τu , and τv were compared to the SOC Atlas climatology (Josey et al., 1998). 14 15

16



1 2

0 1 2 3Standard Deviation (mm/day)

0

1

2

3

Sta

ndar

d D

evia

tion

(mm

/day

)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.95

0.99

1.0

Correlat ion

0.1

0.20.3

0.4

0.5

0.6

0.7

0.8

0.9

obs.

PrecipitationAll Seasons Climatology

global statistics

median

3 4 Figure 8.3.30. Changes in precipitation statistics for nineteen individual models (and the composite median 5 field derived from the multi-model ensemble). The statistics are the same as those in Figure 8.3.29 and are 6 defined in the text. The observations are from Xie and Arkin (1997). 7 8

9



1 2

3 4 5 6 Figure 8.4.1. Hindcasts of globally averaged annual mean surface temperature, obtained from ensembles of 7 HadCM3 simulations started from analyses of observed ocean and atmosphere anomalies and including 8 anthropogenic and natural forcings (major volcanic eruptions are assumed not to be known about in 9 advance). Simulations were started from 1st March, June, September and December from 1979 to 2001, with 10 three additional ensemble members started from consecutive days preceding each start date. These four 11 simulations were combined with the four simulations started a season earlier to form eight member 12 ensembles. Panel (a) shows hindcast skill as a function of lead time, where skill (S) is defined as one minus 13 the normalised error variance between hindcast and observed anomalies. S = 1 for a perfect hindcast and 14 zero for a hindcast no better than one of zero anomaly. The dashed curve shows the component of skill 15 attributable to internal climate variations, estimated by removing the mean global warming trend from the 16 hindcasts. Panels (b) and (c) show time series of hindcast and observed values for hindcasts one and nine 17 years ahead respectively. The red shading shows the hindcast confidence interval diagnosed from the 18 ensemble standard deviation assuming a t-distribution centred on the ensemble mean (white curve). 19 20

21



1 2

3 4 Figure 8.6.1. Comparison of GCM climate feedback parameters for water vapour (WV), cloud (C), surface 5 albedo (A), lapse rate (LR) and the combined water vapour + lapse rate (WV+LR) in units of W m–2K–1. 6 ‘ALL’ represents the sum of all feedbacks. Results are taken from Colman (2003) (blue), Soden and Held 7 (2005) (red) and Winton (2005) (green). Closed and open symbols from Colman (2003) represent 8 calculations determined using the PRP and the RCM approaches respectively. Crosses represent the water 9 vapour feedback computed for each model from Soden and Held (2005) assuming no change in RH. Vertical 10 bars depict the estimated uncertainty in the calculation of the feedbacks from Soden and Held (2005). 11 12

13



1 2

3 4 Figure 8.6.2. Estimates of water vapour feedback from the cooling associated with Mt Pinatubo, derived 5 from observations and from an ensemble of experiments using HadCM3. The histogram denotes 82 monthly 6 model estimates, shown in terms of probabilities. The shaded curve is a fitted normal distribution to model 7 estimates with the 5% and 95% represented by darker shading. Observed monthly estimates are indicated by 8 the vertical lines. (From Forster and Collins, 2004) 9 10

11



1 2

3 4 Figure 8.6.3. Change in the NET (left panel), SW (middle panel) and LW (right panel) CRF normalized by 5 the change in global mean surface air temperature predicted by AR4 mixed-layer ocean atmosphere models 6 in 2 × CO2 equilibrium experiments. For each panel, results (in W m–2K–1) are shown for global (GL), 7 tropical (30S–30N, TR) and extratropical (EX) areas. The intermodel spread of the CRF response to climate 8 warming primarily arises from different model predictions of the change in tropical SW CRF. Adapted from 9 Webb et al. (2005). 10 11

12



1 2

3 4 Figure 8.6.4 Sensitivity (in W m–2K–1) of the tropical SW cloud radiative forcing to sea surface temperature 5 changes associated with climate change (in 1% per year CO2 increase experiments), derived from 15 AR4 6 ocean-atmosphere models in different regimes of the large-scale tropical circulation (the 500 hPa vertical 7 pressure velocity is used as a proxy for large-scale motions, negative values corresponding to large-scale 8 ascending motion, and positive values to large-scale subsidence). Results are presented for two groups of 9 models: models that predict a positive anomaly of the tropical NET CRF in climate change (in red, 8 models) 10 and models that predict a negative anomaly of the tropical NET CRF (in blue, 7 models). The large 11 intermodel spread of the tropical CRF response to climate change primarily arises from different predictions 12 of the radiative response of boundary-layer clouds in regimes of large-scale subsidence. From Bony and 13 Dufresne (2005). 14 15

16



1 2

3 4 Figure 8.6.5. Scatterplots based on AR4 model output of the simulated springtime snow albedo feedback 5 parameter in the context of external forcing (ordinate) vs. the springtime snow albedo feedback parameter in 6 the context of the seasonal cycle (abscissa) for the Eurasian (left) and North American (right) land masses. 7 The external forcing snow albedo feedback parameter is calculated by dividing the difference in mean April 8 surface albedo averaged over the continents poleward of 30°N between the 22nd and 20th centuries by the 9 difference in mean April surface air temperature between the 22nd and 20th centuries averaged over the 10 same regions. The seasonal cycle snow albedo feedback parameter, based on 20th century climatological 11 means, is calculated by dividing the difference between April and May northern hemisphere continental-12 mean surface albedos between April and May surface air temperature averaged over the same area. A least-13 squares fit regression line for the simulations is also shown. The seasonal cycle feedback parameter was also 14 calculated based on the surface albedo climatology of the 1984–2000 ISCCP data set and the surface air 15 temperature climatology of the ERA40 reanalysis from the same time period. This value is plotted as a 16 vertical line, with the associated shaded region indicating the 95% confidence interval of the estimate due to 17 the shortness of the time series. Numbers, used as plotting symbols, correspond to the following AR4 18 transient climate change experiments: (1) cnrm_cm3, (2) mri_cgcm2_3_2a, (3) giss_model_e_r, (4) 19 csiro_mk3_0, (5) ncar_pcm1, (6) ukmo_hadcm3, (7) cccma_cgcm3_1, (8) iap_fgoals1_0_g, (9) 20 mpi_echam5, (10) ukmo_hadgem1, (11) miub_echo_g, (12) ipsl_cm4, (13) ncar_ccsm3.0, (14) 21 miroc3_2_medres, (15) inmcm3.0, (16) gfdl_cm2_0, (17) gfdl_cm2_1. The Northern Hemisphere snow 22 albedo feedback’s magnitude in the context of the present-day climatological springtime rise in temperatures 23 is highly correlated with its magnitude in the context of centennial-scale human-induced climate change. 24 Therefore if the strength of snow albedo feedback in the present-day seasonal cycle is known for any 25 particular model, its strength in the climate change context can be accurately predicted. Since the real 26 world’s seasonal cycle is well-sampled, the strength of snow albedo feedback in the context of the real 27 seasonal cycle is easily measured and compared to the simulated values. Adapted from Hall and Qu (2005). 28 29

30



1 2

3 4 Figure 8.8.1. Latitudinal distributions of the zonally averaged surface temperature (a, b) and precipitation 5 (c,d) for present-day boreal winter (December, January, February; DJF) (a,c) and boreal summer (June, July, 6 August; JJA) (b,d) as simulated by some of the EMICs used in Chapter 10 of the present report (see Table 7 8.8.2). Observational data are represented by circles and crosses. The vertical gray bars indicate the range of 8 GCM results (see text). Note that the version of LOVECLIM employed in this intercomparison exercise has 9 no biosphere and inland ice components. The version of MIT-IGSM2 also somewhat differs from the one 10 described in Table 8.8.2. (Adapted from Petoukhov et al., 2005). 11 12 13



1 2

3 4 Figure 8.8.2. Differences in globally averaged, annual mean surface temperature (dTg) and precipitation 5 (dPg) between an equilibrium climate adjusted to a doubling of atmospheric CO2 concentration and the pre-6 industrial climate. The coloured dots refer to results obtained by some of the EMICS used in Chapter 10 of 7 the present report (Table 8.8.2), while the grey crosses correspond to results from GCMs published in Le 8 Treut and McAveney (2000). The CLIMBER-2 and CLIMBER-3α results are represented by closed and 9 open red circles, respectively. Note that the version of LOVECLIM utilised in this intercomparison exercise 10 has no biosphere and inland ice components. The version of MIT-IGSM2 also somewhat differs from the 11 one described in Table 8.8.2. (Adapted from Pethoukhov et al., 2005.) 12 13

14



1

2 3 Question 8.1, Figure 1. Globally averaged surface air temperature, from observations, and as simulated by 4 climate models for the instrumental record climate in response to major forcings, natural and anthropogenic. 5 The multiple model lines represent an ensemble of model runs, which together indicate that internal 6 (unforced) variability is unlikely to be a good explanation for the trends simulated (Source: IPCC, 2001). 7 8

1 2 chapter 8: climate models and their evaluation 3 4 ... · ipsl-cm4 miroc3.2(hires)...

Documents