Northern Hemisphere blocking simulation in current climate … · 2020. 6. 15. · Schiemann et...

Supplement of Weather Clim. Dynam., 1, 277–292, 2020https://doi.org/10.5194/wcd-1-277-2020-supplement© Author(s) 2020. This work is distributed underthe Creative Commons Attribution 4.0 License.

Supplement of

Northern Hemisphere blocking simulation in current climatemodels: evaluating progress from the Climate ModelIntercomparison Project Phase 5 to 6 and sensitivity to resolutionReinhard Schiemann et al.

Correspondence to: Reinhard Schiemann ([email protected])

The copyright of individual parts of the supplement might differ from the CC BY 4.0 License.

Schiemann et al.: Blocking in GCMs (Supplement) 1

Supplement

Northern Hemisphere blocking simulation in current climatemodels: evaluating progress from CMIP5 to CMIP6 and sen-sitivity to resolution, Weather and Climate Dynamics, 2020,by Reinhard Schiemann et al.5

List of Figures

S1 Blocking frequency bias for the ANOM in-dex, boreal winter, and (a) high-resolutionforced, (b) high-resolution coupled, (c) low-resolution forced, (d) low-resolution coupled10

PRIMAVERA simulations, and (e) CMIP5,(f) CMIP6 simulations. Stippling showsagreement on the sign of the bias by at least(a,c) 6 of 6, (b,d) 6 of 7, (e) 19 of 29, and (f)10 of 13 simulations. . . . . . . . . . . . . . 415

S2 As Fig. S1 but for boreal summer. (Blockingfrequency bias for the ANOM index and (a)high-resolution forced, (b) high-resolutioncoupled, (c) low-resolution forced, (d) low-resolution coupled PRIMAVERA simula-20

tions, and (e) CMIP5, (f) CMIP6 simula-tions. Stippling shows agreement on the signof the bias by at least (a,c) 6 of 6, (b,d) 6 of7, (e) 19 of 29, and (f) 10 of 13 simulations.) . 5

S3 Metrics of blocking performance (a,b -25

blocking frequency, c,d - spatial correla-tion, e,f - root-mean-square error) for theANOM index and boreal winter, for theATL domain (-90-90E, 50-90N). The left-hand side of each panel shows metrics for30

PRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the right-hand side show distributions of the metricacross CMIP5 and CMIP6 simulations. The’*’ symbol in the column ’ERA/IV’ shows35

the reanalysis estimate and the boxplot is anestimate of the expected agreement given in-ternal variability (see text). . . . . . . . . . . 6

S4 As Fig. S3 but for boreal summer. (Met-rics of blocking performance (a,b - block- 40

ing frequency, c,d - spatial correlation, e,f- root-mean-square error) for the ANOMindex and for the ATL domain (-90-90E,50-90N). The left-hand side of each panelshows metrics for PRIMAVERA simulations 45

at different grid spacings (resolutions). Box-plots on the right-hand side show distribu-tions of the metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate and 50

the boxplot is an estimate of the expectedagreement given internal variability (see text).) 7

S5 As Fig. S3 but for the AGP index and thePAC domain (90-270E, 50-75N). (Metricsof blocking performance (a,b - blocking fre- 55

quency, c,d - spatial correlation, e,f - root-mean-square error) for boreal winter. Theleft-hand side of each panel shows met-rics for PRIMAVERA simulations at dif-ferent grid spacings (resolutions). Boxplots 60

on the right-hand side show distributionsof the metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expected 65

agreement given internal variability (see text).) 8

S6 As Fig. S3 but for the AGP index, the PACdomain (90-270E, 50-75N) and boreal sum-mer. (Metrics of blocking performance (a,b- blocking frequency, c,d - spatial correla- 70

tion, e,f - root-mean-square error). The left-hand side of each panel shows metrics forPRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the right-hand side show distributions of the metric 75

across CMIP5 and CMIP6 simulations. The’*’ symbol in the column ’ERA/IV’ showsthe reanalysis estimate and the boxplot is anestimate of the expected agreement given in-ternal variability (see text).) . . . . . . . . . . 9 80

2 Schiemann et al.: Blocking in GCMs (Supplement)

S7 As Fig. S3 but for the PAC domain (120-240E, 40-90N). (Metrics of blocking perfor-mance (a,b - blocking frequency, c,d - spatialcorrelation, e,f - root-mean-square error) forthe ANOM index and boreal winter. The left-5

hand side of each panel shows metrics forPRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the righ-hand side show distributions of the metricacross CMIP5 and CMIP6 simulations. The10

’*’ symbol in the column ’ERA/IV’ showsthe reanalysis estimate and the boxplot is anestimate of the expected agreement given in-ternal variability (see text).) . . . . . . . . . . 10

S8 As Fig. S3 but for the PAC domain (120-15

240E, 40-90N) and boreal summer. (Met-rics of blocking performance (a,b - block-ing frequency, c,d - spatial correlation, e,f -root-mean-square error) for the ANOM in-dex. The left-hand side of each panel shows20

metrics for PRIMAVERA simulations at dif-ferent grid spacings (resolutions). Boxplotson the righ-hand side show distributions ofthe metric across CMIP5 and CMIP6 sim-ulations. The ’*’ symbol in the column25

’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expectedagreement given internal variability (see text).) 11

S9 Persistence of blocking events (a,b - me-dian, c,d - 90th percentile) for the ANOM30

index and boreal winter, for the ATL domain(-90-90E, 50-90N). The left-hand side ofeach panel shows metrics for PRIMAVERAsimulations at different grid spacings (res-olutions). Boxplots on the right-hand side35

show distributions of the persistence metricacross CMIP5 and CMIP6 simulations. The’*’ symbol in the column ’ERA/IV’ showsthe reanalysis estimate and the boxplot is anestimate of the expected agreement given in-40

ternal variability (see text). . . . . . . . . . . 12

S10 As Fig. S9 but for boreal summer. (Per-sistence of blocking events (a,b - median,c,d - 90th percentile) for the ANOM indexand for the ATL domain (-90-90E, 50-90N).45

The left-hand side of each panel shows met-rics for PRIMAVERA simulations at differ-ent grid spacings (resolutions). Boxplots onthe right-hand side show distributions of thepersistence metric across CMIP5 and CMIP650

simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expectedagreement given internal variability (see text).) 13

S11 As Fig. S9 but for the AGP index and for 55

the PAC domain (90-270E, 50-75N). Per-sistence of blocking events (a,b - median,c,d - 90th percentile) for boreal winter. Theleft-hand side of each panel shows metricsfor PRIMAVERA simulations at different 60

grid spacings (resolutions). Boxplots on theright-hand side show distributions of the per-sistence metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate and 65

the boxplot is an estimate of the expectedagreement given internal variability (see text).) 14

S12 As Fig. S9 but for the AGP index, for thePAC domain (90-270E, 50-75N), and for bo-real summer. (Persistence of blocking events 70

(a,b - median, c,d - 90th percentile). Theleft-hand side of each panel shows metricsfor PRIMAVERA simulations at differentgrid spacings (resolutions). Boxplots on theright-hand side show distributions of the per- 75

sistence metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expectedagreement given internal variability (see text).) 15 80

S13 As Fig. S9 but for the PAC domain (120-240E, 40-90N). (Persistence of blockingevents (a,b - median, c,d - 90th percentile)for the ANOM index and boreal winter. Theleft-hand side of each panel shows metrics 85

for PRIMAVERA simulations at differentgrid spacings (resolutions). Boxplots on theright-hand side show distributions of the per-sistence metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column 90

’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expectedagreement given internal variability (see text).) 16

S14 As Fig. S9 but for the PAC domain (120-240E, 40-90N) and boreal summer. (Persis- 95

tence of blocking events (a,b - median, c,d- 90th percentile) for the ANOM index. Theleft-hand side of each panel shows metricsfor PRIMAVERA simulations at differentgrid spacings (resolutions). Boxplots on the 100

right-hand side show distributions of the per-sistence metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column’ERA/IV’ shows the reanalysis estimate andthe boxplot is an estimate of the expected 105

agreement given internal variability (see text).) 17


S15 Decomposition of blocking bias into (i) aterm due to the bias in the number of events(E), (ii) a term due to the bias in persis-tence (P), and (iii) a cross term (C). Boxplotsshow the ensemble spread across the CMIP5,5

CMIP6, and all coupled PRIMAVERA mod-els, respectively. The AGP index with 5-daypersistence is used and results are shown forthe ATL and PAC domains, and for borealwinter (DJF) and summer (JJA). The block-10

ing bias is normalised by the reanalysis esti-mate of blocking frequency. . . . . . . . . . . 17

List of Tables

S1 CMIP5 simulations used in this study. Ther1i1p1 ensemble member from the historical15

experiment is used for each simulation. . . . 18S2 CMIP6 simulations used in this study. All

simulations are for the historical experimentduring 1950–2014. . . . . . . . . . . . . . . 18


PRIMAVERA HF(x6) ANOM

a DJF

PRIMAVERA HC(x7) ANOM

b DJF

PRIMAVERA LF(x6) ANOM

c DJF

PRIMAVERA LC(x7) ANOM

d DJF

CMIP5(x29) ANOM

e DJF

CMIP6(x13) ANOM

f DJF

-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06

Figure S1. Blocking frequency bias for the ANOM index, boreal winter, and (a) high-resolution forced, (b) high-resolution coupled, (c) low-resolution forced, (d) low-resolution coupled PRIMAVERA simulations, and (e) CMIP5, (f) CMIP6 simulations. Stippling shows agreementon the sign of the bias by at least (a,c) 6 of 6, (b,d) 6 of 7, (e) 19 of 29, and (f) 10 of 13 simulations.


PRIMAVERA HF(x6) ANOM

a JJA

PRIMAVERA HC(x7) ANOM

b JJA

PRIMAVERA LF(x6) ANOM

c JJA

PRIMAVERA LC(x7) ANOM

d JJA

CMIP5(x29) ANOM

e JJA

CMIP6(x13) ANOM

f JJA

-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06

Figure S2. As Fig. S1 but for boreal summer. (Blocking frequency bias for the ANOM index and (a) high-resolution forced, (b) high-resolution coupled, (c) low-resolution forced, (d) low-resolution coupled PRIMAVERA simulations, and (e) CMIP5, (f) CMIP6 simulations.Stippling shows agreement on the sign of the bias by at least (a,c) 6 of 6, (b,d) 6 of 7, (e) 19 of 29, and (f) 10 of 13 simulations.)


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IVatm. grid spacing at 50°N (km)

0.00

0.02

0.04

0.06

0.08

0.10

0.12

frequ

ency

of b

lock

ed d

ays

ANOM, ATL, DJF, coupled

a


a

AWICMCCCNRMEC-EarthECMWFMOHCMPI-M

0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV

atm. grid spacing at 50°N (km)

0.00

0.02

0.04

0.06

0.08

0.10

0.12

frequ

ency

of b

lock

ed d

ays

ANOM, ATL, DJF, AMIP

b


b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2sp

atia

l cor

rela

tion


d


d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.00

0.01

0.02

0.03

0.04

0.05

0.06

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.00

0.01

0.02

0.03

0.04

0.05

0.06

root

-mea

n-sq

uare

erro

r


f


f


Figure S3. Metrics of blocking performance (a,b - blocking frequency, c,d - spatial correlation, e,f - root-mean-square error) for the ANOMindex and boreal winter, for the ATL domain (-90-90E, 50-90N). The left-hand side of each panel shows metrics for PRIMAVERA simu-lations at different grid spacings (resolutions). Boxplots on the right-hand side show distributions of the metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreementgiven internal variability (see text).


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/


0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

frequ

ency

of b

lock

ed d

ays

ANOM, ATL, JJA, coupled

a


a


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

frequ

ency

of b

lock

ed d

ays

ANOM, ATL, JJA, AMIP

b


b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

nANOM, ATL, JJA, AMIP

d


d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

root

-mea

n-sq

uare

erro

r


f


f


Figure S4. As Fig. S3 but for boreal summer. (Metrics of blocking performance (a,b - blocking frequency, c,d - spatial correlation, e,f -root-mean-square error) for the ANOM index and for the ATL domain (-90-90E, 50-90N). The left-hand side of each panel shows metricsfor PRIMAVERA simulations at different grid spacings (resolutions). Boxplots on the right-hand side show distributions of the metric acrossCMIP5 and CMIP6 simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of theexpected agreement given internal variability (see text).)


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/


0.000

0.005

0.010

0.015

0.020

0.025

0.030

frequ

ency

of b

lock

ed d

ays

AGP, PAC, DJF, coupled

a


a


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

frequ

ency

of b

lock

ed d

ays

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.70

0.75

0.80

0.85

0.90

0.95

1.00

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.70

0.75

0.80

0.85

0.90

0.95

1.00

spat

ial c

orre

latio

nAGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.0000

0.0025

0.0050

0.0075

0.0100

0.0125

0.0150

0.0175

0.0200

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.0000

0.0025

0.0050

0.0075

0.0100

0.0125

0.0150

0.0175

0.0200

root

-mea

n-sq

uare

erro

r

AGP, PAC, DJF, AMIP

f

AGP, PAC, DJF, AMIP

f


Figure S5. As Fig. S3 but for the AGP index and the PAC domain (90-270E, 50-75N). (Metrics of blocking performance (a,b - blockingfrequency, c,d - spatial correlation, e,f - root-mean-square error) for boreal winter. The left-hand side of each panel shows metrics forPRIMAVERA simulations at different grid spacings (resolutions). Boxplots on the right-hand side show distributions of the metric acrossCMIP5 and CMIP6 simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of theexpected agreement given internal variability (see text).)


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

frequ

ency

of b

lock

ed d

ays

AGP, PAC, JJA, coupled

a


a


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

frequ

ency

of b

lock

ed d

ays

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00sp

atia

l cor

rela

tion

AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

root

-mea

n-sq

uare

erro

r

AGP, PAC, JJA, AMIP

f

AGP, PAC, JJA, AMIP

f


Figure S6. As Fig. S3 but for the AGP index, the PAC domain (90-270E, 50-75N) and boreal summer. (Metrics of blocking performance (a,b- blocking frequency, c,d - spatial correlation, e,f - root-mean-square error). The left-hand side of each panel shows metrics for PRIMAVERAsimulations at different grid spacings (resolutions). Boxplots on the right-hand side show distributions of the metric across CMIP5 and CMIP6simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreementgiven internal variability (see text).)


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/


0.00

0.02

0.04

0.06

0.08

0.10

0.12

frequ

ency

of b

lock

ed d

ays

ANOM, PAC, DJF, coupled

a


a


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.00

0.02

0.04

0.06

0.08

0.10

0.12

frequ

ency

of b

lock

ed d

ays

ANOM, PAC, DJF, AMIP

b


b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

nANOM, PAC, DJF, AMIP

d


d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

root

-mea

n-sq

uare

erro

r


f


f


Figure S7. As Fig. S3 but for the PAC domain (120-240E, 40-90N). (Metrics of blocking performance (a,b - blocking frequency, c,d -spatial correlation, e,f - root-mean-square error) for the ANOM index and boreal winter. The left-hand side of each panel shows metrics forPRIMAVERA simulations at different grid spacings (resolutions). Boxplots on the righ-hand side show distributions of the metric acrossCMIP5 and CMIP6 simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of theexpected agreement given internal variability (see text).)


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/


0.00

0.01

0.02

0.03

0.04

0.05

0.06

frequ

ency

of b

lock

ed d

ays

ANOM, PAC, JJA, coupled

a


a


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.00

0.01

0.02

0.03

0.04

0.05

0.06

frequ

ency

of b

lock

ed d

ays

ANOM, PAC, JJA, AMIP

b


b


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

n


c


c


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.2

0.4

0.6

0.8

1.0

1.2

spat

ial c

orre

latio

nANOM, PAC, JJA, AMIP

d


d


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

root

-mea

n-sq

uare

erro

r


e


e


0 25 50 75 100

125

CMIP

5CM

IP6

ERA/

IV


0.000

0.005

0.010

0.015

0.020

0.025

0.030

root

-mea

n-sq

uare

erro

r


f


f


Figure S8. As Fig. S3 but for the PAC domain (120-240E, 40-90N) and boreal summer. (Metrics of blocking performance (a,b - blockingfrequency, c,d - spatial correlation, e,f - root-mean-square error) for the ANOM index. The left-hand side of each panel shows metrics forPRIMAVERA simulations at different grid spacings (resolutions). Boxplots on the righ-hand side show distributions of the metric acrossCMIP5 and CMIP6 simulations. The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of theexpected agreement given internal variability (see text).)


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


b


b


b


b


b


b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, ATL, DJF, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, ATL, DJF, AMIP

d


d


d


d


d


d


Figure S9. Persistence of blocking events (a,b - median, c,d - 90th percentile) for the ANOM index and boreal winter, for the ATL domain(-90-90E, 50-90N). The left-hand side of each panel shows metrics for PRIMAVERA simulations at different grid spacings (resolutions).Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations. The ’*’ symbol in thecolumn ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internal variability (see text).


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


b


b


b


b


b


b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, ATL, JJA, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, ATL, JJA, AMIP

d


d


d


d


d


d


Figure S10. As Fig. S9 but for boreal summer. (Persistence of blocking events (a,b - median, c,d - 90th percentile) for the ANOM indexand for the ATL domain (-90-90E, 50-90N). The left-hand side of each panel shows metrics for PRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations.The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internalvariability (see text).)


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.8

2.0

2.2

2.4

2.6

2.8

3.0

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.8

2.0

2.2

2.4

2.6

2.8

3.0

med

ian

bloc

king

per

siste

nce

(day

s)

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b

AGP, PAC, DJF, AMIP

b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) AGP, PAC, DJF, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) AGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d

AGP, PAC, DJF, AMIP

d


Figure S11. As Fig. S9 but for the AGP index and for the PAC domain (90-270E, 50-75N). Persistence of blocking events (a,b - median, c,d- 90th percentile) for boreal winter. The left-hand side of each panel shows metrics for PRIMAVERA simulations at different grid spacings(resolutions). Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations. The ’*’symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internalvariability (see text).)


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


2.0

2.2

2.4

2.6

2.8

3.0

3.2

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


2.0

2.2

2.4

2.6

2.8

3.0

3.2

med

ian

bloc

king

per

siste

nce

(day

s)

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b

AGP, PAC, JJA, AMIP

b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) AGP, PAC, JJA, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d

AGP, PAC, JJA, AMIP

d


Figure S12. As Fig. S9 but for the AGP index, for the PAC domain (90-270E, 50-75N), and for boreal summer. (Persistence of blockingevents (a,b - median, c,d - 90th percentile). The left-hand side of each panel shows metrics for PRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations.The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internalvariability (see text).)


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

med

ian

bloc

king

per

siste

nce

(day

s)


b


b


b


b


b


b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, PAC, DJF, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


6

8

10

12

14

16

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, PAC, DJF, AMIP

d


d


d


d


d


d


Figure S13. As Fig. S9 but for the PAC domain (120-240E, 40-90N). (Persistence of blocking events (a,b - median, c,d - 90th percentile) forthe ANOM index and boreal winter. The left-hand side of each panel shows metrics for PRIMAVERA simulations at different grid spacings(resolutions). Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations. The ’*’symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internalvariability (see text).)


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

med

ian

bloc

king

per

siste

nce

(day

s)


a


a


a


a


a


a


a


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

med

ian

bloc

king

per

siste

nce

(day

s)


b


b


b


b


b


b


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


4

6

8

10

12

14

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, PAC, JJA, coupled

c


c


c


c


c


c


c


0 25 50 75 100

125

CMIP

5

CMIP

6

ERA/

IV


4

6

8

10

12

14

90th

per

cent

ile b

lock

ing

pers

isten

ce (d

ays) ANOM, PAC, JJA, AMIP

d


d


d


d


d


d


Figure S14. As Fig. S9 but for the PAC domain (120-240E, 40-90N) and boreal summer. (Persistence of blocking events (a,b - median,c,d - 90th percentile) for the ANOM index. The left-hand side of each panel shows metrics for PRIMAVERA simulations at different gridspacings (resolutions). Boxplots on the right-hand side show distributions of the persistence metric across CMIP5 and CMIP6 simulations.The ’*’ symbol in the column ’ERA/IV’ shows the reanalysis estimate and the boxplot is an estimate of the expected agreement given internalvariability (see text).)

E P C E P C E P C E P C E P C E P C E P C E P C E P C E P C E P C E P C0.8

0.6

0.4

0.2

0.0

0.2

0.4

0.6

0.8

rel.

bloc

king

bia

s

ATL, DJF ATL, JJA PAC, DJF PAC, JJA

CMIP5CMIP6PRIMAVERA

Figure S15. Decomposition of blocking bias into (i) a term due to the bias in the number of events (E), (ii) a term due to the bias inpersistence (P), and (iii) a cross term (C). Boxplots show the ensemble spread across the CMIP5, CMIP6, and all coupled PRIMAVERAmodels, respectively. The AGP index with 5-day persistence is used and results are shown for the ATL and PAC domains, and for borealwinter (DJF) and summer (JJA). The blocking bias is normalised by the reanalysis estimate of blocking frequency.


Table S1. CMIP5 simulations used in this study. The r1i1p1 ensemble member from the historical experiment is used for each simulation.

No. Centre Model Period

1 CSIRO-BOM ACCESS1-0 1950–20052 CSIRO-BOM ACCESS1-3 1950–20053 BCC BCC-CSM1-1 1950–19974 BCC BCC-CSM1-1-M 1950–20055 BNU BNU-ESM 1950–20056 CCCma CanESM2 1950–20057 CMCC CMCC-CESM 1950–20058 CMCC CMCC-CM 1950–20059 CMCC CMCC-CMS 1950–200510 CNRM-CERFACS CNRM-CM5 1950–200511 EC-Earth-Consortium EC-EARTH 1950–200512 LASG-CESS FGOALS-g2 1950–200513 NOAA-GFDL GFDL-CM3 1950–200514 NOAA-GFDL GFDL-ESM2G 1950–200515 MOHC HadCM3 1960–200516 MOHC HadGEM2-CC 1950–200517 MOHC HadGEM2-ES 1981–200518 IPSL IPSL-CM5A-LR 1950–200519 IPSL IPSL-CM5A-MR 1950–200520 IPSL IPSL-CM5B-LR 1950–200521 MIROC MIROC5 1950–200522 MIROC MIROC-ESM-CHEM 1950–200523 MIROC MIROC-ESM 1950–200524 MPI-M MPI-ESM-LR 1950–200525 MPI-M MPI-ESM-MR 1950–200526 MPI-M MPI-ESM-P 1950–200527 MRI MRI-CGCM3 1950–200528 MRI MRI-ESM1 1950–200529 NCC NorESM1-M 1950–2005

Table S2. CMIP6 simulations used in this study. All simulations are for the historical experiment during 1950–2014.

No. Centre Model Ensemble member

1 BCC BCC-CMS2-MR r1i1p1f12 BCC BCC-ESM1 r1i1p1f13 CCCma CanESM5 r1i1p1f14 CNRM-CERFACS CNRM-CM6-1 r2i1p1f25 CNRM-CERFACS CNRM-ESM2-1 r3i1p1f26 IPSL IPSL-CM6A-LR r1i1p1f17 MOHC HadGEM3-GC31-LL r1i1p1f38 MOHC UKESM1-0-LL r1i1p1f29 MRI MRI-ESM2-0 r1i1p1f110 NASA-GISS GISS-E2-1-G r1i1p1f111 NCAR CESM2 r1i1p1f112 NCAR CESM2-WACCM r2i1p1f113 NOAA-GFDL GFDL-CM4 r1i1p1f1

Northern Hemisphere blocking simulation in current climate … · 2020. 6. 15. · Schiemann et...

Documents

Transcript of Northern Hemisphere blocking simulation in current climate … · 2020. 6. 15. · Schiemann et...