a revision of the earth’s energy balance framework - nasa · a revision of the earth’s energy...

A revision of the Earth’s energy balance framework

A. E. DesslerDept. of Atmospheric Sciences

Texas A&M University

T. Mauritsen and B. StevensMax Planck Institute for Meteorology

abrupt 4xCO2 GFDL-CM3

Year

∆R∆T

s

abrupt 4xCO2 GFDL-CM3

Year

∆R∆T

s

abrupt 4xCO2 GFDL-CM3

Year

∆R∆T

s

abrupt 4xCO2 GFDL-CM3

Year

∆F∆R∆T

s

abrupt 4xCO2 GFDL-CM3

∆Ts

∆R

Year

Year

abrupt 4xCO2 GFDL-CM3

∆R∆T

s

Year

abrupt 4xCO2 GFDL-CM3

∆R

∆R = ∆F + λ ∆Ts

∆Ts

∆R = ∆F + λ ∆Ts = 0

At equilibrium:

∆R = ∆F + λ ∆Ts

∆Ts = -∆F/λ

= 0

At equilibrium:

∆R = ∆F + λ ∆Ts

∆Ts = -∆F/λ

= 0

At equilibrium:

∆T2xCO2 = -∆F2xCO2/λ = ECS

∆R = ∆F + λ ∆Ts

∆Ts = -∆F/λ

= 0

At equilibrium:

∆T2xCO2 = -∆F2xCO2/λ = ECS

Q: is this the right way to describe our planet’s energy balance?

∆R = ∆F + λ ∆Ts

∆Ts = -∆F/λ

= 0

At equilibrium:

∆T2xCO2 = -∆F2xCO2/λ = ECS

Q: is this the right way to describe our planet’s energy balance?

Q: is the response of the TOA flux linear in global avg. surface

temperature?

6

satellite obs. record (2000-2016)

7

satellite obs. record (2000-2016)

∆R = ΔF + λ ΔTS

7

satellite obs. record (2000-2016)

∆R = ΔF + λ ΔTS CERES

7

satellite obs. record (2000-2016)

∆R = ΔF + λ ΔTS obs.CERES

7

satellite obs. record (2000-2016)

∆R = ΔF + λ ΔTS obs.CERES

small

∆R

∆Ts

∆R = ΔF + λ ΔTS

∆R

∆Ts

∆R = ΔF + λ ΔTS

∆R

∆Ts

Slope = λ (W/m2/K)

∆R = ΔF + λ ΔTS

CER

ES E

d. 4

, EBA

F

ERAi

Global, monthly avg., detrended, 3/2000-9/2016

CER

ES E

d. 4

, EBA

F

ERAi

Global, monthly avg., detrended, 3/2000-9/2016

CER

ES E

d. 4

, EBA

F

ERAi

Global, monthly avg., 3/2000-9/2016Lesson: λ∆T is not a very good parameterization for ∆R (for

interannual variability)

CER

ES E

d. 4

, EBA

F

ERAi

Global, monthly avg., 3/2000-9/2016Lesson: λ∆T is not a very good parameterization for ∆R (for

interannual variability)

Should not be surprising: 90% of TOA LW radiation emitted

by atmosphere, not surface

12

Derive ECS & λ from 20th century record

∆R = ΔF + λ ΔTS

12

Derive ECS & λ from 20th century record

∆R = ΔF + λ ΔTS obs.

12

Derive ECS & λ from 20th century record

∆R = ΔF + λ ΔTS climate model

obs.

12

Derive ECS & λ from 20th century record

∆R = ΔF + λ ΔTS climate model

obs.OHC

13

∆R = ΔF + λ ΔTS climate model

obs.

λ ≈ -2 W/m2/K ECS ≈ 1.5-2.0 K

OHC

Derive λ from 20th century record

13

∆R = ΔF + λ ΔTS climate model

obs.

λ ≈ -2 W/m2/K ECS ≈ 1.5-2.0 K

OHC

This anchors the low end of the IPCC's ECS range to 1.5 K

Derive λ from 20th century record

14

∆R = ΔF + λ ΔTS

ΔF obs.

ΔR ΔTS

modelssame same same

λ ≈ -2 W/m^2 λ ≈ -1.3ECS < 2 K ECS 2-4.5 K

14

∆R = ΔF + λ ΔTS

ΔF obs.

ΔR ΔTS

modelssame same same

λ ≈ -2 W/m^2 λ ≈ -1.3ECS < 2 K ECS 2-4.5 K

Perfect model experiment

• 100-member MPI-ESM1.1 ensemble – model producing output for CMIP6

• 1850-2005; identical historical forcing • only difference is the initial conditions

15

16

17

For each ensemble member …

∆R = ΔF + λ ΔT

17

For each ensemble member …

∆R = ΔF + λ ΔT

solve for this

18

λ = ∆(R-F)/∆Ts W/m2/K

18

λ = ∆(R-F)/∆Ts W/m2/K

internal variability

19

λ = ∆(R-F)/∆Ts ECS = -∆F2xCO2/λ

19

λ = ∆(R-F)/∆Ts ECS = -∆F2xCO2/λ

19

λ = ∆(R-F)/∆Ts ECS = -∆F2xCO2/λ

We only have one realization of the 20th century, so there’s no guarantee that λ and ECS based on these

calculations are good estimates

20

revised framework

∆R = ΔF + Θ ΔTA

20

revised framework

∆R = ΔF + Θ ΔTA 500-hPa tropical

temperatures

20

revised framework

∆R = ΔF + Θ ΔTA 500-hPa tropical

temperatures

following Trenberth, Murphy, Spencer

20

revised framework

∆R = ΔF + Θ ΔTA 500-hPa tropical

temperaturesConverts change

in TA to flux; replaces λ

following Trenberth, Murphy, Spencer

CERES Ed. 4 TOA fluxes & ERA-interim temperatures monthly avg. detrended anomalies

Ts = global avg. surface temperature TA = tropical avg. 500-hPa temperature

22

λ = ∆(R-F)/∆TSΘ = ∆(R-F)/∆TA

22

λ = ∆(R-F)/∆TSΘ = ∆(R-F)/∆TA

confirmed in CMIP5 control runs

23

∆R = ΔF + Θ ΔTA

• Can we use our new framework to better predict ECS?

23

∆R = ΔF + Θ ΔTA

• Can we use our new framework to better predict ECS?

• Use measurement of Θcontrol from CERES to constrain ECS

23

∆R = ΔF + Θ ΔTA

• Can we use our new framework to better predict ECS?

• Use measurement of Θcontrol from CERES to constrain ECS

• In CMIP5 ensemble, Θcontrol ≈ Θabrupt4xCO2

25

∆R = ΔF + Θ ΔTA

3.45 K

3.45 K

13 of 15

3.45 K3 of 13

3.45 K

3.45 K

ECS < 3.5 K

• the conventional energy balance framework has problems — TOA flux is not strongly related to surface temperature

• creates problems inferring ECS from historical data

• new framework using atmospheric temperature: ECS < 3.5 K

31

Conclusions

backup slides

32

CERES ∆R at each latitude vs. ∆Ts at that same latitude detrended monthly avg. anomalies

global vs. tropical T500

35

20th century obs. record

∆R = ΔF + λ ΔT

35

20th century obs. record

∆R = ΔF + λ ΔT based on fixed SST runs

35

20th century obs. record

∆R = ΔF + λ ΔT based on fixed SST runs

ensemble

35

20th century obs. record

∆R = ΔF + λ ΔT based on fixed SST runs

ensembleensemble

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

∆T

∆R

CMIP5 control runs (NorESM1-M)

months

37

38

CCSM4 CMIP5 control run

λ = ∆(R-F)/∆Ts

Θ = ∆(R-F)/∆T500t

39

CMIP5 control runs

5-95% percentile range

2.4 K

3.45 K

2.4 K

3.45 K

2.4 K

3.45 K

2.4 K

3.45 K 3 of 13 models

2.4 K

3.45 K

11 of 11 models