Assessing Radiative Forcings in
CMIP5
Brian Soden
Rosenstiel School for Marine and Atmospheric Science
University of Miami
CMIP5 Radiative Concentration Pathways
Abrupt 4x CO2
Motivation

Models do not routinely archive (or compute) radiative
forcing for idealized experiments (e.g., 2xCO2) or 21st
Century projections.

There is little understanding of how much uncertainty
in model projections of climate change results from
differences in radiative forcing.
Methodology

Decompose net radiative flux response at TOA into
contributions from troposphere and stratosphere using
radiative kernels.

Separate feedbacks from forcings by regressing
against global mean temperature.
 “Radiative Feedbacks” are correlated to temperature.


“Radiative Adjustments” are not (e.g., stratospheric cooling).
Use CMIP5 “Abrupt 4xCO2” Scenario to uncorrelate
the evolution of forcing from surface warming.
Decomposing TOA Radiative Flux Changes
TOA
=
Imbalance
(from GCM)
Radiative Feedbacks
(correlated to Ts)
+
Radiative
+
Adjustments
(not correlated to Ts)

Compute the radiative contributions for feedbacks from
Temperature (T), Water Vapor (W), Surface Albedo (a),
Clouds (C) using “Radiative Kernels”

As well as non-feedback “Fast” adjustments e.g., from
stratospheric (St) temperature or water vapor.

The remaining term required to reproduce the imbalance in
TOA radiation (R) is the instantaneous radiative forcing (G).
Forcing
R  RT  RW  Ra  Rc  RSt,T  RSt,W  G.
Radiative Kernels
R dT
R dW R dC R d




T dTs W dTs C dTs  dTs
R: Net TOA Flux
T/C/W/: Feedback variables
Radiative Kernel: Sensitivity of TOA Flux to FB variable
Climate Response: Sensitivity of Feedback variable to Ts
The climate response dX/dTs is further separated into the climate feedback
X/Ts using linear regression:
X=a + b Ts
X/Ts = b
•Define the “feedback” as the component which is correlated to Ts.
•Define the “adjustment” as any remaining non-correlated change in X.
Comparison of 2xCO2 Forcing: Kernel vs Tropopause
Comparison of 4xCO2 Forcing: Kernel vs Gregory
ENSEMBLE MEAN (clear-sky): Abrupt 4xCO2
Finite Differencing
Regression
8
8
6
6
6
4
4
4
2
2
2
TOA Imbalance
2
Radiative Flux (W/m )
8
0
0
0
30
60
90
120
150
Feedback
20
60
90
120
150
30
Strat T
Strat WV
20
10
10
0
0
-10
-10
-10
-20
-20
-20
10
0
-30
0
30
60
90
120
150
3
-30
0
30
60
90
120
150
3
3
Adjustment
2
Radiative Flux (W/m )
30
T
WV
Albedo
20
-30
2
2
2
Trop T
Trop WV
Albedo
Strat T
Strat WV
1
1
0
0
0
-1
-1
-1
-2
1
-2
0
30
60
90
120
150
9.0
-2
0
30
60
90
120
150
9.0
9.0
8.5
8.5
8.0
8.0
8.0
7.5
7.5
7.5
7.0
7.0
7.0
6.5
6.5
6.5
Instantaneous Forcing
2
Radiative Flux (W/m )
0
0
30
2
Radiative Flux (W/m )
30
8.5
6.0
6.0
0
30
60
90
Elapsed Time (years)
120
150
6.0
0
30
60
90
Elapsed Time (years)
120
150
ENSEMBLE MEAN (clear-sky): Abrupt 4xCO2
Finite Differencing
Regression
8
8
6
6
6
4
4
2
2
TOA Imbalance
2
Radiative Flux (W/m )
8
0
30
60
90
120
150
Feedback
2
Radiative Flux (W/m )
2
0
0
30
60
90
120
150
30
20
30
T
WV
Albedo
20
Strat T
Strat WV
20
10
10
0
0
-10
-10
-10
-20
-20
-20
-30
10
0
-30
0
30
60
90
120
150
3
-30
0
30
60
90
120
150
3
3
Adjustment
2
Radiative Flux (W/m )
4
0
0
30
2
2
2
Trop T
Trop WV
Albedo
Strat T
Strat WV
1
1
0
0
0
-1
-1
-1
-2
1
-2
0
30
60
90
120
150
9.0
-2
0
30
60
90
120
150
9.0
9.0
8.5
8.5
8.0
8.0
8.0
7.5
7.5
7.5
7.0
7.0
7.0
6.5
6.5
6.5
Instantaneous Forcing
2
Radiative Flux (W/m )
No significant stratospheric
feedbacks
8.5
6.0
6.0
0
30
60
90
Elapsed Time (years)
120
150
6.0
0
30
60
90
Elapsed Time (years)
120
150
ENSEMBLE MEAN (clear-sky): Abrupt 4xCO2
Finite Differencing
Regression
8
8
6
6
6
4
4
4
2
2
2
TOA Imbalance
2
Radiative Flux (W/m )
8
0
0
0
30
60
90
120
150
Feedback
20
10
10
0
0
-10
-10
-20
-20
60
90
120
150
30
30
60
90
120
150
Strat T
Strat WV
20
10
Fast Adjustments dominated by
stratospheric cooling  ~25% of total
forcing.
0
-10
-20
-30
0
3
-30
0
30
60
90
120
150
3
3
Adjustment
2
Radiative Flux (W/m )
30
T
WV
Albedo
20
-30
2
2
2
Trop T
Trop WV
Albedo
Strat T
Strat WV
1
1
0
0
0
-1
-1
-1
-2
1
-2
0
30
60
90
120
150
9.0
-2
0
30
60
90
120
150
9.0
9.0
8.5
8.5
8.0
8.0
8.0
7.5
7.5
7.5
7.0
7.0
7.0
6.5
6.5
6.5
Instantaneous Forcing
2
Radiative Flux (W/m )
0
0
30
2
Radiative Flux (W/m )
30
8.5
6.0
6.0
0
30
60
90
Elapsed Time (years)
120
150
6.0
0
30
60
90
Elapsed Time (years)
120
150
ENSEMBLE MEAN (total-sky): Abrupt 4xCO2
Finite Differencing
Regression
8
Methodology
6
4
2
0
30
60
90
120
Feedback
20
4
2
2
0
0
30
60
90
120
150
30
T
WV
Albedo
Cloud
20
Strat T
Strat WV
20
10
10
0
0
-10
-10
-10
-20
-20
-20
10
0
-30
0
30
60
90
120
150
3
-30
0
30
60
90
120
150
3
3
2
2
1
1
0
0
-1
-1
Adjustment
2
Radiative Flux (W/m )
4
30
2
Radiative Flux (W/m )
6
150
-30
2
Strat T
Strat WV
1
0
Trop T
Trop WV
Albedo
Cloud
-1
-2
-2
0
30
60
90
120
150
8.0
-2
0
30
60
90
120
150
8.0
8.0
7.5
7.5
7.0
7.0
7.0
6.5
6.5
6.5
6.0
6.0
6.0
5.5
5.5
5.5
Instantaneous Forcing
2
Radiative Flux (W/m )
6
0
0
30
7.5
5.0
5.0
0
30
60
90
120
0.2
150
5.0
0
30
60
90
120
150
0.2
0.2
0.1
0.1
0.1
0.0
0.0
0.0
-0.1
-0.1
-0.1
Residual
2
Radiative Flux (W/m )
8
TOA Imbalance
2
Radiative Flux (W/m )
8
-0.2
0
30
60
90
Elapsed Time (years)
120
-0.2
150
-0.2
0
30
60
90
Elapsed Time (years)
120
150
ENSEMBLE MEAN (total-sky): Abrupt 4xCO2
Finite Differencing
Regression
8
Methodology
6
4
2
0
30
60
90
120
Feedback
20
4
2
2
0
0
30
60
90
120
150
30
T
WV
Albedo
Cloud
20
Strat T
Strat WV
20
10
10
0
0
-10
-10
-10
-20
-20
-20
30
60
90
120
150
10
0
-30
0
3
0
Cloud adjustments are small.
30
60
90
120
-30
150
3
3
2
2
1
1
0
0
-1
-1
Adjustment
2
Radiative Flux (W/m )
4
30
2
Radiative Flux (W/m )
6
150
-30
2
Strat T
Strat WV
1
0
Trop T
Trop WV
Albedo
Cloud
-1
-2
-2
0
30
60
90
120
150
8.0
-2
0
30
60
90
120
150
8.0
8.0
7.5
7.5
7.0
7.0
7.0
6.5
6.5
6.5
6.0
6.0
6.0
5.5
5.5
5.5
Instantaneous Forcing
2
Radiative Flux (W/m )
6
0
0
30
7.5
5.0
5.0
0
30
60
90
120
0.2
150
5.0
0
30
60
90
120
150
0.2
0.2
0.1
0.1
0.1
0.0
0.0
0.0
-0.1
-0.1
-0.1
Residual
2
Radiative Flux (W/m )
8
TOA Imbalance
2
Radiative Flux (W/m )
8
-0.2
0
30
60
90
Elapsed Time (years)
120
-0.2
150
-0.2
0
30
60
90
Elapsed Time (years)
120
150
Estimates of Instantaneous Radiative Forcing from CMIP5 Scenarios
Ensemble Mean Estimates of CMIP5 Instantaneous Radiative Forcing
Estimates of Radiative Forcing from CMIP3 (IPCC AR4 Models)
Estimates of Radiative Forcing from CMIP3 (IPCC AR4 Models)
Estimates of Radiative Forcing from CMIP3 (IPCC AR4 Models)
Estimates of Radiative Forcing from CMIP3 (IPCC AR4 Models)
Assessing Intermodel Spread of Radiative Forcing
Conclusions

Virtually all of stratospheric changes are “adjustments” to CO2
forcing. No significant feedback from temperature or water vapor in
stratosphere.

Stratospheric adjustments provide only about 25% of total radiative
forcing. Lower than previous estimates (of which there are few).

Temperature has the largest “fast” adjustment (larger than clouds)
in the troposphere (larger than clouds).

Adjustments roughly double the intermodel spread in forcing for
CO2 scenarios, but are smaller in RCPs. Cause unclear ATT.
Assessing Intermodel Spread of Radiative Forcing