p. friedlingstein and ic prentice paris/bristol/exeter/sidney + inputs from v. masson-delmotte

P. Friedlingstein and IC PrenticeParis/Bristol/Exeter/Sidney

+ inputs from V. Masson-Delmotte

The magnitude of the problem

Brussels, May. 27th 2009

Uncertainty due to the carbon cycle uncertainty

2.6 – 4.1 °C2.4 – 5.6 °C

830 ppm

730 – 1000 ppm

Higher [CO2], larger climate change

IPCC, 2007

Climate-Carbon Cycle FeedbackCO2 = EMI - Fao - Fab (1) T = CO2 + Tind (2)with: Fao = ao CO2 +ao T (3) Fab = ab CO2 +ab T (4)

(3) and (4) in (1), then (1) in (2) gives:

T = 1/(1-g) Tunc

with:g = (ao + ab )/(1+ ao + ab)

Climate-Carbon Cycle FeedbackT = 1/(1-g) Tunc = f Tunc

g = (ao + ab )/(1+ ao + ab)

g is the gain of the climate-carbon cycle feedback

f = 1/(1-g) f is the feedback factor

and is the carbon cycle sensitivity to climate (C/T)

Climate-Carbon Cycle Feedback

Carbon cycle sensitivity

to climate

gClimate carbon cycle

gain

0.04 – 0.30

30 – 200 GtC/K

What are the available observations ?

Glacial interglacial CO2 – Temperature

€

g =ΔT

ΔCO2

×ΔCO2

ΔT

Climate sensitivity is estimated here from 2xCO2 GCMs estimates,in the absence of physical feedbacks (black body response only).

Two caveats

Glacial interglacial CO2 – Temperature

1. Physical feedbacks

€

g =ΔT

ΔCO2

×ΔCO2

ΔT

€

g = α ×γ

1+ β

Torn and Harte, 2006

Friedlingstein et al., 2006

is the climate sensitivity, accounting for all physical feedbacks

€

TTH 06 =ΔF

λ BB

λ BB = 3.8Wm−2K −1

ΔTF 06 =ΔF

λ i∑λ i∑ =1.3 ± 0.3Wm−2K −1

gG-IG= 0.04*3.8/1.3= 0.12

Using the Full EPICA record

Glacial interglacial CO2 – Temperature

€

CO2

ΔT= 7.8633 ppm/K and taking from AR4 gG-IG= 0.08

€

T

ΔCO2

2. Does this help for future projections?

Last Millennium and LIA

Last Millennium and LIA

dCO2/dT= 50.6 ppm/K dCO2/dT= 39.9 ppm/K

Last Millennium and LIA

Last Millennium and LIA

dCO2/dT= 7.7 [ 1.7 – 21.4] ppm/K

Confusion in terminology …

dCO2/dT is neither g no …

Last Millennium and LIA

dCO2/dT= 7.7 [ 1.7 – 21.4] ppm/K

One could derive the gain g:(again, taking dT/dCO2

from 2xCO2 sensitivity)

€

gLIA = 7.7 ×3[2 to 4.5]

286= 0.08 [0.05 to 0.12]

€

T

biosphere

Ocean

€

CO2 = γΔT − βΔCO2

i.e.

€

CO2

ΔT=

γ

1+ β

€

Cout = γΔT

€

CO2

time

€

Cin = βΔC

Last Millennium and LIAOr one could derive

But one needs to know on millenium time scales …

Interannual variability of CO2G

t. C

per

yea

rS

OI

1955 1960 1965 1970 1975 19851980 1990 1995 2000

8

6

4

2

30

0

-30

CO2 Annual Growth Rate

Interannual variability of CO2

dCO2/dT= 2.9 ppm/K

= -90 GtC/K gG-IG= 0.03

Summary

gain Carbon sensitivity to climate (GtC/K)

G-IG 0.08 ≈ -110*

LIA 0.08 ≈  -110*

IAV 0.03** -90

C4MIP models average

0.15 -109

*assuming ≈ 5.5, i.e. AF≈ 0.15**assuming equilibrium response

Palaeo and historical CO2 variability could help to constraintClimate carbon cycle feedbackEstimate of seems to be more robust than g across timescales

SummaryPalaeo and historical CO2 variability might help

to constrain Climate carbon cycle feedbackHowever, large uncertainties on data and on

use of dataEstimate of seems to be more robust than g

across timescales. Is this accidental ?Do we get the “right” number for the right

reason (right process) ?Best way is certainly not what I just

presented...

We should simulate the past rather thanplay with past data to infer future response