Uses of the paleoclimate record in understanding climate sensitivity

1. Uses of the paleoclimate record in understanding climate sensitivity • Extrapolation and interpolation • Simple example • Inferring climate sensitivity: time issues • Vostok CO2 and tropical temperature • A test for covariance • Correcting for bias in the records • Greenhouse gas feedbacks • A different control upon glacial/interglacial atmospheric CO2

2. An eminent philosopher among my friends, who can dignify even your ugly furniture by lifting it into the serene light of science, has shown me this pregnant little fact. Your pier-glass or extensive surface of polished steel made to be rubbed by a housemaid, will be minutely and multitudinously scratched in all directions; but place now against it a lighted candle as a centre of illumination, and lo! the scratches will seem to arrange themselves in a fine series of concentric circles round that little sun. It is demonstrable that the scratches are going everywhere impartially and it is only your candle which produces the flattering illusion of a concentric arrangement, its light falling with an exclusive optical selection. (George Eliot, Middlemarch)

3. Extrapolation 1

4. Extrapolation 2

5. Extrapolation 3

6. Extrapolation 4

7. Extrapolation 5: T=  f+0.1 ( f)2+0.01 ( f)3

8. 10 years

9. 100 years

10. 1000 years

11. 10,000 years

12. 100,000 years

13. 1,000,000 years

14. 10,000,000 years

15. 100,000,000 years

16. CO2 and delta18O over the last 60 million years. (Zachos et al, 2008)

17. Spectral estimates of surface temperature from monthly to 100,000 year time-scales High latitude continental Tropical marine 41,000yr 1yr

18. Sensitivity of the climate system as a function of forcing timescale

19. Climate sensitivity estimates (from Tamsin Edwards et al, 2007)

21. Climate forcing, f, and the temperature response, r, as a function of time, t… and an estimate of the climate sensitivity, S.

22. But its actually more complicated than that because of time-uncertainty Time errors take random walks

23. Case 1: Random age errors and perfectly correlated Vostok CO2 records

24. Case 1: Regression between CO2 and age corrupted versions true slope slope with age errors

25. Case 2: tuning noise to the Vostok CO2 record A random record and CO2, r=-0.1 After tuning, r=0.4

26. Case 2: tuning noise to the Vostok CO2 record untuned tuned histogram of many tuned slopes Maximum tuned cross-correlation

27. Comparing Vostok CO2 and the Cocos Ridge tropical temperatures untuned tuned

28. Regression of tuned and untuned CO2 forcing and temperature “response” S=1.9 oC/(W/m2) (r=0.75, p<0.01) S=0.45 oC/(W/m2) (r=0.1)

29. A test for the presence of covariance between records which are time-uncertain (Eddie Haam)

30. A means to correct for regression suppression: Covariance can be written as, Bias correction can be estimated as,

31. Estimation of slope with age errors and bias correction (true slope is 1) bias corrected slopes raw slopes

32. Regression between Vostok CO2 and tropical temperature bias correction, b=0.4 S=2 oC/(W/m2) S=1.1 oC/(W/m2) S=0.45 oC/(W/m2)

33. Temperature records available for constraining sensitivity

34. Conclusions for parts 1 and 2 Paleoclimate advantages: The paleoclimate record can provide guidance in extrapolating from our current climate state and, in a sense, changes the problem to one of interpolation as both warmer and colder climates are bracketed. This past climate record is the only means of observationally constraining slow feedbacks. The very different climate states in the past permit for testing the accuracy and completeness of our models. Disadvantages: Large and novel uncertainties arise , particularly errors in timing. Proxies usually cannot be directly related to model variables. Requires climate dynamicists to learn some geology.