CO 2 Lead/Lag

1. CO2 Lead/Lag Marcy Laub, Emily Malkin

2. “The Great Global Warming Swindle” CO2 lag seen as dismantling it as a cause of global warming - deniers say climate change not anthropogenic https://www.youtube.com/watch?v=bi2QKY3zW8Q&t=0m26s

3. Initial Perturbation Eccentricity Tilt/Obliquity Milankovich Cycles: changes in Earth’s tilt and orbit around the sun cause changes in radiative forcing Their Effect: • Warmer oceans release CO2 • CO2 amplifies • Lag vs. warming duration Precession CO2 Solubility vs. Temp.

4. Ice Structure Influence on Pressure and Transport Convective Zone Air mixed by surface wind that flows freely through very porous snow Diffusive Column Stagnant air mixed by movement of atoms and molecules Close-Off/LID, Gas trapped

5. Gas-Ice Age Difference • Top of snowpack porous to bottom of firn, air diffuses throughout • Air at bottom of firn trapped as pressure closes pores → younger air enclosed in older ice • Not the same as time lag

6. Measures of Synchronizing Ice and Gas Δage the difference in age between the ice and gas phases at any given depth Δdepth the depth difference between gas and ice of the same age

7. TIII TI TI vs. TIII Deglaciation • TI - Termination 1, last deglaciation, 21,000 ybp • Dust and isotope change in phase • TIII - 240,000 ybp, deglaciation • Changes in dust before changes in isotope • Different lead/lag conclusion for each δ18O (‰) Age (1000 ybp) old new

8. Paper Summaries Caillon, et al. (2003): CO2 has significant lag in TIII deglaciation, on a timescale consistent with Southern Ocean temperature-CO2 feedbacks Parrenin, et al. (2013): CO2 and temperature are synchronous from TI to present

9. Vostok Ice Drilling Station

10. Location of Vostok Allows for Age Model Uncertainties Due to Ice Flow Complications in measuring accumulation rates: • Wind • Upstream flow from west Acknowledged by Caillon but unaccounted for

11. Slower Densification at Low Accumulation Rates • Antarctic ~5,000 year ice-gas age difference • Greenland ~500-600 years Why? Antarctic accumulation rates are much lower at 166 mm/year vs. 2,500 mm/year Sahara: 25 mm/year, Boston: 1050 mm/year

12. Well-Established Temperature Proxies Exist • Caillon uses D, deuterium • “Heavy hydrogen” isotope: 2 neutrons • Fractionation ofisotopes • dependson • temperature warm cold Isotopes in icereflect temperature at time of deposition

13. Gas bubble trapped in ice containing Ar Caillon Strives to Prove Ar as a Temp. Proxy • Heavy element subject to gravitational fractionation • Measurements from gas • δ40Ar fractionation extent varies with DCH • Will show DCH-temp relationship T proxy in gas eliminates need to determine Δage

14. D, Ar Exhibit Similar Patterns at Different Depths • δD record from ice, well-established • δ40Ar record from gas • Same features • 2-step increase δD δ40Ar • δ40Ar hypothesized temperature proxy new old

15. D, Ar Relationship Offers Data Validation D measured in ice Ar measured in gas Use phasing to find Δdepth Convert Δdepth to Δage, using time resolution

16. Calculated Δage Allows Continuous Record Comparison • FDM: Deeper firn depth for higher accumulation rates and colder temperatures • Their results: high correlation between all 3 Temp Accumulation δ40Ar thick Firn depth Primary driver of firn depth unclear thin TIII new old

17. δ40Ar RecordCreates More Accurate Firn Model • Conclude no convective zone at TIII • Direct relationship at max DCH– DCH varies only with temp Error in total firn depth estimate From data DCH influenced more by temp. than accumulation

18. Axis Shift to Produce Best Temp-CO2 Fit Shows Time Lag • Shift CO2 record back 800 yrs → R2 = 0.88 for CO2 and δ40Ar • Plausible lag for S. Ocean feedback Note time axis difference CO2 δ40Ar CO2 lags temperature by 800 years, lag significant

19. δ40Ar in the Context of Well-Documented Records Provides Sequence of TIII Deglaciation Events • CH4,δ18O global signals • δ40Arincreases before δ18O • Antarctic warms ~6000 yr before N. hemisphere • Note shallower CO2peak CH4 δ18O δ40Ar CO2 from 2nd source Northern warming lags, CO2 leads global warming CO2 old new

20. Caillon Conclusions • Temperature leads CO2 by 800±200 years • CO2 is an amplifier, and is exacerbated further by anthropogenic forcing

21. Controversial Bipolar Seesaw Hypothesis Links Greenland and Antarctic Ice Records T increase, Greenland • Unconfirmed • T changes in N & S hemisphere out of phase • Rapid T increase in Greenland = T maximum in Antarctica new old time T max, Antarctica Comparing T records synchronizes cores

22. Estimating Timing of Antarctic Temp. Variations from Global CH4 Signal EDC ice depth (m) • Tie points: “signature” rapid CH4 change • Determine Δdepth between temp. proxy extremaand tie points → Only 3 Δdepth estimates, consistent with bipolar seesaw δD (‰) CH4 (ppbv) Detectable CH4 shift new old years before present

23. Determining EDC Δdepth in Spite of Uncertainties Due to Slow Accumulation Rate • Ice records synced by volcanic ash layers • Synchronized EDC CH4 record with EDML/TALDICE → Only 3 approximations of Δdepth for EDC At time=t Known from firn densification model

24. δ15N Confirmed as Reliable Δdepth Metric Firn densification model δ15N data, N best fit EDML, TALDICE, Seesaw Δdepth Most N data within dashed line, 1σ • δ15N agrees with other metrics → conv. zone assumption upheld → δ15N varies directly with DCH • δ15N data more continuous Core depth (m) Δdepth can be estimated for entire core

25. Assumptions Allow for Computation of a Continuous Δdepth from δ15N δ15N Record Lock-In Depth, continuous record no convective zone firn models Δdepth, continuous record convert to time CO2 and T records on same time axis

26. Noise Reduced by Stacking Core Data Antarctic Temperature Stack (ATS) comprised of EDC, Vostok, Dome Fuji, TALDICE and EDML cores • Synchronize core records • Convert δD and corrected δ18O record to T record • Average the five cores for each point in time (resampled every 20 years) Standard deviation for 220-year moving average: SD, EDC = .52°C SD, ATS = .20°C >

27. Core Records as a Six-Point Linear Function • 10,000 Monte Carlo simulations give 4 tie points → probability distribution fits data around points • Linear be start, tie and end points Forces linear fit and phasing within a given window

28. Younger Dryas TI Bolling Osc. Temp. and CO2 Phasing Without Ice-Gas Age Difference Holocene • Temp change,°C • Atmospheric CO2, ppm Rad. forcing of aCO2, W/m2 CH4, ppb • Pearson corr, aCO2/ATS = .993 • 4 points where linear slopes shift • CO2 break leads ATS twice, • CO2 break lags ATS twice, Lead/lag magnitudes insignificant age new old

29. Chosen Linear Fit Influences Phasing Original breakpoint New linear fit New breakpoint Magnified temperature change data at TI Synchronicity potentially an artifact of the data fitting ~800 years

30. Underestimation of CO2 During Abrupt Spikes and Using Linear aCO2 Measure • TI: 10 ± 160 years, CO2 leads • Bolling Oscillation: -260 ± 130 years, CO2 lags • Corrected for “fast increases,” -10 ± 130 yrs • Younger Dryas: 60 ± 120 years, CO2 leads • Holocene: -500 ± 90 years, CO2 lags • Corrected, -130 ± 90 years • Corrections maintain insignificance of lead/lag

31. Parrenin Conclusions • Temperature and CO2 are synchronous • CO2 acts as an amplifier for weak ~0.6°C global warming by rCO2 • Effective δ15N method - adopted by outside studies

32. Independent Data Find T and CO2 Coupled Pedro, et al. Presented by Brook (2013) • Pedro uses coastal cores, existing CO2, temp proxies • Same conclusionsof synchronicitydespitedifferent methodology T Index CO2 (ppm) Parrenin, et al. T Anomaly Age

33. Ice Impurities Thought to Influence Fractionation TIII: Dust peaks before Ar EDC depth, m TI: Same trend btw dust and N Conclusions not necessarily conflicting Vostok depth, m