Our basic understanding of past ocean circulation Geoffrey (Jake) Gebbie Research Associate, Harvard University Visiting Scientist, MIT ACDC Summer School, 5 June 2009 The view down a borehole (Helen Freeman, U. Wales Swansea)
Ultimately, we wish to make predictions about the future state of the meridional overturning circulation (MOC). Understanding the present and past MOC may be a prerequisite.
The state of LGM modeling
We expect that many of the relevant coupled dynamical processes occur on decadal and longer timescales. The instrumental record is not long enough to well document this lower frequency variability.
Uptake of anthropogenic carbon by the ocean More than just predictions of the MOC, we want to understand the impact on quantities of societal importance. Hall et al. 2004
We know that there have been large changes in the conditions of the past seafloor. Lisiecki et al. 2008
The unique difficulties of the observations of the past ocean circulation: 1. Proxy data 2. Mostly limited to seafloor sediments, which allows some information about the sea surface as well 3. Seafloor sediments don't exist everywhere along the seafloor 4. Chronology of sediment accumulation rate 5. Dissolution of calcium carbonate below certain depths
The sparsity of sediment cores courtesy O. Marchal
A form of proxy data?
By the length of the mercury, I can deduce that this thermometer must have been somewhere rather cold (Norway, perhaps?).
A focused approach to understanding past ocean circulation The full 3-D, time-evolving circulation is too complex to be resolved by our limited data. 1. Highlight a particular time period, for example, the Last Glacial Maximum (19-24 kyr ago) 2. Focus on the static problem what is the water mass configuration and the differences between modern and LGM water properties? POP QUIZ 3. What can be said about the dynamic problem estimating rates of past ocean transport? 4. Revisit the full complexity of past ocean circulation
The Static Problem
δ13c is a tracer that is thought to be nearly conservative Curry and Oppo, 2005
Interpreting C13 From Lynch-Stieglitz 2003
Oxygen Isotopes (δ18o) - 10 + - - 20 + 0 O18 of seawater is conservative in the ocean interior Courtesy lorrainelisiecki.com
Oxygen Isotopes (δ18o) - 10 + - - 20-30 + - 30 1 Courtesy lorrainelisiecki.com CaCO3 f(t) 4
δ18o dependencies δ18ocalcite = f( δ18owater, T) = δ18owater + at' + c (a,c determined by chemistry of calcification) δ18owater = bs' + d (empirical) b=0.5 Data from Schmidt 1999
Phase space of LGM data Use Curry-Oppo compilation of d18o, d13c. Plus Adkins-Schrag pore water dataset.
Water mass decompositions A conservative tracer A non-conservative tracer Mass conservation
Add Cadmium/Calcium ratio. Can solve for amount of NADW, AABW, and AAIW. Set of simultaneous equations with 4 unknowns and 4 constraints.
LGM water-mass decomposition The fraction of the volume of the Atlantic water column by water mass
The relative importance of the 3 major water masses during the LGM Mass fraction
Preliminary information about rates of flow: the nonconservative signal in C13 The nonconservative nature of d13c -- a clock for the glacial age of the ocean. Given the modern C13 calibration, the Atlantic Ocean has ages up to 1000 years old.
A modern-day water-mass decomposition Volume fraction from 5 surface patches Latitude-depth sections of the Atlantic, 50W-10W
Modern-day water-mass decomposition Volume fraction from 5 surface patches Latitude-depth sections of the Pacific, 180-150W
Some words of caution: 1. How many water masses are there? 2. Is the ocean in equilibrium?
The composition of the deep Pacific Conservation of phosphate, oxygen and mass Stommel and Arons 1962 phosphate: 2.8 = 1.1 m1 + 2.4 m2 + P oxygen: 90 = 290 m1 + 250 m2-175 P mass: 1 = m1 + m2 Solution: m1 = 0.49, m2 =0.51, P = 1.06 Units of micromole/kg where applicable Same answer as phosphate-star method of Broecker et al. 1998
Temperature-salinity volumetric census for the modern ocean NADW AABW Worthington, 1976, 1979
The surface origin of ocean waters -T dv/db = A v, Adjoint Green function
THE SPECTRUM OF WATER MASSES
The core path of North Atlantic water through the world ocean
The core of North Atlantic water throughout the world ocean Atlantic Indian Pacific
A volumetric census of North Atlantic Water below 1500 meters volume Min: 500 years Max: 1100 years
Getting temperature and salinity for the LGM Adding pore-water salinity measurements
A first estimate of the LGM hydrography
Quiz/Discussion: What is the overturning circulation?
The transport problem
Similarities between Proctactinium/Thorium ratio and do18 McManus et al. 2004
Additional Pa/Th data points Gherardi et al. 2009
Sortable silts Rockall Trough McCave et al. 1995
A Proxy with a built-in timescale: 14C
Radiocarbon-based estimates of the age of the ocean
No reservoir correction: Solve for t given Co= Catm and Cf = ocean data
Standard computation of radiocarbon age
Nonlinear mixing of radiocarbon ages Consider a sample composed of two equally-mixed end members. The green, red, and magenta samples all have C=0.55, and a radiocarbon age of 5000 yrs. However, the true mean ages are all greater than 5000 years. In the case where all end members have the same initial concentration, the radiocarbon age is the lower limit on the true age.
Data-based estimation method Define TTD(t) to be the transit time distribution from the surface to the ocean interior. Following e.g. Hall and Plumb, 1996, Holzer & Hall, 2000 The radiocarbon equation is now: The mean age is: Solve for TTD(t) that satisfies the data equation while minimizing or maximing the age. Use Karmarkar's linear programming method which accounts for non-negative nature of TTD(t).
An Inverse Gaussian Form for the Transit Time Distribution
The Florida Straits
δ18o as a paleo-density tracer δ18ocalcite = f( δ18owater, T) = δ18owater + at' + c (a,c determined by chemistry of calcification) δ18owater = bs' + d δ18ocalcite = at' + bs' + c + d Linearized Eq. Of State: ρ = (z) T' + ß(z) S' + Ω(z) (empirical) b=0.5 Data from Schmidt 1999
δ18o LGM δ18o Holocene Instrumental
Florida Straits during the last 2,000 years Lund et al. 2006
Lund et al. 2006: a 10-15% reduction during the Little Ice Age
Modern-day time-mean temperature and velocity fields in the Florida Strait From Leaman et al. 1987
South Atlantic: 30 S Lynch-Stieglitz et al 2006
Inverse problem with a steady-state geostrophic model thermal wind conservation of mass equation of state LGM observations δ18ocalcite = at' + bs' + c z a,b determined by modern values Mass conservation necessary to resolve reference level problem Solved by Gauss-Markov Method x
Estimated modern reference circulation Error bars are optimistically small because only small deviations from thermal wind balance are allowed. Max overturning: 22 Sv +- 3 Sv
Water-mass properties of the inverse solution Gebbie and Huybers 2006 Water-mass changes are enough to explain the altered O18 values without any change in the overturning. New data shows that crossbasin differences in temperature and salinity may have been large. Is such a situation dynamically tenable?
3. Wrapping it up
Interpreting the time-variable data sets Skinner and Shackleton, 2004
GCM/State estimate approach Wunsch and Heimbach, 2008, found that equilibrium times are very long. 1975 m Based upon a 2000-year tracer integration with the 12-year repeating ECCO state estimate physical fields.
The lag in the observations is potentially explained by shifts in the endmembers and the consequent ocean transit times.
Advanced topics on the table: Southern Ocean overturning The ocean state with enhanced sea ice cover Explaining proxy records with abrupt changes in ocean dynamics
Paleoceanographic proxies primarily inform us about water-mass geometry in the past. Finding a reliable method of estimating transport rates remains an ongoing goal. New paleoceanographic proxies are getting more and more accurate information about temperature and salinity, and hence density, the most important physical variable of all. For discussion: how far can we get without knowing the paleo-wind field? And what was the paleo-wind field anyway?
Water mass decompositions
Constraints Conservation of mass Vorticity balance Thermal wind Conservation of δ13c, δ18o, and 14C