Paleoceanography - Oxygen Isotope Paleothermometry

1 PALEOCEANOGRAPHY lecture 3 Oxygen-isotope paleothermometry I. 1940's: Kullenberg's invention of the piston core. Previously, deep-sea coring was done by gravity coring, where a pipe with an in-line stopper at the top and "core catcher" at the bottom (interleaved flexible metal fingers) was dropped into the seafloor. Friction against the walls of the pipe limited the length of core that could be obtained. Modern gravity cores can get cores ~ 5 m in length; older ones obtained only a meter or two. Kullenberg introduced a piston into the pipe; the piston was rigged with a cable that prevented it from moving down with the pipe, creating "suction" which helped force the sediment into the pipe as it moved down into the sediment. Modern piston cores can obtain as much as ~50 m of sediment (although more typically, ~20 m). II. 1950's: Emiliani A. Using Urey's mass spectrometer, which required about 5 mg of calcium carbonate (about 100-200 individuals of foraminifera), Emiliani analyzed a variety of species of planktonic foraminifera from the Caribbean and found an apparent depth stratification, with some species (G. sacculifer, G. ruber) recording isotopic temperatures close to that of surface seawater; others recording colder temperatures. B. Emiliani analyzed downcore records of apparent surface dwellers throughout the Atlantic; making a correction for changes in the isotopic composition of seawater (more about this later), he calculated a 6-8° decrease in tropical ocean surface temperatures during glacial periods (i.e. about 1.5‰ δ18O increase). He found evidence for many glacial/interglacial cycles over the last half million years; he coined the isotope stage stratigraphy system (now commonly referred to as "MIS" (Marine Isotope Stage); and he argued that the data supported the Milankovitch mechanism of climate change. 2 1. Emiliani developed an isotope stage numbering scheme, based on periods of warmer (odd) and colder (even) that he could recognize reliably in his records (smaller changes were ignored). 2. This work created quite a stir, and was quickly criticized on several grounds: a. It violated the prevailing 4-ice-age theory from continental stratigraphy. b. Meteorologists though that the tropical temperature change seemed excessive. c. Micropaleotologists found discrepancies between their initial micropaleontological work (G. menardii stratigraphy) and the down-core O-18 record. d. Biologists (e.g. Bé) argued that foraminiferal ecological shifts may have altered the depth habitat of organisms (and hence temperatures). e. The time scale (based on 230Th/231Pa) was criticized. f. Various statistical errors were pointed out. 3. Despite all of this criticism, with 1996 hindsight we can say that Emiliani was right about most of these points, with the major exception being the time scale and the amplitude of tropical cooling. • the problem of the time scale: Emiliani derived his time scale from a core which had been dated by the "231Pa/230Th" method. This method assumes that the initial value of the ratio at zero age is the production rate from the 235U and 234U in seawater. We now know that this assumption is incorrect: Th and Pa are fractionated by their differential particle reactivity in the ocean. 3 III. Oxygen isotope hydrology A. As it turns out, the major problem with Emiliani's interpretation is his correction for the change in the isotopic composition of seawater. B. Oxygen (and hydrogen) isotopes in the atmosphere and precipitation. 1. Water vapor in equilibrium with water has δ18O ~ -9-11‰, varying slightly with temperature: α =1.01113 - 1.06x10-3 T + 2.88x10-7 T2. 2. If an air mass initially equilibrated with δ18O = 0 ‰ water is then isolated, cooled, and the resulting condensation continuously drawn off, the vapor becomes progressively more depleted in 18O: exponential dependence of vapor pressure on temperature and Rayleigh (1896) fractionation: R = f (α (T ) −1) Ri where and T = temperature α (Τ) = Rliq/Rvap isotope fractionation (slightly temperature dependent) f = the fraction of the initial water vapor remaining (set by the equilibrium vapor 18 16 R/Ri = O/ O ratio of the remaining water vapor compared to its initial value Fraction Remaining Vapor 1.0 0.50 0.75 0.25 0 4 18 δ O (SMOW) Condensate 9% -5 -10 -15 11% Vapor -20 -25 -30 20 15 10 0 Cloud Temperature O -20 C 18 δ O in cloud vapor and condensate plotted as a function of the remaining vapor in the cloud for a Rayleigh 18 process. The temperature of the cloud in degrees Celsius is shown on the lower axis. δ O values are relative to SMOW. The increase in fractionation with decreasing temperature is taken into account. After Dansgaard (1964). 3. Similar fractionation occurs for hydrogen isotopes, except that since the mass difference is proportionately larger, the isotope fractionation for D/H is larger (by about a factor of 8) 4. Since most evaporation occurs in the tropical and subtropical ocean (highest vapor pressures) and since most precipitation results from the transport of this tropical water vapor towards cooler polar regions, there is a strong correlation between average annual temperature and the average annual δ18O. 0 Gough Is. Valentia Dublin -10 o 61 N Grennedal 18 75 N Upepnavik -20 Copenhagen Goose Bay, Labrador o δ O (SMOW) Barbados Is. o 71 N Umanak S. Greenland -30 N. Greenland -40 o 85 S Horlick MTNS. South Pole -50 -50 -40 -30 -20 -10 0 10 20 30 Mean Annual Air Temperature C O 18 Mean annual δ O of precipitation as a function of the mean annual air temperature at 18 the earth's surface. Note that δ O values are progressively lighter as the mean annual temperature becomes lower. After Dansgaard (1964). 5 5. Although Emiliani knew that modern ice was relatively light and had an isotopic composition of about -25‰, he argued that the bulk of snowfall during the glacial maximum (where ice is accumulating at lower latitudes) must have an isotopic composition of about -15 ‰, because this is the composition of precipitation at ~-5 to 0°C (where the bulk of snowfall will occur because of the exponential dependence of vapor pressure on temperature). From then-current estimates of sea-level change (about 100m) and the assumption that ice sheets must have an isotopic composition of about -15‰, Emiliani calculated that the contribution of ice volume change to changes in the isotopic composition of the ocean corresponded to about 1/3 of the total change: Moδo + Miδi = constant Mo = 1.4 x 1021 kg Ao = 361 x 106km2 C. One major problem with Emiliani's reasoning is that ice sheets don't represent average snowfall; wastage of the sheet is highest near the southern margin where snowfall is also the highest; glacial flow results in the bulk of the ice sheet representing the isotopic composition of the northern interior, with much more negative δ18O (-30 to -50 ‰). 6 y I Surface C Flow lines x C I Bed rock Vertical cross section of an ice sheet resting on a horizontal subsurface. Ice particles deposited on the snow surface will follow lines that travel closer to the base the farther inland the site deposition. An ice mass formed around the divide (I-I) will be plastically deformed (thinned) with depth as suggested by the lined areas. The dashed curve along the vertical ice core (C-C) shows the calculated horizontal velocity profile Vx (Weertman 1968b). The horizontal arrows along C-C show the adopted approximation to Vx (Dansgaard et al. 1969). 1. This was pointed out by Dansgaard (1968), who made measurements of δ18O in the Camp Century ice core in northwestern Greenland, finding typical values of -28‰ near the surface and typical values of -40‰ in the glacial maximum. 2. It might seem that the isotopic composition of the Laurentide and Fennoscandian glacial ice is forever unknowable (because the ice is gone). However two ingenious ways of establishing isotopic composition of glacial ice have been suggested: a. Hillaire-Marcel and Causse (1989) investigated the isotopic composition of carbonates precipitated from sub-glacial streams. The temperature is constrained to a narrow range (melting point as a function of pressure); modern analogues suggest that a range of -1.5 to 5 °C is reasonable. If the carbonates are at equilibrium, we can calculate the isotopic composition of the meltwater to better than 1‰. These results suggest that ice near the southern margins of ice Laurentide ice sheet was about -16 to -19‰, and that of the interior of the ice sheet about -30‰. In view of the likely flow lines, this result support the notion that the bulk of the ice sheet had an isotopic composition closer to -30‰ than to 15‰. b. As pointed out by McDuff (1984), the delay in diffusion of salt and isotopes into the pore waters of oceanic sediments should lead to a maximum of salinity and a minimum in δ18O near 40 m depth, because of the time scale of diffusion: __ t~ x2 /D ~ (16x106 cm2) / (2x10-6 cm2/sec) ~ 250,000 years So pore waters at that depth are closer to the average salinity and δ18 __ O over the past several glacial cycles rather than to the current oceanic values which are less salty and isotopically lighter. Schrag and DePaolo (1993) applied this technique 7 and estimated that seawater δ18O shifted by -1.0±0.25‰ since the last glacial maximum; assuming a sea-level change of 120 m (most recent estimate) would imply that the isotopic composition of the latest Pleistocene ice sheet was about – 32 ± 8‰. Later, Adkins et al. (2002) extended this approach to paleosalinity estimation, which allows for estimation of sea level change. Taken at face value, this data implies ~150 m fall in sea level at the Last Glacial Maximum (LGM), which is 20-30 m more than estimated from sea level observations. 3. Thus, depending somewhat on the exact δ18O of the glacial ice sheets and the actual sea level change, it appears that about 2/3 of the glacial-interglacial δ18O difference would be due to ice volume (the reverse of Emiliani's estimate). IV. Shackleton benthic foram δ18O. A. This same conclusion also was proposed by N.J. Shackleton, who suggested that temperature change in the cold deep ocean must have been small (some waters certainly could not have been colder than the freezing point), so that δ18O in benthic foraminifera would represent mainly ice-volume changes. Problem: benthic foraminifera are scarce (only a few percent of total foraminifera), so in general one requires 100x the raw sample size to get enough benthic foraminifera. Shackleton collaborated with workers who developed an improved mass spectrometer that required 10x less sample, which made benthic foram analysis practicable. B. Shackleton's first paper compared mixed benthic species with surface planktonic species and found a 1:1 correlation. So Shackleton argued that most ("at least 2/3") of the planktonic and benthic δ18O change was due to ice volume. This would imply that tropical surface temperatures did not change very much, contrary to Emiliani's conclusion of 8-10°C cooling. As this idea came to be accepted, it was common to find "at least 2/3" converted to "more than 2/3" or even "most of the benthic isotopic change is due to ice volume changes". 1. Work over recent years indicates that the 2/3 number was probably just about on the mark. The arguments require that you know more about the physical evidence for paleo sea level [to be covered later]. 2. Shackleton pointed out that if most of the oxygen isotope signal is due to ice volume, 18O records from throughout the ocean are "globally synchronous" (within the ~1000 year mixing time of the ocean) and can serve to establish a globally synchronous relative chronology. He also established several "sub-stages" to Emiliani's stratigraphy that enhanced the corrrelation precision of this stratigraphy. 3. Broecker and van Donk (1970) noted that in many data sets, deglaciations were often much more extreme than glacial inceptions ("sawtooth" records) and introduced the concept of "Terminations" which are the large abrupt deglaciations such as the transition out of the last glacial maximum, which was dubbed "Termination I". 4. Shackleton (1969) introduced the concept of “substages” to include cool/warm events within oxygen isotope stage 5: Warm 1 2 3 4 5 6 5a 5b 5c 5d 5e Isotope stages (Emiliani) Isotope sub-stages (this paper) Overall range 1.8 per mil Cold Postglacial Last interglacial 100 ka B.P. Generalized oxygen isotope palaeotemperature record through last six isotope stages, after Emiliani (1961). C. One of the first complications for these arguments was the observation that the oxygen isotope composition of shells of benthic foraminifera can differ from one another by substantial amounts. J.C. Duplessy et al. (1970) found that many species of benthic foraminifera have differences of up to 0.7 ‰ from one another. This weakened Shackleton's conclusion somewhat, and required further work on single-species benthics. 10 0 Pyrgo sp. +5 Various benthic species Planulina wuellestorfi +4 δ 18 +3 Globorotalia hirsuta Orbulina universa +2 +1 Globorotalia truncatulinoides 100 200 300 400 500 600 700 800 Depth in the core (mm) 18 Variation of δ O in three benthic and three pelagic species of Foraminifera in the core. D. Shackleton showed that Uvigerina spp. appears to be in equilibrium (compared to the 1970 O'Niel inorganic equation), and focused on developing benthic isotope stratigraphies based on this species. The conclusions remained much the same as his previous work, and people (other than an angry Emiliani) began to view benthic δ18O records as proxy sea-level records. For a period in the 1970's (and unfortunately, even today…), it was common for people to label benthic δ18O records as "ice volume", virtually neglecting any possible role for deep ocean temperature variations. E. The argument against Emiliani's tropical paleotemperature interpretation was strengthened by John Imbrie's paleo-ecological temperature method, which also indicated that that tropical sea surface temperatures had not changed much. This method needs to be covered in some detail before we proceed.