How warm were the tropical oceans at the Last Glacial Maximum (LGM), 21 thousand years ago? A good estimate would be valuable for several reasons. First, it would help us to understand LGM climates and the ecological response to this climate. Second, it would provide a constraint by which to judge climate models when they are used to simulate LGM climates – if the climate model output resembles the past conditions, it increases our confidence that they give good projections of future climate with increased greenhouse gases. Third, the change in temperature between the LGM and the pre-industrial can be used to estimate climate sensitivity as the forcings (greenhouse gases, albedo and orbital) are known.
The first global estimate of SST changes since the LGM, which came from the CLIMAP project in the 1970s and 1980s, mainly based on planktonic foraminifera, showed limited cooling in the tropics and warming in the subtropical Pacific.
The MARGO project, a more recent compilation of global SST proxies for the LGM, found more extensive tropical cooling, typically of 1-2°C, with an east-west gradient in anomalies that climate models do not replicate.
Some other proxy data suggest that the tropics were much cooler than reconstructed by MARGO or CLIMAP. Strontium/calcium ratios of corals suggest SSTs were up to 6°C cooler, which might imply a much higher climate sensitivity than the 2.4°C (1.4-3.5°C) estimated using the MARGO data. However there are few coral data, perhaps partly because the oceans were too cold for widespread coral growth, and partly because most LGM corals are difficult to sample beneath 120 m of water).
This discrepancy has interested me and is part of the motivation for the reanalysis of the MARGO foraminiferal assemblage data I have started.
A paper in a recent issue of Nature Geosciences by Tripati et al adds some more data that suggests that CLIMAP and MARGO underestimated tropical cooling. Tripati et al use clumped isotope palaeothermometry to reconstruct sea surface conditions off Papua New Guinea since the LGM. Clumped isotope palaeothermometry is a relatively new stable isotope method. Traditional methods have used the ratio of 18O to 16O in calcite, for example from foraminifera tests, but this ratio is sensitive to both temperature and the isotopic composition of the source water, which complicates interpretations. Clumped isotope palaeothermometry uses the temperature sensitive-tendency of the heavier carbon and oxygen isotopes to clump together in the same calcite molecule. Carbonate molecules with a heavy carbon atom and an heavy oxygen atom, 13C18O16O, are more common than expected by chance, but become rarer as the temperature at which the calcite formed increases. About half a percent of oxygen atoms are 18O, making the ratio of 18O to 16O much easier to measure than the excess of 13C18O16O, which is present in 44 ppm of carbonate molecules. Importantly, clumped isotope palaeothermometry, abbreviated as Δ47 (the sum of the isotopic weights), is independent of the source water composition and seems to show few species-specific vital effects.
Tripati et al find that the SST of Papua New Guinea was 4-5°C cooler than modern, and that the depression of LGM snowlines in the mountains of Papua New Guinea by almost a kilometre is not consistent with the smaller SST cooling reconstructed by MARGO and by many of the climate models.
If the work of Tripati et al is supported by new reconstructions, and perhaps by a reanalysis of the MARGO foraminifera assemblage data, then questions will need to be asked about the climate model performance. Any discrepancy could be because the models have underestimated climate sensitivity, alternatively the increases in dust and aerosols at the LGM, which have not been incorporated into the climate models, may have helped cool the climate. If the dust can explain the discrepancy, then perhaps current estimates of climate sensitivity do not need to be greatly increased.
Tripati et al 2014. Modern and glacial tropical snowlines controlled by sea surface temperature and atmospheric mixing. Nature Geoscience, 7, 205–209 doi:10.1038/ngeo2082