Research School of Earth Sciences, Australian National University, Canberra

Oxygen isotope data from deep sea cores and stratigraphic data from coral terraces in Papua New Guinea indicate significant sea level fluctuations that cannot be related to orbital forcing of climate. Four events with amplitudes of 10-25 m occurred between 30 and 63 ka. Earlier events are recognised at about 72, 90 and 115 ka, and an apparently very large event has been reported at 130 ka, during the Penultimate deglaciation. TIMS U-series dates from the coral terraces suggest that sea level rises between 30 and 63 ka correlate with isotopic interstadials 8, 12, 14 and 18 in Greenland but the amplitudes, rates, and peak timing of these events cannot be determined from dating alone.

A computer model of reef growth was used to search for the sea level curve that generates the best match to topographic, stratigraphic and age data from Huon Peninsula. Starting from a user-defined initial surface, the model generates successive reef profiles at 200-year time steps in response to given sea level changes and tectonic movements. Coral growth rate is assumed to be constant (G) from sea level to a critical depth Zc and decreases linearly below that to zero at a cut-off depth. Input data include user-defined rates for G, tectonic uplift (U) and coastal cliff retreat (E). Uplift can be uniform or intermittent. Model sequences were compared with dated, closely surveyed sections at Bobongara and Kanzarua for 30-75 ka period; for the period 75-140 ka, the Kwambu and Kwangam sections were used.

The search for a sea level curve commenced with the HP2 curve of Chappell and Shackleton (1986). Key parameters were varied, including timing, maximum, minimum and shape of each sea level oscillation. Oscillations also were deleted or inserted. Best-fit simulations are those that match the topographic profiles, facies structure and age measurements from observed terrace sequences. No simulations match all data perfectly but good-fit models are generated by sea level curves that define a narrow envelope with prominent peaks at 36, 44, 52 and 60 ka, and lesser peaks at 32.5, 49, 72, 90 and 115 ka. It proved difficult to match observations with simulations that included a large excursion at 130 ka. Peaks at 36 and 44 ka coincide Heinrich events H4 and H5; the peak at 60 ka and H6 may also coincide. Good-fit simulations indicate that sea level was rising relatively slowly (1-3 m.ka^-1) for several thousand years before these events, suggesting that H-events may be triggered when rising sea level destabilises the ice sheets.