ABSTRACT

The aims of this thesis are to evaluate the coral Sr/Ca -SST relationship, and the reliability of coral-based seawater 18O reconstructions as indicators of SSS variations on seasonal to centennial time scales covering more than hundreds years. Furthermore, the potential of the corals to record major climate phenomena such as the El Niņo Southern Oscillation (ENSO) will be evaluated. For this purpose, we have analyzed coral cores from the western Pacific warm Pool (Timor, Indonesia) and the central Pacific (Tahiti, French Polynesia).

(1) Coral Sr/Ca-SST calibrations are calculated and compared using several coral Sr/Ca records and various instrumental SST datasets. The Tahiti Sr/Ca records are used as examples, because there are several coral Sr/Ca records from different coral cores available in this area. Usually only one Sr/Ca record from a single coral core is calibrated with an SST dataset selected from the different data products currently available. SST datasets from global SST data products refer to a defined area and have grid-box resolution. The SST record inferred from coral Sr/Ca is local and represents a point measurement. Therefore, using the regression equation by calibrating Sr/Ca with grid-SST data may lead to a bias in reconstructed SST. I calibrate Sr/Ca with grid-SST using several Sr/Ca records from Tahiti, i.e., three single Sr/Ca records and average Sr/Ca records from several cores. Different SST datasets are used. The results show that averaging proxy measurements from multiple coral colonies improves the correlation coefficients of the proxy and SST, provides a better estimate of the SST variance, and minimizes the residual SST. SST reconstructions using average proxy records are suggested to be more representative of regional SST variations.

(2) The methods of coral-based seawater 18O reconstructions from paired coral 18O and Sr/Ca measurements proposed by Gagan et al. (1994, 1998) and Ren et al. (2002) are evaluated and a new simpler and statistically correct method is proposed. Coral 18O and Sr/Ca measurements at Tahiti are used as examples. I find that the Gagan et al. (1994, 1998) method contains a methodological error, because coral 18O is calibrated with SST only, and if SST and SSS co-vary, there will be a bias in the slope estimate of coral 18O vs. SST. The Ren et al. (2002) method requires the calculation of the first derivatives from measured coral 18O and Sr/Ca and can be simplified using the centering method (i.e., by removing the mean values from the variables) proposed in this study. 18Osw variations calculated using either the Ren et al. (2002) or Centering method are the same. Ideally, a multiple linear regression (MLR) of coral 18O vs. SST and SSS should be used for the calculation of 18Osw and SSS. The MLR can account for covariant SST and SSS changes in the regression. To omit the regression constant in the MLR, I propose to center the MLR equations. The statistical error propagation of reconstructed 18Osw is calculated to assess our ability to resolve past variations in 18Osw (SSS). Only variations that are larger than the combined analytical uncertainty of coral 18O and Sr/Ca can be resolved.

(3) Further, the climatic variations recorded in the coral proxy records from Tahiti and Timor are carefully examined.
(a.) Tahiti is located in the south western tropical Pacific (SWTP) (24°S-10°S and 160°E-140°W) which is characterized by a horizontal SSS gradient between a south-eastward oriented tongue of fresh water featuring the warm pool and a westward oriented tongue of high salinity water advected by the southern branch of the south equatorial current coming from the central south Pacific. SSS variations in this area arise from the displacement of this salinity front. The salinity front separates the high salinity waters formed in the subtropical region (20°S, 120°W), where evaporation exceeds precipitation, and the low salinity waters of the warm pool area, where precipitation exceeds evaporation. The frontal region is located under the south Pacific convergence zone (SPCZ) which plays a significant role in global atmospheric circulation. The movement of SPCZ also influences the climate in this region. In the SWTP, ENSO related SST anomalies are an order of magnitude smaller than seasonal anomalies, while the magnitude ENSO-related SSS anomalies is twice as large as the seasonal signal. Tahiti is located exactly at the eastern margin of the SWTP.

Coral Sr/Ca records of the Tahiti cores show local SST. However, the coral Sr/Ca based SST reconstruction from Tahiti shows much larger variation on interannual times scales than grid-SST, i.e., the variance of the annual mean SST reconstruction based on the monthly Sr/Ca-SST calibration differs from instumental grid-SST. This reflects the different spatial scales of the proxy and grid-SST and is a serious problem to any attempt to assess the magnitude of regional SST variations from fossil corals. Seasonal SSS variations at Tahiti can not be resolved because the signal is smaller than the combined analytical uncertainty of coral 18O and Sr/Ca. At Tahiti, coral Sr/Ca and 18O are poor indicators of ENSO variability. This is not surprising, since Tahiti is located close to a zero line of the ENSO-related SST anomaly pattern in the tropical Pacific. However, the coral Sr/Ca records from Tahiti appear to record the SST anomalies in the Niņo 4 region at interannual scales. In the Niņo 4 region, the leading SST mode is a decadal mode, which may be captured in the proxy records from Tahiti.

The results of this thesis contribute significantly to coral paleoclimatology studies at Tahiti which will follow the IODP Tahiti expedition in 2005.

(b.) Timor is located in the western Pacific warm pool (WPWP) which is characterized by the warmest ocean water on the earth and is an important driver of the tropical atmospheric circulation. The system of currents transporting warm water out of the WPWP and into the Indian ocean is called the Indonesian Throughflow (ITF). The ITF plays an important role in the global climate system and also in modulating and terminating warm ENSO events and it is also influences the Australasian monsoon, which in turn drives the climate in the WPWP. The Timor coral site is located in an ITF exit passage (Ombai strait). The ITF, the Australasian monsoon, and the ENSO phenomenon are all influencing the climatic conditions in this area.

At Timor, coral Sr/Ca shows SST, while 18O is influenced both by SST and 18Osw (SSS). Seasonal SSS variations can be reconstructed from paired 18O and Sr/Ca measurements. Coral Sr/Ca and 18O do not show a clear correlation with El Niņo. Filtering of the Timor records in the ENSO frequency bands does not improve the correlation of the Timor coral proxy records with Niņo 3.4. However, spectral analysis shows significant coherence between Timor coral Sr/Ca and the Niņo 3.4 index at interannual periods. At decadal time scales, coral Sr/Ca and SST show similar trends. Spectral analysis of the Timor Sr/Ca record shows significant power at decadal periods. A running correlation between coral Sr/Ca and the Pacific Decadal Oscillation (PDO) index shows a high correlation for November to February average months.

Dissertation abstrak of Sri Yudawati Cahyarini
Email: yudawati@yahoo.com