PALEOTEMPERATURE ESTIMATION METHODS - SENSITIVITY TEST ON 2 WESTERN EQUATORIAL PACIFIC CORES

This study examines three methods of estimating past sea surface temperature (SST): Imbrie and Kipp's transfer function method (IKM), the Modern Analogue Technique (MAT) and the Weighted Averaging Method (WAM). By testing a number of experimental transfer functions on two western equatorial Pacific cores, RC17-177 and ERDC-93P, it is shown that SST estimation is sensitive to the number of factors and to the selection of core-top samples in the calibration. The most plausible result was given by an equation which is calibrated from a restricted region which is sufficiently large to fully cover the potential range of down-core SST estimates. The MAT gave spurious SST estimates, probably due to the failure of the dissimilarity coefficient. The WAM persistently gave lower SST estimates than the IKM because of the averaging effect of the method, but the SST time series estimated by the two methods show almost identical fluctuation patterns. The tolerance weighting approach in the WAM may be inappropriate because tolerances of different species may have different implications. All the above methods are affected by the distribution of calibration samples.