Not logged in
Data Publisher for Earth & Environmental Science

Hayes, Angela; Kucera, Michal; Kallel, Nejib; Sbaffi, Laura; Rohling, Eelco J (2005): Glacial Mediterranean sea surface tempertures based on planktonic foraminiferal assemblages. PANGAEA,, Supplement to: Hayes, A et al. (2005): Glacial Mediterranean sea surface temperatures based on planktonic foraminiferal assemblages. Quaternary Science Reviews, 24(7-9), 999-1016,

Always quote citation above when using data! You can download the citation in several formats below.

RIS CitationBibTeX Citation

We present a new reconstruction of Mediterranean sea surface temperatures (SST) during the last glacial maximum (LGM). A calibration data set based on census counts of 23 species of planktonic foraminifera in 129 North Atlantic and 145 Mediterranean core top samples was used to develop summer, winter and annual average SST reconstructions using artificial neural networks (ANNs) and the revised analogue method (RAM). Prediction errors determined by cross-validation of the calibration data set ranged between 0.5 and 1.1 °C, with both techniques being most successful in predicting winter SSTs. Glacial reconstructions are based on a new, expanded data set of 273 samples in 37 cores with consistent minimum level of age control.
The new LGM reconstructions suggest that the east-west temperature gradient during the glacial summer was 9 °C, whereas during the glacial winter, the gradient was 6 °C, both some 4 °C higher than that existing today. In contrast to earlier studies, our results tend to suggest much cooler SST estimates throughout the glacial Mediterranean, particularly in the eastern basin where previous SST reconstructions indicated a decrease of only 1 °C. Our new SST reconstructions will provide the modelling community with a detailed and updated portrayal of the Mediterranean Sea during the LGM, setting new targets on which glacial simulations can be tested.
MARGO_0000 * Method/Device: Literary studies (LIT)
3 datasets

Download Data

Download ZIP file containing all datasets as tab-delimited text — use the following character encoding: