The elemental composition of lake sediment cores is often the result of several independent processes. In this study we attempt to extract statistically independent climate related signals from µ-XRF multi element data of a core drilled from Lake Hazar in Eastern Anatolia, using the independent component analysis (ICA) method. In addition, we analysed ostracod shells for oxygen and carbon isotopes. The ICA method has advantages over traditional dimension reduction methods, such as principal component analysis or factor analysis, because it is based on maximal statistical independence rather than uncorrelatedness, where independence is a stronger property.
The Hz11-P03 core, which represents the last 17.3 ka, was recovered from Lake Hazar which, at times, formed the headwaters of the Tigris. Applying the ICA method, we selected two out of six independent components by measuring distance correlation similarity. We propose that one of the selected components can be read as a proxy for temperature and the other for precipitation in this region.
Our results indicate that the region was relatively cold and wet during the late glacial, between 17.3 and 14.8 ka BP, and wet and warm during Bølling-Allerød. The lake level dropped below today's level during the Younger Dryas stadial (12.49 and 11.76 ka BP), forming a marked hiatus in the core's stratigraphic record. During the beginning of the Holocene, while precipitation values were high, the temperature gradually increased until 8 ka BP. Between 8 and 5 ka BP, the region was warm but extremely dry. After 5 ka BP, around 3.5 ka BP temperatures suddenly fell, and three abrupt dry phases are observed around 3.5 ka and 2.8 ka and 1.8 ka BP.
application/zip, 3 datasets