Quantitative reconstruction of past climate is crucial to an understanding of climate mechanisms, and provides a means of defining model boundary conditions and of evaluating palaeoclimate simulations. In recent decades, transfer functions have been used to quantitatively reconstruct past environment from biological data (e.g. pollen, diatoms, insects, molluscs, foraminifera and dinoflagelates). There are two main approaches based on the principle of calculations: a regression-based technique and an analogue-based technique. The regression-based technique is also known as multivariate statistical analysis. With this method, transfer function equations are derived by modelling the responses of modern assemblages in relation to the environmental variable(s), and then by applying them to fossil samples to estimate environmental variable(s) for the geological past. The analogue-based technique includes the best modern analogues and response surfaces, with modern analogues being used to assign climate values to fossil assemblages. However, each method has its own limitations.

Wenying Jiang (Key Laboratory of Cenozoic Geology and Environment, IGGCAS) and her colleagues present an improved methodology of the best modern analogues technique (PFT-MAT) in which environmental proxies and a moisture index (α, i.e. ratio of actual evapotranspiration to potential evapotranspiration) are used to constrain the selection of modern analogues. The method is tested using high-resolution, precisely dated palaeorecords (pollen, Pediastrum and δ18O of authigenic carbonate) from Lake Bayanchagan, northern China. Results show that the improved PFT-MAT generally yields smaller error bars for the reconstructed climate parameters than the traditional PFT-MAT. In addition, three prominent climatic events are identified from the improved reconstructions; namely, a cold event around 8400 cal. yr BP and two warm events around 6000 and 2000 cal. yr BP, which is consistent with other regional palaeoclimatic records. Our data also show that changes in tree component correlate well with α variations during the entire Holocene, with the highest tree components and highest α values between 8000 and 5500 cal. yr BP, indicating the dominant role of α in the growth of trees in northern China rather than single temperature or precipitation. The improved PFT-MAT is therefore an efficient method for quantitative reconstructions of palaeoclimate in arid and semi-arid regions.

Download Here