A transfer-function model developed from an extensive surface-pollen data set in northern China and its potential for palaeoclimate reconstructions

A total of 237 surface soil pollen samples were collected from less disturbed zonal communities, and a new surface pollen data set has been established in northern China. Pollen—climate parameter transfer functions using weighted-averaging regression and calibration partial least squares (WA-PLS) model as a tool to reconstruct past climate changes from pollen data, were tested with this new surface pollen data set. The results of cross-validation showed that the method worked very well with precipitation ( R = 0.94, R 2 jack = 0.89, RMSEP jack = 69 mm, the safe dynamic range of the reconstruction was 40 mm~800 mm) and worked well with temperature ( T ann : R = 0.84, R 2 jack = 0.71, RMSEP jack = 2.0°C, the safe dynamic range of the reconstruction was −4°C~13°C; MT wa : R = 0.79, R 2 jack = 0.63, RMSEP jack = 2.5°C, and the safe dynamic range of the reconstruction was 8°C~26°C), indicating that the surface pollen data set and WA-PLS reconstruction technique in this paper have strong potential for palaeoclimatic reconstruction.