Maximum July–September temperatures derived from tree-ring densities on the western Loess Plateau, China

Tree-ring densitometric studies conducted in semi-arid regions are exceedingly rare. In this study, we evaluate a new maximum latewood density chronology from a collection of Picea purpurea trees growing in semi-arid northwestern China for climatic reconstruction purposes. We find statistically significant correlations between annual maximum density variations and July–September maximum temperatures over the 1957–2014 calibration period. Using a linear model to transfer the density variations to temperature estimates, we then reconstruct maximum summer temperatures from 1840 to 2014. Statistical analyses and verification tests demonstrate that the resulting transfer function is both reliable and stable. The reconstructed temperature series shows strong inter-annual and decadal variability. Several extremely cool and warm years are apparent on an inter-annual scale, while three extended cool periods and a warm period are observed on a decadal scale. In addition, the reconstructed temperature series exhibits a pattern synchronous with that of the Indian monsoon index, as well as a moderate relationship to volcanic eruptions. This densitometric study reveals typical climatic variation characteristics and provides useful data for deepening our understanding of climate history in the semi-arid regions of China.