[Reconstruction of summer NDVI over the past 339 years based on the tree-ring width of Picea schrenkiana in Bayinbuluke, Central Tianshan, China].

The normalized difference vegetation index (NDVI) is widely used in various fields of vegetation research. Due to the short observation time, however, it is difficult to meet the research needs at long time scale. Here, we established a tree-ring width chronology (STD) based on Picea schrenkiana in Bayinbuluke, and calculated the correlation coefficient of chronology and NDVI with meteorological data. The results showed that both tree-ring width index and NDVI were significantly correlated with meteorological data. Combined with the significant positive correlation between width chronology and NDVI in June-August (r=0.7, P<0.01, n=38), summer NDVI (from June to August) was reconstructed over the past 339 years using a regression model. During 1680-2018, the reconstruction series had four dense vegetation periods (1738-1765, 1786-1798, 1964-1973 and 2000-2018) and five sparse vegetation periods (1690-1714, 1825-1834, 1850-1880, 1895-1920 and 1945-1955). The reconstruction reflected the hydrological signals in the central Tianshan Mountains. The comparison with the surrounding reconstructions revealed that when the runoff of Kaidu River increased and the local environment was humid, the vegetation coverage was high; otherwise the vegetation coverage was low. The extreme value of the reconstruction series also captured a series of natural disasters recorded in historical documents. Results of HYSPLT (Hybrid Single-Particle Lagrangian Integrated Trajectory Model) backward trajectory model and wind field analysis showed that NDVI anomalies were affected by the precipitation from Westerlies.