Improved El Nino Southern Oscillation signals extracted by principal component analysis of tree-ring oxygen isotope records from the East Asian monsoon region of China

There is a great need for the reconstruction of independent El Nino Southern Oscillation (ENSO) from different archives and locations to provide a robust understanding of its driving and feedback mechanisms on long time scales. Though tree-ring cellulose delta O-1(8) (delta O-1(8)TR) records from the East Asian monsoon (EAM) region of China have the potential to reconstruct ENSO on an annual time scale, delta O-1(8)TR ENSO signals may be masked by local climate effects since any individual delta O-1(8)TR record is influenced by both large-scale atmospheric circulation and local climate. Few attempts have been made to reduce the uncertainties of the reconstructed ENSO index by delta O-1(8)TR records from the EAM region of China by ensemble data analysis. In this study, twelve published delta O-1(8)TR records spanning over 100 years (1902-2003 CE) have been collected and processed using principal component analysis (PCA). The first principal component (PC1) extracted from the delta O-1(8)TR records shows a positive/negative relationship with the annual Nino 3.4 index/southern oscillation index (SOO, with correlation coefficients higher than those between single delta O-1(8)TR record and the Nino 3.4 index/SOL suggesting the ENSO signal is improved by PCA. A 31-year running correlation on PC1- Nino 3.4 index/SOI reveals an increasingly stronger correlation between delta O-1(8)TR and ENSO since the last century, inferring that, in the EAM region, the delta O-1(8)TR-ENSO correlation may become much stronger under continuous global warming in the future. Spectrum and spatial correlation analysis further support the idea that PC1 responds to ENSO. The significant relationships between observational precipitation delta O-1(8) and Nino 3.4 SST anomaly and PC1 from 1973-2003 CE suggest that precipitation delta O-1(8) is a key link to connect delta O-1(8)TR with ENSO. This study shows the advantage of ensemble data analysis on delta O-1(8)TR records in EAM region of China, which could better extract ENSO signals and reduce the uncertainties of reconstructed ENSO variations by individual delta O-1(8)TR records from the EAM region of China. It provides an alternative independent annual indicator for ENSO reconstruction on long time scales when more delta O-1(8)TR records from this region are extended in the near future.