Abstract
Considerable attention has recently been devoted to the linear trend of drought at the decadal to inter-decadal time scale; however, the nonlinear variation of drought at multi-decadal scales and its relation to atmospheric circulation need to be further studied. The linear and nonlinear variations of the Palmer drought severity index (PDSI) in Shandong from 1900 to 2012 and its relations to the Pacific decadal oscillation (PDO), El Niño-Southern Oscillation (ENSO), Siberian high (SH) and Southern Oscillation (SO) phase changes from multi-scale are detected using linear regression, the Mann-Kendall test, ensemble empirical mode decomposition (EEMD) and the Pearson correlation analysis method. The results indicate that the PDSI shows no statistically significant linear change trend from 1900 to 2012; however, before (after) the late 1950s, PDSI shows a significant upward (downward) trend (P < 0.01) with a linear rate of 0.28/decade (-0.48/decade). From 1900 to 2012, the PDSI also exhibits a nonlinear variation trend at the inter-annual scale (quasi-3 and quasi-7-year), inter-decadal scale (quasi-14-year) and multi-decadal scale (quasi-46 and quasi-65-year). The variance contribution rate of components from the inter-annual scale is the largest, reaching 38.7%, and that from the inter-decadal scale and multi-decadal scale are 18.9% and 19.0%, respectively, indicating that the inter-annual change exerts a huge influence on the overall PDSI change. The results also imply that the effect of the four atmospheric circulations (PDO, ENSO, SH, SO) on PDSI at the multi-decadal variability scale are more important than that at the other scales. Consequently, we state that PDSI variation at the inter-annual scale has more instability, while that at the inter-decadal and multi-decadal scale is more strongly influenced by natural factors.