ALBERT

Description: A major problem in modern dendrochronology is that the methods traditionally used for linking tree-ring growth data to climate records are not well suited to reconstructing low-frequency climatic variations. In this study, we explored the alternative ensemble empirical mode decomposition (EEMD) to detrend tree-ring records and extract climate signals without removing low-frequency information. Tree cores of Pseudotsuga menziesii var. glauca (Mayr.) Franco were examined in a semi-arid forest in southern interior British Columbia, western Canada. Ring width data were decomposed into five oscillatory components (intrinsic mode functions, IMFs) of increasingly longer periodicities. IMF 1 was considered white noise, IMF 2 was used to create the first diameter growth index (DGI-1), and IMF 3 and IMF 4 were combined to create the second diameter growth index (DGI-2), whereas IMF 5 and the residual term together were considered as the trend term. The highest significant cross-correlations between DGI-1 and the NAOAugust, NIÑO12May, and PDOJanuary indices were found at 1-year lags. DGI-2 had positive and persistent correlations with NAOJune and PDOMay at 0- to 3-year lags and with NAOMay at 2- and 3-year lags. Our results indicate that periods of slow growth in the tree-ring record matched periods of drought in the North American Pacific Northwest. Such water-limiting conditions are likely caused by oscillatory patterns in the Pacific Ocean sea surface temperatures that influence precipitation in the Pacific Northwest. These drought events are likely exacerbated by changes in winter precipitation (snowpack) related to oscillations of the Atlantic Ocean sea surface temperatures, highlighting the ecological effects of both oceans on terrestrial ecosystems. Such relationships could not be easily found by traditional tree-ring analyses that remove some of the low-frequency signal, and therefore, we suggest EEMD as an additional tool to establishing tree growth – climate relationships.