Ensemble standardization constraints on the influence of the tree growth trends in dendroclimatology

Tree growth trends can affect the interpretation of the response of tree-ring proxies (especially tree-ring width) to climate in the low-frequency band, which in turn may limit quantitative understanding of centennial-scale climate variability. As such, it is difficult to determine if long-term trends in tree-ring measurements are caused by age-dependent growth effects or climate. Here, a trend similarity ranking method is proposed to define the range of tree growth effects on tree-ring width chronologies. This method quantifies the inner and outer boundaries of the tree growth effect following two extreme standardization methods: curve fitting standardization and regional curve standardization. The trend similarity ranking method classifies and detrends tree-ring measurements according to the ranking similarity between the regional growth curve and their long-term trends through curve fitting. This standardization process mainly affects the secular trend in tree-ring chronologies, and has no effect on their inter-annual to multi-decadal variations. Applications of this technique to the Yamal and Torneträsk tree-ring width datasets and the maximum latewood density dataset from northern Scandinavia reveals that multi-centennial and millennial-scale temperature variations in the three regions provide substantial positive contributions to the linear warming trends in the instrumental period, and that the summer warming rate during the 20th century is not unprecedented over the past two millennia in any of the three regions.