Millennia-long Tree-ring Chronologies as Records of Climate Variability in Finland

There have been great advances in the field of dendroclimatological research during the past decades. Tree-ring data have been used to derive various climate variables for certain localities and for larger regions. Characteristics of tree-rings provide measurements which can be used as records of climate variability prior to any instrumental weather observations. This is how dendroclimatology contributes to the urgent need for greater understanding of the nature of the fluctuations in climate that have occurred in the past and that are occurring in the present. Present study focus on the long tree-ring material collected from Finland and adjacent areas. Four regional datasets of tree-rings were used here to study their climatedependent variability and feasibility for palaeoclimate research. Datasets of carefully cross-dated ring-widths provided the annual increments of Scots pine (Pinus sylvestris L.) at the northern forest-limit, northern, middle and southern boreal forest zones in Finland. Ring-widths from northern forest-limit were used in the comparison of tree-ring standardization methods. Ring-width chronologies from forest-limit and south-east Finland were used to reconstruct past variations in climate, in temperature and precipitation, respectively. Characteristics of tree-rings from north to south exhibited a clear environmental and climatological gradient. Three northernmost regional chronologies showed significant and positive dependence on mid-summer temperatures (July). All four chronologies correlated significantly and positively with May precipitation. The latter relationship was greatest in the southernmost region where also June precipitation had similar influence on tree-rings. Three out of four regional chronologies correlated significantly with winter-time North Atlantic Oscillation. Comparison of standardization methods included two types of splines, negative exponential function with regression line as well as regional curve standardization (RCS). In addition, indices were computed by division and subtraction, as well as subtraction after power transformation. This part of the study focused on the preservation of the growth variability at long time scales. Yet, the behaviour of typical forest-limit ring-widths in the standardization process was thoroughly studied. It was shown that although RCS bears potentiality for preservation of great amount of lowfrequency variability in the resulting chronology, it is also the most sensitive method for temporal irregularities in the data. This fact points to the uncertainties in the behaviour of tree-ring indices at time scales longer than the mean segment length, to the issue which is still not completely understood. Tree-ring chronology from northern forest-limit was used in the reconstruction of mid-summer temperatures for past seven and half millennia. Due to wide inter-tree spacing, original tree-ring material was expected to contain non-climatic bias of tree ageing only, and ring-widths for this work were standardized using RCS method. Reconstruction explained 37 percent of the observed variability and was successfully verified using several statistical tests. Tree-ring chronology from south-east Finland was used in the reconstruction of early-summer (May-June) precipitation for past eleven centuries. Due to clear closedcanopy signature of tree-rings, standardization in this work was made using 67 %n splines in order to remove non-climatic growth disturbances prior to any palaeoclimatic interpretation. Reconstruction explained 31 percent of the observed rainfall variability; it was verified and correlated successfully with independent climatic series. Both of the tree-ring based palaeoclimate reconstructions provided evidence for greatly fluctuating Holocene climate. Anomalous changes in the reconstructed temperature occurred from annual to multicentennial time scales. Negative temperature anomalies of single years were shown to be associated with volcanic forcing. Due to methodological restrictions, the reconstructed precipitation was limited to exhibit fluctuations at interannual to multidecadal time scales. Uncertainties in both of the reconstructions were further discussed in the context of literature review.