The construction of oxygen isotope chronologies from tree-ring series sampled at different temporal resolution and its use as climate proxies: statistical aspects

Tree-ring oxygen stable isotope data series from conifers growing on the Dachstein Plateau (Austrian Alps) were selected to demonstrate the applicability of the serial pooling method using shifted 5-year tree-ring blocks for summer temperature reconstruction. The addressed method allows the construction of long isotope chronologies with significant climate correlation and well preserved climate sensitivity applying the irreducible sample replication of five trees. The linear regression model for temperature reconstruction is verifiable and the predicted data are well correlated with instrumental data, especially reproducing the long-term temperature trend. However, the reduced mean variance leads to loss of extreme years, which can be regulated by the combination of one data series in annual resolution with five shifted 5-year block data series. This significantly improves the variance of the mean chronology, sufficiently to identify extremes. Therefore, we recommend the use of mixed data sets as a compromise between essential sample replication and economic considerations.