<abstract xmlns="http://www.w3.org/1999/xhtml">Tree-ring data is commonly used in forest science and dendrochronology. As the collected datasets represent restricted populations of theoretical infinite sample size, an interesting question deals with the sample selection that is carried out during the fieldwork and through the data analyses. This paper considers the latter issue, by statistically examining a recently completed Scots pine dataset of timberline tree-rings from Lapland (northern Finland). Following the detrending of individual ring-width series, the composition of the data was restricted using a pre-determined criteria of linear correlativity between the individual sample series and the master chronology (Rmaster). This procedure reduced both the number of sites and the sample size (i.e. the number of individual tree-ring series) and altered the tree-ring statistics of the remaining subset of the data in a systematic fashion. It was seen that the first-order autocorrelation, mean sensitivity and standard deviation all ascended with the uplifted Rmaster criterion. Conspicuously, such filtering also reduced the correlation between the resulting tree-ring chronology and climate parameter. The results indicated that the screening of the data will alter the chronology statistics in a way that may be artificially generated, irrelative to the predetermined sample selection criteria. We remain to assume that the most fundamental selection of data is attained through the cross-dating process.</abstract>

Tree-ring data is commonly used in forest science and dendrochronology. As the collected datasets represent restricted populations of theoretical infinite sample size, an interesting question deals with the sample selection that is carried out during the fieldwork and through the data analyses. This paper considers the latter issue, by statistically examining a recently completed Scots pine dataset of timberline tree-rings from Lapland (northern Finland). Following the detrending of individual ring-width series, the composition of the data was restricted using a pre-determined criteria of linear correlativity between the individual sample series and the master chronology (Rmaster). This procedure reduced both the number of sites and the sample size (i.e. the number of individual tree-ring series) and altered the tree-ring statistics of the remaining subset of the data in a systematic fashion. It was seen that the first-order autocorrelation, mean sensitivity and standard deviation all ascended with the uplifted Rmaster criterion. Conspicuously, such filtering also reduced the correlation between the resulting tree-ring chronology and climate parameter. The results indicated that the screening of the data will alter the chronology statistics in a way that may be artificially generated, irrelative to the predetermined sample selection criteria. We remain to assume that the most fundamental selection of data is attained through the cross-dating process.

Timberline tree-ring statistics examined through chronology stripping

Finnish Forest Research Institute, P.O. Box 16, 96301 Rovaniemi, Finland

Finnish Forest Research Institute, P.O. Box 18, 01301 Vantaa, Finland

Forestry Studies

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

{"article-title":"Timberline tree-ring statistics examined through chronology stripping"}

Tree-ring data is commonly used in forest science and dendrochronology. As the collected datasets represent restricted populations of theoretical infinite...

Timberline tree-ring statistics examined through chronology stripping

Publié en ligne: 19 sept. 2014

Pages: 5 - 15

Reçu: 03 févr. 2012

Accepté: 11 mai 2012

DOI: https://doi.org/10.2478/v10132-012-0001-9

© 2014

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.