New Publication
Adjusting significance of tree-ring
Exciting New Publication: Adjusting the Significance of Daily Climate Responses in Tree-Ring Proxies
Posted by Jim Stagge on November 06, 2024
Access the paper: Adjusting the Significance of Daily Climate Responses in Tree-Ring Proxies
I am thrilled to announce the publication of our recent study in the Journal of Forestry Research titled “Adjusting the Significance of Daily Climate Responses in Tree-Ring Proxies”, with Max Torbenson as the lead author.
Overview of the analysis presented in this paper exemplified by a July–August precipitation signal where the Ax is r = 0.50.
Study Abstract
Correlating tree-ring parameters with daily resolved climate data is becoming increasingly common for understanding the complex relationships between tree growth and the surrounding environment. However, with an increased number of calculated correlations, there is an inherent risk of spurious significance.
In this study, we present an analysis using synthetic weather and tree-ring data mimicking the statistical properties of ten real-world sites across Europe to quantify the extent to which numerous comparisons may inflate maximum correlations. Comparisons of different tree-ring proxies, considering varying overlapping period lengths and seasons, revealed 95th percentile correlation differences reaching 0.25 by chance.
Using synthetic tree-ring chronologies with an assigned non-signal (r = 0.00), spurious correlations can reach statistical significance in over 60% of tests. Correlation inflation is greater when: (1) the climate-proxy relationship is weaker; (2) comparison periods are shorter; and (3) the length of seasonal windows is longer. Autocorrelation in the proxy records does not appear to have a major effect.
These findings indicate that caution should be exercised when computing high numbers of correlations with limited observations. We provide tables listing correlation inflations for precipitation- and temperature-sensitive tree-ring chronologies that can inform interpretations of significance.
Why Does This Matter?
This study highlights the need for careful statistical analysis in dendroclimatic research. By providing concrete examples of how statistical significance can be misinterpreted when numerous correlations are computed, we hope to raise awareness of the limitations and challenges inherent in using tree-ring data as climate proxies.
Our findings stress the importance of using caution when drawing conclusions from climate-proxy relationships, especially when dealing with large datasets and numerous comparisons. We provide tools and tables for researchers to assess and adjust for correlation inflation, helping to refine interpretations in future studies.
Diagram of the general patterns of correlation inflation drivers suggested by the results.
Read the Full Study
The full publication is available in the Journal of Forestry Research.