A double bootstrap approach to Superposed Epoch Analysis to evaluate response uncertainty

Mukund P. Rao, Edward R. Cook, Benjamin I. Cook, Kevin J. Anchukaitis, Rosanne D. D'Arrigo, Paul J. Krusic, Allegra N. LeGrande

Research output: Contribution to journalArticlepeer-review

41 Scopus citations


The association between climate variability and episodic events, such as the antecedent moisture conditions prior to wildfire or the cooling following volcanic eruptions, is commonly assessed using Superposed Epoch Analysis (SEA). In SEA the epochal response is typically calculated as the average climate conditions prior to and following all event years or their deviation from climatology. However, the magnitude and significance of the inferred climate association may be sensitive to the selection or omission of individual key years, potentially resulting in a biased assessment of the relationship between these events and climate. Here we describe and test a modified double-bootstrap SEA that generates multiple unique draws of the key years and evaluates the sign, magnitude, and significance of event-climate relationships within a probabilistic framework. This multiple resampling helps quantify multiple uncertainties inherent in conventional applications of SEA within dendrochronology and paleoclimatology. We demonstrate our modified SEA by evaluating the volcanic cooling signal in a Northern Hemisphere tree-ring temperature reconstruction and the link between drought and wildfire events in the western United States. Finally, we make our Matlab and R code available to be adapted for future SEA applications.

Original languageEnglish (US)
Pages (from-to)119-124
Number of pages6
StatePublished - Jun 2019


  • Bootstrap
  • Dendrochronology
  • Dendroclimatology
  • Disturbance
  • Fire-scars
  • Paleoclimate
  • Resampling
  • Tree-rings
  • Uncertainty
  • Volcanos
  • Western US
  • Wildfire

ASJC Scopus subject areas

  • Ecology
  • Plant Science


Dive into the research topics of 'A double bootstrap approach to Superposed Epoch Analysis to evaluate response uncertainty'. Together they form a unique fingerprint.

Cite this