Ensemble Empirical Mode Decomposition As An Alternative For Tree-Ring Chronology Development

Abstract

Since its establishment, tree-ring analysis has benefitted several scientific fields. Because of its many advantages, dendrochronology is a first choice to reconstruct past environmental variability. Two major concerns about the current tree-ring reconstruction paradigm are the subjective choices of detrending functions and the lack of fidelity to data of chronology generation methods. It is difficult to recover the original tree-ring data once they have been detrended and standardized. In this study, ensemble empirical mode decomposition (EEMD) is introduced as an objective high-fidelity stand-alone approach for developing tree-ring chronologies. Basic concepts of EEMD, recommended steps in developing chronologies, and available public domain programs are discussed. To demonstrate the potentials of EEMD for chronology development, two examples are provided, one for climate and the other for streamflow reconstructions. In both examples, EEMD chronologies show higher correlations with the instrumental data and have more power in their spectra than the ones developed based on the current tree-ring reconstruction approach. General usage concerns and cautions are also addressed.