Abstract Tree rings provide a natural archive of past environmental variability and are a valuable resource for paleoclimate reconstructions of the Common Era. However, tree rings typically only provide climate information during a portion of the year and uncertainty in the seasonality of climate influence on the proxy formation adds to the challenge of separating climatic signal from non‐climatic noise. We propose a Bayesian hierarchical model for the simultaneous estimation of a target reconstruction season and reconstruction of local climate. We estimate the target reconstruction season through the introduction of latent monthly weights, whose priors can be adjusted to reflect expert knowledge. Model behavior is explored using pseudoproxy experiments and applications to tree‐ring chronologies with known seasonal biases. Our proposed model provides meaningful information about the true growth‐determining season of the proxy and leads to improved uncertainty quantification while retaining overall model skill.

Read original article

Read original article