Palaeoclimate Reconstruction

Bayesian inverse regression and spatio-temporal uncertainty quantification for reconstructing past climates from proxy data.

Reconstructing past climates from biological and geological proxies (such as fossil pollen counts) poses challenging statistical inverse problems involving zero-inflated compositional responses, non-linear ecological transfer functions, radiocarbon dating uncertainty, and stochastic climate volatility.

Our research develops fast approximate Bayesian inversion techniques and hierarchical spatio-temporal models that jointly propagate chronological, proxy, and model uncertainties when inferring palaeoclimate trajectories over thousands of years.

Project Publications

Sweeney, James, Michael Salter-Townshend, Tamsin Edwards, Caitlin E Buck, and Andrew C Parnell. 2018. “Statistical Challenges in Estimating Past Climate Changes.” WIREs Computational Statistics 10 (5): e1437.

Parnell, Andrew C, James Sweeney, Thinh K Doan, Michael Salter-Townshend, Judy RM Allen, Brian Huntley, and John Haslett. 2015. “Bayesian Inference for Palaeoclimate with Time Uncertainty and Stochastic Volatility.” Journal of the Royal Statistical Society Series C: Applied Statistics 64 (1): 115–38.

Parnell, Andrew C, James Sweeney, Thinh K Doan, Michael Salter-Townshend, Judy RM Allen, Brian Huntley, and John Haslett. 2012. “On Bayesian Modelling of the Uncertainties in Palaeoclimate Reconstruction.” arXiv Preprint arXiv:1206.5009. https://arxiv.org/abs/1206.5009.

Salter-Townshend, Michael, and John Haslett. 2012. “Fast Inversion of a Flexible Regression Model for Multivariate Pollen Counts Data.” Environmetrics 23 (7): 595–605.

Parnell, Andrew C, John Haslett, Michael Salter-Townshend, Brian Huntley, and Judy RM Allen. 2009. “Quantifying Uncertainty in Palaeoclimate Reconstructions.” In EGU General Assembly Conference Abstracts, 785.

Salter-Townshend, Michael. 2009. “Fast Approximate Inverse Bayesian Inference in Non-Parametric Multivariate Regression with Application to Palaeoclimate Reconstruction.” PhD thesis, Trinity College Dublin.

Haslett, John, Andrew Parnell, and Michael Salter-Townshend. 2006. “Modelling Temporal Uncertainty in Palaeoclimate Reconstructions.” In Proceedings of the 21st International Workshop on Statistical Modelling, 26–37.

Haslett, John, Matt Whiley, Sudipto Bhattacharya, Michael Salter-Townshend, Simon P Wilson, Judy RM Allen, Brian Huntley, and Fraser JG Mitchell. 2006. “Bayesian Palaeoclimate Reconstruction.” Journal of the Royal Statistical Society Series A: Statistics in Society 169 (3): 395–438.

Salter-Townshend, Michael, and John Haslett. 2006. “Modelling Zero Inflation of Compositional Data.” In Proceedings of the 21st International Workshop on Statistical Modelling, 448–56.