Reconstructing Past Climate from Natural Proxies and Estimated Climate Forcings Using Long Memory Models

• Activity Number: 509
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: ASA Advisory Committee on Climate Change Policy
• Author(s): Bo Li*+
• Companies: Purdue University
• Keywords: external forcings ; long-memory dependence ; proxies ; temperature reconstruction

Abstract

We produce new reconstructions of Northern Hemisphere temperature anomalies over the last millennium, based on a model which includes the effects of external climate forcings and accounts for the long-memory features displayed in the data sets. Our reconstruction is based on two linear models, one in which the latent temperature series is linearly related to three main external forcings, and another in which the observed temperature proxy data is linearly related to the unobserved temperatures. We have carefully investigated the correlation structure in regression errors using rigorous statistical tests, and we find that a long memory fractional Gaussian noise is proper for both linear models. We show that the long memory model helps to quantify the uncertainty of the reconstruction more precisely, and the use of external climate forcings is crucial in substantially reducing uncertainty levels. Our reconstruction compares favorably with previous results, as measured via the validation metric of empirical coverage probability for 20th-century observations. Finally, we provide a measurement of equilibrium climate sensitivity over the last millennium based on our reconstruction.