Applications of Gaussian Subordination Models in Climate Research

• Activity Number: 133
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: IMS
• Author(s): Sucharita Ghosh*+
• Companies: Swiss Federal Research Institute WSL
• Keywords: long memory ; kernel smoothing ; rapid climate change ; changing distributional patterns

Abstract

We consider estimating the mean function, its derivatives and rapid change points for unevenly spaced time series generated by Gaussian subordination models Y(i)=G(Z(i),t(i)) where G has Hermite rank q. Asymptotic results are derived under suitable conditions on the spacing of time points. The results are illustrated by examples from paleoclimatology where the observations date back several thousand years. A related problem is estimation of time itself which is also discussed briefly. A special case is when observations are evenly spaced and these have applications in modeling instrumental climate records. We discuss asymptotics and applications to fossil oxygen isotopes and pollen records and other instrumental data from the 20th century. This has been a joint collaboration with several colleagues and partially funded by the Swiss National Science Foundation.