Lena Ringstad Olsen

Accumulation variability derived from an ice core from coastal Dronning Maud Land, Antarctica

Abstract A 100 m long ice core was retrieved from the coastal area of Dronning Maud Land (DML), Antarctica, in the 2000/01 austral summer. The core was dated to AD 1737 by identification of volcanic horizons in dielectrical profiling and electrical conductivity measurement records in combination with seasonal layer counting from high-resolution oxygen isotope (δ18O) data. A mean long-term accumulation rate of 0.29 ma–1w.e. was derived from the high-resolution δ18O record as well as accumulation rates during periods in between the identified volcanic horizons. A statistically significant decrease in accumulation was found from about 1920 to the present. A comparison with other coastal ice cores from DML suggests that this is a regional pattern.

Multiscale spectral analysis for detecting short and long range change points in time series

Identifying short and long range change points in an observed time series that consists of stationary segments is a common problem. These change points mark the time boundaries of the segments where the time series leaves one stationary state and enters another. Due to certain technical advantages, analysis is carried out in the frequency domain to identify such change points in the time domain. What is considered as a change may depend on the time scale. The results of the analysis are displayed in the form of graphs that display change points on different time horizons (time scales), which are observed to be statistically significant. The methodology is illustrated using several simulated and real time series data. The method works well to detect change points and illustrates the importance of analysing the time series on different time horizons.

Bayesian multiscale feature detection of log-spectral densities

A fully-automatic Bayesian visualization tool to identify periodic components of evenly sampled stationary time series, is presented. The given method applies the multiscale ideas of the SiZer-methodology to the log-spectral density of a given series. The idea is to detect significant peaks in the true underlying curve viewed at different resolutions or scales. The results are displayed in significance maps, illustrating for which scales and for which frequencies, peaks in the log-spectral density are detected as significant. The inference involved in producing the significance maps is performed using the recently developed simplified Laplace approximation. This is a Bayesian deterministic approach used to get accurate estimates of posterior marginals for latent Gaussian Markov random fields at a low computational cost, avoiding the use of Markov chain Monte Carlo techniques. Application of the given exploratory tool is illustrated analyzing both synthetic and real time series.