Online Program Home
My Program

Abstract Details

Activity Number: 619 - Spatial and Spatial-Temporal Statistics
Type: Topic Contributed
Date/Time: Thursday, August 2, 2018 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics and the Environment
Abstract #327214
Title: Extended Ensemble Kalman Filters for High-Dimensional Hierarchical State-Space Models
Author(s): Jonathan Stroud* and Matthias Katzfuss and Christopher K. Wikle
Companies: Georgetown University and Texas A&M University and University of Missouri
Keywords: Filtering; Smoothing; Sequential Monte Carlo; MCMC; Spatio-Temporal Model; Covariance

The ensemble Kalman filter (EnKF) is a computational technique for approximate inference in spatio-temporal state-space models. It has been successfully used in many real-world nonlinear data-assimilation problems with very high dimensions, such as weather forecasting. However, the EnKF is most appropriate for linear Gaussian state-space models with known parameters. Here, we consider a broader class of hierarchical state-space models, which includes two additional layers: The parameter layer handles unknown variables that cannot be easily included in the state vector, while the transformation layer allows for non-Gaussian observations. For Bayesian inference in these models, we propose a general class of extended EnKFs, which approximate inference on the state existing Bayesian methods (e.g., Gibbs sampler or particle filter) using the EnKF or the ensemble Kalman smoother. Extended EnKFs enable approximate, computationally feasible filtering and smoothing in many high-dimensional, nonlinear, and non-Gaussian space-time models with unknown parameters. We highlight several examples including assimilation of heavy-tailed and discrete data and online or offline parameter estimation.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program