Online Program Home
My Program

Abstract Details

Activity Number: 68
Type: Topic Contributed
Date/Time: Sunday, July 31, 2016 : 4:00 PM to 5:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #319043 View Presentation
Title: Modeling Spatial Compositional Data: Reconstructions of Past Land Cover and Uncertainties
Author(s): Behnaz Pirzamanbein* and Johan Lindström and Anneli Poska and Marie-José Gaillard-Lemdahl
Companies: Lund University and Lund University and Lund University and Linnaeus University
Keywords: Gaussian Markov Random Field ; Dirichlet observation ; Adaptive Metropolis adjusted Langevin ; Fossil pollen records ; Confidence regions

We construct a hierarchical model for spatial compositional data which is used to reconstruct past land-cover compositions (coniferous forest, broadleaved forest, and unforested/open land) for 5 time periods during the past 6000 years over Europe. The model consists of a Gaussian Markov Random Field (GMRF) with Dirichlet observations. A block updated Markov chain Monte Carlo (MCMC), including an adaptive Metropolis adjusted Langevin step, is used to estimate model parameters. The sparse precision matrix in the GMRF provides computational advantages leading to a fast MCMC algorithm. Reconstructions are obtained by combining sparse pollen-based estimates of vegetation cover with output from a dynamic vegetation model and scenarios of past deforestation. To evaluate uncertainties in the predictions a novel way of constructing joint confidence regions for the composition at each prediction location is proposed. The model is evaluated through cross validation for 5 time periods, and by comparing reconstructions for the recent past to a present day European forest map. The evaluation results are promising and the model is able to capture known structures in past land-cover compositions.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association