Geostatistical Modeling

Geostatistical models are used for the analysis of geographic data consisting of measurements of a continuous, real-valued variable taken at various locations in two-dimensional space. Geostatistical modeling involves estimation of spatial dependencies and prediction of values at un-measured locations. Bayesian geostatistical modeling provides realistic estimation of prediction error, as well as other advantages, including the ability to combine information from disparate sources. However, Bayesian geostatistical modeling, which uses Markov chain Monte Carlo (MCMC) methods, is computationally intensive, especially when a large number of measurement locations are treated. We use high performance and Grid computing to develop parallel MCMC algorithms for Bayesian geostatistical modeling. We also apply our modeling methods into atmospheric science and public health applications.

Project Team: Marc Armstrong, Kate Cowles, Brian Smith, Shaowen Wang, Jun Yan

For testing purpose, download this sample dataset and upload it in your portlet.