Spatial Statistics


Ongoing resaerch projects in SAMSI Program on Model Uncertainty. 

To be announced in May, 2019.

More info about this research can be found  here.


  • I will present work in July, 2019 at Joint Statistical Meetings in Denver, Colorado [link]
  • I will present work in May, 2019 at IMS/ASA Spring Research Conference in Virginia Tech [link]
  • I will present work in May, 2019 at Statistical (and mathematical) Perspectives on Uncertainty Quantification (SPUQ) workshop in UNC Chapel Hill [link]
  • I presented Initial Exploratory Analysis of Synthetic Storm Tracks in SAMSI Storm Surge of Model Uncertainty Working Group in Nov, 2018

Relevant Book

Hierarchical modeling and analysis for spatial data (2nd Edition)

This book is written by Alan Gelfand and his students Sudipto Banerjee and Bradley Carliln. The book keeps up to date with the evolving landscape of space and space-time data analysis and modeling. Since the publication of the first edition, the statistical landscape has substantially changed for analyzing space and space-time data. More than twice the size of its predecessor, Hierarchical Modeling and Analysis for Spatial Data, Second Edition reflects the major growth in spatial statistics as both a research area and an area of application.

Alan Gelfand is now Emeritus Professor in Duke University. I was lucky enough to take Alan's last Spatial Statistics course in 2018 Fall, right before he retired. Before the first class of this course I met him in his office, and we had a nice little talk there. He mentioned the scope of this course (Modeling data with spatial structure; point-referenced (geo-statistical) data, areal (lattice) data, and point process data; stationarity, valid covariance functions; Gaussian processes and generalizations; kriging; Markov random fields (CAR and SAR); hierarchical modeling for spatial data; misalignment; multivariate spatial data, space/time data specification), and some prerequisites this course expected. Then we talked about his MCMC history, related to the current big data burst, and about his recent exciting research focus on environment and ecology. Alan is super nice as a teacher who always encourages questions in class, and never fail to give satisfying answers. No matter how naive questions I poked, he always began with "Good one" when answering, and I could even see the twinkle in his eye. I was not Duke student but he still sent me his slides and homework and some other relevant paper via email every week. This course deepened my understanding towards spatial statistics, which facilitated my contributions to SAMSI Storm Surge research project later.

The Interview

David Warton interviews Alan Gelfand at the Statistics in Ecology and Environmental Monitoring conference in Queenstown, NZ. Alan was the keynote speak of SEEM. Later David posted the interview on Methods Blog of the British Ecological Society. Alan is best known for proposing Bayesian estimation of a posterior distribution using Gibbs sampling, in his classic papers 'Sampling-Based Approaches to Calculating Marginal Densities' [link] and 'Illustration of Bayesian Inference in Normal Data Models Using Gibbs Sampling' [link]. He discusses the origins of the idea that revolutionised Bayesian statistics, his current research, and his passion for ecology.


Other Research Work

Using Format