Pan-American Advanced Study Institute

on Spatio-Temporal Statistics

June 16-26, 2014

Búzios, RJ, Brazil


This Pan-American Advanced Study Institute consists of a winter school with two workshops. The winter school, which will have about 40 participants,  covers a broad range of theoretical and applied areas of spatial and spatio-temporal statistics, with emphasis on climate, air pollution, and ecology. In parallel with the winter school there will be two advanced workshops. The winter school students are expected to also participate in the workshops. One workshop will be dealing with nonstationary covariance structures, and is aiming at a comparison of the various approaches available in the literature. The other workshop covers methods for multivariate spatial statistics. It will be possible to participate only in the workshops, but such participation is by invitation only. .

The schedule will leave the afternoons open for outdoor activities, as winter in this part of Brazil generally is quite pleasant.

Here are the results of a questionnaire before the conference started about the background of the participants.

Description of institute

Data and software page

Winter school

Note: this schedule has been updated. Latest update June 23.

Sunday World Cup: 13 SWI-ECU 16 FRA-HON 19 ARG-BIH

Monday, June 16

World Cup: 13 GER-POR 16 IRN-NIG 19 GHA-USA

8:30-9 Registration and introduction.
9-9:45 Geostatistics (Guttorp) Basic introduction to stationary spatial prediction (kriging) with focus on the Gaussian situation. Classical and Bayesian approaches.
0 1 4 8 17

9:45-10 Coffee break

10-11 Practicum: geoR (Ribeiro Jr.)
0 2 3 10 14
11-11:45 Application: Air quality (Rodrigues)
0 1 6 14 8

12 Lunch

17-17:45 Covariance functions (Jun) Bochner's theorem and positive definite functions. Representation of isotropic covariance. Classes of covariance functions. Compactly supported covariance. Spherical models. Estimation of covariances.
0 3 6 14 9

17:45-18 Coffee break

18-19:45 Practicum and Application: Housing data analysis in geoR (Schmidt)
0 1 11 11 7

Tuesday, June 17

World Cup: 13 BEL-ALG 16 BRA-MEX 18 RUS-KOR

9-9:45 Nonstationary models I (Sampson) Deformation approach to spatial nonstationarity. Thin-plate splines. Bayesian estimation.
0 4 6 14 9

9:45-10 Coffee break

10-10:45 Application: French precipitation (Meiring)
0 2 10 12 7
10:45-11:45 Practicum (Sampson and Meiring)
0 2 10 12 7

12 Lunch

18-18:45 Nonstationary models II (Calder) The kernel approach. Weighted averages. Basis function expansions.
1 4 6 13 7

18:45-19:00 Coffee break

19-20 Practicum (Calder)
1 14 10 7 9
20-20:45 Application: Ozone and wind: including covariates in covariances. (Schmidt)
0 2 8 10 10

Wednesday, June 18 

World Cup: 13 AUS-NED 16 ESP-CHI 18 CAM-CRO

9-9:45 Gaussian Markov Random Fields (Lindström) Sparse precision matrices. Gaussian MRF simulation and estimation. The stochastic PDE approach. Integrated nested Laplace approximations.
0 4 4 11 12

9:45-10 Coffee break

10-11 Practicum: R-INLA (Bolin & Lindström)
0 3 8 8 11
R code
11-11:45 Application: Confidence intervals for contour lines (Bolin)
0 5 5 11 9

12 Lunch

Nonstationary workshop
17-18:20  Opening session: Overview of nonstationary covariance models

18:20-18:40 Coffee break

18:40-20:00 :  Group discussions.
Group 1: Space-time means
Group 2: Visualization
Group 3: Space-varying variogram coefficients
Group 4: Method performance metrics

Thursday, June 19

World Cup: 13 COL-CIV 16 URU-ENG 19 JPN-GRC

9-9:45 Spatiotemporal processes (Gamerman) Singular value decomposition. Space-time covariance. Dynamic linear models.
0 1 7 17 7

9:45-10 Coffee break

10-11 Practicum: SpatioTemporal (Lindström)
R code
0 1 8 13 8
11-11:45 Application: Wetlands assessment (Hoeting)
0 1 8 12 9

Nonstationary workshop
17-18:20 Housing data (spatial)
Bolin: Non-gaussian fit

18:20 Coffee break

18:40-20:00 Air quality data (space-time)
Jun and Jeong

Friday, June 20 

World Cup: 13 ITA-CRI 16 SWI-FRA 19 HON-ECU

9-9:45 Multivariate models (Banerjee) Kronecker model. Co-regionalization with spatially varying coefficients. Hierarchical approach.
0 1 8 13 9

9:45-10 Coffee break

10-11 Practicum: RandomFields (Schlather)
0 1 13 7 8
11-11:45 Application: Sea surface temperature estimation (Sansó)
0 1 9 13 8

12 Lunch

Nonstationary workshop
17-18:20 Temperature (global space-time)

Jun: Global analysis.
Furrer: Global analysis
18:20-18:40 Coffee break

18:40-20:00 Simulated data sets

20:00-??? PARTY (pool area)

Saturday, June 21

World Cup: 13 ARG-IRN 16 GER-GHA 18  NIG-BIH

Nonstationary workshop
9-10:00 Simulation data results

Meiring: Deformation simulation
Risser: Process convolution simulation
Lindström: Simple analysis of Higdon set. R code
Sampson: Warps and biorthogonal grid for irregularly spaced Higdon simulation

10:00-10:20 Coffee break

10:20-11:30 Group discussions
White paper structure
Simulation design
Estimation and comparison
11:30-12:00 Group reports and conclusions

12 Lunch

Saturday afternoon through Sunday are open for excursions etc. There will be a boat tour (for R$70) on Sunday. We meet at the reception by 9:30; the boat leaves at 10:00 from the pier at the near edge of downtown.
(CANCELLED--possibly Tuesday afternoon)

Sunday World Cup: 13 BEL-RUS 16 KOR-ALG 19 USA-POR

Monday, June 23


9-9:45 Extremes (Cooley) Generalized extreme value distributions. Generalized Pareto distribution. Max-stable processes. Composite likelihood.

0 0 4 6 21
9:45-10 Coffee break

10-11 Practicum: SpatialExtremes (Ribatet)
0 0 9 9 11
11-12 Application: Risk assessment and extreme weather (Bravo)
0 1 9 9 11

12 Lunch

Multivariate workshop
19-19:50 Opening session:  Sudipto Banerjee: On complex spatial dependencies: Low-rank cross-covariance models

19:50-20:10 Coffee break

20:10-21 Bo Li: An approach to modeling asymmetric multivariate spatial covariance structures

Tuesday, June 24


Boat tour in the morning (We meet at the reception by 9:30). Lunch follows at the hotel.

14:00 Jony Pinto Jr. Point process with spatio-temporal heterogeneity
14:15 Abhirup Datta: Hierarchical nearest neighbor Gaussian process models for large spatial datasets
14:30 Farzaneh Safavimanesh: Bayesian inference for the Poisson line cluster point processes
14:45 Henrike Häbel: Spatial statistical analysis of the micro-structure of nano-silica particle gels
15:00 Somak Dutta: Matrix-free conditional simulations of Gaussian Markov random fields and their applications
15:15 Soojin Roh: Multivariate localization methods for ensemble Kalman filtering

15:30 Coffee break

15:45 Michael Horrel: Half-Spectral Space-Time Models and a Natural Condition
16:00 Jorge Mazzeo: Spatial Organization of Neuroepithelial Cells during Embryo Development
16:15 Zach Weller: Hurricane Forecast Assessment
16:30 Behnaz Pirzamanbein: Reconstructing the past land cover using fossil pollen estimates
16:45 Erin Schliep: Climate-driven individual tree growth modeling fusing two data sources

17:00-17:10 Short break.

Multivariate workshop
17:10-18:00  Martin Schlather: Conditional Modelling of Extreme Wind Gusts by Bivariate Brown-Resnick Processes
18:00-18:50 Tanya Apanasovich: Matérn Class of Cross-Covariance Functions for Multivariate Random Fields

18:50-19:10 Coffee break

19:10-20:00  Xiangping Hu: Multivariate Gaussian random fields with SPDEs

Wednesday, June 25


9-9:45 Spatial point processes (Särkkä) Cluster models. Self-exciting models. Markovian point processes. Residual analysis.
1 0 4 10 17

9:45-10 Coffee break

10-11 Practicum: Spatstat (Särkkä and Guttorp)
1 2 3 10 13
11-11:45 Application: Forestry (Dean)
0 2 7 9 9

12 Lunch

15:00-15:30 Johan Lindström: Spatio-Temporal Smoothing for Very Large Datasets
15:30-16:00 Thaďs Fonseca: Non-Gaussian spatiotemporal modeling through scale mixing.
16:00-16:30 David Bolin: Geostatistical modeling using non-Gaussian Matérn fields
16:30-17:00 Break

Multivariate workshop
17-17:50  Paul Sampson: A novel dimension reduction approach for spatially-misaligned multivariate air pollution data

17:50-18:40  Reinhard Furrer: Multivariate modelling and efficient estimation of random fields with application to roller data

18:40-19 Coffee break

19-19:50 Open discussion on strategies for large multivariate spatial data sets including estimation and co-kriging

Thursday, June 26


9-9:45 Misalignment problems (Craigmile) Hierarchical models. Downscaling. Upscaling. Change of support.
0 0 3 6 21

9:45-10 Coffee break

10-11 Practicum: Chesapeake Bay (Craigmile)
Solutions R code
0 0 5 11 9
11-11:45 Application: Up- and downscaling of regional climate models (Guttorp)
0 0 6 11 8

12 Lunch

Multivariate workshop
17-17:50 Alexandra Schmidt: Modelling multivariate counts varying continuously in space
17:50-18:10 Coffee break

18:10-19  Open discussion on theoretical treatments of multivariate models--are there any ways forward to formalizing the benefit of co-kriging under a multivariate model. Also strategies and ideas/approaches for including multivariate modeling in a spatial statistics class.

Workshop on nonstationary covariance models

W 6/18 - Sa 6/21

In most papers in the nonstationary covariance model literature, comparisons are only made to stationary (or even isotropic) models, not to other nonstationary approaches. There is a need for an organized comparison of these models, and an assessment of what models work in what circumstances. Furthermore, few of these models have easily accessible software available, whence a comprehensive software library is needed. The purpose of the workshop is to compare the different methods, develop a library of software, and produce a white paper with recommendations as to what methods are most useful in what situations. P`articipants in this workshop will have prepared in advance by analyzing a few given data sets using their particular approach. In addition, the method(s) of comparison of the approaches will have been predetermined in discussion between the participants. The workshop will last four days. The first two days will be presentations of results. The third day will be devoted to producing an outline of a paper detailing the comparison of the methods, while the final day will be spent discussing how to set up the software library.

Workshop on multivariate spatial models

M 6/23 - Th 6/26

The intent of this workshop is to make participants aware of different approaches to multivariate spatial models, how to analyze multivariate spatial data, and what software is available or needs to be developed. The goal of the workshop is to produce a white paper on multivariate spatial statistics. Like the other workshop, it will last four days. This will be a more traditional type of workshop, with lectures and discussion groups.

List of school instructors

Sudipto Banerjee, U. Minnesota, USA
David Bolin, Chalmers U., Sweden
Lelys Bravo de Guenni, Simón Bolívar U., Venezuela
Catherine Calder, Ohio State U., USA
Dan Cooley, Colorado State U., USA
Peter Craigmile, Ohio State U., USA
Charmaine Dean, U. Western Ontario, Canada
Dani Gamerman, UFRJ, Brazil
Peter Guttorp, U. Washington, USA
Jennifer Hoeting, Colorado State U., USA
Mikyoung Jun, Texas A&M, USA
Johan Lindström, Lund U., Sweden
Wendy Meiring, U. California at Santa Barbara, USA
Mathieu Ribatet, U. Lyon 1, France
Paulo Ribeiro Jr., UFPR, Brazil
Eliane Rodrigues, UNAM, Mexico
Paul Sampson, U. Washington, USA
Bruno Sansó, UCSC, USA
Aila Särkkä, Chalmers U., Sweden
Martin Schlather, U. Mannheim, Germany
Alexandra Schmidt, UFRJ, Brazil


Lelys Bravo de Guenni, Universidad Simón Bolívar, Venezuela
Charmaine Dean, University of Western Ontario, Canada
Peter Guttorp, University of Washington, USA
Eliane Rodrigues, Universidad Nacional Autónoma de México
Alexandra Schmidt, Universidade Federal do Rio de Janeiro, Brazil

Local organization:
Thaďs Fonseca, UFRJ, Rio de Janeiro
Nancy Garcia, UNICAMP, Campinas
Alexandra Schmidt,UFRJ, Rio de Janeiro

Nonstationary covariance models
Peter Guttorp, University of Washington, USA
Wendy Meiring, University of California at Sant Barbara, USA
Paul Sampson, University of Washington, USA
Alexandra Schmidt, Universidade Federal do Rio de Janeiro, Brazil

Multivariate spatial models
Will Kleiber, University of Colorado Boulder, USA
Eliane Rodrigues, Universidad Nacional Autónoma de México

Local information

Hotel Atlântico Búzios
The hotel registration page is now closed. Please do not register directly with the hotel, as you may not be able to get the negotiated rates. Please note prices have been updated November 12, 2013.
Room costs includes registration fee, breakfast and lunch. Daily prices

The cost of rooms is calculated assuming that participants do stay at the hotel. The charge would be higher otherwise, as we would have to pay for conference facilities. Registration fee for those not staying at the conference hotel is 2500 R$.

A bus will transport participants from the Rio de Janeiro International Airport to the hotel on June 15, and back on June 27. Details of the bus(es) will be announced by late May. Note that nationals from some countries (including the US) will need a visa in order to enter Brazil. This can not be obtained at the port of entry, but must be secured in advance. See this site for details.

Information about Búzios, and some information about what to do.