TIES working groups

The TIES membership committee coordinates three TIES working groups launched in September 2021, under the leadership of Matthew Wheeler (National Institute of Environmental Health Science), Marta Blangiardo (Imperial College London), and Christopher Wikle (University of Missouri). The aim of this initiative is to improve networking across the Society’s members and develop creative collaboration.

The members of the three TIES working groups are currently working towards advancing statistical and computational methods in different directions, motivated by emerging environmental and computational problems. Here are their exciting research themes:

(i) Matthew Wheeler’ workgroup - The group Bayesian Methods for Complex Environmental Systems is developing multivariate spatial models for large datasets. By extending Moran and Wheeler (2022), the group has constructed a spatial process and Bayesian sampling algorithms that can model arbitrary continuous surfaces for large n problems. In addition to this work, the group is currently developing a software interface that allows researchers to model these surfaces and include these near-exact GP approximations into their models. 

Tutors: Jonathan Stroud & Monica Pirani

(ii) Marta Blangiardo’s workgroup - The group Statistical methods for Source Apportionment is working on methodologies for identifying ultrafine particle (UFP) sources, an active topic on Environmental Health research. A relevant aspect of this analysis is that particle size distribution is measured during the sampling, which is considered indirect for UFP sources. Then, an accurate characterization of UFP sources is crucial for understanding air pollution health impacts. The team has been working on two feasible approaches: (1) A Bayesian Dirichlet Process that facilitates uncertainty quantification about the identified UFP sources; (2) Functional data representation of UFP sources that allows for dependencies among different particle sizes.

Tutors: Melanie Meis & Carolina Euan

(iii) Christopher Wikle’s workgroup - The group AI methods in Environmental Science is currently focused on bringing a statistical perspective to the field of explainable AI, i.e., to what extent neural networks and other data-driven approaches can attribute the relative importance of different inputs, by testing some metrics of explainability to a weather data set and comparing them. Future efforts will focus also on the aspects of uncertainty quantification in AI, more specifically on the role of stochastic, possibly Bayesian methods in neural networks in space and/or time. 

Tutors: Susan Simmons & Stefano Castruccio

The TIES working groups bring together researchers from North America, India, and Europe! The participants that are sharing this collaborative research experience for the 2021-22 edition are:

Abhi Datta (USA), Lance Waller (USA), Won Chang (USA), Edward Boone (USA), K. Indulekha (India), Wesley Burr (Canada), Ben Swallow (UK), Israel Martinez-Hernandez (UK), Oliver Barenbod (UK).

The TIES membership committee

Stefano Castruccio (USA), Carolina Euan (UK), Melanie Meis (Argentina), Monica Pirani (UK), Susan Simmons (USA), Jonathan Stroud (USA)