Proposal Description

The field of spatial and spatiotemporal statistics is rapidly evolving, driven by the need to analyze increasingly complex and large-scale datasets. This session brings together four distinguished speakers who will present their latest research on innovative methods for spatial and spatiotemporal data analysis.

Pulong Ma from Iowa State University will introduce Residual Tree Gaussian Processes (ResTGP), a new Bayesian framework that combines the strengths of treed and multi-resolution Gaussian processes. ResTGP offers computational efficiency and flexibility, which is particularly suitable for spatially heterogeneous and big data. The method’s recursive message-passing strategy for Bayesian inference marks a significant departure from traditional Metropolis-Hastings algorithms, promising more efficient and scalable computations.

Wei-Ting Wu from National Dong Hwa University, Taiwan, will discuss a novel methodology for tail estimation of the spectral density under fixed domain asymptotics. This approach focuses on the spectral domain rather than the traditional spatial domain, allowing for a broader class of spatial dependence models.

ShengLi Tzeng from National Chung Hsing University, Taiwan, will explore the application of geographically constrained spectral clustering to multivariate spatiotemporal data. This method leverages spatial constraints to improve the clustering of complex datasets, providing insights into patterns and dependencies that are critical for environmental studies.

Hao-Yun Huang from National Dong Hwa University, Taiwan, will present multi-resolution spatial methods on the sphere. His research focuses on spatial prediction in the presence of measurement errors, with a particular emphasis on irregularly located data. By developing a special class of basis functions in the thin-plate spline function space, Hao-Yun Huang offers a multi-resolution representation that enhances the accuracy and efficiency of spatial predictions.

Together, these presentations will provide a comprehensive overview of the latest advancements in spatial and spatiotemporal statistics, highlighting innovative approaches that address current challenges in the field and offer new opportunities for research and applications.