Spatial Blind Source Separation – An Overview – Klaus Nordhausen (University of Helsinki)

Spatial Blind Source Separation – An Overview

Modeling multivariate geostatistical spatial data presents numerous challenges due to its inherent complexity and spatial dependencies. Spatial blind source separation (SBSS) offers a promising solution by adopting a latent modeling framework that enables univariate modeling of latent components, simplifying the analysis. However, the recovery of these latent components depends on the specific assumptions made, giving rise to a variety of SBSS methodologies. In this talk, we will provide a comprehensive overview of the different approaches to spatial blind source separation and highlighting their underlying assumptions.