In today’s era of big data, advances in sensor technology have enabled the collection of massive datasets that are used to gain deeper insights into the wonders of biological systems. However, accurate statistical models are essential for answering applied research questions and testing hypotheses based on such big data.  Constructing statistical models that accurately capture the complexities of real-world biological systems is a challenging task. This difficulty arises primarily from the complex interactions among system components and the inherent dynamic nature of biological processes.

This research focuses on modelling the spatio-temporal correlations and dynamics inherent in many biological processes such as climate systems and biodiversity patterns. In particular, we will address the following challenges: (i) Developing data-driven models to capture complex spatio-temporal correlations, (ii) Crafting sample designs specifically tailored to analyse spatio-temporal data effectively and efficiently and (iii) Modelling spatio-temporal extremes to better understand their patterns and implications.

At the core of our investigation are Bayesian and statistical methods, which play a crucial role in developing the techniques we employ. To capture the complex spatio-temporal structures inherent in our data, we adopt a so-called hierarchical approach. This approach organizes the complex interactions within the system into smaller, interconnected components, or layers. One key advantage of this method is its flexibility, particularly in modelling the extremes of a process by reshaping the final layer that represents the data-generating mechanism. Additionally, hierarchical models facilitate Bayesian inference, enabling the incorporation of prior knowledge to compute sample designs and guide decision-making. Furthermore, this framework also supports continuous learning, allowing the model to improve iteratively as new data becomes available.

Overall, the framework we aim to develop will cover the entire data-to-decision pipeline, beginning with efficient data acquisition, progressing through robust data modelling, and resulting in informed conclusions through hypothesis testing. This integrated approach can play a crucial role to improve our understanding of the climate and biodiversity crises, two of the most urgent challenges facing humanity today.