Title: Spatio-temporal modelling in environmental and ecological systems
Authors: Claire Miller - University of Glasgow (United Kingdom) [presenting]
Abstract: Multiple potential data sources exist to aid in the monitoring and management of environmental and ecological systems. Data include, for example, those collected from long-term monitoring programmes, citizen science data, automatic sensor monitoring data and data from processed satellite retrievals. However, information gaps still exist making global challenges such as water pollution difficult to address. For example, the 2018 UN Water SDG 6 Synthesis Report suggests that the global data currently collected through the SDG process do not reflect the general state or trends known about freshwater ecosystems from other data sources. Where data do exist, the challenge lies in appropriately combining the available data streams to fill the knowledge gaps by providing improved estimation and prediction of, for example, water quality. The purpose is to give examples of the statistical methodological and computational challenges presented in such a context including combining data streams which are of different spatial and temporal support, identifying and accounting for potential bias and uncertainty in data retrievals, reducing dimensionality (while accounting for sparse data) and accounting for highly correlated nested spatial scales. Methodological developments and examples from recent work, including the GloboLakes and Hydroscape projects, will be presented.