Uncertainties are a key factor in empirical science when describing the state of the world, its change, and relationships between driving forces. Spatial statistics is the branch of statistics that tries to model and understand uncertainties in spatial and spatio-temporal data. Spatio-temporal modelling involves building theories, testing them against available data, quantifying the uncertainties remaining, and informing about subsequent modelling and measurement requirements.
Measurement and modelling are the the pillars under empirical research. In both,geoscientific and geographic research, space and time form two of the main indexes along which data sets can be ordered, combined and compared. In the era of big dataand data deluges, some suggest that the size of the data replaces the need to test hypothesis or develop theories. In reality, hypothesis testing and theory development even get harder when available data are collected before the hypotheses are formulated, when the data generating process is not well understood, or when the data does not easily relate to the theory under question. New questions are how to sensibly analyse such big data, how to meaningfully combine them with classical, well designed experimental data, and how to realistically quantify uncertainties.
Work in the Spatio-temporal Modelling Lab includes the following themes:
- meaningful spatial statistics
- spatio-temporal copulas for phenomena across space and time
- uncertainty management in geoscientific workflows
- monitoring network design and evaluation
- trajectory planning for mobile sensors
- statistical models for land use change
- object-based image analysis for change detection
- spatio-temporal GIS
- statistical analysis of trajectories
A continuing theme underlying most of these topics is to improve R as an environment for spatio-temporal modelling. Whenever possible, we collaborate with national and international authorities or industry, and support them to take better decisions.