Scottish Sensor Systems Centre

University: Professor Marian Scott (Prof of Environmental Statistics), Professor Adrian Bowman, Dr Claire Miller, University of Glasgow

Collaborative Partner: Scottish Environment Protection Agency (SEPA)

Project Type: Investigative (£10K Grant)

Project Duration: 4 months (Sept - Dec 2011)

Project Abstract

Motivated by a need to improve existing modelling and visualisation technologies for high dimensional, spatially distributed data, this was a proof-of-concept designed to show that statistical models and analyses from Glasgow University could be implemented in SEPA offices. The focus of the work was on the application of reporting on the state of a river network (River Tweed), in particular looking at nitrate concentrations collected at a set of fixed monitoring locations.

A vital aspect of the work involved identifying the types of statistical models and visualisation tools of most use to SEPA scientists. Of particular importance was determining the data requirements for network modelling – an example being the need for a ‘connectivity matrix’ that describes the river branching structure in a way that is convenient for model fitting and is a crucial component for network models. SEPA was specifically interested in understanding:
•where measures aimed at improving environmental quality were most effective and why;
•what constraints and controls the physical environment may place upon the effectiveness of intervention measures;
•how SEPA could communicate and share its understanding of river networks with its stakeholders via improved visualisation tools.

The project outcomes included:
•Acquisition of consumables and data from the River Tweed specifically, as well as documenting data required for general river network modelling.
•Developing and debugging statistical models including a fast, novel and completely general method for estimating pollutant levels over a sparse network of monitoring stations on a river.
•Developing quantification and visualisation of uncertainties, including a visualisation tool with interactive graphics tailored to the needs of SEPA.
•Test visualisation on focus groups at SEPA, and training for SEPA staff, outlining the main benefits of new models and types of visualisations and how they could be used to better understand changes in pollutant levels.
•Documentation: new R code which can be used on other river catchments where data are available; a scientific report produced detailing statistical and computational concepts as well as data requirements; and an academic paper comparing new method with other established approaches.

Project Impact

The project demonstrated the potential to extend these types of river network models to include variables related to surrounding land usage and meteorology that could increase predictive capabilities, and further lend itself to even more informative visualisation. As well as having particular importance in the environment reporting developments, the project outcomes may have alternative relevance and applicability in providing a better understanding of the wider impacts of environmental change.