Sensor Networks Data Analytics
The environmental statistics group at the University of Glasgow has longstanding expertise in data analytics applied to sensor networks and satellite remote sensing, including modelling water quality and its impact on human and ecosystem health, greening vegetation, and flooding. We have on-going relationships with Agencies such as the Scottish Environment Protection Agency (SEPA), Health Protection Scotland (HPS) and Environment Agency (EA). Our work has developed new statistical modelling methodologies to deal with data streams coming from different sensors (differing spatial and temporal resolutions), and fusion where appropriate with models. Some examples of the various strands of this work is given below.
Types of sensor investigations
- Modelling of AURN and diffusion tube data to investigate significant trends in air quality over time, in spatial patterns, and to find evidence of break points in change. Statistical downscaling to areal units (such as datazone), with uncertainty quantification.
- Data and model fusion (using the point location data and gridded model (or satellite, including MERIS, AATSR, OLCI, Sentinel-3) data to produce a refined estimate of the pollution field and its uncertainty.
- Model emulation and decision support, to investigate the management of traffic (numbers and fleet characteristics) in combination with weather conditions on modelled air quality.
- Big data analytics, and functional data analytic methods to examine spatial coherence in water quality globally identifying lakes showing similar in time and space.
- Calibration of low cost sensors, with design issues concerning their use.
- Network design to ensure efficient sampling (including in river quality monitoring, flooding).
- Response of vegetation to changes in climate using satellite observations of the earth's surface (MODIS)
- Development of a sensor data framework investigating water quantity and soil moisture (data linkage of local sensors, national sensor networks and satellite observation).