“Hybridizing big data, sensing and modelling at the landscape scale”
In this presentation we will explore three key concepts for data analysis and decision support at landscape scales in agriculture, forestry and related areas. The first of these relates to the mixing of data-driven approaches, mathematical modelling, and bio-physical scientific understanding, as a way to make practical and truly impactful progress in complex and contentious situations. We will illustrate this mainly through examples of prescriptive analytics and quantitative risk management pertaining to wildfires, and well as in real options valuation as applied to routing roads through ecologically sensitive areas. The second concept is the holistic “end to end” design of agricultural and environmental analytics systems, from remote sensing and field sensors through to decision support tools. We provide some key examples including in aquaculture and in broadacre crop forecasting. The third concept centres on the notion of “the model as a sensor”, that is, where scientifically-validated models are used as part of a layered approach to sensing and prediction that spans from aerial photogrammetry and satellite-based remote sensing, through to expensive high-accuracy sensor systems, models, then densely-deployed inexpensive sensing based on the Internet of Things. Together, these concepts help give us the means of tackling important challenges by way of a fusion of cyberphysical systems, big data, mathematics and the natural sciences.