Abstract:
Computer models are essential tools in the earth system sciences. They underpin our search for understanding of earth systems functioning and support decision-making across spatial and temporal scales. Ever growing computing power and data availability enable the construction of increasingly complex and coupled models of human-environment interactions. Yet while these progresses make models capable (in principle) of addressing new questions, they rarely help reducing the uncertainties associated with model responses. Indeed, gaps and errors in input data and epistemic uncertainty about the best way to characterise many natural and anthropogenic processes, keeps challenging our ability to establish if a model is "valid" and suitable to inform decisions - even more so when informing long-term decision under a changing climate.
In this talk I will show how we can design tailored computational experiments to (i) gain a deeper understanding of model behaviour, hence leading to more robust model evaluation; (ii) make predictions for large spatial scales with limited data; and (ii) learn about system response to uncertain drivers such as climate and land-use change. I will argue that we can use a common methodology (namely “Global Sensitivity Analysis”) for running these computer experiments in a structured way, and provide application examples from my recent research in hydrology and natural hazard assessment.
Short Bio: Dr Francesca Pianosi is a senior lecturer in water and environmental engineering at the University of Bristol. Her research and teaching activities focus on improving the way we use data analysis and mathematical modelling for hydrology and water engineering applications. Currently her two key research interests are: (1) the use of simulation and optimisation methods to support decision-making, particularly for water resource management; (2) the quantification and attribution of uncertainty in mathematical models, and in particular models used for natural hazard assessment. She is the lead developer of the open-source SAFE Toolbox for the quantification and attribution of uncertainty in mathematical models, used by thousands of researchers and practitioners worldwide across engineering and natural sciences. She currently holds an Early Career EPSRC Fellowship delivering new simulation and optimisation models to support water resource management planning in the UK, and a co-Investigator of the UK Centre for Greening Finance and Investment (CGFI), a national centre established in 2021 to accelerate the adoption and use of climate and environmental data and analytics by financial institutions.