BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Talks.cam//talks.cam.ac.uk// X-WR-CALNAME:Talks.cam BEGIN:VEVENT SUMMARY:Multi-model and model structural uncertainty quantification with a pplications to climate science - Nathan Urban (Los Alamos National Laborat ory) DTSTART:20180306T160000Z DTEND:20180306T164500Z UID:TALK101869@talks.cam.ac.uk CONTACT:INI IT DESCRIPTION:A common approach to quantifying the uncertainty in computer m odel predictions is to calibrate their tuning parameters to observational data. However\, the largest uncertainties may not lie in the models'\; parameters\, but rather in their "structures": modelers make different cho ices in numerical schemes\, physics approximations\, sub-grid closures\, a nd included processes. These choices result in different models that all c laim to represent the same system dynamics\, but may disagree in their pre dictions. This talk is aimed at presenting concepts and motivation concer ning how to address such multi-model or structural uncertainty challenges. I present three methods. The first method\, Bayesian multi-model combinat ion\, converts structural uncertainties in multiple computer models into p arametric uncertainties within a reduced model. A hierarchical Bayesian st atistical approach combines these parametric uncertainties into a single d istribution representing multi-model uncertainty\, which can be updated wi th observational constraints to dynamically bias-correct the multi-model e nsemble. The second method uses system identification techniques to learn the governing equations of a PDE system. A non-intrusive model reduction a pproach is developed to rapidly explore uncertainties in alternate model s tructures by perturbing the learned dynamics. The third method is aimed at integrated uncertainty problems that require propagating uncertainties th rough multiple system components. It constructs a Bayesian network or grap hical model where each node in the network quantifies uncertainties in a p articular physical process\, which can be informed by multiple types of mo del and data. LOCATION:Seminar Room 1\, Newton Institute END:VEVENT END:VCALENDAR