Thursday 29 Oct 2015: Statistical Science Seminar: Modeling Lagrangian trajectories as stochastic processes
Sofia Olhede - University College London
Fluid dynamics are often modeled and estimated from the Lagrangian perspective. In this talk I will discuss basic stochastic processes that mimic the behavior of Lagrangian trajectories observed in many environments - for example the motion of satellite-tracked ocean surface drifters. Although these basic processes do not exactly replicate the complex behavior of the ocean surface, fitting their parameters to observed data provides useful summaries of structure. The challenge of Lagrangian data is that it moves both in time and space, and as such I will discuss how to extend our stochastic processes to account for the inherent nonstationarity and heterogeneity of ocean dynamics.