Constrained ensemble simulation for inverse Lagrangian prediction problem
Mike Chin, Arthur Mariano, Ashwanth Srinivasan
RSMAS / U.Miami
(Abstract received 08/14/2009 for session C)
ABSTRACT
Stochastic dynamic models are often used in drifter track simulations due partly to uncertainty in subgrid scale circulation velocity. The random flight model is an example of such. The nature of Lagrangian trajectory forecast using such models is that an ensemble of trajectories with a common release location would diverge in time. Numerically, divergence of trajectories is not convenient for inverse Lagrangian prediction problems such as drifter array deployment and source estimation problems, in which a common (or very small number of) release location(s) is sought given the knowledge of the final distributions of drifter locations/concentrations. In this talk, we explore application of the particle filter to constrain the drifter trajectories to "contract" while the Lagrangian simulations progress backward in time. The initial goal of the experiment is to improve localization of the source (or deployment site) of the drifters.
<< Previous Abstract | C101 | C102 | C103 | C104 | C401 | C402 | C403 | Next Abstract >>