HR: 09:10h
AN: NG21A-05 [Abstracts]
TI: An Exploration of Gradient Descent Nonlinear Noise Reduction in Atmospheric Prediction
Problems
AU: * Khare, S P
EM: khare@ucar.edu
AF: National Center for Atmospheric Research, 1850 table mesa drive, Boulder, CO 80305
United States
AB:
The use of so-called gradient descent nonlinear noise reduction techniques in atmospheric prediction problems is explored.
Gradient descent nonlinear noise reduction techniques can be used to deterministically compute trajectories of a dynamical
forecast model consistent with observational evidence at multiple time levels (shadowing trajectories). The basic theory of
gradient descent nonlinear noise reduction will be reviewed, along with a number of illustrative results in a suite of
low-order dynamical systems. Comparisons to a state of the art ensemble Kalman filter assimilation system will also be made.
DE: 4400 NONLINEAR GEOPHYSICS (3200, 6944, 7839)
DE: 4410 Bifurcations and attractors
DE: 4420 Chaos (7805)
DE: 4450 Nonlinear maps
SC: Nonlinear Geophysics [NG]
MN: Fall Meeting 2005