Home > Events > 2011 Seminars & Colloquia > JAN MANDEL - University of Colorado Denver

JAN MANDEL - University of Colorado Denver

Main Content

Morphing and wavelet EnKF data assimilation
When
03 November 2011 from 4:00 PM to 5:00 PM
Where
111 Tyson Bldg.
Add event to calendar
vCal
iCal

Based on joint work with J. D. Beezley, L. Cobb, A. Krishnamurthy, A.
K. Kochanski, K. Eben, P. Jurus, and J. Resler.

Data assimilation for problems with coherent features, such as rain
fronts and fire lines, require a position correction in addition to
the classical amplitude correction. A morphing method is developed,
which augments the state space by a displacement field, and it
combines both the position and amplitude corrections into one step; in
effect, the morphing method replaces linear combinations in the
analysis update by more suitable intermediate states.

A variant of the Ensemble Kalman Filter (EnKF) with the diagonal
approximation of the covariance in a wavelet space is presented, which
allows to decrease significantly the ensemble size and to evaluate the
EnKF update at a very low computational cost, for a class of
observation functions, including image data.

The methods and their applications are under development, and their
advantages and limitations will be discussed. Applications include
assimilation of radar precipitation data into WRF, and assimilation of
fire position information into a coupled WRF and wildland fire model.

Filed under: ,