Skip to content
 
CIRA Fellow

Dr. Milija Zupanski

Senior Research Scientist, CIRA, Colorado State University

Mailing Address:
Dr. Milija Zupanski
Cooperative Institute for Research in the Atmosphere
Colorado State University
1375 Campus Delivery
Fort Collins, CO 80523-1375
  • Office Location:
    CIRA Room 113
  • 970-491-8298
About Me:

Milija Zupanski received his BSc in Meteorology from University of Belgrade, Serbia, and MS (1987) and PhD (1990) in Meteorology from the University of Oklahoma, with Prof. Yoshi Sasaki as advisor. His area of interest include ensemble data assimilation, hybrid variatonal-ensemble data assimilation, nonlinear and non-differentiable optimization and preconditioning, coupled data assimilation, non-Gaussian probability assumptions, predictability and chaos theory, and applied mathematics focusing on weather and climate.
He worked at the NOAA National Centers for Environmental Prediction (NCEP) from 1990-2001, where he was a principal developer of the four-dimensional variational (4D-Var) data assimilation with then operational Eta model. In 2001 Dr. Zupanski joined CIRA, where he was one of the principal developers of the 4D-Var with RAMS model, and is the principal developer of the Maximum Likelihood Ensemble Filter (MLEF) – an ensemble assimilation/prediction system with estimation of uncertainties. In recent years his work is focusing on various applications of the MLEF to weather and climate, and to coupled data assimilation. Dr. Zupanski collaborates with research groups at Colorado State University, NOAA, NASA, DOD, as well as from U.S. and international Universities and Research Labs.
Presently, Dr. Zupanski is a Senior Research Scientist and CIRA Fellow. He leads CIRA Data Assimilation Team. (Research web-page at http://da.cira.colostate.edu)

Research Interests:

His areas of interest include ensemble data assimilation, nonlinear and non-differentiable optimization and preconditioning, non-Gaussian probability assumptions, predictability and chaos theory, and applied mathematics focusing on weather and climate. Dr. Milija Zupanski is a principal developer of the Maximum Likelihood Ensemble Filter (MLEF).

My Work/Projects:

MLEF as a non-differentiable minimization algorithm

Wednesday, March 5, 2014

Zupanski_plotsThe MLEF can be derived without common differentiability and linearity assumptions (Zupanski et al. 2008). As a consequence, non-differentiable minimization algorithms can be derived as generalization of gradient-based methods, such as the nonlinear conjugate gradient (CG) and quasi-Newton (QN) methods. The non-differentiable aspect of the MLEF algorithm is illustrated in an example with one-dimensional Burgers model and simulated observations. A comparison between the generalized non-differentiable CG and the standard differentiable CG methods is shown, in an example with cubic non-differentiable observation operator. Both the cost function and the gradient norm show a superior performance of the MLEF-based non-differentiable minimization algorithm. These results indicate important advantage of the MLEF for assimilation of cloud observations and processes, which are particularly challenging due to their discontinuous nature.

Images: The cost function (left panel) and the gradient norm (right panel) for non-differentiable CG (solid blue line) and for the standard, differentiable CG (dashed red line) method.