[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

SEMINAR: MPO Seminar:Dr. Mohamed Iskandarani, Wednesday, September 7, at 3:00 p.m., in S/A 103, Slab Seminar room

From: Sandrine Apelbaum <sapelbaum@rsmas.miami.edu>
Subject: SEMINAR: MPO Seminar:Dr. Mohamed Iskandarani, Wednesday, September 7, at 3:00 p.m., in S/A 103, Slab Seminar room
Date: Fri, 2 Sep 2011 15:17:56 -0400

MPO Seminar


Dr. Mohamed Iskandarani, 



"Application of Polynomial Chaos Expansions for Uncertainty Quantification in Oceanic Simulations"


Room: Slab Seminar Room, S/A 103

Date: Wednesday,September 7, at 3:00 p.m.



Authors: Mohamed Iskandarani, Ashwanth Srinivasan, Carlisle Thacker, Omar Knio and Alen Alexanderian

Abstract

Oceanic simulations are subject to numerous sources of uncertainties that include initial and boundary conditions, and model errors. Robust
and reliable oceanic forecasts require an accurate and realistic assessment of these uncertainties. Here we present a methodology, based on Polynomial Chaos Expansions, to do just that.  The method is akin to spectral expansion methods, and takes as input the distribution of the uncertain data. The output data is then expanded in a suitable orthogonal series whose coefficients are the Fourier-like stochastic modes. These modes can be computed through ensemble simulations. Once the coefficients are computed, statistical moments of any quantity, whether directly represented in the model or derived, can be readily calculated. We will 
present briefly the broad outlines of the methodology, and we will present two sample calculations.  The first is concerned with quantifying the uncertainties in the Gulf of Mexico circulation stemming from uncertainties in the nesting boundary conditions. The other is concerned with quantifying parametric uncertainties in HYCOM's
subgrid scale parameterizations, specifically those related to the drag coefficients and the KPP-parameters, on the cooling under hurricane Ivan.We will present various tools for variance and sensitivy analyses to identify the main contributors to the output uncertainties. Finally the strength and limitations of the PC-approach will be discussed.


Meteorology and Physical Oceanography 
Rosenstiel School of Marine and Atmospheric Science
University of Miami
4600 Rickenbacker Causeway
Miami, FL 33149-1098
Tel     (305) 421-4057
Fax     (305) 421-4696