Previous Abstract | Back to Abstracts Page | Next Abstract
Application and validation of polynomial chaos methods to quantify uncertainties in simulating the Gulf of Mexico circulation using HYCOM.
Mohamed Iskandarani, Matthieu Le Henaff, Guotu Li , W. Carlisle Thacker, Omar M. Knio, Ashwanth Srinivasan
University of Miami
(Abstract received 05/05/2015 for session X)
ABSTRACT
We present an analysis of uncertainties in a 30-day forecast of the circulation in the Gulf of Mexico due to input uncertainties in initial conditions and wind forcing. The variability modes of the circulation and the winds were identified separately via an empirical orthogonal function decomposition; a dominant subset of these modes were retained and multiplied by a stochastic amplitude to generate the initial conditions and wind perturbations. An ensemble calculation was then used to construct a polynomial chaos based proxy for the HYCOM-predicted sea surface height (SSH) and mixed layer depth. This proxy encapsulates the dependence of these quantities on the uncertain inputs. The proxy was validated to ensure its reliability and was used to generate a much larger ensemble. The large ensemble was in turn used to analyze the statistics of the forecast and to generate the probability density functions of Sea Surface Height and mixed layer depth. Our analysis show that the standard deviation in the ensemble's Sea Surface Height exceeds that observed from AVISO after about 3 weeks. This loss of predictability is concentrated in the Loop Current region and is associated with an eddy shedding event. Furthermore, the uncertainties over the Gulf of Mexico shelves during that period is dominated by uncertainties in the winds whereas the uncertainties in the Loop Current region and the deep Gulf are dominated by initial condition uncertainties.
Previous Abstract | Back to Abstracts Page | Next Abstract
2015 LOM Workshop, Copenhagen, Denmark June 2nd - 4th, 2015