Bayesian Hierarchical Model Characterization of Model Error in Ocean Data Assimilation and Forecasts
Abstract
Objectives: A sequence of project objectives build upon experience gained under prior Office of Naval Research (ONR) support. First, we will extend time- and space-dependent error covariance BHM from the Mediterranean Forecast System (MFS) to Regional Ocean Model System (ROMS) applications in the California Current System (CCS). Second, reduced-dimension error process models will be developed from ensembles of ROMS analyses and forecasts wherein selected model parameterizations (e.g. diffusion) are treated as random. Monte Carlo sampling algorithms will be developed to obtain posterior distributions for prescribed error models (e.g. additive, multiplicative, etc.). Third, based on the experience gained in the first and second sets of objectives, we will develop an ocean forecast model error process BHM to evolve distributions for model error. Funding for this research arrived at the cooperating institutions in the latter half of the fiscal year (NWRA/CoRA funding in place as of late May 2010, University of Missouri funding arrived as late as August 2010). In this report, we elaborate plans and progress in pursuit of the first set of objectives.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 2010
- Accession Number
- ADA597803
Entities
People
- Christopher K. Wikle
- L. M. Berliner
- Radu Herbei
- Ralph F. Milliff
Organizations
- Northwest Research Associates