Data Assimilation for Ocean Acoustic Tomography
Abstract
This research evaluates the effectiveness of assimilating ocean acoustic tomography data into numerical ocean models. Acoustic tomography uses sound to remotely sample integrated or averaged properties of the ocean. These properties include sound speed and current velocity. Three different models are used in the assimilation: 1) a linear Rossby wave model without advection, 2) a Rossby wave model with advection, and 3) a non-linear quasi-geostrophic model. We examine the tomographic data's effectiveness by assimilating non- averaging point measurements, such as temperature data, and integral data with the Rossby wave model. First, simulated data are used in benchmark experiments. These are followed by a second series of experiments in which actual tomographic data collected during the Acoustic Mid-Ocean Dynamics Experiment (AMODE) are used. We compare the value of the integral measurements in terms of forecasting accuracy in both physical and spectral space. Although both forms of data constrain the model equally as well in an average sense (the number of data were chosen to do this), the tomographic data are found to constrain low wavenumber components of the model more accurately Differences between the model estimates and data are within expected error bars. These error bars are based on prior statistical assumptions about errors in the model and observations, boundary conditions and initial conditions. The Rossby wave models with and without advection account for 62.1% and 60.7% of the data variance respectively and the quasi-geostrophic model accounts for 66.0% of the variance. The purpose of this research is not to evaluate any particular numerical model, but rather to better understand the usefulness of assimilating tomographic measurements into numerical models.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2000
- Accession Number
- ADA381847
Entities
People
- Chris G. Walter
Organizations
- University of Washington