Accession Number:

ADA383797

Title:

NAVDAS Source Book 2000: NRL Atmospheric Variational Data Assimilation System

Descriptive Note:

Final rept.,

Corporate Author:

NAVAL RESEARCH LAB MONTEREY CA MARINE METEOROLOGY DIV

Personal Author(s):

Report Date:

2000-01-01

Pagination or Media Count:

157.0

Abstract:

The development of the NRL Atmospheric Variational Data Assimilation System NAVDAS began in 1996. At that time, both the regional and global atmospheric data assimilation requirements of Fleet Numerical Meteorological and Oceanographic Center FNMOC were met with versions of a multivariate optimal interpolation algorithm MVOI, originally developed in the mid-1980s Barker, 1992 Goerss and Phoebus, 1992. The NRL MVOI algorithm was based on the most powerful formulation of the problem available at that time, that of Andrew Lorenc 1981 at the European Centre for Medium Range Forecasting ECMRF. In particular, the NRL MVOI used a box or volume formulation that permitted a few hundred observations located in the same region to be processed simultaneously, thus minimizing but not eliminating data selection. Moreover, the forecast error covariance was in conception, but not in execution reasonably general, permitting the geostrophic and nondivergence constraints to be imposed weakly, if desired. However, in the years since the implementation of the NRL MVOI system, great strides had been made in atmospheric data assimilation-both in the academic world and in other operational centers. Firstly, the OI algorithm had been generalized to the three-dimensional variational 3DVAR algorithm. Like the OI algorithm, this was a static three-dimensional algorithm in which all the observations over a particular time window were processed simultaneously as if they were all valid at exactly the same time and the time evolution was entirely handled by the evolution of the forecast or background field. Compared to the OI algorithm, the 3DVAR algorithm had several advantages 1 A global solution was obtained-there was no data selection 2 Many observation types that are difficult to handle properly with the OI algorithm could be handled properly in 3DVAR.

Subject Categories:

  • Meteorology
  • Physical and Dynamic Oceanography
  • Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE