Skip Top Navigation
only rendered in printing; banner identifies JCSDA logo and the NOAA, NASA, USAF, and US Navy organizations
JCSDA Logo and banner

JCSDA News and Newsletters

Center News

Navy Implements 4-Dimensional Variational Data Assimilation System

Impact tests indicate that 5-day forecast skill is extended about 9 hours in the Southern Hemisphere and 4 hours in the northern Hemisphere, and tropical cyclone track forecast errors are reduced out to 120 hours.

Comparison of 500 hPa forecast height anomaly correlations for the 30-level 3D NAVDAS and 42-level 4D NAVDAS-AR assimilation systems for the Northern and Southern Hemispheres, February 1 – March 28, 2009

Comparison of 500 hPa forecast height anomaly correlations for the 30-level 3D (NAVDAS) and 42-level 4D (NAVDAS-AR) assimilation systems for the Northern and Southern Hemispheres, February 1 – March 28, 2009 (top). Homogeneous comparison of Tropical Cyclone forecast track error (nm) for the 4D-Var (red) and 3D-Var (blue) systems (bottom). Number of verifying forecasts is given below each forecast length (hour), February 1 - March 31, 2009.

On September 23, 2009, the Navy's Fleet Numerical Meteorology and Oceanography Center initiated operational use of the Naval Research Laboratory's (NRL) 4-Dimensional variational data assimilation system to provide the analysis for its Navy Operational Global Atmospheric Prediction System (NOGAPS). The 4D NRL Atmospheric Variational Data Assimilation System – Accelerated Representer (NAVDAS-AR) replaced the global 3D NAVDAS, which had been operational since 2003. As part of this transition, a NOGAPS/L42 (42 vertical levels) model replaced NOGAPS/L30, increasing the model top to 0.04 hPa, for better assimilation of satellite radiances.

The figure illustrates the positive impact of NAVDAS-AR on 5-day NOGAPS forecasts, using the 500 hPa geopotential height anomaly correlation and Tropical Cyclone (TC) forecast track error. 5-day forecast skill is extended about 9 hours in the Southern Hemisphere and 4 hours in the Northern Hemisphere, and TC track forecast errors are reduced out to 120 hours.

An adjoint-based observation impact monitoring system, NAVDAS-AR adjoint Observation monitoring System (NAVOBS), became operational at the same time. NAVOBS shows the impact of individual observations (or satellite channels) and observation types on global forecast error, aiding decisions regarding quality control and observation selection. NAVOBS was based on theoretical work by Baker, Langland, and Daley, with the web-based product development supported by JCSDA.

Daley, along with Xu and Rosmond laid the foundation for NAVDAS-AR. Its successful deployment marks the beginning of advanced data assimilation methods for the Navy. Unlike the model-space 4D-Var algorithms implemented elsewhere, NAVDAS-AR is an observation-space algorithm based on a variant of the representer algorithm pioneered by Prof. A. Bennett. This system can optimally determine the most likely 4D state of the atmosphere from observations, numerical forecasts, error statistics, and atmospheric dynamics through minimizing a generalized 4D cost function. It is advantageous to search for the minimum of the 4D cost function in observation-space in the presence of model errors, because the size of the control variables (equal to the number of assimilated observations) is always the same--whether we assume the model is perfect (strong constraint) or imperfect (weak- constraint). Besides assimilating the conventional in situ and satellite observations (including geostationary rapid-scan and feature-tracked winds; winds from QuikScat, WindSat, ASCAT, ERS-2, AVHRR, MODIS, SSM/I and SSMIS; and total precipitable water from WindSat, SSM/I and SSMIS), NAVDAS-AR assimilates remotely-sensed microwave and infrared sounder radiances from AMSU-A, SSMIS, and AIRS. Additional sensors, such as IASI and AMSU-B/MHS are routinely assimilated in the research versions of NAVDAS-AR, and will be added to the operational system over the next several months.

In addition to the late Dr. Roger Daley's innovative work while a UCAR distinguished visiting scientist at NRL, the following scientists helped to make this development and transition a success: Drs. L. Xu, N. Baker, B. Ruston, T. Hogan, P. Pauley, R. Langland, and Mr. S. Swadley, NRL; Drs. T. Rosmond (NRL, retired) and B. Chua, both SAIC; Dr. R. Pauley and Mr. L. Lyjak, FNMOC. (L. Xu and N. L. Baker, NRL)

JCSDA Quarterly Newsletters

Send items for the JCSDA Quarterly to: george.ohring@noaa.gov. Deadline: 2 weeks before the end of each calendar quarter.

Newsletters are posted as PDFs below.


 
 
Modified October 6, 2009 11:10 AM
NOAA | NWS | NASA
Navy Weather | Air Force Weather
Privacy |  Link & Product Disclaimers | Information Quality
Accessibility | JCSDA webmaster
Hosted for JCSDA at STAR
Level A conformance icon, W3C-WAI Web Content Accessibility Guidelines 1.0Valid HTML 4.01