Image Generation

1
State of the Art and Evolution of Weather
Simulation Technology
December 6, 2006 Robert Reynolds Principal
Simulation Engineer Northrop Grumman Corporation
2
The Uniform Atmosphere Model

• Formulated by STOW-97
• Not entirely uniform
• Vertical variation
• Horizontally uniform
• Implemented for ModSAF
• Environmental Editor GUI
• Only 16 values, but
• They are highly correlated
• Too easy to create non-physical conditions
• What about time variation?
• Look to the Physics!

3
Numerical Weather Prediction

• Real weather models implement these ?
• Direct implementation for simulation is beyond
  state-of-the-art
• Too much SME required
• Boundary conditions?
• Since STOW-97
• Leverage operational METOC data centers
• NWP models are being run 24x7
• Throw no NWP model output away!

4
Master Environmental Library
Operational Weather Forecast Centers
mission planning

• FNMOC
• AFWA
• STENNIS

Gridded Digital Forecasts

• NOGAPS
• COAMPS
• MM5
• NCOM
• MODAS
• WaveWatch

Numerical Weather Models
archive
5
Environmental Scenario Generator
NCEP/NCAR 2.5 x 2.5 deg Global Archive
Fuzzy-Logic Search Engine

• Region of Interest
• Desired conditions
• High winds
• Light precipitation
• (correlated clouds,
• waves, acoustics)

ACMES 10 and 40 km Regional Archive
Regional NWP Models
Gridded Digital Forecasts
6
Defining a Uniform Atmosphere
Gridded Digital Forecast
Simulation Object Model
7
Ocean, Atmosphere and Space Environment Simulation
HLA Federation
OASES
Editor
Ingestor
Publisher
persistent object DB
Transformer
ESG
8
Example Hurricane Katrina Impact on Keesler
AFB, MS
9
The Gridded Weather Model

• Ocean
• Vertical Profiles (1D)
• Wave Spectra (2D)
• Acoustics (3D)

• Atmosphere
• Vertical Profiles (1D)
• Surface Weather (2D)
• Cloud / Fog Layers (2D)
• Radar Ducts (2D)
• Sun / Moon Position (2D)
• Illumination (2D)
• Precipitation (2D 3D)
• EM Propagation (3D)
• Haze / Aerosols (3D)
• Ambient Weather (3D)

Space Terrain
10
Looking Forward

• Distribution of environmental Features via
  Publish/Subscribe has received most attention
• But approach does not scale well (see paper)
• Pre-distribution offers scalable solution
• But doesnt support use of live environments
• And simulated weather cannot be modified at
  run-time
• Distributed modeling of environmental Effects has
  received comparatively little attention
• AFCCCs HyperCube offers one strategy
• OASES Query Server an effects model framework?

11
Thank You for Your Interest!Questions?