Data Assimilation Systems Overview

This section provides a comprehensive overview and comparison of major operational and research data assimilation systems around the world.

Major Systems

  1. JEDI (Joint Effort for Data assimilation Integration)

    • Developer: JCSDA

    • Modern, unified framework with object-oriented design

    • Supports multiple models and observation types

    • High-performance computing capabilities

  2. GSI (Gridpoint Statistical Interpolation)

    • Developer: NCEP

    • Operational system for weather forecasting

    • Supports 3D/4D-Var and hybrid EnVar

    • Direct assimilation of satellite radiances

  3. DART (Data Assimilation Research Testbed)

    • Developer: NCAR

    • Research-focused ensemble DA system

    • Educational tools and comprehensive documentation

    • Multiple model interfaces

  4. PDAF (Parallel Data Assimilation Framework)

    • Developer: AWI

    • High-performance parallel framework

    • Focus on ensemble methods

    • Model-agnostic design

  5. WRF-Data Assimilation

    • Developer: NCAR

    • Specific to WRF model

    • Supports various DA methods

    • Includes both 3D/4D-Var and hybrid approaches

Comparison Table

Data Assimilation Systems Comparison

Feature

JEDI

GSI

DART

PDAF

WRF-DA

Architecture

Object-oriented C++

Fortran 90

Fortran 90

Fortran 90/95

Fortran 90

License

Apache 2.0

Public Domain

Apache 2.0

LGPL-3

Public Domain

Methods
3D/4D-Var
EnVar
LETKF
3D/4D-Var
Hybrid EnVar
EnKF
EAKF
DART
EnKF
LETKF
ESTKF
3D/4D-Var
Hybrid
EnKF

Parallelization
MPI
OpenMP
GPU
MPI
OpenMP

MPI

MPI

MPI

Models

Multiple

Multiple

Multiple

Multiple

WRF only

Observations
Conventional
Satellite
Radar
Conventional
Satellite
Radar
Conventional
Some satellite

User-defined
Conventional
Radar

Documentation

Excellent

Good

Excellent

Good

Good

Active Development

Very Active

Active

Active

Active

Moderate

Learning Curve

Steep

Steep

Moderate

Moderate

Steep

Community Support

Strong

Strong

Strong

Moderate

Strong

Key Features Comparison

  1. Modularity

    • JEDI - Highly modular, object-oriented design

    • GSI - Monolithic with some modularity

    • DART - Modular design with clear interfaces

    • PDAF - Highly modular for ensemble methods

    • WRF-DA - Tightly coupled with WRF

  2. Performance

    • JEDI - Excellent, modern HPC support

    • GSI - Good, operational performance

    • DART - Good for research applications

    • PDAF - Excellent parallel efficiency

    • WRF-DA - Good for regional applications

  3. Extensibility

    • JEDI - Highly extensible through OO design

    • GSI - Moderate, requires system knowledge

    • DART - Good through interface layers

    • PDAF - Good for new ensemble methods

    • WRF-DA - Limited to WRF framework

  4. Use Cases

    • JEDI - Research to operations

    • GSI - Operational forecasting

    • DART - Research and education

    • PDAF - Research and development

    • WRF-DA - Regional weather prediction

References

See Data Assimilation Systems for detailed information about each system.

See also