How noaa-gsl/mpas_app works
Core System Components
• MPAS-Model (Main computation engine)
• Build System
• Workflow Management
• Data Processing Pipeline
• Platform-Specific Configurations
Main Architectural Layers
a) Infrastructure Layer:
• Conda environment management
• Module system integration
b) Configuration Layer:
• Default configuration system
• User configuration override system
• Machine-specific configurations
• YAML-based configuration management
c) Workflow Layer:
• Rocoto workflow management
• Task scheduling and dependencies
• Data retrieval system
• Model initialization
• Boundary conditions processing
d) Processing Components:
• Ungrib data processing
• Initial conditions creation
• Lateral boundary conditions
• MPAS atmosphere executable
• Output conversion system
Use a top-down approach with clear separation of concerns
• Top: User Interface (Scripts and Configuration)
• Middle: Workflow Management and Task Orchestration
• Bottom: Computation and Data Processing
Component Grouping
• Build System (build.sh and related components)
• Configuration Management (YAML files and config processing)
• Workflow Management (Rocoto integration)
• Data Processing Pipeline
• Platform-Specific Modules
Configuration Flow
default_config.yaml → user_config.yaml → experiment.yaml
Build Process
build.sh → platform-specific modules → MPAS-Model
Workflow Pipeline
Data Retrieval → Ungrib → Initial Conditions → Boundary Conditions → Forecast
Output Processing
MPAS Output → convert_mpas → Lat/Lon Grid
• Component types
• Platform-specific elements
• Configuration hierarchy
• Task dependencies
Level of Detail:
• Keep high-level workflow visible
• Abstract internal MPAS-Model details
• Clear interfaces between components
• Configuration and platform flexibility
• File locations and naming conventions