Methodology¶
Pipeline Overview¶
The data processing pipeline transforms raw FVCOM model outputs into standardized, publicly accessible datasets:
Raw Data (Level 00)¶
Original FVCOM NetCDF outputs containing:
- 3D velocity fields (u, v components) at 10 sigma layers
- Sea surface elevation (zeta)
- Bathymetry
- Grid connectivity (node/element relationships)
Standardization (Level a1)¶
Raw outputs are standardized with:
- Consistent variable naming following CF Conventions
- Coordinate transformation to WGS84 (EPSG:4326) where necessary (see Coordinate Systems)
- Temporal validation and formatting to ISO 8601
- Global attribute metadata following ACDD conventions
VAP Derivation (Level b1)¶
Value-Added Products (VAPs) are computed from the standardized data. Each variable in the Variables section documents its specific derivation equation. General methods include:
Depth-Averaging¶
For mean variables (current speed, power density), values are averaged across all 10 sigma layers at each timestep, then averaged over the full hindcast year:
Depth-Maximum¶
For 95th percentile variables, the maximum value across sigma layers is taken at each timestep, then the 95th percentile is computed over the full time series:
Current Speed¶
Velocity magnitude is computed from eastward (\(u\)) and northward (\(v\)) components:
Power Density¶
Kinetic energy flux per unit area [@hass_2011_assessment]:
where \(\rho = 1025\) kg/m\(^3\) (nominal seawater density).
Summary Statistics (Level b4)¶
Summary-level Parquet files contain one row per grid point with all VAP variables. These enable efficient analytical queries without loading full time-series data.
Atlas Data (Level b5)¶
A subset of summary statistics formatted for the Marine Energy Atlas visualization platform, including spatial metadata for map rendering.
Quality Control¶
Quality control is applied at each processing stage. See Quality Assurance for verification details.