i.hyper.smac - SMAC Atmospheric Correction for Hyperspectral Data

DESCRIPTION

This module implements the Simplified Method for Atmospheric Correction (SMAC) for hyperspectral data in GRASS GIS. It provides two methods for atmospheric correction:

1. Simple SMAC: A faster, less accurate method suitable for quick processing
2. Libradtran SMAC: A more accurate method using the libRadtran radiative transfer model

The module is based on the original SMAC algorithm developed by Rahman and Dedieu (1994) and the implementation by Olivier Hagolle.

FEATURES

• Supports multiple hyperspectral sensors (AVIRIS, HYPERION, PRISMA, etc.)
• Two correction methods: simple and libradtran-based
• Automatic estimation of atmospheric parameters:
  • Aerosol Optical Depth (AOD) using the Dense Dark Vegetation (DDV) method
  • Water Vapor Content (WVC) using atmospheric absorption features
• Uses libRadtran for accurate atmospheric simulations (for libradtran method)
• Handles both per-pixel and scene-average atmospheric correction
• Preserves the original data structure and metadata

REQUIREMENTS

• GRASS GIS 8.0 or later
• Python 3.x
• NumPy
• PyEphem (for solar zenith angle estimation)
• libRadtran (for atmospheric simulations)
• scipy (for interpolation and optimization)
• pandas (for data handling)
• scikit-learn (for regression models in AOD estimation)

USAGE

# Simple SMAC (default)
i.hyper.smac input=name output=name elevation=name [sensor=string]
             [aod=float] [water_vapor=float] [ozone=float] 
             [solar_zenith=float] [solar_azimuth=float]
             [view_zenith=float] [view_azimuth=float]
             [--overwrite] [--verbose] [--quiet] [--keep-temp]

# Libradtran SMAC
i.hyper.smac input=name output=name elevation=name method=libradtran
             sensor=string [aod=float] [water_vapor=float] [ozone=float]
             [solar_zenith=float] [solar_azimuth=float]
             [view_zenith=float] [view_azimuth=float]
             [aot_lut=name] [visibility=float]
             [aerosol_type=string] [--overwrite] [--verbose] [--quiet] [--keep-temp]

PARAMETERS

Input/Output

• input - Name of input 3D raster map
• output - Name for output 3D raster map
• elevation - Name of elevation raster map (default: DEM from GRASS location)

Method Selection

• method - Atmospheric correction method (simple, libradtran; default: simple)

Atmospheric Parameters (both methods)

• sensor - Sensor type (AVIRIS, HYPERION, PRISMA, etc.)
• aod - Aerosol Optical Depth at 550nm (if not provided, will be estimated)
• water_vapor - Water vapor content in g/cm² (if not provided, will be estimated)
• ozone - Ozone content in cm-atm (default: 0.3)
• solar_zenith - Solar zenith angle in degrees (required)
• solar_azimuth - Solar azimuth angle in degrees (default: 0)
• view_zenith - View zenith angle in degrees (default: 0)
• view_azimuth - View azimuth angle in degrees (default: 0)
• aerosol_type - Aerosol type (continental, maritime, urban, desert; default: continental)

Libradtran-specific Parameters

• aot_lut - Look-up table for AOD estimation (if not provided, will be generated)
• visibility - Visibility in km (if not provided, will be estimated from AOD)

EXAMPLES

Simple SMAC with automatic parameter estimation

# Automatically estimate AOD and water vapor
i.hyper.smac input=hyperspectral output=corrected elevation=dem sensor=AVIRIS

Simple SMAC with manual parameter specification

# Manually specify AOD and water vapor
i.hyper.smac input=hyperspectral output=corrected elevation=dem sensor=AVIRIS \
  aod=0.2 water_vapor=2.5

Simple SMAC with specific atmospheric parameters

i.hyper.smac input=hyperspectral output=corrected elevation=dem sensor=HYPERION \
  aod=0.2 water_vapor=2.5 solar_zenith=30

Libradtran SMAC with custom parameters

i.hyper.smac input=hyperspectral output=corrected elevation=dem method=libradtran \
  sensor=AVIRIS aod=0.15 water_vapor=2.0 solar_zenith=25 \
  aerosol_type=maritime visibility=50

Using an AOT look-up table with Libradtran

i.hyper.smac input=hyperspectral output=corrected elevation=dem method=libradtran \
  sensor=PRISMA aot_lut=aot_lookup.txt solar_zenith=30

NOTES

• Automatic Parameter Estimation:

  • AOD is estimated using the Dense Dark Vegetation (DDV) method, which identifies dark vegetation pixels and uses their reflectance in the blue region
  • Water vapor is estimated using atmospheric absorption features around 940nm and 1130nm
  • If estimation fails, default values are used (AOD=0.15, Water Vapor=2.0 g/cm²)

• Performance Considerations:

  • The simple method is faster but less accurate than the libradtran method
  • AOD and WVC estimation add computational overhead but improve accuracy
  • The module creates temporary files in the current mapset (use --keep-temp to preserve them)

• Requirements:

  • The libradtran method requires the libRadtran software to be installed and properly configured
  • Additional Python packages (scipy, pandas, scikit-learn) are required for parameter estimation

• Supported Sensors:

  • PRISMA, AVIRIS, AVIRIS_NG, HYPERION, ENMAP, OSK_GHOST, PIXEL, ESPER, IPERLITE, KUVASPACE_23, KUVASPACE_32, WYVERN_23, WYVERN_32, HYP4U, TANAGER

• Best Practices:

  • For best results, provide as much metadata as possible
  • Verify the estimated parameters using the module's output messages
  • Consider using the --keep-temp flag for debugging if results are unexpected

REFERENCES

• Rahman, H., & Dedieu, G. (1994). SMAC: a simplified method for the atmospheric correction of satellite measurements in the solar spectrum. International Journal of Remote Sensing, 15(1), 123-143.
• Original SMAC implementation: https://github.com/olivierhagolle/SMAC

AUTHORS

• Original SMAC algorithm: H. Rahman and G. Dedieu
• GRASS GIS implementation: [Yann Chemin] dr.yann.chemin@gmail.com

SOURCE CODE

Available at: GRASS GIS Addons Repository

SEE ALSO

• r.smap
• i.atcorr
• i.vi