A sensor invariant Atmospheric Correction (SIAC)

Feng Yin

Department of Geography, UCL

ucfafyi@ucl.ac.uk

This atmospheric correction method uses MODIS MCD43 BRDF product to get a coarse resolution simulation of earth surface. A model based on MODIS PSF is built to deal with the scale differences between MODIS and Sentinel 2 / Landsat 8. We uses the ECMWF CAMS prediction as a prior for the atmospheric states, coupling with 6S model to solve for the atmospheric parameters. We do not have topography correction and homogeneous surface is used without considering the BRDF effects.

🐳 Docker:

Please refer to the SIAC Docker usage instruction inside Docker folder.

Citation:

Yin, F., Lewis, P. E., and Gómez-Dans, J. L.: Bayesian atmospheric correction over land: Sentinel-2/MSI and Landsat 8/OLI, Geosci. Model Dev., 15, 7933–7976, https://doi.org/10.5194/gmd-15-7933-2022, 2022.

Auxiliary data needed (Automatically downloaded by SIAC):

• MCD43 :
  • 16 days before and 16 days after the Sentinel 2 / Landsat 8 sensing date.
  • This has been updated to automatically download data from Google Earth Engine (GEE), which is much faster than the previous way. This means you will need to register to get access to GEE at here.
  • Or you can still use the previous way to download the data by adding the Gee = False in the SIAC_S2 or SIAC_L8 class, i.e. SIAC_S2(**kwargs, gee=False) or SIAC_L8(**kwargs, gee=False).
• ECMWF CAMS Near Real Time prediction:
  • Time step of 3 hours or 1 hour with the start time of 00:00:00 over the date, and data from 01/01/2015 are mirrored in UK Jasmin server at: https://gws-access.jasmin.ac.uk/public/nceo_ard/cams/
• Global DEM:
  • Global DEM VRT file built from ASTGTM2 DEM, and most of the DEM over land are mirrored in UK Jasmin server at: https://gws-access.jasmin.ac.uk/public/nceo_ard/DEM_V3/global_dem.vrt
• Emulators:
  • Emulators for atmospheric path reflectance, total transmittance and single scattering Albedo, and the emulators for Sentinel 2 and Landsat 8 trained with 6S.V2 are packed in the current repository.

Installation:

You will need to have gdal and Lightgbm installed, and it is suggested to install them with:

• conda:

  conda install -c conda-forge gdal lightgbm

• mamba:

  mamba install -c conda-forge gdal lightgbm

Then you can install SIAC:

• Directly from GitHub

  pip install https://github.com/MarcYin/SIAC/archive/master.zip

GEE authenticate:

If you have not used GEE python API before, you will need to authenticate to GEE first after you installed SIAC:

• In terminal:

  earthengine authenticate --auth_mode=notebook

• Or in python:

  import ee
  ee.Authenticate()

Usage:

The typical usage of SIAC for and Landsat 8&9:

• Sentinel 2

  from SIAC import SIAC_S2
  global_dem = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/DEM_V3/global_dem.vrt'
  cams_dir = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/cams/'
  SIAC_S2('/directory/where/you/store/S2/data/', global_dem = global_dem, cams_dir=cams_dir)

• Landsat 8

  from SIAC import SIAC_L8
  global_dem = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/DEM_V3/global_dem.vrt'
  cams_dir = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/cams/'
  SIAC_L8('/directory/where/you/store/L8/data/', global_dem = global_dem, cams_dir=cams_dir) 

• Landsat 9

  from SIAC import SIAC_L8
  global_dem = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/DEM_V3/global_dem.vrt'
  cams_dir = '/vsicurl/https://gws-access.jasmin.ac.uk/public/nceo_ard/cams/'
  SIAC_L8('/directory/where/you/store/L9/data/', global_dem = global_dem, cams_dir=cams_dir)

Outputs from SIAC_S2

All the outputs from SIAC are specified in the siac_output.json within the original S2 L1C folder:

An example of the siac_output.json

The following table specifies a list of the outputs from SIAC and their corresponding meanings:

Abbreviation	Description	Scale	Comments
siacLog	Siac log file	N/A
toaOvrs	Toa reflectance RGB overviews	N/A
boaOvrs	Surface reflectance RGB overviews	N/A
toaOvrFull	Toa reflectance RGB overview full resolution	N/A
boaOvrFull	Surface reflectance RGB overviews	N/A
viewAngles	View angles for each band	0.01	2 bands GeoTiff: B1 view azimuth, B2 view zenith
sunAngles	Sun angles for each band	0.01	2 bands GeoTiff: B1 sun azimuth, B2 sun zenith
SurfaceReflectance	Surface reflectance for each band	0.0001
SurfaceReflectanceUncertainty	Surface reflectance uncertainty for each band	0.0001
atmoParas	Atmospheric parameters	N/A	aerosol optical depth[-], total column of water vapour [ $g/cm^2$ ] and total column of ozone [ $cm-atm$ ]
atmoParasUncs	Atmospheric parameter uncertainties	N/A
Cloud probability	Cloud	0.01
Version	Version of the SIAC software	N/A
CleanPixelPercentage	Clean pixel percentage	N/A
ValidPixelPercentage	Valid pixel percentage	N/A

Outputs from SIAC_L8

All the outputs from SIAC are within the original L8/L9 L1C folder.

• An example of correction for Landsat 5 for a more detailed demonstration of the usage is shown here

Examples and Map:

A page shows some correction samples.

A map for comparison between TOA and BOA.

LICENSE

GNU GENERAL PUBLIC LICENSE V3