Awesome
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.
:whale: 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 theSIAC_S2
orSIAC_L8
class, i.e.SIAC_S2(**kwargs, gee=False)
orSIAC_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