Skip to content

Optical Products Calibration

Description

The calibration of Optical EO data is done in the CopernicusLAC Platform via a dedicated service, named Optical Products Calibration (OPT-Calib), which derives in systematic calibrated images from ingested optical EO data products acquired from multiple EO-missions. Output optical calibrated single-band assets of TOA/BOA reflectance can be used as input for further thematic processing (e.g. co-location, change detection).

Workflow

The OPT-Calib service implements the workflow depicted below.

graph TB i[(CDSE)] style i fill:#ffde86,stroke:#333,color:#282828,stroke-width:2px i --> a1(EO data product) subgraph Input style Input fill:#e8e8e8,stroke:#333,color:#282828 a1[/Optical Dataset/] style a1 fill:#acc8ff,stroke:#333,color:#282828,stroke-width:2px end subgraph Optical Products Calibration a1 --> dntor[DN to radiance] a1 -.-> dntobt[DN to brightness temperature] style dntobt fill:#ffcccc,stroke:#cc0000 style a1 fill:#cfdfff,stroke:#333,color:#282828 dntor --> rtor[Radiance to reflectance] rtor --> p1(TOA/BOA Reflectance) style p1 fill:#cfdfff,stroke:#333,color:#282828 dntobt -.-> p2(Brightness Temperature) style p2 fill:#cfdfff,stroke:#333,color:#282828 p1 -.-> opt_index[Spectral Index generation] style opt_index fill:#ffcccc,stroke:#cc0000 opt_index -.-> p3(NDVI, MIRBI, NBR, NBR2) style p3 fill:#cfdfff,stroke:#333,color:#282828 p1 -.-> lcae[Land cover assets extraction] style lcae fill:#ffcccc,stroke:#cc0000 lcae -.-> p4(cloud_mask, SCL, land_cover) style p4 fill:#cfdfff,stroke:#333,color:#282828 p1 --> cog[Convert to COG] p2 -.-> cog[Convert to COG] p3 -.-> cog[Convert to COG] p4 -.-> cog[Convert to COG] cog --> stac[Create STAC item] end subgraph Output style Output fill:#e8e8e8,stroke:#333,color:#282828 stac --> o1[/Reflectance asset for each CBN: red, green, etc./] style o1 fill:#acc8ff,stroke:#87afff,color:#282828,stroke-width:2px stac -.-> o2[/Brightness Temperature asset for each CBN: lwir11, etc./] style o2 fill:#acc8ff,stroke:#87afff,color:#282828,stroke-width:2px stac -.-> o3[/Spectral indexes: NDVI, NBR2, MIRBI/] style o3 fill:#acc8ff,stroke:#87afff,color:#282828,stroke-width:2px stac -.-> o4[/Land cover, cloud mask, SCL/] style o4 fill:#acc8ff,stroke:#87afff,color:#282828,stroke-width:2px end

Note

Steps highlighted in red in the OPT-Calib flowchart indicate the ones which are executed only if/when necessary according to the specific source EO data product to be calibrated.

The processor employs the Optical Calibration application of Orfeo Toolbox1 or plain matrix calculations to apply the conversion of DN to radiance and reflectance.

A detailed description of each step of the OPT-Calib chain is provided below.

Digital Numbers to Radiance or Brightness Temperature

Top Of Atmosphere (TOA) Radiance in Wμm-1 m-2 sr-1 is derived from DN values using the following formula:

\[ L_\lambda = gain \times DN + offset \]

where: \(L_\lambda\) is TOA Radiance in Wμm-1 m-2 sr-1.

Radiance to Reflectance

TOA reflectance \(\rho_\lambda\) in spectral band \(\lambda\) is then derived from:

\[ \rho_\lambda = \frac {L \times d^2_{E,S} \times \pi} {ESUN \times cos(θ_{S})} \]

where: \(L_\lambda\) is the radiance in spectral band \(\lambda\), \(d^2_{E,S}\) is the is the earth-Sun distance in AU at given time, \(ESUN\) is the band averaged Exo Atmospheric Solar Irradiance at 1AU in mW m^-2 nm^-1, and \(θ_{S}\) is the solar zenith angle, and \(L\) is TOA Radiance in Wμm-1 m-2 sr-1.

ESUN is derived from reference Solar Spectral Irradiance and depends on radiometric resolution and Filter Spectral Response Profiles for each band of the optical EO data. This information is usually provided in the mission handbooks or manuals. For those missions where the ESUN values were not provided, these are derived from Thuillier 20022, over nominal band spectral ranges.

DN to to Radiance or Brightness Temperature for Sentinel-3 SLSTR L1B

Sentinel-3 SLSTR Top Of Atmosphere (TOA) radiance (mW/m2/sr/nm) are derived from L1B products using the SNAP software. Brightness temperature (K) from Sentinel-3 SLSTR thermal infrared bands at 3742, 10854, 12023 nm are also derived with the SNAP software.

CBN Description Band Name Wavelength centre (nm) Resolution
green Cloud screening, vegetation monitoring, aerosol S1 554.27 500
red NDVI, vegetation monitoring, aerosol S2 659.47 500
nir NDVI, cloud flagging, pixel co-registration S3 868 500
cirrus Cirrus detection over land S4 1374.80 500
swir16 Cloud clearing, ice, snow, vegetation monitoring S5 1613.40 500
swir22 Vegetation state and cloud clearing S6 2255.70 500
mwir38 SST, LST, Active fire S7 3742 1000
lwir11 SST, LST, Active fire S8 10854 1000
lwir12 SST, LST S9 12022.50 1000
fire1 Active fire F1 3742 1000
fire2 Active fire F2 10854 1000
Table 2 - Spectral bands included into a Sentinel-3 SLSTR L1B Calibrated Dataset.

DN to Reflectance for Sentinel-2 MSI L2A

Sentinel-2 Bottom of Atmosphere (BOA) reflectance for a band i is derived from the DN value of the L2A product using the formula below3,4:

\[ reflectance_{(i)} = { {DN_{(i)} + addoffset_{(i)}} \over quantificationvalue_{(i)} } \]

where :

  • \(reflectance_{(i)}\) is the BOA reflectance for a L2A band i L2A_BOAi.

  • \(DN_{(i)}\) is the digital number for a L1C or L2A band i. The digital number DN=0 represents the “NO_DATA” value.

  • \(addoffset_{(i)}\) is the offset for the band i. This value is extracted from BOA_ADD_OFFSET in the L2A metadata.

  • \(quantificationvalue_{(i)}\) is the scaling factor for the band i. This value is extracted from QUANTIFICATION_VALUE in L2A metadata.

CBN Description Band Name Wavelength centre (nm) Resolution
coastal Coastal aerosol B01 442.7 (S2A), 442.3 (S2B) 60m
blue Blue B02 492.4 (S2A), 492.1 (S2B) 10m
green Green B03 559.8 (S2A), 559.0 (S2B) 10m
red Red B04 664.6 (S2A), 665.0 (S2B) 10m
rededge70 Vegetation red edge B05 704.1 (S2A), 703.8 (S2B) 20m
rededge74 Vegetation red edge B06 740.5 (S2A), 739.1 (S2B) 20m
rededge78 Vegetation red edge B07 782.8 (S2A), 779.7 (S2B) 20m
nir NIR B08 832.8 (S2A), 833.0 (S2B) 10m
nir08 Narrow NIR B8A 864.7 (S2A), 864.0 (S2B) 20m
nir09 Water vapour B09 945.1 (S2A), 943.2 (S2B) 60m
swir16 SWIR B11 1613.7 (S2A), 1610.4 (S2B) 20m
swir22 SWIR B12 2202.4 (S2A), 2185.7 (S2B) 20m
Table 3 - Spectral bands included into a Sentinel-2 MSI L2A Calibrated Dataset.

Note

Starting from the Processing Baseline 04.00 (upgrade of the 29/09/2021), Sentinel-2 L1C and L2A products are provided with negative radiometric values (implementing an offset). In particular, the dynamic range is shifted by a band-dependent constant with the introduction of the RADIO_ADD_OFFSET in the product annotation. This evolution allows avoiding the loss of information due to clamping of negative values in the predefined range [1-32767] occurring over dark surfaces5.

Spectral index generation

In the spectral index generation step a selection of spectral indexes are derived from the multispectral calibrated single band assets. The spectral indexes already included in an optical calibrated dataset are listed in Table 4.

Asset name Index
ndvi NDVI - Normalised Difference Vegetation Index
nbr NBR - Modified Normalized Burn Ratio index
nbr2 NBR2 - Modified Normalized Burn Ratio 2 index
mirbi MIRBI - Mid-Infrared Burn Index index
Table 4 - List of spectral indexes potentially available in an Optical Calibrated Dataset.

NDVI - Normalised Difference Vegetation Index

\[ NDVI = { (nir - red) \over (nir + red) } \]

Reference: Rouse et al. (1973)3.

MIRBI - Mid-Infrared Burn Index

\[ MIRBI = { (swir22 * 10) - (swir16 * 9.8) } + 2 \]

Reference: Trigg and Flasse (2001)4.

NBR - Normalized Burned Ratio

\[ NBR = { (nir - swir22) \over (nir + swir22) } \]

Reference: Key and Benson (2006)5.

NBR2 - Normalized Burned Ratio 2

\[ NBR2 = { (swir16 - swir22) \over (swir16 + swir22) } \]

Reference: Storey et al. (2016)6.

Warning

Spectral indexes are derived from Sentinel-2 L2A calibrated datasets only.

Land cover

When available the calibrated dataset also includes single band assets providing valuable information on the land cover of the areas covered by image footprint.

Cloud mask

In the CopernicusLAC Platform each Sentinel-3 SLSTR calibrated dataset contains the cloud mask band derived from the basic cloud masking of Sentinel-2 SLSTR L1B processing.

Scene classification layer

In the CopernicusLAC Platform each Sentinel-2 L2A calibrated dataset contains the Scene Classification Layer (SCL) layer which is offered as single band asset at 20m resolution. The SCL contained within the Sentinel-2 L2A data is derived with the Sen2Cor processor and identifies areas as clouds, snow, cloud shadows, vegetation, bare soil and water. Description, bit values, and color key of each SCL class are defined in the below table 5.

Value Classification HTML Color code
0 No data #000000
1 Saturated or defective #ff0000
2 Dark area pixels #2f2f2f
3 Cloud shadows #643200
4 Vegetation #00a000
5 Bare soils #ffe65a
6 Water #0000ff
7 Unclassified #808080
8 Cloud medium probability #c0c0c0
9 Cloud high probability #ffffff
10 Thin cirrus #64c8ff
11 Snow or ice #ff96ff
Table 5 - Value and color code for each class of the SCL layer.

For more information find the Sentinel-2 L2A Algorithm documentation here

External land cover data

A Sentinel-3 SLSTR L1B calibrated dataset also contains also a single band land cover asset derived from the ESA CCI 2020 land cover dataset at 300 m resolution. This asset is derived after a cropping of the dataset over the image footprint and a warping of the raster to have it resampled and co-located to the same grid of Sentinel-3 SLSTR VIS/SWIR multispectral bands.

Outputs

The outputs of the Optical Products Calibration service are the following products given in COG format:

  1. TOA/BOA reflectance or brightness temperature single band assets for all multispectral bands,

  2. Single band assets representing NDVI, MIRBI, NBR, and NBR2 spectral indexes,

  3. Cloud mask single band asset as bitmask,

  4. SCL single band asset as discrete raster,

  5. Land cover single band asset as discrete raster.

The output of OPT-Calib is a STAC Item with the all the output Assets included.

Product specifications for the Optical Products Calibration service can be found in the following tables.

Attribute Value / description
Long Name TOA/BOA Reflectance from calibrated Panchromatic or Multispectral Calibrated Optical data
Short Name r-coastal, r-blue, etc. for all optical CBN for VIS, NIR, SWIR
Description TOA/BOA Reflectance single band asset for VIS, NIR, SWIR CBNs rescaled to 10000 from calibrated optical data
Processing level As source product (L1/L2)
Data Type Unsigned 16-bit Integer
Band Single
Format COG
Projection Native
Units Dimensionless
Valid Range [0 - 10,000]
Scale Factor *0.0001
Attribute Value / description
Long Name TOA brightness Temperature from calibrated optical data
Short Name bt-lwir11, bt-lwir12
Description Brightness Temperature (K) single band asset rescaled to 100 from calibrated optical data
Processing level As source product (L1/L2)
Data Type Unsigned 16-bit Integer
Band Single band
Format COG
Projection Native
Units K
Scale Factor *0.01
Attribute Value / description
Long Name Spectral Indexes
Short Name ndvi, mirbi, nbr, nbr2
Data type Float 32
Band Single
Format COG
Projection Native
No data value NaN
Attribute Value / description
Long Name Cloud Mask from Sentinel-3 SLSTR L1B data (1=cloud, 0=no-cloud))
Short Name cloud_mask
Data type UInt8
Band Single
Format COG
Projection Native
Spatial Resolution 1km
No data value 0
Attribute Value / description
Long Name Sentinel-2 L2A Scene Classification from sen2cor
Short Name scl
Data type Int16
Band Single
Format COG
Projection Native
Spatial Resolution 20m
No data value Nan
Attribute Value / description
Long Name ESA CCI Land Cover 2020
Short Name land-cover
Data type Int16
Band Single
Format COG
Projection Native
Spatial Resolution 500m
No data value Nan

  1. Orfeo Toolbox, Optical Calibration, Available at: https://www.orfeo-toolbox.org/CookBook/Applications/app_OpticalCalibration.html. ↩

  2. Thuillier, G., Hersé, M., Labs, D. et al. (2003), “The Solar Spectral Irradiance from 200 to 2400 nm as Measured by the SOLSPEC Spectrometer from the Atlas and Eureca Missions”. Solar Physics 214, 1–22. DOI: https://doi.org/10.1023/A:1024048429145. ↩

  3. Rouse J., Haas R. H., Schell J. A., Deering D. (1973), “Monitoring vegetation systems in the great plains with ERTS”, NASA. Goddard Space Flight Center 3d ERTS-1 Symp., Vol. 1, Sect. A. Available at: ntrs.nasa.gov ↩↩

  4. Trigg, S.; Flasse, S., "An evaluation of different bi-spectral spaces for discriminating burned shrub savanna", Int. J. Remote Sens. 2001, 22, 2641–2647. DOI: 10.1080/01431160110053185. ↩↩

  5. Key, C. H. and Benson, N. C. (2006), “Landscape Assessment (LA): Sampling and Analysis Methods”, USDA Forest Service Gen Tech. Rep RMRS-GTR-164-CD. FIREMON Fire effects monitoring and inventory System. Available at: fs.fed.us.) ↩↩

  6. Storey E.A., Stow D.A., O’Leary J.F. (2016), "Assessing postfire recovery of chamise chaparral using multi-temporal spectral vegetation index trajectories derived from Landsat imagery". Remote Sens. Environ. 2016, 183, 53–64. DOI: 10.1016/j.rse.2016.05.018. ↩