Optical data

You will find an optical tutorial here.

Implemented optical constellation

The product resolution is the one given in Data Access Portfolio (2014-2022, section 6.2). The Data Access Portfolio Document presents the offer of the datasets and data access services that are made available to the Copernicus Users in response to their Earth Observation data requirements. However, especially for SAR data, the default pixel size of GRD bands is different from the product resolution!

Constellations	Class	Use archive
Sentinel-2	S2Product	✅
Sentinel-2 on AWS cloud	S2CloudProduct	-
Sentinel-2 Theia	S2TheiaProduct	✅
Sentinel-3 SLSTR	S3SlstrProduct	✅
Sentinel-3 OLCI	S3OlciProduct	✅
Landsat 8/9 OLI-TIRS	LandsatProduct	✅
Landsat 7 ETM	LandsatProduct	COL1*: ❌, COL2: ✅
Landsat 4/5 TM	LandsatProduct	COL1*: ❌, COL2: ✅
Landsat 1-5 MSS	LandsatProduct	COL1*: ❌, COL2: ✅
Harmonized Landsat-Sentinel	HlsProduct	✅
PlanetScope	PlaProduct	✅(❌ if multiple subdatasets)
SkySat	SkyProduct	✅(❌ if multiple subdatasets)
RapidEye	ReProduct	✅(❌ if multiple subdatasets)
Pleiades-Neo	PneoProduct	✅
Pleiades	PldProduct	✅
Vision-1	Vis1Product	✅
SPOT 4/5	Spot45Product	✅
SPOT 6/7	Spot67Product	✅
Maxar**	MaxarProduct	✅
SuperView-1	Sv1Product	✅
GEOSAT-2	Gs2Product	✅

* Archived Landsat Collection-1 are not managed because of the tar.gz format, which is too slow to process. It is better to work on the extracted product. Landsat-9 Collection-1 products don’t exist.

** *Maxar satellites: Worldview Legion, 2, 3, 4 and GeoEye-1 (but WorldView-1, QuickBird should work, but they don’t have been tested).

Warning

Satellites products that cannot be used as archived have to be extracted before use.

Constellations	Product Types	Default Pixel Size
Sentinel-2	L1C & L2A & L2Ap	10 m
Sentinel-2 on AWS Cloud	L2A (for now)	10 m
Sentinel-2 Theia	L2A	10 m
Sentinel-3 SLSTR	RBT	300 m
Sentinel-3 OLCI	EFR	500 m
Landsat 8/9 OLI-TIRS	Level 1 & 2	30 m
Landsat 7 ETM	Level 1 & 2	30 m
Landsat 4/5 TM	Level 1 & 2	30 m
Landsat 1-5 MSS	Level 1	60 m
Harmonized Landsat-Sentinel	L30 & S30	30 m
PlanetScope	L3A & L3B	3 m
SkySat	(Collect Product) ortho_*	0.5 m
RapidEye	L3A	5 m
Pleiades-Neo	SEN**, PRJ, ORT & MOS	0.3 m (PAN), 1.2 m (MS)
Pleiades	SEN**, PRJ, ORT & MOS	0.5 m (PAN), 2 m (MS)
Vision-1	PRJ & ORTP	0.9 m (PAN), 3.5 m (MS)
SPOT 4/5	L1A, L1B, L2A	10 m (PAN), 20 m (MS)
SPOT 6/7	SEN**, PRJ, ORT & MOS	1.5 m (PAN), 6 m (MS)
Maxar*	Standard & Ortho	0.3 to 0.5 m (PAN), 1.2 to 2.0 m (MS)
SuperView-1	L1B & L2A	0.5 m (PAN), 2 m (MS)
GEOSAT-2	L1B & L1C & L1D & L1S	0.4 m (PAN), 2 m (MS) for L1D data

* *Maxar satellites: Worldview Legion 2, 3, 4 and GeoEye-1 (but WorldView-1, QuickBird should work, but they don’t have been tested).

** ⚠ DIMAP SEN products are orthorectified using RPCs and not the rigorous sensor model. Shift can occur. Please refer to this issue.

Note

Note that for the rest of this documentation:

• HLS.S30 and Sentinel-2 data are equivalent (with the only difference of a 30 m gsd).

• HLS.L30 and Landsat-OLI-TIRS data are equivalent.

The goal of EOReader is to implement every constellation that can be used in the Copernicus Emergency Management Service. The constellations that can be used during CEMS activations are (as of 09/2021):

Optical bands

The following bands are available in EOReader but may not be available for all constellations.

Warning

EOReader loads bands in UTM. VHR bands can be orthorectified but EOReader needs a DEM for that. Be sure to position the environment variable EOREADER_DEM_PATH to the DEM you want.

Satellite bands

Note

Optical bands are always loaded in reflectance, except for brilliance temperature bands that are left as is.

However, the user can ask for bands “as is” by using the TO_REFLECTANCE keyword with the load and stack functions. It will give the bands in the provider format (either DN, radiance or reflactance in uint16…). See the optical notebook to see how to implement it.

These bands will be cleaned according to the user’s needs. By default, only the pixels outside the detectors footprints are set to nodata. Methods to clean optical bands are best described in the notebook about the optical cleaning methods.

Band names

These bands are mainly based on Sentinel-2 bands. The names are the ones from Sentinel-2, even if sometimes it can be confusing (especially for Coastal Aerosol and SWIR Cirrus bands).

Some additions to the Sentinel-2 bands have been made (YELLOW, PAN…):

• CA: Coastal Aerosol

• BLUE

• GREEN_1: First Green band of PlanetScope 8 bands, mapped to the fifth band of Sentinel-3 OLCI

• GREEN

• YELLOW

• RED

• VRE_1: Vegetation Red Edge 1

• VRE_2: Vegetation Red Edge 2

• VRE_3: Vegetation Red Edge 3

• NIR: Near Infrared

• NARROW_NIR: Spectrally resolved Narrow Near Infrared (band 8A for Sentinel-2, same as NIR for other sensors)

• WV: Water vapour

• SWIR_CIRRUS: Cirrus band

• SWIR_1: Short Wave Infrared 1

• SWIR_2 Short Wave Infrared 2

• PAN: Panchromatic

• TIR_1: Thermal Infrared 1

• TIR_2: Thermal Infrared 2

Note

Note that bands that are not mapped can still be used (i.e. for Sentinel-3 OLCI and SLSTR). They will be called by their true name (i.e. F1 or Oa05). However, the mapped band will have to be called with their mapped name, i.e. BLUE rather than Oa04 for Sentinel-3 OLCI. But the bands that are not present in the bands.__init__ file won’t be recognized.

Optical band mapping between constellations

Thermal bands are not represented in this plot. Here is the table of the mapping:

Bands names (alias)	Thermal IR (TIR_1)	Thermal IR (TIR_2)
Sentinel-3 SLSTR	8 (1km)	9 (1km)
Landsat OLI-TIRS (8-9)	10 (100m)	11 (100m)
Landsat ETM (7)	6 (60m)	6 (60m)
Landsat TM (5-4)	6 (120m)	6 (120m)
Landsat MSS (only Landsat-3)	8 (240m)	8 (240m)

Not all bands of these constellations are mapped in EOReader. However, they can still be used by using directly their name (i.e. F1 or Oa05)*

Sentinel-2 L2A specific bands

EOReader can now load (since v0.22.0) Sentinel-2 L2A specific bands (quality assurance bands) as any other band:

• Aerosol optical thickness (AOT)

• Scene average water vapour (WVP)

• Scene classification layer (SCL)

Cloud bands

Maximum 5 cloud bands are available, according to the files provided in the data. All the bands are rasterized and orthorectified if needed (for Sentinel-2 or 3 data for example), ready to be stacked.

The only difference with the other bands is that the cloud bands are provided in uint8 and have a nodata equal to 255.

• RAW_CLOUDS: Raw Cloud file as provided (the only changes are the orthorectification and rasterization). Can provide other flags, or cloud probability.

• CLOUDS: Cloud presence (1) or absence (0). If clouds are provided in probabilities, their presence is determined according to Landsat definition (proba> 67%)

• CIRRUS: Cirrus presence (1) or absence (0). If clouds are provided in probabilities, their presence is determined according to Landsat definition (proba> 67%)

• SHADOWS: Shadows presence (1) or absence (0). If clouds are provided in probabilities, their presence is determined according to Landsat definition (proba> 67%)

• ALL_CLOUDS: Cloud OR Cirrus OR Shadows presence (1) or absence (0).
  Do not take into account missing bands (i.e. for Landsat MSS constellation, ALL_CLOUDS == CLOUDS)

Sensors	Cloud files	Clouds Bands
Sentinel-2	✅	RAW_CLOUDS, CLOUDS, CIRRUS, ALL_CLOUDS
Sentinel-2 Theia	✅	RAW_CLOUDS, CLOUDS, SHADOWS, CIRRUS, ALL_CLOUDS
Sentinel-3 OLCI	❌
Sentinel-3 SLSTR	✅	RAW_CLOUDS, CLOUDS, CIRRUS, ALL_CLOUDS
Landsat 8/9	✅	RAW_CLOUDS, CLOUDS, SHADOWS, CIRRUS, ALL_CLOUDS
Landsat 7	✅	RAW_CLOUDS, CLOUDS, SHADOWS, ALL_CLOUDS
Landsat 4/5 TM	✅	RAW_CLOUDS, CLOUDS, SHADOWS, ALL_CLOUDS
Landsat 1-5 MSS	✅	RAW_CLOUDS, CLOUDS, ALL_CLOUDS
Harmonized Landsat-Sentinel	✅	RAW_CLOUDS, CLOUDS, SHADOWS, CIRRUS, ALL_CLOUDS
PlanetScope	✅	RAW_CLOUDS, CLOUDS, SHADOWS, CIRRUS, ALL_CLOUDS
SkySat	✅	RAW_CLOUDS, CLOUDS, SHADOWS, CIRRUS, ALL_CLOUDS
RapidEye	✅	RAW_CLOUDS, CLOUDS, ALL_CLOUDS
Pleiades	✅	RAW_CLOUDS, CLOUDS, ALL_CLOUDS
Pleiades-Neo	✅	RAW_CLOUDS, CLOUDS, ALL_CLOUDS
SPOT-4/5	❌
SPOT-6/7	✅	RAW_CLOUDS, CLOUDS, ALL_CLOUDS
Vision-1	❌
Maxar (GeoEye-1, WorldViews)	❌
SuperView-1	❌
GEOSAT-2	❌

DEM bands

Optical constellations can all load DEM, SLOPE and HILLSHADE bands. The SLOPE band is given in degrees. Please post an issue if you need this band in percent.

Use the environment variable EOREADER_DEM_PATH to position your worldwide DEM. You can use both a local path e.g. /mnt/dataserver/dems/srtm_30_v4/index.vrt or \\dataserver\DEMS\srtm_30_v4\index.vrt or a URL pointing to a web resources hosted on a S3 compatible storage e.g. https://s3.storage.com/dem-bucket/srtm_cog.tif (not available on Windows for now).

Available spectral indices

EOReader uses (from version 0.18.0) the indices described in the awesome spectral indices (ASI) project.

⚠ You may need to install the last spyndex directly from GitHub latest version to have all available indices: pip uninstall spyndex && pip install -U --no-deps git+https://github.com/awesome-spectral-indices/spyndex.git

In addition to all these indices, EOReader kept the following indices for legacy purposes:

Spectral Indices	ASI equivalent	Formulas (with ASI syntax)	Name
NDMI21 (or NDMI2100)	❌	(N - S2) / (N + S2)	Normalized Difference Moisture Index (with 2100 nm)
NDRE2	❌	(N - RE2) / (N + RE2)	Normalized Difference Red-Edge Index with RE2
NDRE3	❌	(N - RE3) / (N + RE3)	Normalized Difference Red-Edge Index with RE1
CI32	❌	(RE3 / RE2) - 1	Chlorophyll Index RedEdge RE3/RE2
CI21	❌	(RE2 / RE1) - 1	Chlorophyll Index RedEdge RE2/RE1
SBI	❌	(R**2 + N**2)**0.5	Soil Brightness Index
SCI	❌	3*G - R - 100	Soil Cuirass Index
SRSWIR	❌	S1/S2	Simple Ratio SWIR16/SWIR21 Clay Minerals
TCBRI	❌	0.3037 *B + 0.2793 * G + 0.4743 * R + 0.5585 * N + 0.5082 * S1 + 0.1863 * S2	Tasseled Cap Brightness
TCGRE	❌	0.2848 *B + 0.2435 * G + 0.5436 * R + 0.7243 * N + 0.0840 * S1 + 0.1800 * S2	Tasseled Cap Greenness
TCWET	❌	0.1509 *B + 0.1973 * G + 0.3279 * R + 0.3406 * N + 0.7112 * S1 + 0.4572 * S2	Tasseled Cap Wetness
WV_VI	❌	(WV - R) / (WV + R)	WorldView Vegetation Index
WV_WI	❌	(WV - A) / (WV + A)	WorldView Water Index

Note

WorldView-1 is not concerned by spectral indices as it only has a panchromatic band

Some indices are parametric (see the list here). To add any parameter you can pass it to your load or stack functions:

>>> prod.load("WDRVI", alpha=1)

Masks

Since v0.22.0, EOReader can load masks directly in the load function. However, the band masks are not exposed directly into eoreader.bands, but through their enums. They can also be called by their string representation.

# HLS FMASK mask, called through the enum
>>> from eoreader.bands import HlsMaskBandNames
>>> prod.load(HlsMaskBandNames.FMASK)

For Sentinel-2 et Sentinel-2 Theia data, some masks are band-specific and should be called with the associated_bands keywords. This leads to a side effect that can be disorienting: these masks should be retrieved from the dataset with an unusual key ({band}_{associated_band}).

>>> # Sentinel-2 Theia, called with mask strings
>>> masks = prod.load(["DFP", "EDG", "SAT", "MG2", "IAB", "IAB"], associated_bands={"SAT": "RED", "DFP": ["GREEN", "RED"]})

>>> # Show all variables
>>> masks.data_vars
Data variables:
    DFP_GREEN                 (band, y, x) 
    DFP_RED                   (band, y, x)
    S2TheiaMaskBandNames.EDG  (band, y, x)
    SAT_RED                   (band, y, x)
    S2TheiaMaskBandNames.MG2  (band, y, x)
    S2TheiaMaskBandNames.IAB  (band, y, x)

# Open saturation mask for green band
>>> masks["DFP_GREEN"]
<xarray.DataArray 'DFP_GREEN' (band: 1, y: 10980, x: 10980)> Size: 482MB
dask.array<where, shape=(1, 10980, 10980), dtype=float32, chunksize=(1, 1024, 1024), chunktype=numpy.ndarray>
Coordinates:
  * x            (x) float64 88kB 5e+05 5e+05 5e+05 ... 6.098e+05 6.098e+05
  * y            (y) float64 88kB 5.5e+06 5.5e+06 5.5e+06 ... 5.39e+06 5.39e+06
    spatial_ref  int64 8B 0
  * band         (band) int64 8B 1
Attributes:
    ...

See the different MaskBandNames enums to discover all the available masks and how to call them. ⚠ This feature is experimental.

Documentary Sources

• Data Access Portfolio (2014-2024)

Copernicus

• Copernicus Contributing Missions

Landsat

• Collection 1 vs Collection 2

• Quality assessment Collection 1

• Quality assessment Collection 2

• MSS Collection 2 Data Format

• TM Collection 2 Data Format

• ETM Collection 2 Data Format

• OLCI Collection 2 Data Format

Sentinel-2

• Cloud masks

• Product Specification (Processing Baseline < 4.0)

• Product Specification (Processing Baseline > 4.0)

Harmonize Landsat-Sentinel

• Product Guide v2.0

Sentinel-2 Theia

• Product Format

Sentinel-3

• OLCI main page

• SLSTR main page

• OLCI Handbook

• SLSTR Handbook

• OLCI Product Format

• SLSTR Product Format

• SLSTR Clouds

PlanetScope, SkySat and RapidEye

• Product Specification

• Instruments

• Usable Data Mask

Vision-1

• Vision-1 on Airbus Website

• Presentation of the satellite

• User guide

DIMAP V1: SPOT-4/5

• Product Format (a bit too old)

• Preprocessing levels

• Resolution and spectral modes

• Convert to reflectance

DIMAP V2: Pleiades, SPOT-6/7

• DIMAP V2 Format

• Pleiades User Guide (2019)

• SPOT-6/7 User Guide (2019)

• Pleiades Neo

Maxar (ex Digital Globe)

• Worldview Legion

• Radiometric Use of WorldView Legion 1 and WorldView Legion 2 imagery

• Worldviews

• System-Ready Imagery

• View-Ready Imagery

• Map-Ready Imagery

• ISD

• Digital Globe Standard Imagery (old)

• Digital Globe Products Guide (old)

• Conversion from DN to TOA reflectance

SuperView-1

• Imagery Product Guide

GeoSat-2 (ex Deimos-2)

• Imagery User Guide

Instruments

Most instruments can be found on OSCAR or on ESA Earth Online.

Band mapping

You can find a magnificent band comparison chart on the Imagico blog.

Index

• Awesome Spectral Indices

• Index consistency

• WorldView Index

• Specific sources inside the index function documentation in indices

Documentation last accessed on 12/03/2025