The Sen2VM core is a standalone tool designed to generate geolocation grids that will be included in the Level-1B (L1B) product. Its primary function is to create direct location grids, mapping the L1B product in sensor geometry down to the ground. Additionally, the tool supports the generation of inverse location grids, enabling mapping from a specific ground area back to the corresponding area in sensor geometry.
Please note that Sen2VM exists implemented as a SNAP plugin, which calls the Sen2VM standalone tool during execution.
This documentation is split into 4 parts:
- Instructions to compile and/or run Sen2VM with some examples: $ Quickstart
- Description and format of expected inputs: § Inputs
- Details on generated outputs: § Outputs
- Validation process, including test procedures and data used: § Validation
First, download the jar of Sen2VM core, then run the following command to launch it:
java -jar target/sen2vm-core-<NN.NN.NN>-jar-with-dependencies.jar -c [configuration_filepath] [-p [parameters_filepath]]
Where:
- <NN.NN.NN> is the version number of Sen2VM launched,
- configuration_filepath: configuration file containing all inputs related to product or grids that are required by Sen2VM (see §2.1 Configuration file for further information). Please note that this input is Mandatory.
- parameters_filepath: file to configure the detectors/bands to process. If not available, all detectors/bands will be processed (see §2.2 Parameters file for further information).This input is Optional.
Note
Sen2VM core can also be rebuild from sources. Please refer to §1.3 How to compile sen2vm-core
Example from current repository:
java -jar target/sen2vm-core-0.0.1-jar-with-dependencies.jar -c src/test/resources/configuration_example.json -p src/test/resources/params.json
Note
To understand the configuration, please refer to §2 Inputs
Caution
gdal version shall be compatible with the new Sen2VM grids. Official gdal does not yet include this driver/possibility. A Pull Request is currently opened, but in the meantime, this gdal version can be find here
Direct location grids can be used to preform a resampling. It can be done using gdal or using OTB resampler. To see the geometric validation of those 2 methods, please refer to the Validation Document
L1B with geolocation grids are seamlessly handled by gdal through the development of a dedicated S2 L1B gdal driver.
# If needed, point to the local gdal version handling Sen2VM grids
# export PATH=~/code/senv2vm/bin/bin/:$PATH
# export LD_LIBRARY_PATH=~/code/senv2vm/bin/lib:$LD_LIBRARY_PATH
#Initialisation of the product for gdal
gdalinfo /PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml
# Creation of a mosaic for D09_B01 (D - detector; B - band)
gdal_translate SENTINEL2_L1B_WITH_GEOLOC:"/PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml":S2B_OPER_GEO_L1B_DS_2BPS_20241019T153411_S20241019T120215_D09_B01 /PATH_TO_DATA/working/madeire_D09_B01.tif
# Resampling/orthorectification using gdal
gdal_warp SENTINEL2_L1B_WITH_GEOLOC:"/PATH_TO_DATA/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml":S2B_OPER_GEO_L1B_DS_2BPS_20241019T153411_S20241019T120215_D09_B04 /PATH_TO_DATA/working/projected_D09_B04.tif -t_srs EPSG:32628 -tr 10 -10This method can be resumed into three main steps:
- Creation of a mosaic of all images,
- Convertion of the direct location grid into an inverse location grid using scipy,
- Computation of the otb resampling using the mosaic and the inverse location grid.
# If needed, point to the local gdal version handling Sen2VM grids
# export PATH=~/code/senv2vm/bin/bin/:$PATH
# export LD_LIBRARY_PATH=~/code/senv2vm/bin/lib:$LD_LIBRARY_PATH
# To get the name of the grid for your band/detector either
# - look inside teh DATASTRIP/S*/GEO_DATA folder
# - launch a gdalinfo to get the list
gdalinfo /PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml
# Creation of a mosaic of images of all granules (for D09_B01)
gdal_translate SENTINEL2_L1B_WITH_GEOLOC:"/PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml":S2B_OPER_GEO_L1B_DS_2BPS_20241019T153411_S20241019T120215_D09_B01 /PATH_TO_DATA/working/madeire_D09_B01.tif
# Conversion of Sen2VM direct location grid into an inverse location grid with a 45m step
python sen2vm_invloc_from_dir_loc_grid.py /PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/DATASTRIP/S2B_OPER_MSI_L1B_DS_2BPS_20241019T153411_S20241019T120215_N05.11/GEO_DATA/S2B_OPER_GEO_L1B_DS_2BPS_20241019T153411_S20241019T120215_D09_B01.vrt --loglevel INFO 45.0 /PATH_TO_DATA/working/grid_D09_B01.tif
# Use OTB for resampling (default BCO interpolator) with
# - io.in: Mosaic of all images
# - grid.in: an inverse location grid
otbcli_GridBasedImageResampling -io.in /PATH_TO_DATA/working/madeire_D09_B01.tif -io.out /PATH_TO_DATA/working/warp_otb_D09_B01.tif -grid.in /PATH_TO_DATA/working/grid_D09_B01.tif -grid.type loc -out.ulx 293050 -out.uly 3697900 -out.spacingx 60 -out.spacingy -60 -out.sizex 933 -out.sizey 2040
# Add georeferencing to the image
gdal_translate -a_srs EPSG:32628 /PATH_TO_DATA/working/warp_otb_D09_B01.tif /PATH_TO_DATA/working/warp_otb_D09_B01_georef.tifPlease note that the script sen2vm_invloc_from_dir_loc_grid.py used by this method is directly available on this git, at this location.
Necessary prerequisites:
- numpy
- rasterio
- utm
- scipy
- argparse
- pathlib
Caution
Please note that there is currently an issue on the OTB side. Until corrected, the grid information must be adjusted by half the resolution of the target pixel (spacing) in both directions, to be synchronised as following:
- out.ulx shall be updated to 293080 ( = 293050 + 60/2)
- out.uly shall be updated to 3697870 ( = 3697900 + (-60)/2)
# If needed, point to the local gdal version handling Sen2VM grids
# export PATH=~/code/senv2vm/bin/bin/:$PATH
# export LD_LIBRARY_PATH=~/code/senv2vm/bin/lib:$LD_LIBRARY_PATH
# To get the name of the grid for your band/detector either:
# - look inside the DATASTRIP/S*/GEO_DATA folder
# - launch a gdalinfo to get the list
gdalinfo /PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml
# Creation of a mosaic of images of all granules (for D09_B01)
gdal_translate SENTINEL2_L1B_WITH_GEOLOC:"/PATH_TO_DATA/S2B_MSIL1B_20241019T120219_N0511_R023_20241022T154709.SAFE/S2B_OPER_MTD_SAFL1B_PDMC_20241022T154709_R023_V20241019T120217_20241019T120235.xml":S2B_OPER_GEO_L1B_DS_2BPS_20241019T153411_S20241019T120215_D09_B01 /PATH_TO_DATA/working/madeire_D09_B01.tif
# Use OTB for resampling (default BCO interpolator) with:
# - io.in: Mosaic of all images
# - grid.in: an inverse location grid
otbcli_GridBasedImageResampling -io.in /PATH_TO_DATA/working/madeire_D09_B01.tif -io.out /PATH_TO_DATA/working/warp_otb_D09_B01.tif -grid.in /PATH_TO_DATA/OUTPUT_INV_GRID/XXX.tif -grid.type loc -out.ulx 293050 -out.uly 3697900 -out.spacingx 60 -out.spacingy -60 -out.sizex 933 -out.sizey 2040
# Add georeferencing to the image
gdal_translate -a_srs EPSG:32628 /PATH_TO_DATA/working/warp_otb_D09_B01.tif /PATH_TO_DATA/working/warp_otb_D09_B01_georef.tifBefore compiling/installing sen2vm-core, make sure to install the required dependencies. To do so, please refer to https://github.com/sen2vm/sen2vm-build-env/tree/main
Then, inside sen2vm-core folder, run the next commands:
mvn clean install
java -jar target/sen2vm-core-<NN.NN.NN>-jar-with-dependencies.jar -c [configuration_filepath] [-p [parameters_filepath]]
Inputs required by Sen2VM are:
- L1B Product,
- Some GIPP files (parameters files used in operational production),
- Digital Elevation Model (DEM),
- GEOID model to measure precise surface elevations,
- IERS bulletin that provides data and standards related to Earth rotation and reference frames,
- Additional information for configuration.
All inputs are described in this section.
The configuration file will contain all information about the product, the auxiliarry data and the operations to be performed. It is a file in JSON format and an example is available at: https://github.com/sen2vm/sen2vm-core/blob/main/src/test/resources/configuration_example.json
Each parameter description can be found in the table below:
| Name | Type | Required | Description |
|---|---|---|---|
| l1b_product | string | Mandatory | Path to L1B_PRODUCT folder, where L1B_Product format is in SAFE format (described in § L1B Product, including DATASTRIP + GRANULE) |
| gipp_folder | string | Mandatory | Path to a folder containing at least the 3 types of GIPP required by Sen2VM (other will be ignored). For more information, refer to § GIPP. |
| gipp_check | boolean | Optional | If true (default), check of GIPP version activated (see GIPP section § GIPP) |
| dem | string | Mandatory | Path to the FOLDER containing a DEM in the right format (cf § Altitude/DEM). |
| geoid | string | Mandatory | Path to the FILE containing a GEOID in the right format (cf § Altitude/GEOID) |
| iers | string | Optional | Path to the IERS folder containing the IERS file in the right format (cf § IERS) |
| operation | string | Mandatory | In term of operation you can select the following Sen2VM configurations:
|
| deactivate_available_refining | boolean | Optional | If set to false (default), refining information (if available in Datastrip Metadata) are used to correct the model before geolocation, cf. product description in § L1B Product |
| export_alt | boolean | Optional | If set to false (default), direct location grids will include only two bands: Long/Lat. If set to true, a third band representing the Altitude will also be exported, increasing the output grid size. See product description in §[Direct location grids](#31-direct location grids) |
| steps | double | Mandatory | The step is mandatory and must be specified as one per resolution: “10m_bands”, “20m_bands” & “60m_bands””. Please note that only floating numbers in the format NNNN.DDD are accepted and that the unit is given in pixel for direct location and in metrics of referential system for inverse location. |
| inverse_location_additional_info | Mandatory if “inverse”, else useless. | For the inverse location additional information please refer to the dedicated table below |
The field “inverse_location_additional_info” is not required and will be ignored if direct location grids are asked. However, it is mandatory for inverse location grids generation and Sen2VM will raise an error if this information is missing.
| Name | Type | Required | Description |
|---|---|---|---|
| ul_x | double | Mandatory | X coordinates of the upper left point defining the area of desired resampled product. |
| ul_y | double | Mandatory | Y coordinates of the upper left point defining the area of desired resampled product. |
| lr_x | double | Mandatory | X coordinates of the lower right point defining the area of desired resampled product. |
| lr_y | double | Mandatory | Y coordinates of the lower right pointdefining the area of desired resampled product. |
| referential | string | Mandatory | A string defining the referential used. Please note, ul and lr points’ coordinates are given in this referential. Example: EPSG:4326 |
| output_folder | string | Mandatory | Path where the inverse location grids are written (see § Inverse location grids) |
Note
Inverse location grids footprint will enclose desired product footprint [ul_x, ul_y, lr_x, lr_y].
Note
L1B products can be downloaded at https://browser.dataspace.copernicus.eu/. Please note that special access for L1B products might be required by submitting a request via the FAQ section.
Important
The expected format is compatible with the SAFE format, i.e. a folder structured as illustrated in the following sections.
The mandatory inputs for Sen2VM are the Datastrip and the Granules metadata.
These metadata must be organized in a specific directory structure. Within the L1B folder, only the following subdirectories will be considered:
- DATASTRIP
- GRANULES
Additional files (as in the SAFE format) may be present in the folders, but they will simply be ignored by Sen2VM core.
As for the DATASTRIP folder, it must contain a subfolder named after the Datastrip reference and inside this subfolder, the Datastrip Metadata (also named after the Datastrip reference). This structure is compatible with the SAFE format and others files/folders will be ignored:
Finally the GRANULE folder shall contain one folder per granule, each with the GRANULE naming convention, and each containing the GRANULE metadata. This format is compatible with SAFE and others files/folders will be ignored:
[...]
Refining for Sentinel-2 refers to geolocation refinement that aims at correcting the satellite geometric models relative to the Global Reference Image.
Refining information can be found in the Datastrip metadata. If refined, the model of the satellite shall be adjusted accordingly. This information is stored in the field “Level-1B_DataStrip_ID/Image_Data_Info//Geometric_Info/Refined_Corrections_List/Refined_Corrections/MSI_State”. It is present only if the flag “Image_Refining” is set to “REFINED” and absent if set to “NOT_REFINED”.
An optional boolean argument is available in the configuration file (see §2.1 Configuration file): deactivate_available_refining.
Warning
By default, it is set at false, meaning that the refining information shall be taken into account if available in the Datastrip Metadata. However, if set at true, the datastrip shall be considered as NOT_REFINED, meaning ignoring the refining information.
GIPP are configuration files used in operation to:
- represent the stable satellite information,
- configure calibration parameters of the satellite,
- configure several algorithms of the processing chain.
By nature, GIPP are versionnable. It is important to process with the version used to generate the L1B product.
Warning
A check is implemented to verify that the version used is the same than the one listed in the Datastrip metadata (check on the name). This check can be deactivated through "gipp_check" parameter of the configuration file (cf §2.1 Configuration file). This parameter is optional, and by default, its value is set to true..
The versions of the GIPP used in operation are listed in the L1B Datastrip Metadata of the L1B product (see §2.1.1 L1B Product), in the tag Level-1B_DataStrip_ID/Auxiliary_Data_Info/GIPP_LIST, as illustrated below:
Caution
Please note that the GIPP are not directly available in L1B products; they must be downloaded beforehand by users on this dedicated repository.
The GIPP required are the following ones:
- GIP_VIEDIR: contains Viewing Direction required by Rugged to create viewing model based on TAN_PSI_X/Y_LIST tags. There is one GIP_VIEDIR file per band and each file contains information per detector (in the following tags: [DATA/VIEWING_DIRECTIONS_LIST/VIEWING_DIRECTIONS/TAN_PSI_X_LIST] and [DATA/VIEWING_DIRECTIONS_LIST/VIEWING_DIRECTIONS/TAN_PSI_Y_LIST])
- GIP_SPAMOD: contains transformations to apply to viewing direction from tags, available in the [DATA] field:
- PILOTING_TO_MSI,
- MSI_TO_FOCAL_PLANE,
- FOCAL_PLANE_TO_DETECTOR XML, available per detector
- GIP_BLINDP: contains information on blind pixel, contained in BLIND_PIXEL_NUMBER tag: [DATA/BAND/BLIND_PIXEL_NUMBER], available per band
The main purpose of the tool is to add geolocation to the L1B product images. Hence to be precise, the altitude shall be taken into account, as its importance is far from neglectable on final geolocation, as illustrated below:
For this, as Sen2VM uses SXGEO (OREKIT/RUGGED), a GEOID and a DEM shall be used. So, path of both shall be provided in input (cf §2.1 Configuration file).
Access to the DEM is provided via a path to a folder containing the dataset. The DEM must meet the following requirements:
- it should be split into files or folders (dynamically read) per square degrees,
- each DEM file (per square degree) shall be readable by gdal.
Examples of DEM structures can be found in /src/test/resources/DEM/
The GEOID shall be readable by gdal. One example of GEOID is available at S2__OPER_DEM_GEOIDF_MPC__20200112T130120_S20190507T000000.gtx
The IERS represents the "International Earth Rotation and Reference Systems". It is important to have a valid and precise one to get a precise geolocation. During operational processing of the Sentinel-2 data, IERS information are integrated in the L1B metadata datastrip, hence IERS information are available in the L1B product used in input, in the field Level-1B_DataStrip_ID/Auxiliary_Data_Info/IERS_Bulletin as illustrated below:
IERS bulletins are used seamlessly with OREKIT, leading in very precise EOP (Earth Orientation Parameters) handling. However, the IERS available in the L1B Datastrip metadata is not in the right format, hence EOP entries are initialized directly with the information, as “custom EOP”, skipping the IERS bulletins reader parts.
However, a very slight (neglectable) difference will be observed in geolocation compared to operational processing which uses the full IERS bulletin for its ortho-rectification process.
Hence an IERS bulletin can be provided in input by users (as optional).
Important
Sen2VM behaviour regarding IERS will then be:
- If an IERS bulletin is provided in input:
- Verify the date of the IERS bulletin compared to the Datastrip acquisition dates (using DATASTRIP_SENSING_START and DATASTRIP_SENSING_STOP in “Level-1B_DataStrip_ID/General_Info/Datastrip_Time_Info"
- Go in error if the IERS does not contains the full datastrip duration,
- Use the IERS bulletin if it contains the full datastrip duration,
- If not, use the information available in the L1B Datastrip Metadata.
Format of the IERS bulletin that can be provided is Bulletin A
Sen2VM calls SXGEO which is a mono-thread software. Sen2VM is also designed to be a mono-threaded software. However, as computations can be long, and because each couple detector/band is an independent model, user might want to process only parts of the Datastrip per thread. This capability is handled by SXGEO and was propagated to Sen2VM. This way, users can parallelize the process by itself, outside of Sen2VM.
The selection of detectors and bands to process is defined in a JSON parameters file. If this file is not provided, Sen2VM will process all detectors/bands.
The parameters file contains 2 fields:
- “detectors”: detectors are passed through string representing Sentinel-2 detectors encoded in 2 digits, separated by “-”. Detectors indexes are from “01” to “12”.
- “bands”: bands are passed through string representing Sentinel-2 bands encoded in 2 digits with a “B” before and separated from “-”. Bands are going from “B01” to “B12”, including a “B8A”.
If a field (“detectors” or “bands”) is missing in the params.json file, all items of the missing field will be processed. For example, if “bands” if absent from the previous example in figure above, Sen2VM will process all bands of detectors 1, 3 and 4.
Tip
It is to be noted that a small optimization in SXGEO is done not to reload DEM tiles when processing bands of the same resolution for the same detector.
The output of the Sen2VM tool can be either direct location grids or inverse location grids. Their computation depends on the following parameters:
- selected detectors and bands,
- grid step,
- computation options (e.g. refining).
Please note that only the direct location grids will be included in the input product and handled by gdal. Inverse location grids, as they represent a particulat area on the ground will be exported outside the product, in a folder selected by the user (which can be inside the input product if wanted).
A direct location grid is a grid which maps sensor coordinates with ground ones in WGS84 coordinates (EPSG:4326). Direct location grid is regular and in sensor reference frame (for one band/detector couple).
Sen2VM direct location grid computation takes as input the L1B product, the auxiliary information (see L1B product, GIPP, Altitude, IERS) and the grid parametrization:
- Bands/detectors to process (Parameters File),
- 3 steps, one per band resolution (10m, 20m, 60m) in pixels (double) Configuration File.
As output:
- At granule level: geolocation grids will be written (per granules/bands).
- At datastrip level:
- several .vrt (virtual dataset) will be written (per detectors/bands).
- The configuration file used in input with the date/time will be added in with the vrt files
Output grids will be integrated directly in the input product.
Caution
Please note that writing permissions in the L1B input folder are mandatory.
Before processing, a verification will be done to determine whether direct location grids are already available in the input L1B product folder, for the detectors/bands selected. If at least one is present for one couple detector/band, Sen2VM will raise an error and stop. Both granules and datastrip folder will be inspected (see output grids format and location in the following sections).
Grids’ location and naming is at granules level:
- 1 grid per couple “L1B granule”/”Sentinel-2 band”,
- Grids include 2 (optionally 3) bands (Long/Lat/alt)
- Grids are in geotiff format with float32 coding positions which allow approximately centimetre precision for lat/lon coordinates. JP2000 is not suitable, as it does not support Float32 encoding, resulting in insufficient precision.
- Grids location will be inside a GEO_DATA folder which will be inside each granules folders (at the same level than IMG_DATA and QI_DATA folders),
- Grids naming conventions will respect the corresponding image data inside the IMG_DATA folder with:
- GEO instead of MSI
- .tif instead of .jp2 extension as jp2 encoding is not possible for float32 data.
As example, for an image of the IMG_DATA folder, named:
- S2B_OPER_MSI_L1B_GR_DPRM_20140630T140000_S20230428T151505_D02_B01.jp2
The direct location grid will be generated in the GEO_DATA folder, and named:
- S2B_OPER_GEO_L1B_GR_DPRM_20140630T140000_S20230428T151505_D02_B01.tif
Note
The configuration file used in input with the date/time will be added in with the vrt files
At datastrip level grids’ location and naming is:
- 1 vrt per couple “detector”/”Sentinel-2 band”,
- Grids include 2 (optionally 3) bands (Long/Lat/alt)
- Grid location will be located inside a GEO_DATA folder, which resides within the DATASTRIP directory, at the same level as the QI_DATA folder.
- Grids naming conventions will respect the corresponding datastrip metadata convention with:
- GEO instead of MTD
- _DXX_BYY.vrt instead of .xml extension.
As example, for an datastrip metadata of the DATASTRIP folder, named:
- S2B_OPER_MTD_L1B_DS_DPRM_20140630T140000_S20230428T150801.xml
A folder named GEO_DATA, beside the QI_DATA folder and datastrip metadata will contain 156 vrt files (12 detectors x 13 bands) named:
- S2B_OPER_GEO_L1B_DS_DPRM_20140630T140000_S20230428T150801_DXX_BYY.vrt
Example of product with grid inside it:
Note
To be consistent with granules convention with first pixel center, outputs grids have the same first pixel centres.
Grids characteristics are:
-
Convention: grid raster type is "point" with first grid cell centre at first pixel centre of the first granule (referred by PIXEL_ORIGIN in granule geometric info). Grid position given an image position (row, col), is
with (grid row,grid col) = (Pixel Origin, Pixel Origin) = (1,1) at center of first grid cell
Example:
-
Grid Metadata: Grids will contain Metadata information of GDAL geolocation grids:
- SRS: WGS84 WKT format
- PIXEL_OFFSET
- LINE_OFFSET
- PIXEL_STEP: GRID STEP
- LINE_STEP: GRID STEP
- GEOREFERENCING_CONVENTION: PIXEL_CENTER
-
Overlap: to ensure grid granule continuity, one pixel overlap will be added. Taking advantage of this overlap, one granule can be individually resampled, or multiple granules (whole Datastrip for example) using the vrt grid concatenation without extra disk cost). Thus PIXEL/LINE OFFSET of the Grid Metadata will vary for each grid.
Direct location grids are intended to be used with bilinear interpolation operation. Direct location (i.e lon/lat positions) should be as follow:
-
Given an image position (row/col) compute grid fractional position (grid row, grid col):
-
Use bilinear interpolation on (grid row, grid col) to retrieve lon/lat/(alt).
Caution
Note that based on the used convention, the [1,1] coordinate is the first pixel of the grid
Tip
If user wants to perform direct location of a position outside the granule footprint, a bilinear extrapolation is possible.
Grids should at least have 2x2 cells
An inverse location grid is a grid which maps ground coordinates with sensor ones. Inverse location grids are georeferenced in geographic or cartographic reference frame. Inverse location grid is regular in ground reference frame (for one band and one detector).
To define the extend of the inverse location grid, parameters are described in section §2.1 Configuration file, but it can be resumed at:
- A referential system,
- A square defined by:
- One Upper Left point (UL),
- One Lower Right point (LR),
- 3 steps, one per band resolution (10m, 20m, 60m) in the referential metrics,
- An output folder.
As output, a geolocation grid will be created for each band and detector to process.
Outputs will be written in the folder provided by the user. For inverse location grids, granule level is not foreseen, since granule footprint can result in large margins in projected data.
Output grids’ convention will be:
- 13 grids (1 per Sentinel-2 band) per detector intersecting the area,
- Grids are in geotiff format with float32 coding positions which allow decimal information on col/row positions. JP2000 is not suitable, as it does not support Float32 encoding, resulting in insufficient precision.
- Grids naming conventions will respect the corresponding datastrip metadata convention with:
- INV instead of MTD,
- _DXX_BYY.tif instead of .xml extension.
As example, for a Datastrip metadata of the DATASTRIP folder:
- S2B_OPER_MTD_L1B_DS_DPRM_20140630T140000_S20230428T150801.xml
Output will be named:
- S2B_OPER_INV_L1B_DS_DPRM_20140630T140000_S20230428T150801_DXX_BYY.tif
Example:
Note
The configuration file used in input with the date/time will be added in with the vrt files
Inverse location grids will give the coordinates in the “detector” image reference frame, as if the granules from the same detector were concatenated into a single image.
- Grid metadata:
- are written as GDAL GEO information (SRS and geotransform) and Metadata keys, containing ROW_BAND and COL_BAND to specify row col index and PIXEL_ORIGIN which specify granule first pixel center convention, and NoData value.
- Metadata keys are:
- the SRS
- NoData fill value
- the grid footprint
- the size of the grid,
- the step of the grid,
- margins,
To ensure input bounding box coverage:
- Center of the first grid point is aligned with UL of the bounding box,
- Center of the last grid pixel is computed to ensure that it covers at least LR of the input bounding box.
As for example:
Grids are intended to be used with bilinear interpolation operation. Inverse locations (i.e image position given a ground position) should be as follow:
-
Given an ground position (lon/lat) or (x, y) compute grid fractional position (grid row, grid col) using grid geotransform for GRID_ORIGIN and GRID_STEP:
-
Use bilinear interpolation on (grid row, grid col) to retrieve row/col.
Caution
Beware of convention [PIXEL ORIGIN,PIXEL ORIGIN] is the first pixel of the grid..
Grids should at least have 2x2 cells.
In a nutshell, validation is split into 2 main parts:
- Functional tests:
- Integrated into the CI/CD process,
- Validate that Sen2VM takes into account the different input provided, and that output grids are consistent with reference ones.
- Quality tests:
- Launched manually,
- Proving, using several TDS, that grids generated by Sen2VM are consistent with the ones generated with the legacy,
- Proving that those grids can be used to reach equivalent quality at L1C,
- These processings will also be used to assess the processing performances.
Tests are more detailed in:
- In /src/test/java/esa/sen2vm/ for functionnal tests,
- In dedicated Document, for quality tests (which also includes functionnal tests description)