Skip to main content
Skip table of contents

SYNERGY Processing

Processing Baseline

The following table summarises the evolutions of the processing baseline used to process Sentinel-3 Synergy data since October 2018 for Sentinel-3A and February 2019 for Sentinel-3B.

Related Synergy Processing Baseline documents:

Processing Baseline

IPF Version

Changes

Date of deployment

SYN_L2_.002.20.00

SY2: 06.28

  • Handling partial saturation in cloud masking

  • Corrections of bugs

S3A: 28/02/2024

S3B: 27/02/2024

SYN_L2_.002.18.01
SYN_L2V.002.09.01
AOD_NTC.002.08.01

SY2 06.26
SY2_VGS 06.13
SY2_AOD 01.09

  • Inclusion of the SLSTR calibration factors and the OLCI offset

  • Adaptation to the OLCI frame offset and partial Saturation evolutions

  • Update of the VGT Spectral response function to be consistent with PROBA-V

  • Correction of several bugs

25-July-2023 for S3A
18-July-2023 for S3B

SYN_L2_.002.18.01 SYN_L2V.002.09.01 AOD_NTC.002.08.01

SY2: 06.25

SY_2_VGS: 06.12

SY2_AOD: 1.08

  • Minor update to account for S3C and S3D

S3A: 25/07/2023

S3B: 18/07/2023

SYN_L2_.002.15.00
SYN_L2V.002.07.00
AOD_NTC.002.06.00

SY2: 06.23
SY_2_VGS: 06.11
SY2_AOD: 1.06

Corrections of bugs

23-Aug-2022 for S3A
9-Sep-2022 for S3B

SYN_L2_.002.15.00
SYN_L2V.002.07.00
AOD_NTC.002.06.00

SY2: 06.22
SY_2_VGS: 06.10
SY2_AOD: 1.06

Insertion of the degradation flag from L1

27-Jan-2022

Processing baseline number in the manifest and products

S3A: 2.77
S3B: 1.55

SY2: 06.21
SY_2_VGS: 06.09

Bug corrections

14-Jun-2021

S3A: 2.66
S3B: 1.40

SY2: 06.20

Correction of rectangular patterns observed in T550 & SDR values

23-Jun-2020

Correction of negative AOD values found in the SYNERGY T550 product

2.63 / 1.36

SY2: 06.19

Bug corrections

30-Jan-2020

2.56 / 1.28

SY2: 06.18
SY2_VGS: 06.08

The SYN L2 VGT-like processing includes similar algorithms than PROBA-V concerning projection on 1 km grid and composite methods. As a result of this improvement, more geographical details are available on VGT-like product, providing a dataset more consistent with the actual geographical area.

15-Jan-2020

Concerning VGT composite methods, several selection rules have been added in addition to the "maximum NDVI" one. These rules are the same than the one applied on PROBA-V datasets and are driven by configuration parameters.

The cloud and snow/ice flags defined at SYNERGY L2 processing are now duplicated in VGT flags for consistency reason. In addition, several issues detected related to the cloud flag, such as blockiness detected at the edge of the cloud mask have been corrected.

2.51 / 1.23

SY2: 06.17
SY2_VGP: 06.16
SY2_VGS: 06.07

Corrections of

  • Synergy wrong generation of time.nc values

  • Typo in some SYN VGT-P /VGT-S attributes

  • SY_2_SYN products missing SLSTR oblique scans

  • SY_2_VGK products with wrong footprint

06-Jun-2019

2.44 / 1.16

SY2: 06.16
SY2_VGP: 06.16

New IDEPIX cloud flags now used in VGT-P/K products

16-Jan-2019
21-Jan-2019

Correction of AG variable (T550) over ocean set to zero instead of fill value

Correction of NDVI set to 0 instead of _FillValue over ocean in VG products

Improving VGS composite method

2.40

SY2: 06.15
SY2_VGS: 06.06

First version

 

Sentinel-3A Synergy Processing Baseline Timeline

Sentinel-3B Synergy Processing Baseline Timeline

L1 Algorithms

SYN Level-1 processing aims to retrieve OLCI and SLSTR radiances and brightness temperature in their acquisition geometry. The same computation is done for their associated annotations. This processing also aims to compute the correspondence grids between the OLCI reference channel and all other OLCI and SLSTR channels.

SYN Level-1 processing is divided into the following parts:

  1. A pre-processing step which reads and adapts the Level-1B data for subsequent processing.

  2. The main processing steps:

    1. Inter-instrument mis-registration estimation aims to correspond, for each OLCI camera module, the reference SLSTR visible reference channel with the reference OLCI channel.

    2. Computation of correspondence between the OLCI reference channel and all other OLCI and SLSTR channels, based on the mis-registration estimated at the previous step and on intra-instrument mis-registration estimation stored in a characterisation ADF.

    3. Projection of all OLCI/SLSTR and subsampled parameters on the same SYN reference grid (i.e. the OLCI acquisition grid)

  3. Post-processing to define and write the Level-1 product, SY_1_MISR

The pre-processing focuses on reading and adaptation of OLCI and SLSTR Level-1B product and is divided into two sub-steps.

  1. An extended land/sea mask is applied to focus only on land surface a)  (Note that this distinction ocean/land can be taken into account to now include ocean pixels in global aerosol processing)

  2. The OLCI retrieval section retrieves the full images of the five OLCI camera modules in their acquisition geometry. This processing involves radiometric measurement and associated annotations. The TOA radiance from OLCI reference channels is converted to TOA reflectance. A first selection Ground Control Point (GCP) is also included. This Ground Control Point grid can be computed from an external database or regularly created.

  3. A similar process, except GCP selection, is done for each channel of the SLSTR nadir and oblique views during SLSTR retrieval. The conversion is done for TOA radiance from SLSTR reference channels.

The main processing section is divided into four sections, gathering inter and intra-measurement spatial mis-registration.

  1. The extraction of OLCI/SLSTR imagette pairs extracts imagettes around each GCP, one in the OLCI reference channel (called the context imagette) and one larger imagette in the SLSTR reference channel (called the search imagette). The search imagette is projected onto the OLCI geometry. This section also includes rejection tests on selected Ground Control Points.

  2. The sub-pixel shift estimation at GCP sub-step computes a correlation surface between the two imagettes that is shifted around the GCP according to shift vectors and finds its sub-pixel maximum. This sub-step also includes final rejection tests on selected GCP. These two last sub-steps are performed for each selected GCP.

  3. The deformation model estimation sub-step computes the parameters of a piece-wise deformation model giving the mis-registration of the reference SLSTR channel at each pixel of the OLCI selected channel.

  4. The correspondence between reference OLCI and other channels sub-step uses all previously computed parameters and mis-registration ADFs to compute the correspondence between the OLCI reference channel image and all other OLCI channels, and SLSTR nadir channels. The correspondence between the OLCI reference channel and the SLSTR oblique channel is computed by estimating the common swath between these two images and establishing a superimposition in the gridded Level-1B product. These results are used to project all OLCI/SLSTR parameters on the same grid, the SYN reference grid, i.e. the OLCI acquisition one.

The post-processing step gathers all de-registration information, inter and intra-instrument and writes this information to the Level-1 product.

L2 Algorithms

SYN AOD Processing

SYN Level 2 AOD processing starts, once the whole OLCI and SLSTR datasets have been projected on the OLCI reference grid, with the creation of a super-pixel dataset.

A super-pixel is defined on the OLCI image pixel on a 15x15 pixels basis. All parameters are checked regarding surface classification and cloud detection before being averaging. Then, on this super-pixel dataset, an aerosol retrieval module is performed to retrieved aerosol optical thickness at 550 nm.

This aerosol retrieval is close to the one applied on SYN L2 SDR products (i.e. based on a double optimization of the error model), but with important differences:

  • An ocean transfer radiative model has been included, taken into account SLSTR radiances and providing aerosol characteristics over the whole swath

  • The inversion process and associated Look-up-tables have been adapted to retrieve continuous aerosols components from 35 aerosols mixtures

  • Prior climatology of aerosol properties has been included and uncertainty estimate per retrieval is provided for each pixel

An additional post-processing filtering has been also added to detect missed cloudy pixels. It is based on a simple image processing considering all neighbors of a retrieved super-pixel, i.e. boxes of 9 (3x3) super-pixels. For an individual AOD retrieval to successfully pass this step of filtering the following criteria need to be met:

  • at least 3 neighboring super-pixels with successful retrievals are required

  • the sample corrected standard deviation of the valid super-pixels in the 3x3 box needs to be smaller than the minimum of 0.15 or 80% of the mean AOD value + 0.04

A filtering regarding Sun Zenith Angle is also performed to exclude all AOD values associated with a SZA higher than 78°

SYN VGT-P and VGT-S Processing

SYN Level-2 VGT processing aims to combine information from the OLCI and SLSTR instruments to provide comparable products to SPOT-VGT to allow continuity of data delivery to the existing user community of SPOT-VGT.

The products selected for continuity are the P, S1 and S10 products named SY_2_VGP, SY_2_VGS1 and SY_2_VGS10 , respectively.

The P projection is defined at top of atmosphere (TOA), while the S1 and S10 are composite products defined at top of canopy (TOC), i.e. after an atmospheric correction. S1 is defined over a daily time period and S10 over a 10-day (decade) period.

The starting point of the processing is the co-located TOA L2 product from the SYN branch. It is divided into four steps:

  1. Spectral band mapping to simulate the SPOT-VGT spectral bands (four bands) at top of atmosphere from the OLCI and SLSTR bands.

  2. Pixel flagging, indicating land, water, cloud cover, ice/snow are transferred from SYN L2 datasets.

  3. Projection from the ortho-geolocation grid to the SPOT VGT Plate-Carrée grid at 1 km resolution. At the end of this sub-step, the SY_2_VGP product can be written.

  4. Atmospheric correction of the VGT-P product to obtain TOC reflectance and computation of the NDVI. This last step aims to produce the VGS1&10 products.

VGT S1 products 
give 
a
 single 
‘best’
 value
 for
 TOC
 reflectance 
at
 the
 four
 VGT channels (B0, B2, B3, MIR),
 based
 on
 those giving the
 maximum 
of 
Normalised Difference Vegetation Index (NDVI). 
The NDVI
 is
 a
 simple
 numerical
 indicator
 that
 is
 widely
 used
 to
 give
 an
 indicator
 of
 vegetation
 amount, 
derived
 from 
the 
 NIR 
and
 RED VGT 
channels:

The 
method
 and
 principals  
for 
generation 
of 
the 
VGT S10 
product 
are 
identical 
to 
the 
VGT S1, 
except 
that 
S10 
is formed
 using
 the
 maximum
 value
 composite
 over
 a
 10‐day
 period.


S1
 and
 S10
 can
 either
 be
 generated
 in
 parallel,
 as
 is
 the
 case
 in
 CTIV, (Centre de Traitement des Images VEGETATION)
 or
 alternatively
 S10
 can
 be
 generated
 from
 S1.
 The
 advantage 
of 
the 
former 
approach 
is 
that 
the 
same
 compositing
 program
 can 
be 
used 
to 
generate 
S1 and
 S10.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.