Wildfires. Credit: lassedesignen

Release of new Sentinel-3 SLSTR processing baseline in January


Wildfires. Credit: lassedesignen
Wildfires. Credit: lassedesignen

A new processing baseline (PB), v2.61, for the Sentinel-3 SLSTR products is planned to be released on 15 January 2020.

Last Updated

24 March 2022

Published on

19 December 2019

This update is primarily driven by the forthcoming release of the SLSTR L2 Fire Radiative Power (FRP) product (further information to come). In addition, several improvements have also been implemented in the SLSTR L2 SST PB.

This release will increase the baseline collection of SLSTR L1B product from version 003 to 004. The baseline collection of the SLSTR L2 SST products remains the same (003).

Release of new Sentinel-3 SLSTR Processing Baseline in January
Figure 1: Brightness Temperature (BT) differences between S7 and S8 channel are shown in the left panel and F1 BT measurements are shown in the right panel.

Summary of changes to SLSTR L1

  • Revised ortho-regridding of SLSTR L1 radiances and BTs (selecting nearest neighbour pixel instead of the first one)
  • Revised geo-referencing of SLSTR F1 fire channel BT (introducing dedicated “f’ grid)
  • Improved geometric calibration for oblique view
  • Improved S7 BT calibration
  • Temporal interpolation of ECMWF meteorological fields
  • Improved quality checks during decontamination and black body crossover tests
  • Removal of the c stripe (time domain integrated) from product and update of several netCDF variable attributes

Summary of changes to SLSTR L2 SST

  • Improved WST quality level

Further details of changes and impacts

The main changes to the SLSTR L1 processor are related to revised ortho-regridding of radiance and brightness temperature (BT) measurements, to improve the quality of radiometric imagery and geo-referencing of the SLSTR F1 fire channel. This is important for improving the spatial co-registration between the S7 and F1 channels. Previously, the first available pixel was projected to the corresponding image location, in the new version the closest instrument pixel to the centre of the image location is projected on the image grid. This modification improves SLSTR L1 radiometric imagery in general, but especially over the elevated terrain. It also improves the co-registration of the nadir and oblique view.

As part of the evolution, a new F1 geolocation grid ('f') has been introduced due to the specific detector geometry of F1 channel, which helps to improve the co-registration between F1 and S7 channels. The new geometric calibration has been implemented to resolve existing oblique view geolocation offsets and to address general updates related to the new ortho-regridding. The maximum valid brightness temperature of the S7 channel has been increased beyond 305K to support the SLSTR FRP algorithm. All S7 BTs above 305K are, however, flagged as 'S7 invalid radiances' as the values are beyond the nominal limits of the channel and, therefore, should be used with extreme caution depending on the user application and associated quality requirements. These changes are a prerequisite to allow the forthcoming use and deployment of the SLSTR L2 FRP processor.

There are also some smaller improvements:

  • Temporal interpolation of the meteorological fields is implemented to improve calculation of probabilistic cloud mask over land, but without impact on Bayesian cloud mask over sea.
  • New quality checks are introduced to improve flagging of data during decontamination and black body crossover test.
  • The time domain integrated ('c') stripe has been removed from the product.
  • Several netCDF variable attributes have been corrected.

The impact of the above changes on the SLSTR L2 SST product is minor. The main change in SLSTR L2 SST processing is an update to the algorithm that assigns the L2P quality level. In particular, the inclusion of a check of the pointing flag to identify (as QL = 3) poor quality data during satellite manoeuvres.

Results showing the impact of the processing changes on the SLSTR L2 SST product can be seen in Figures 2 and 3. Here, one day of daytime differences between SLSTR and OSTIA for QL=5 are shown for the current processing baseline (Figure 2) and for the new processing baseline (Figure 3). Overall, the changes are negligible, with small differences due to the ortho-regridding affecting the Bayesian cloud mask. However, the benefit of the additional test on the pointing flag is clear — degraded data during an out-of-plane manoeuvre visible in the current processing baseline are no longer present with the new processor.

 Daytime SST differences between SLSTR-A and OSTIA for 11.12.2019 for current processing baseline.
Figure 2: Daytime SST differences between SLSTR-A and OSTIA for 11.12.2019 for current processing baseline.
 As Figure 2 but for the new processing baseline.
Figure 3: As Figure 2 but for the new processing baseline.

Additional details about the corrected issues is be available in Level 1 and Level 2 Product Notices.

For more information, contact our User Service Helpdesk.