Day 4

Detailed paper information

Back to list

Paper title Sen2Cor Version 3.0 Processor Applied to Landsat-8 Data: Implementation and Preliminary Results
  1. Francesco Cristiano Pignatale Telespazio Germany GmbH Speaker
  2. Uwe Müller-Wilm
  3. Jerome Louis Telespazio France
  4. Vincent Debaecker Telespazio France
  5. Bringfried Pflug German Aerospace Center
  6. Leonard Kohlhepp Telespazio Germany GmbH
  7. Bodo Werner Telespazio Germany GmbH
  8. Carine Quang CS GROUP France
  9. Enrico Giuseppe Cadau Serco Italia SpA
  10. Valentina Boccia ESA - European Space Agency
  11. Ferran Gascon ESA - ESRIN
  12. Rosario Quirino Iannone RHEA for ESA/ESRIN, Frascati (Roma) Italy
Form of presentation Poster
  • Open Earth Forum
    • C5.03 Open Source, data science and toolboxes in EO: Current status & evolution
Abstract text Sen2Cor (latest version 2.10) is the official ESA Sentinel-2 processor for the generation of the Level-2A Bottom-Of-Atmosphere reflectance products starting from Level-1C Top-Of-Atmosphere reflectance. In this work, we introduce Sen2Cor 3.0, an evolution of Sen2Cor 2.10 able to perform the processing of Landsat-8 Level-1 products in addition to Sentinel-2 Level-1C products.
In this study, we test the resulting capability of the Sen2Cor 3.0 algorithms (also updated to work in a Python 3 environment) such as the scene classification and the atmospheric correction, to process Landsat-8 Level-1 input data. This work is part of the Sen2Like framework that aims to support Landsat-8-9 observations and to prepare the basis for future processing of large set of data from other satellites and missions. Testing and measuring the capacity of Sen2Cor 3.0 to adapt to different input and reliably produce the expected results is, thus, crucial.
Sentinel-2 and Landsat-8 have seven overlapping spectral bands and their measurements are often complimentary used for studying and monitoring, for example, the status and variability of the Earth’s vegetation and land conditions. However, there are also important differences between these two sensors, such as the spectral-band response, spatial resolution, viewing geometries and calibrations. These differences and quantities are all reflected in their resulting L1 products. A dedicated handling process for those differences is, thus, needed. Moreover, contrary to Sentinel-2, Landsat-8 does not have the water-vapour band that is used by Sen2Cor to perform the atmospheric correction of Sentinel-2 products. Therefore, important information is missing and further implementation is required in order to retrieve the necessary data from external sources to prepare the scene for the Landsat-8 processing. Moreover, new set of Look-Up Tables had to be prepared.
In this work, we address the modifications applied to Sen2Cor and the uncertainty due to the Level 1 to Level 2 processing methodology. Further, we present a qualitative comparison between Sen2Cor 3.0 generated Sentinel-2 and Landsat-8 L2 products and Sen2Cor 2.10 generated Sentinel-2 L2A products. Finally, we list foreseen optimizations for future development.