Abstract
From spaceborne remote sensing platforms, Deep Convective Clouds (DCC) appear as bright and spectrally stable objects. Thus, they are ideal targets for the radiometric calibration monitoring of optical sensors and hence used for this purpose for more than two decades. A method based on DCC reflectances is applied to the Earth observation sensors Sentinel-3 OLCI (Ocean and Land Colour Instrument) A and B, Sentinel-2 MSI (MultiSpectral Instrument) A and B, Landsat-8 OLI (Operational Land Imager) and Landsat-9 OLI-2 for intercomparison of their relative radiometric uncertainty.
Originally developed at ACRI-ST for the long-term post-tandem phase analysis of the cross-calibration of OLCI-A/B, the DCC method has later been adapted to monitor the relative uncertainty of the Sentinel-2 MSI-A/B payloads. The method has been used operationally to follow the radiometry of MSI sensors since January 2022 within the Optical Mission Performance Cluster (OPT-MPC) activities, part of the Copernicus program and managed by the European Space Agency (ESA). Then, as a comparison purpose, the method has also been adapted to Landsat-8 OLI and Landsat-9 OLI-2.
The data selection is performed over the inter-tropical region, where convection movements provide a constant formation of DCCs. The DCC pixels selection relies on thresholds on NIR (Near InfraRed) and cirrus bands – B08, B10 and B5, B9 respectively for MSI A/B and OLI/OLI-2 – and on the SLSTR (Sea and Land Surface Temperature Radiometer) A/B thermal infrared (TIR) 10.85 μm S8 band for OLCI A/B. Indeed, OLCI sensors fly together with SLSTRs onboard Sentinel-3 and provide the opportunity to take advantage of the SLSTR TIR band for DCC detection. Selected data are converted to top-of-atmosphere reflectances, and the two-way gaseous transmission is handled to correct the absorption occurring above the clouds. For each sensor and spectral band, histograms of the corrected DCC reflectances are extracted and accumulated. After a month of data collection (hundreds to thousands of products depending on the mission), a skewed- Gaussian is fit over the accumulated histograms. The second inflexion point of the distribution is extracted and used as the reflectance indicator.
After a detailed description of the above-mentioned methodology, the time-evolution of the reflectance indicators and of the inter-calibration ratios between different couples of sensors will be presented for 2022. In addition to the already known VNIR (Visible and Near InfraRed) bias of 1-2 % for OLCI-A/B and the good agreement between MSI-A/B following the harmonization of early 2022 (ratios lower than 1 % for bands 1 to 8A), an emphasis will be put on the comparisons between Sentinel and Landsat missions on the VNIR/SWIR (Short Wavelength InfraRed) spectral range. Preliminary insights on the application of the method over VNIR – and eventually TIR (Thermal InfraRed) – SLSTR bands will be proposed as well as the potential evolutions of the method and perspectives on the topic of multi-mission radiometric intercomparisons.
Radiometric Calibration Intercomparison of Sentinel-2 MSI, Sentinel-3 OLCI and Landsat OLI Using Deep Convective Clouds
From spaceborne remote sensing platforms, Deep Convective Clouds (DCC) appear as bright and spectrally stable objects. Thus, they are ideal targets for the radiometric calibration monitoring of optical sensors and hence used for this purpose for more than two decades. A method based on DCC reflectances is applied to the Earth observation sensors Sentinel-3 OLCI (Ocean and Land Colour Instrument) A and B, Sentinel-2 MSI (MultiSpectral Instrument) A and B, Landsat-8 OLI (Operational Land Imager) and Landsat-9 OLI-2 for intercomparison of their relative radiometric uncertainty.
Originally developed at ACRI-ST for the long-term post-tandem phase analysis of the cross-calibration of OLCI-A/B, the DCC method has later been adapted to monitor the relative uncertainty of the Sentinel-2 MSI-A/B payloads. The method has been used operationally to follow the radiometry of MSI sensors since January 2022 within the Optical Mission Performance Cluster (OPT-MPC) activities, part of the Copernicus program and managed by the European Space Agency (ESA). Then, as a comparison purpose, the method has also been adapted to Landsat-8 OLI and Landsat-9 OLI-2.
The data selection is performed over the inter-tropical region, where convection movements provide a constant formation of DCCs. The DCC pixels selection relies on thresholds on NIR (Near InfraRed) and cirrus bands – B08, B10 and B5, B9 respectively for MSI A/B and OLI/OLI-2 – and on the SLSTR (Sea and Land Surface Temperature Radiometer) A/B thermal infrared (TIR) 10.85 μm S8 band for OLCI A/B. Indeed, OLCI sensors fly together with SLSTRs onboard Sentinel-3 and provide the opportunity to take advantage of the SLSTR TIR band for DCC detection. Selected data are converted to top-of-atmosphere reflectances, and the two-way gaseous transmission is handled to correct the absorption occurring above the clouds. For each sensor and spectral band, histograms of the corrected DCC reflectances are extracted and accumulated. After a month of data collection (hundreds to thousands of products depending on the mission), a skewed- Gaussian is fit over the accumulated histograms. The second inflexion point of the distribution is extracted and used as the reflectance indicator.
After a detailed description of the above-mentioned methodology, the time-evolution of the reflectance indicators and of the inter-calibration ratios between different couples of sensors will be presented for 2022. In addition to the already known VNIR (Visible and Near InfraRed) bias of 1-2 % for OLCI-A/B and the good agreement between MSI-A/B following the harmonization of early 2022 (ratios lower than 1 % for bands 1 to 8A), an emphasis will be put on the comparisons between Sentinel and Landsat missions on the VNIR/SWIR (Short Wavelength InfraRed) spectral range. Preliminary insights on the application of the method over VNIR – and eventually TIR (Thermal InfraRed) – SLSTR bands will be proposed as well as the potential evolutions of the method and perspectives on the topic of multi-mission radiometric intercomparisons.