1.1

Instrument characteristics

The Sentinel-4/UVN instrument is a hyperspectral spectrometer operating with designated spectral bands in the solar reflectance spectrum. The prime Sentinel-4/UVN parameters are listed in Table 1.

Table 1:

Sentinel-4/UVN instrument main design and performance parameters [2].

The requirement specifications of S4-UVN are amongst others set for level-0 and for level-1b data at End of Life (EOL) for the full signal dynamic range, for unpolarised spatially uniform scenes with a confidence level of one sigma. The requirement specifications have to be verified, partly on the ground and partly in orbit, during the commissioning phase (phase E1). ESA has implemented a number of requirements to be met before launch to assure a minimum of values to be met as apportionment of the total calibration. Some requirements, such as the instrument polarisation sensitivity, can on the other hand only be verified during an on-ground calibration phase, because the full set of required linearly polarized input light scenarios can only be offered to the instrument on the ground.

Figure 2 shows the S4 PFM instrument in its test configuration (at the Airbus OTN test facility prior to shipment to the RAL TVAC facility)

Fig. 2:

S4 PFM instrument before shipment to the RAL TVAC facility.

1.2

Instrument Characterization and Calibration

After completion of the assembly, alignment and environment test phases in the 2020-2021 timeframe, the S4-UVN PFM instrument is ready for the crucial phase of Characterization and Calibration (C&C phase) at the RAL (UK) facility since early summer 2022. A full calibration is being performed on ground under flight representative thermal-vacuum conditions (i.e. thermal-vacuum environment with flight representative conditions for pressure, temperatures for detectors and optical bench).

Fig 3 shows the S4-UVN PFM integrated into the TVAC facility in RAL.

Fig. 3 :

S4 PFM instrument inside RAL TVAC chamber.

The aim of the calibration and on-ground characterization activities for the S4-UVN instrument performed during the combined characterization and calibration campaign is twofold: the verification of the performance related customer requirements at L0 and L1b and the provision of measurements needed to derive the calibration keydata. The latter ones are required by the science processor for the L0 to L1b processing [1].

The strategy and rationale to verify the S4-UVN requirements are based on the on ground measurements list specified in the S4-UVN requirement specification. This is a comprehensive list of measurements established by ESA based on past experience from similar missions like SCIAMACHY, OMI, GOME & GOME 2. This list is provided in Table 2. The on-ground measurements list is completed with requirements related to specific characterizations of the instrument such as straylight, Instrument Spectral Response Function (ISRF), absolute radiometry traceability, …

Table 2:

list of the instrument parameters that shall be calibrated on-ground with sufficient accuracy to meet all requirements on the accuracy of the level 1b data products.

The C&C campaign logic is such that in advance to the calibration phase, directly after the TB/TV phase, a series of Performance Verification measurements (occurring during the so-called debugging phase) are being performed on one hand as instrument and OGSE health check and on the other hand to ensure that the anticipated L0 instrument performances (for instance polarization sensitivity) are as expected. From this point in time the instrument configuration is frozen (i.e. no possible change in configuration anymore unless all acquired Calibration Key Data will have to be measured again) and the start of the calibration with the generation of the corresponding instrument key data is kicked off.

This paper describes the performance verification and calibration measurement program, its rationale and presents preliminary results from the S4-UVN PFM C&C campaign.

2.1

Radiometric calibration

The radiometric calibration includes the calibration of the radiometric goniometries for Earth radiance and sun irradiance measurements prior to launch with an accuracy that is compliant with the in-flight absolute radiometric accuracy requirements of the Earth spectral radiance, on absolute irradiance and absolute Earth reflectance, and the relative viewing angle dependencies thereof. The viewing properties of the instrument (i.e. lines of sight, fields of view, intra-channel co-alignment and inter-channel co-alignment) will be included, as well as the illumination angles of the diffusers, in the calibration prior to launch to accuracies that are compliant to the in-flight accuracy requirements of these viewing properties. The absolute radiometric radiance and irradiance are traceable to primary radiometric standards used during on-ground calibration.

The radiometric requirements of S4-UVN are per nature difficult to achieve. For instance the instrument response in Earth observation mode shall be calibrated on the ground to an accuracy better than 1.0% and the instrument response in sun calibration mode shall be calibrated on-ground to an accuracy better than 0.8% (all values apply on a one sigma confidence level).

Based on experience from previous similar instrument calibrations, in order to meet such demanding requirements it is required to combine the results of the measurements obtained with different illumination sources (Optical Ground Support Equipment - OGSE) in order to increase the accuracy of the calibration keydata and in order to minimize errors originating from peculiarities associated with each illumination source, such as light flux level, distance to instrument, homogeneity of illumination, generated straylight, representativeness of in-orbit illumination, temporal stability, polarization, absolute radiometric calibration status. Each of these parameters contributes to the overall error for that specific illumination source, and by combining the results from different sources the final errors can be reduced considerably. In order to achieve this target, it is necessary to properly commission the different light sources and perform measurements where the radiometric dependencies on the parameter that is varied are investigated. For this purpose the following set of OGSE is used in combination for the radiometric calibration measurements (see Table 3 and 4). In accordance with this approach the absolute radiometric calibrations and their angular dependencies are calibrated separately.

Table 3:

OGSE used during Radiometric measurements.

Table 4:

Radiometric measurements and their objectives.

The used OGSE are designed to optimally calibrate both the absolute Earth radiance and sun irradiance, as well as calibrate the Earth reflectance via the instrument Bi-directional Scattering Distribution Function (BSDF). The requirements on the instrument BSDF (reflectance) are most stringent, because the earth reflectance is used for most of the L1b-L2 data processing algorithms, and a number of error contributions in radiance and irradiance absolute radiometric calibrations cancel in the instrument BSDF / earth reflectance.

Examples of the implemented OGSE measurements configurations are presented in Figure 4.

Fig. 4:

Top Left Sun Beam Simulator (SBS) [pink] + OGSE diffuser [yellow] on the Earth port (configuration 2R). Top Right: FEL lamp [yellow/blue] in front of the sun port configuration layout (configuration 1I). Bottom: Integrating Sphere (ISP) [green] on the Earth port configuration (configuration 3R).

Fig. 5 presents the preliminary results for the ISP measurements during the debugging phase for Abs_Rad_2 (config 3R) and Refl_sphere (config. 3I/3R).

Fig. 5:

Top Preliminary Abs_Rad_2 from debugging phase (ISP) [gray] + MIN [blue] + MAX [red] predictions on the Earth port (config 3R) Left: UVVIS channel / Right: NIR channel. Bottom: Preliminary Refl_sphere from debugging phase (ISP) [black] (config 3I/3R). Left: Nominal on-board-diffuser / Right: Redundant on-board diffuser. All preliminary measurements from the debugging phase lie within the predictions accounting for the measurement precision.

Straylight:

Spectral stray light, spatial stray light and combinations thereof are being considered. The spectral stray light calibration extends beyond the optical channel boundaries. For the UV-VIS-NIR source wavelengths in the range from 300 to 1100 nm are considered due to the sensitivity of the detectors in this wavelength range. The accuracy of the pre-launch stray light calibration shall be compliant to the pre-launch accuracy requirements of absolute radiance, absolute irradiance and absolute Earth reflectance, as well as to the absolute radiance, irradiance and reflectance radiometric accuracy requirements for the level 1b data in-orbit. Stray light contributions from inside the instrument field of view and from outside the field of view are considered. For internal stray light the effects from near-field stray light (the area where the spectral and spatial response functions go over into stray light) and far-field stray light are considered. Far-field stray light can be separated into uniform stray light (over spectral and/or spatial pixels) and ghost-type stray light (localised at certain groups of spectral or spatial pixels). For the Sentinel-4/UVN instrument stray light is particularly challenging for the wavelength regions below 320 nm and around 760 nm, due to the low Earth radiance signals originating from ozone and O₂ A-band, respectively, and for unclouded ground scenes with low albedos (<5%), which may collect significant fractions of spectral spatial stray light from clouded ground scenes observed at the same time on the CCD detectors at other wavelengths and viewing positions. On the other hand, the above areas are also the ones that are most interesting from a data usage perspective for tropospheric ozone and NO₂ and cloud and aerosol information from the O₂ A-band spectral region, underlining the importance of proper and sufficient stray light reduction in the level-1b data.

Different measurement approaches and OGSE use for on-ground calibration and keydata generation for the uniform and ghost straylight correction are considered. The rationale of these various measurements and their implementation is explained below:

Additional measurements types are also envisaged for characterization/verification of straylight following the knowledge obtained during the projects evolution.

5-Reflection (5R) ghost measurements: The 5R ghost in NIR was identified via optical simulation and its presence was confirmed in the frame of the lower level Telescope-Spectrometer-Assembly (TSA) optical test. Three dedicated straylight measurements are planned to fully characterize it:

Out of field straylight characterization measurements: although the out of field straylight is not planned to be corrected the following measurements for characterization are planned.

Near slit straylight feature measurements: a characterization of the near slit straylight features is foreseen to measure the illumination dependence of the features observed in the vicinity of the instrument slit (See fig 6)

Fig. 6:

Short range slit straylight observed during the lower level TSA measurements Top UVVIS full slit zoom/rescaled. Bottom: NIR slit zoomed/rescaled

In addition to these measurements that are mandatory based on the current status of analysis and instrument testing, there are a few measurements additionally planned for straylight characterisation and end to end performance verification:

Detection chain:

A complete on-ground calibration of detector properties and electronic properties will be performed. The parameters that are of relevance for the Sentinel-4/UVN instrument are listed in Table 2. Detector exposure smear will be calibrated for non-homogeneous illumination of the spectrometer’s entrance slit.

Polarization

The instrument’s polarization sensitivity is required to be less than 1% at level-0, also because the 0-1b data processing doesn’t foresee radiometric correction steps for polarization. In the instrument design this is realized using a weak polarization scrambler and transmission optics, in combination with small incidence angles for the geostationary mission. The focus of the dedicated on-ground polarization measurements will be to show compliance to this requirement and characterize remaining polarization spectral features.

Spectral features originating in the instrument, e.g. from the on-board diffusers as observed in predecessors missions, as well as the above mentioned remaining polarization spectral features, may hamper the analysis of atmospheric trace gases in the L1b-L2 data processing. These features have as much as possible to be eliminated by the instrument design since correction of the polarization spectral features based on an unknown polarization state of the Earth light is not possible with the Sentinel-4 UVN design. The relative spectral radiometric accuracy (peak-to-peak) are considering small spectral window widths of a couple of nanometers, which for compliance of the requirement incorporate these spectral features next to other relevant errors for the instrument response in sun irradiance and Earth radiance observation modes, as well as for Earth reflectance (ratio radiance over irradiance). As example, in the UVVIS between 315 and 500 nm, the maximum relative radiometric spectral accuracy error over a spectral window width of 3 nm is required to be smaller than 0.05%.

In order to characterize the PFM polarization (sensitivity and polarization spectral features), two types of measurements are implemented with the Polarization (POL) OGSE:

Fig. 7:

Polarization sensitivity in UVVIS and NIR bands (preliminary results). The achieved polarization sensitivity is well below the specified 1%. Note: the feature in NIR around 765nm is an artefact. Assumption is that it is caused by imperfection in compensation of signal variation.

2.2

Spectral calibration

On-ground Spectral calibration will be performed with dedicated spectral calibration sources and in flight with the knowledge of the solar Fraunhofer lines and the atmospheric absorption lines. The spectral calibration will also be checked for its dependence to non-homogeneous illumination of the instrument’s entrance slit (in east-west scan direction): inhomogeneously illuminating the instrument entrance slits will introduce a shape change of the spectral response function, which may in turn introduce an apparent spectral shift of the observed spectral absorption lines in the spectra. In addition, the spectral response function (spectral slit function) of the instrument will be calibrated on the ground. The spectral response functions are used to establish the spectral resolution and for the spectral calibration algorithms in the level-0 to level-1b data processing, as well as for the level-1b to level-2 data retrievals. Most of the level-1b to level-2 data retrievals convolve high-resolution absorption cross section spectra of the atmospheric constituents with the spectral response functions, that have been calibrated accurately prior to launch. The convolved spectra are then compared and fitted to the measured spectra in order to retrieve the atmospheric constituent concentrations. The instrument spectral response functions (ISRF) are thus essential for the accuracy of the level-1b to level-2 data processing and the accuracy of the level-2 data products. For accurate calibration of the spectral response functions a tuneable monochromatic light sources such as a wavelength-tuneable laser can be used.

The spectral calibration comprises the following measurements:

Fig 8 present a first overview of the ISRF measurements for homogeneous scene and for the radiance port.

Fig. 8:

ISRF in UVVIS1 band (preliminary results before post-processing)

2.3

Geometric calibration

The orientation of the instrument’s 3-axis reference frame with respect to the spacecraft’s 3-axis reference frame will be calibrated prior to launch, as this is required for the latitude/longitude geolocation of the observed scenes. This will allow calibrating the relevant instrument geometrical parameters (lines of sight, pixel fields of view, angles on the onboard diffusers) with respect to the instrument alignment cube. The translation to the spacecraft reference frame will then be made via the above relative orientation calibration between the spacecraft and instrument reference frames.

The geometric calibration and characterization includes the derivation of the system’s PSF. Furthermore, besides the intraband co-registration characterization for performance verification purposes (within UVVIS band and within NIR band) the optimized setting of the S4-UVN co-registration offset compensator will be determined and optimized for the UVVIS-NIR interband coregistration. The co-registration offset compensator is a thermally driven method to shift the image on the NIR detector in the N/S direction within one N/S pixel range in order to align its offset with respect to the one of the image on the UVVIS detector. This is one of the first measurements during the on ground campaign in order to perform all other calibration and characterization measurements with this optimized setting.

Additionally, the relative Pixel Line Of Sight (PLOS) map will be measured. This is the angular distance both, the E/W and N/S center positions between two arbitrary detector pixels. These center position(s) (could be more than one, depending on the polarization scrambler performance) will additionally determine the relative line of sight (LOS) calibrations per detector pixel, as well as the co-registration between the various detector pixels, because the measurements will be performed with white light illumination in parallel beams. The width will also determine the spatial resolution and the width and the shape combined will be essential inputs to the integrated energy assessment. The measurement is similar to the star measurements. One or more point sources are scanned across the instrument’s slit in E/W direction. This scanning is performed on ground in a step and stare fashion. A similar sequence is then performed with the point sources centered in the slit and scanned in N/S direction with sub-pixel steps. This measurement sequence will be performed with an unpolarized OGSE white light beam. Even though the polarization scrambler is expected to result in different PSF measurement parameters depending on the input polarization direction, it is assumed that this will not play an important role for the level 0 to 1b and 1-2 data processing or for the performance of the instrument.