1.1

Sentinel-2 mission and MSI instrument

The Copernicus program is a joint initiative of the European Commission (EC) and the European Space Agency (ESA), designed to establish a European capacity for the provision and use of operational monitoring information for environment and security applications. ESA’s role in Copernicus is to provide the definition and the development of the space and ground-related system elements. As part of the Sentinels family, the Sentinel-2 mission provides continuity to services relying on multi-spectral high-resolution optical observations over global terrestrial surfaces. The key mission objectives for Sentinel-2 are: (1) To provide systematic global acquisitions of high-resolution multi-spectral imagery with a high revisit frequency, (2) to provide enhanced continuity of multi-spectral imagery provided by the SPOT series of satellites, and (3) to provide observations for the next generation of operational products such as land-cover maps, land change detection maps, and geophysical variables. Consequently, Sentinel-2 directly contributes to the Land Monitoring, Emergency Response, and Security services. The corresponding user requirements have driven the design towards a dependable multi-spectral Earth-observation system featuring the Multispectral Instrument (MSI) developed by Airbus Defence and Space. It is composed of 13 spectral bands spanning from the visible and the near infrared to the short wave infrared. The spatial resolution varies from 10-meter to 60-meter depending on the spectral band with a 290 km field of view. This unique combination of high spatial resolution, wide field of view and large spectral coverage represents a major step forward compared to previous multi-spectral missions. The mission has already a constellation of two satellites (Sentinel-2A and Sentinel-2B) launched in June 2015 and March 2017, each having a 7.25-year lifetime with consumable sized for 12 years. The two satellites are maintained in the same Sun synchronous orbit with a phase delay of 180° providing a revisit time of five days at the equator. The Sentinel-2 mission is largely detailed in [1]. To cover the 20-year overall mission duration, two additional satellites Sentinel-2C and Sentinel-2D were funded by the European Commission, the first one is expected to be launched at the end of 2023 and the second one will be stored until the replacement of S-2B.

1.2

MSI design description

The MSI instrument is a push-broom multispectral imager. It provides imagery in 13 spectral bands with spatial ranging from 10 to 60 meters. The instrument features an optical telescope providing a wide field of view (FOV) to achieve the required swath width of 290km: 20.0° across-track by 3.5° along-track. An oblong pupil equivalent to 150mm diameter is implemented to achieve a compact design and optimized optical performance.

The instrument is required to operate over a wide spectral range extending form the visible and the near infrared (VNIR, 400-1100nm) to the short wave infrared (SWIR, 1100-2500nm). A concept based on two separate focal planes is implemented. Separated focal planes are used with a dichroic filter which does the splitting between the two spectral domains:

The instrument architecture is composed of:

Figure 1

– Sentinel-2 MSI optical architecture overview

The optical design of the instrument is based on a TMA concept that corrects spherical aberration, coma and astigmatism – it features mirror dimensions up to 600mm. In order to ensure the homogeneity of the filter spectral response, a telecentric configuration is designed. The optical beam is split between VNIR bands (in reflection) and SWIR bands (in transmission) by a splitter close to the focal planes.

The 3 mirrors of the TMA telescope are made of Silicon Carbide (SiC) and supported by a SiC main baseplate structure. This structure is connected to the platform top panel by 3 bipods performing an isostatic mounting for preventing the instrument from being distorted by platform top panel deformations.

The CSM provides for the required observation, calibration and protection functions. It allows the closing of the aperture to protect the instrument from direct Sun illumination. The CSM integrates a diffuser for absolute calibration and allows opening the aperture for nominal Earth imaging. The absolute calibration is performed by means of a Sun diffuser inserted in front of the primary mirror. The focal plane is fully illuminated during this phase.

The primary and secondary structure is a carbon fibered stable structure carrying the telescope assembly, the CSM and the thermal housing of the instrument.

Figure 2

– Sentinel-2 MSI general overview

The main characteristics of the 13 bands are described in the following Tabel 1.

Table 1.

– Sentinel-2 MSI spectral bands description

As any Earth observation instrument, the straylight is one of the major preoccupations. The main optical design drivers on the MSI instrument led to a disadvantageous design regarding the straylight minimization since: (1) there is no intermediate focal plane to include a field stop; (2) the diameter of the entrance pupil, localized behind the M1 mirror, is significantly smaller than the size of this latter mirror – which is unfavorable for the straylight out-of-FOV. That is why the characterization of the straylight due to the Earth albedo becomes mandatory.

2.1

Principle of the test

The principle of this test is to simulate the illumination of the Earth albedo seen from the MSI instrument in orbit. In orbit at 800km altitude, it leads to a field cone of ±63° as shown in the following Figure 3.

Figure 3

- The Earth is seen from the telescope under a field of view of ± 63°

As it is difficult to simulate in one step this configuration, a smaller source shall be moved at several positions around the entrance of the MSI instrument in order to map the whole Earth field cone. Images are recorded for each position of the source allowing us to assess by summation the straylight due to the Earth albedo.

A large light trap is placed in front of the MSI instrument covering its entire field of view (FOV) in order to neutralize the light seen by all pixels inside the FOV. The principle of the light trap is schematized in the Figure 4 below.

Figure 4

- Principle of the light trap

The MSI instrument is installed on a trolley and can be oriented around Z_MSI axis, in such a way that the light source can be placed with the purpose of mapping the Earth albedo. In order to optimize the test duration, a robotized arm is used to precisely and rapidly place the source in every defined positions described in the next section Mapping of the Earth albedo.

Figure 5

– Overview of the straylight setup

2.2

Mapping of the Earth albedo

The positions of the source are calculated in the MSI reference frame. The source plane shall be orthogonal to the direction of the FOV that is why it is positioned on a virtual sphere around the entrance of the MSI instrument at a distance of 3 meters. The relative positions of the source are chosen in order to minimize the overlapping but also to avoid any gap between them.

An exclusion area corresponding to the FOV of MSI instrument with some margins is defined so that no mechanical part of the source interfere with this area. The purpose is to be sure that all the pixels of the instrument can see completely the light trap without mechanical interferences.

The following Figure 1Figure 6 presents the 98 positions defined to cover the full ±63° of Earth field cone during this test.

Figure 6

- Positions of the diffusing source out of the field of view

For the sake of avoiding the interference risks between the FOV of the MSI instrument and the robotized arm, 6 MSI orientations of the trolley (segments of 60°) are defined each including 12 or 14 source positions.

2.3

Characteristics of the source

The following characteristics have been considered for the choice of the source:

A halogen source is consequently a good candidate as it covers all the spectral range but its main drawback is the bad modulation (thermal source and high frequency) without chopper which is not implementable because of the source size. Based on measurements, it appears that only low power halogen lamps have a good modulation: for lamps of 20W, around 25% of the averaged signal is modulated at 100Hz. Therefore, the choice of a matrix of several lamps with a screen diffuser is the best solution.

This modulation will help avoiding the effect of dark signal drift, help reducing the measurement noise and also help avoiding to measure the contribution of parasitic light in the clean room.

The main characteristics of the diffusing source are described the Figure 7 below.

Figure 7

- Characteristics of the diffusing source

2.4

Processing and normalization

The straylight processing is based on the Fourier transform spectral analysis of the temporal measurement at 100Hz for which the FFT peak magnitude at this frequency is directly proportional to the magnitude of the modulated signal (ratio 4). The straylight detection sensitivity depends on:

At the frequency of 100Hz (10ms), the signal is not detectable for the 60-meter bands because of their sampling period (9,3ms) which does not respect the Shannon criteria. These bands are thus not considered for the straylight analysis. Also, as the test is performed at ambient condition the SWIR bands B10 and B12 are not measurable due to high thermal noise.

The measurement noise depends especially on the number of pixels used for averaging and on the acquisition duration. It will be then the lowest for the 10-meter bands. For this test, each pixel line of detector is divided in 10 segments. The number of pixels used for averaging is typically 250 pixels for 10-meter bands.

The Figure 8 above shows:

Figure 8

- Example of FFT processing with straylight

At the end of the processing, the FFT peak frequency is checked if it is included in the narrow range 100Hz ± 0,1Hz otherwise the FFT peak is not considered as straylight.

In order to quantify the straylight, a normalization is applied on the values given by the Fourier transform spectral analysis. The normalization coefficients are obtained simply by measuring the FFT peak magnitude for each band of each detector when the source illuminates directly the FOV of the MSI instrument. This characterization must be performed with reduced integration time in order to avoid signal saturation. The convention used for the quantification is the part per million (ppm) which are well adapted for the low expected signals.

The Figure 9 below presents a synthesis of straylight processing after normalization for one position.

Figure 9

- Example of VNIR straylight processing after normalization for one position

At the end once all positions are processed, a summation is done giving the overall straylight due to the Earth albedo for all the analyzed positions.

2.5

Accuracy of the test

With such normalization, it is easy to estimate the straylight detection sensitivity. A simple dark acquisition will give the threshold under which straylight is not detectable. The following Figure 10 and Figure 11 show that the thresholds are typically 30ppm for VNIR bands and 40ppm for B11.

Figure 10

- Straylight processing with dark in VNIR

Figure 11

- Straylight processing with dark in SWIR

2.6

Light trap efficiency

For the characterization of light trap efficiency, a first radiance measurement is performed with the diffusing source illuminating directly in direction of the light trap. The Figure 12 below shows the acquired spectral radiance.

Figure 12

- Radiance of the light trap (with the diffusing source in direction of the light trap)

Figure 13

- Picture of the robotized arm and the light trap

Finally, the radiance of the light trap is compared to the radiance of the diffusing source (refer to the graph on Figure 7). The result given in Figure 14 shows a ratio around 0,004% (40ppm) over the spectral range, which is equivalent to the measurement accuracy, meaning that the contribution of the light trap is negligible.

Figure 14

- Light Trap efficiency: radiance of the light trap / radiance of the source

Moreover, instrument acquisitions have been performed with the diffusing source illuminating directly in direction of the light trap. This configuration is a worst case because, during the straylight tests, the diffusing source is illuminating towards the instrument. The acquisitions have been done with 2 positions of the instruments (X_MSI axis vertical and X_MSI axis horizontal) in order to detect possible edge effects in the light trap. The results of the light trap efficiency are given by the following Figure 15 and Figure 16. For both MSI orientations, the maximum detected signal is around 30ppm (40ppm for B11) which is still comparable to the measurement accuracy.

Figure 15

- Acquisition with MSI instrument vertical in front of the light trap and with the diffusing source in direction of the light trap

Figure 16

- Acquisition with MSI instrument horizontal in front of the light trap with and the diffusing source in direction of the light trap