During the hot cruise phase (heliocentric distance <1.34 au) the spacecraft -x axis (i.e., the side where the large High Gain Antenna is mounted; ) is maintained pointed to the Sun to use the High Gain Antenna as a sun shield, to protect the spacecraft body from direct high solar flux. The same configuration was used during LEGA. As a consequence of the fixed attitude, surface tracking (nadir pointing) was not possible. Instead, the camera operated with a slanted pointing, with the camera boresight sweeping the target at constant surface phase angle of approximately 90°. The observation window began when the JANUS boresight, approximately aligned with the S/C +z axis (i.e. the nadir direction, the one that will face the target during scientific observations; ), crossed the surface target. The main consequence of the fixed attitude is that the exposure times had to be limited to avoid smearing.

All instruments, including JANUS operated during the Moon and Earth flybys. The Mission Operation Control (MOC) defined two payload operation windows: CA ±1 h for the Moon flyby and from CA -4 h to CA +3 d for the Earth flyby. CA stands for Closest Approach, i.e. the shortest distance between the S/C and the target surface. JANUS was powered ON at CA -1 h for the Moon flyby and CA -2 h for the Earth flyby, respectively 1 and 1.3 h before the beginning of image acquisition. The nominal 12 h for thermal stabilization were not possible, thus, the S/C survival heaters (which can be operated when the instrument is switched off) were used to warm up the JANUS telescope to the -20 °C, allowing to acquire images under thermally stable conditions at the desired temperature. The S/C survival heaters were switched ON about 12 h before the beginning of the imaging sequences.

This approach was first implemented during LEGA (and initially tested during its full rehearsal one month before) and has the advantage of significantly reducing the power consumption. JANUS requires between 10 and 20 W when it is powered ON to reach and/or keep thermal stability, while the survival S/C heaters, which are operated with JANUS off, require significantly less power (about 3 W).

During the science phase, periods of observations will be interleaved with periods of very limited power availability, e.g. downlink windows, in which instruments might be requested to be switched off. In the cold space environment, JANUS drifts away from thermal stability very quickly as soon as the thermal control is switched off. Therefore, different strategies, such as the use of survival heaters, which reduce power consumption assuring image quality, are very important alternatives which shall be tested and optimised to assure the best image quality within the available power envelope.

The Radar for Icy Moon Exploration (RIME) instrument was granted to operate alone for a few minutes around the Moon closest approach to perform some tests of electro-magnetic cleanliness. All other payloads, including JANUS, had to be in quiet mode to ensure that no interference was introduced by the subsystems. JANUS operations began as soon as the “RIME quiet time” was concluded, 30 s before terminator crossing.

In addition to the aforementioned constrains posed by the S/C and the other instruments, JANUS had its own constraints that had to be accounted for during the sequence design.

Smearing. Because of the fixed spacecraft attitude requirement described in Sect. , surface tracking was not possible and the camera boresight swept the target at constant velocity during the flybys. The image exposure times had to be limited to avoid smearing. The smearing limit was set to a motion of the JANUS boresight of below 1/4 pixel. Typical exposure times are summarised in Table .

Filter change time. The time needed to change filter of one position is 0.5 s. This time interval plays a critical role in multifilter sequences for color cubes when the spacecraft is in motion, as in the LEGA case, and the JANUS FOV continuously shifts across the surface. The maximum number of filters that could be used was thus determined by the time that color sequence footprint moved of maximum 60 %–70 % of the JANUS FOV, assuring an overlap of 20 %–30 % between the footprints in all filters. An example is shown in Fig. , where the footprints of F2, F4, F11 and F12 overlap on about 20 %–30 % of the FOV.

During LEGA, JANUS acquired 204 image of the Moon and 173 images of the Earth. All images were in full frame and resolution (not binned/down-sampled) and commanded lossless, to assure maximum data quality. Figures  and show the JANUS footprints plotted over Moon and Earth maps, respectively.

All sequences were rehearsed on the JANUS Ground Reference Model (GRM), which is representative of the flight unit in both hardware and onboard software. The timing between images as well as the duration of the sequences was simulated. This was especially important for flyby operations, where precise time synchronization between the image sequences and the footprint on the surface was critical. No command errors occurred during LEGA, confirming that a full chain rehearsal is a prerequisite for any time critical cruise or science phase campaign.

The images were processed using the nominal JANUS calibration pipeline .

Each flyby was divided in segments within which different instrument settings were tested: single filter imaging, multi-filter imaging and acquisitions with different compression factor. The objectives of each segment are described below.

Moon flyby. Moon observations were acquired over an elongated stripe from the nightside to the dayside (Fig. ).

A total of 204 images were acquired over the target. While observing the nightside, we attempted to observe the Ganymede Laser Altimeter (GALA) spot on the Moon's surface to test the alignment of the two instruments boresighs (Seq. 1, Table ). For this purpose we acquired a number of images in the near-IR filter (F12, λc=1015 nm). This measurement was inherently difficult due to the marginal sensitivity of JANUS at the edge of its spectral range and to the non-ideal geometry and limited time both instruments could operate. In particular, due to the RIME quiet time constraint (Sect. , this observation could only be performed in the 30 s before the terminator crossing.

After the terminator crossing, dayside lunar surface observations were constrained by geometry, relative speed and radiometry, which together determined the usable filters and corresponding exposure times.

The first two sequences in daylight (Seq. 2 and 3, Table ) are characterized by high incidence angle and were acquired in a single filter (F1, λc=650 nm). The cadences were chosen to have overlap between consequent footprints, providing up to fourfold redundancy in surface coverage along the track. These images enable validation of the radiometric performance and post processing techniques, e.g., to correct image smearing and improve resolution.

The subsequent 3 sequences (Seq. 4, 5 and 6, Table ) were multi-filter acquisitions using a variable number of filters. Those sequences allow to extend performance and calibration verifications to the entire wavelength range. While flying over the Mare Fecunditatis all 13 filters were used (Fig. ). The relative uniformity of the Mare Fecunditatis allowed the comparison of images with non-overlapping footprints, enabling the validation of the radiometric performance to the entire wavelength range covered by JANUS. In addition, it enabled the assessment of straylight contributions from areas outside the JANUS FOV.

A compression test was carried on over a cratered area (Seq. 7, Table ): images with increasing compression were acquired to evaluate the information loss as function of the used compression. While flying over the region north of Rimae Gutemberg we acquired a sequence of 14 images in the panchromatic filter (F1, λc=650 nm) with increasing compression ratio from 1:1 to 33:1. The images were acquired as fast as possible, to ensure a significant overlap between the footprints (Fig. ).

Typically, a compression of 3.5:1 is foreseen to be used for well illuminated images. From a first and very preliminary analysis, the information loss is not as drastic as feared as function of the used compression. Small craters and other high-resolution features are preserved up to high compression rates with radiance values also preserved. This result, even though, preliminary, allows to set realistic expectations of the compression that can be used during the science phase, preserving the image quality but at the same time complying with a limited mission-wise data volume availability. Further tests under higher noise levels might be required to predict how compression may affect observations under low illumination in close flybys of the Galilean moons or in night-side observations of Jupiter.

Two final sequences on the Moon surface (Seq. 8 and 9, Table ), one multifilter and one with single filter were carried on while flying towards the Moon limb.

After limb crossing, a number of images with long exposure times (Seq. 10, Table ) were acquired to characterize the straylight originating from an out-of-field source at increasing angular distance and with decreasing apparent size.

Earth flyby. A similar approach was used to plan the Earth observations. Given the higher flyby altitude, image timing was less stringent constraint compared to the Moon flyby, enabling full multifilter observations on several footprints.

Earth observations were acquired over an elongated ground track extending from nightside to dayside with the sub-spacecraft point moving from North of Madagascar to central Pacific over a distance of 18 700 km (Fig. ). JUICE's trajectory over the Earth allowed to test the design of very different exposure times for the changing illumination conditions, test for smearing effects associated to short dwell times (where the dwell time is the time in which the boresight is moving of less than 1/4 of a pixel, thus considered at a fixed position), check the performance of the onboard image compression algorithm and obtain radiometric data to be compared with data from Earth satellites. describe the analyses of these observations. A total of 173 images were acquired, with an average ground sampling resolution of 140 m per pixel. Since the spacecraft velocity over the target varied from 5.0 to 6.8 kms-1 this resulted in dwell times of 18 to 44 ms. For the nightside images, short dwell times limited the possibilities to obtain well-exposed images in other filters than the panchromatic (F1). Thus, nightside images were acquired only with the panchromatic filter (F1) to maximize the visibility of surface features illuminated by the Moon. For the dayside images the high velocity flyby limited the capability to plan for repeated observations of the same areas.

The first two images targeted Earth's limb over Madagascar and the Indian Ocean. Their exposure times were chosen to accommodate to the dwell time of nearly 35 ms of that period and test the visibility of Madagascar's north coast over the dim illumination provided by the Moon. The night-time sequence included 24 images that targeted potential observations of lightning over tropical latitudes and the detection of city lights over Cambodia and Vietnam (Seq. 1, 2, 3 and 4, Table ). Most images were acquired with the panchomatic filter (F1, λc=650 nm) and short exposure times to avoid smearing effects. Two images (Seq. 2, Table ) were instead acquired with the near-IR filter (F12, λc=1015 nm) and longer exposure times (140 ms equivalent to 5 times the dwell time) to test image quality under smearing conditions.

The transition across the terminator was monitored using sequences of short-exposure monochromatic images (Seq. 3 and 4, Table ), including four panchromatic images with 25 ms exposure time to capture the rapid change in illumination. JANUS acquired 116 images of Earth's dayside (Seq. 7 to 12 and 4, Table ). These observations covered tropical latitudes over the ocean and included several multi-filter sequences. Two of the multi-filter sequences targeted an area of the Pacific Ocean also observed with two Earth observing satellites: EnMap and PRISMA both obtaining high-resolution hyperspectral data of the area covered by JANUS to compare the radiometric quality of the observations . Most other daytime observations used the green filter (F3, λc=540 nm) alternating exposure times that varied by a factor of 3–5 to obtain images well exposed for the dark ocean and bright clouds. This filter was preferred over the panchromatic one, as the latter would have saturated even at the minimum exposure time of 0.2215 ms due to its broad spectral bandwidth.

All images were acquired with lossless compression: in addition to preserving the highest images quality, it allowed to investigate the effective compressions under a wide range of illumination conditions.