MAJIS is a dispersion grating imaging spectrometer operating between 500 and 5560 nm by means of two spectral channels (Poulet et al., 2024b). The first channel (VISNIR, 500–2350 nm) is characterized by nominal spectral resolution and sampling of 2.9–4.6 and 3.5–3.8 nm per band respectively, while the second (IR, 2270–5560 nm) works with a spectral resolution of 5.5–7.0 nm and a sampling of 5.9–6.9 nm per band. The nominal instrument's instantaneous field of view (IFOV) is 150 µrad per pixel. MAJIS concept has been optimized for the characterization of the surface and near-surface environment of Jupiter's icy moons (Poulet et al., 2024b), as well as for the investigation of Jupiter's atmosphere (Fletcher et al., 2023). Detailed descriptions of the instrument functioning, operations and calibration are given in Haffoud et al. (2024), Langevin et al. (2024), Poulet et al. (2024a), Filacchione et al. (2024), Rodriguez et al. (2024), Vincendon et al. (2024), and Stefani et al. (2025). Scene geometry is reconstructed via the SPICE-NAIF toolkit (Acton, 1996; Acton et al., 2018) and kernels provided by ESA (ESA SPICE Service, 2019).

Figure 1 and Table 1 summarize footprint locations and main basic properties of the 17 MAJIS EGA data investigated in this work (see Poulet et al., 2026, for further instrumental parameters). Two additional cubes, targeted off-limb for calibration purposes (Poulet et al., 2026), are not considered here. Each MAJIS acquisition consists of hyperspectral cubes (i.e. 2D spatial frames with a third spectral dimension) collected as pushbroom spectral scans via internal mirror rotation, with different widths and lengths.

The first 4 cubes (C1 to C4) pointed to the Earth surface at nighttime and contain a significant signal only in the thermal part of the spectrum (λ>3000 nm). The only exception is C1, where a lightning emission is identifiable at visible wavelengths (D'Aversa et al., 2026). C5 is straddling the terminator and is the first cube containing information on the dayside ocean and clouds. Some coastlines are identifiable in C4 and C5 at thermal wavelengths, as it will be discussed in Sect. 4.4. All the subsequent cubes (C6 to C17) are acquired in daylight and hence the full spectrum can be investigated, even if they only cover the ocean mostly under cloudy/stormy conditions.

Cubes C11, C12 and C13 have been acquired with longer integration times, with the purpose of testing the instrument response. This leads to signal saturation in many regions (especially at visual wavelengths over clouds, see Sect. 2.3.1), that have been removed from our analysis. The spatial resolution in this dataset is quite stable (about 1.4 km per pixel, slightly affected by motion smearing) and is suited for the investigation of both homogeneous and localized cloud structures. On the other hand, the IFOV is affected by unresolved cloudiness (likely widespread) which dilutes the low reflectivity of deep water hence preventing the acquisition of clear-sky ocean (Sect. 2.3.1, Fig. 2).

An observing campaign coordinated to the EGA was conducted by the mission PRISMA (PRecursore IperSpettrale della Missione Applicativa), managed by the Italian Space Agency. The mission hosts a visible and near-infrared imaging spectrometer, covering a range (400–2500 nm) compatible with the MAJIS-VISNIR channel but having a coarser spectral resolution (∼12 nm) in turn compensated by a higher spatial resolution (∼30 m per pixel). Details about the instrument and the mission can be found in Pignatti et al. (2013), while mission characteristics, access, products, calibration, geometry navigation and data policy are fully described in Lopinto et al. (2021).

PRISMA sequences (13 in total, red rectangles in Fig. 1, main parameters summarized in Table 2) consist of a variable number of 30 km × 30 km hyperspectral cubes, each composed of 1000×1000 spatial pixels. Due to the PRISMA orbit (Sun-Synchronous-Low-Earth-Orbit), observations are acquired at a fixed solar local time (∼10:30), making it impossible to achieve spatial/temporal coincidence with MAJIS ones (see next section).

Both MAJIS and PRISMA acquired multispectral data covering the same kinds of structures, offering a useful benchmark for checking MAJIS capabilities in detecting and analyzing specific features of scientific interest. In the following section we investigate how the spectral signatures of the main atmospheric gases and of clouds are affected by the different spatial/spectral resolutions and observing conditions. When reflectances are discussed, they are obtained for both instruments as I/F, where I is the observed spectral radiance and F is Kurucz solar spectral radiance (Kurucz et al., 1984; Kurucz, 1995) available at the website https://earth.gsfc.nasa.gov/climate/projects/solar-irradiance/data (last access: 10 February 2026). For readability, in the following we will refer to spectral radiance simply as radiance.

In principle, Earth observations can encompass different types of surfaces, commonly discriminated spectrally through indices expressed in the formalism of Normalized Difference spectral Indices (NDIs, see Wolf, 2012, for a general review). Useful examples are given in Hurley et al. (2014) (dealing with Rosetta/VIRTIS-M data, Coradini et al., 1999) and in Oliva et al. (2017) (dealing with both Rosetta and Venus Express/VIRTIS-M data, Drossart et al., 2004). Table 3 summarizes these indices (derived from spectral endmembers from Rosetta/VIRTIS-M acquisitions, Fig. 6A–B, since MAJIS observations did not cover surface features in daylight) that we test on PRISMA data (Fig. 6C–D) as a benchmark for the future September 2026 EGA, in which Africa observations are planned. A new ocean index is also defined specifically for MAJIS data, which do not cover all wavelengths of the nominal ocean NDI. It is worth stressing that the ocean class should not be considered as representative of clear-sky conditions as it may actually include some amount of aerosol opacity (Sect. 2.3.1). No specific index has been adopted for generic clouds identification, but we rather assign to this class all pixels that do not meet any of the surface classes' conditions. Indices thresholds can be studied taking advantage of proxy images (e.g. the PRISMA one shown in Fig. 6C, not pertaining to EGA sequence) in which the changing reflecting structures can be clearly identified. The derived values depend on instrument features and require specific tuning when switching between different datasets. Figure 6C–D shows how the different types of spectral classes can be reliably identified, even if, in this case, no ice clouds are present. Other examples of application of the ocean, clouds and ice indices from Table 3 to MAJIS and PRISMA data are discussed in Sect. 4.1. Instead, the application of surface-related indices to MAJIS data did not result in positive identification, since land features in MAJIS data are not seen in daylight illumination, making NDIs not applicable.

In the specific conditions of MAJIS EGA sequence, the most robust land identification must rely on soil/ocean contrast in thermal emission (Sect. 4.4), triggered by the different thermal inertia of the two classes. However, also the presence of clouds in the line of sight induces a decrease of the observed brightness temperature (TB), hence land identification requires matching the shapes of low TB regions within known coastlines. The largest land region emerging in this way is shown in Fig. 7 (cube C4), (Philippines's Busuanga and Coron islands in cube C4), whose identification also allows a refinement of MAJIS pointing reconstruction (Seignovert et al., 2026). The largest brightness temperature contrast for both land/ocean and cloud/ocean cases occurs in the 3500–4000 and 4600–4800 nm spectral ranges, which are less absorbed by atmospheric H₂O and CO₂. The application of this method to other MAJIS data is illustrated in more detail in Sect. 4.4.

MAJIS and PRISMA data allow investigating the distribution of physical properties of ice and how they relate, for example, to the altitude of the clouds where it is identified (see Sect. 4.1 and 4.2). The temperature, crystallinity, grain size, purity, and density affect the shape of ice absorption bands (in particular the main ones at 1500 and 1900 nm) and of the continuum. Since the long wavelength shoulder of the 1900 nm band encompasses the noisy junction between the VISNIR and IR channels of MAJIS, we focus on the 1500 nm band, spectrally well resolved in both MAJIS and PRISMA datasets. This band has a characteristic asymmetry (due to its differential intensity with respect to the 1900 nm one, e.g. Stephan et al., 2021a) affecting the position and shape of the in-between transmission window peak (∼1700 nm) and has been used for the definition of the ice index in Table 3. Within the 1500 nm band, the weaker 1650 nm absorption is present. Its strength is a proxy for the degree of the ice crystallinity and temperature (Fink and Larson, 1975; Filacchione et al., 2016). It is also observable in PRISMA, even if shallower and noisier due to the lower spectral resolution (see zooms in Fig. 8A and B).

The 3000–4000 nm wavelength range, not accessible to PRISMA, hosts two ice reflection peaks at around 3100 nm (the Fresnel peak) and 3700 nm (Fig. 8C). The former varies in shape and intensity as a function of the ice crystallinity (Cartwright et al., 2025) while the latter shifts to longer wavelengths as temperature increases (e.g. Filacchione et al., 2016, see Sect. 3.3.3). Fresnel peak position variations are estimated in the data through cross-correlating each ice spectrum with a constant shape (average peak shape in each cube) which is rigidly shifted with a 0.1 nm sampling (hence allowing the estimation of the peak with a sampling better than the nominal MAJIS one). On the other hand, the 3700 nm peak position is obtained through fitting with a Gaussian function, reliably reproducing its shape.

Another proxy of the ice temperature is the intensity of its thermal emission, becoming significant at wavelengths larger than 4500 nm (Fig. 8D). However, in this range the emitted radiance is absorbed by a plethora of narrow bands of gaseous water, and therefore only a narrow transmission window around 4600 nm is suitable for this purpose. Table 4 summarizes these ice spectral features, identifiable in MAJIS and PRISMA data. The average uncertainties Δ are propagated taking into account the SNR estimates described in Sect. 2.3.1.

As a first investigation of the ice spectral variability in MAJIS and PRISMA observations we take advantage of the unsupervised K-means classification algorithm included in the ENVI software package, version 6.0 (Exelis Visual Information Solutions, Boulder, CO, USA, https://www.nv5geospatialsoftware.com/Products/ENVI, last access: 15 December 2025). This algorithm is capable of grouping the observations into an ensemble of “K” non-overlapping clusters, driven by spectral similarity, whose average spectra are representative of the main signatures in the dataset. This is done through an iterative minimum distance technique whose details are described in Tou and Gonzalez (1974). In this preliminary analysis, we arbitrarily set the algorithm to produce K=5 output average spectra, enough for visualizing the variability of the main ice diagnostic spectral features. It must be noted that, since we are also interested in features pertaining to infrared wavelengths, in MAJIS case the full VISNIR + IR spectral range is considered, and wavelengths longward of 2500 nm contribute to the clustering as well. As we will see in Sect. 4.1, this also has an impact on the spatial distribution of the clusters. The resulting average spectra in the solar range are shown in Fig. 8A and B for MAJIS cube C16 and for one of PRISMA session 04 cubes respectively. These spectra result to be mainly driven by the changing intensity of the continuum, in turn providing information about the opacity of the ice clouds. The color scale is associated with increasing reflectance of the transmission window at 1700 nm (dashed grey, dashed red, orange, blue and green with crosses from low to high, indicating increasing opacity and variable crystal sizes). The same color scheme is retained for the intensity of Fresnel peak at 3100 nm in MAJIS data (Fig. 8C), diagnostic of the ice crystallinity. Instead, spectra with intermediate reflectances at 1700 nm switch order within the 3700 nm ice reflectivity peak (dashed red to blue to orange from low to high, Fig. 8C) indicating the increased weight of thermal emission on the overall signal in this range. At MAJIS wavelengths larger than 4500 nm (Fig. 8D) the initial color scheme is totally disrupted, due to the mixing of information about cloud emissivity, cloud temperature (i.e. the altitude) and gaseous opacity. The combination of high NIR reflectances and low thermal emission (green spectrum with crosses) suggests the presence of optically thick high-altitude clouds, as confirmed by the shallower water absorption bands longward of 4900 nm. On the other hand, large thermal radiance and deep water bands associated with intermediate NIR reflectance (dashed red spectrum) indicate a population of moderate opacity clouds at quite low altitudes. The other spectra present intermediate properties in the thermal range, not strongly correlated with the NIR reflectance, calling for mixed-phase clouds of variable microphysical properties and vertical structure.

The most straightforward method for evaluating cloud altitudes involves the correlation of the brightness temperature at a given wavelength (e.g. 4610 nm, less affected by gaseous absorption in the MAJIS range) with a known vertical temperature profile. For ice clouds, temperatures can be derived from the 3700 nm peak position (Sect. 3.3.3). Other methods that we consider here are based on O₂ absorption bands' variability and on the analysis of clouds' shadows (Sect. 3.3.1 and 3.3.2). In this study, we rely on a fixed average temperature profile (Efremenko and Kokhanovsky, 2021), which may be not representative of the actual thermodynamic conditions of the atmosphere during the observations. As a consequence, all the methods that we adopt yield a range of results, each affected by their own intrinsic limitations. Although they appear quite consistent with each other, more quantitative investigations are postponed to future analyses.

Among the many gaseous features observable in the 4000–5500 nm MAJIS range, two are particularly interesting, being observed as emission bands. These are the CO₂ double-peak at the bottom of the main 4300 nm band and an O₃ signature around 4700 nm. Both are evident above optically thick clouds at high altitudes, blocking the thermal contribution from the surface and lower (hotter) atmospheric layers. The CO₂ peak is radiometrically much more stable than other spectral features against variation of atmospheric structures (see Poulet et al., 2026). It is known to result from the combination of a LTE component induced by temperature increase in the stratosphere, and a non-LTE one due to the CO₂ excitation primarily induced by direct solar pumping occurring at even higher altitudes (where collisional quenching is no longer efficient, e.g. Cassini et al., 2025). The detailed analysis of this emission feature in MAJIS data, implying the evaluation of CO₂ vibrational temperature vertical profiles, is far beyond the purpose of this work. In any case, the spatial distribution of the CO₂ emission intensity can provide interesting insights about the probed layers, and we can indeed use it for detecting atmospheric waves and provide hints about their altitude and propagation (see Sect. 4.3.1). CO₂ emission can be identified already in MAJIS monochromatic frames at 4270 nm (i.e. the position of the main peak of the emission) but the integration of the band in a narrow spectral range is useful for reducing noise and enhancing the contrast in waves' investigation (Sect. 4.3). For the integration we consider wavelengths between 4254 and 4333 nm, which probe high altitudes in the atmosphere and are not affected by the thermal contribution from lower ones. Considering the SNR estimated at these wavelengths (Fig. 3C), we are able to detect waves whose relative intensity between crests and troughs is about 1 %, assuming a 3-sigma uncertainty for the radiance at 4270 nm.

On Earth, ozone has a maximum density in the lower stratosphere but its vertical distribution strongly depends on latitude (see for example Bekki and Lefevre, 2009). It is produced through a very fast and exothermic 3-body recombination reaction that includes O and O₂ in the presence of a catalytic species (either N₂ or O₂). Aside from diagnostic bands at UV (outside MAJIS domain) and VIS wavelengths (the Chappuis band discussed in Sect. 2.3.2), the 4700 nm one is the strongest feature clearly detectable within the MAJIS range. This O₃ band is seen as either an absorption or emission feature in MAJIS nadir-looking observations, depending on the overall thermal emission of the atmospheric column. In clear sky conditions, when the emission from lower warmer layers is dominant, the O₃ 4700 nm band is hardly detectable being overcome by water absorption (as shown in Poulet et al., 2026), unless radiative transfer modeling is performed on the data (e.g. Guerlet et al., 2026). In the presence of mid-altitude clouds, a shallow O₃ band appears in absorption, while the obstruction of the densest part of the atmospheric column due to high-altitude clouds makes the O₃ band appear in emission. Given this phenomenology, in this preliminary study we investigate the O₃ emission amplitude through the difference between brightness temperatures estimated at 4717 nm (strongest O₃ line) and 4660 nm (outside O₃ band). Such a difference is positive when the O₃ is spectrally observed in emission, negative otherwise.

Examples of two MAJIS and PRISMA cubes containing ice clouds, identified through the ice spectral index in Table 3 (threshold <1), are given in Fig. 11. In the MAJIS case, ice is found in localized convective clouds (Fig. 11A and C), so high with respect to the background structures that they even cast well detectable shadows (see Sect. 4.2.2). Instead, in the PRISMA observation ice is detected both in diffuse bright clouds (e.g. at the southern east corner of Fig. 11B and D) and in thinner and less contrasted structures (probably identifiable as high altitude cirrus clouds, e.g. the white regions around longitude 116.4° – latitude 16.5°, Fig. 11B and D) hence proving the effectiveness of the index with different regimes of ice optical depth. Sample spectra from the identified classes are shown in Fig. 11E and F for MAJIS and PRISMA respectively. It must be noted that the very low albedo of the ocean in MAJIS spectrum at visual wavelengths (<1 %) is due to the very slant illumination conditions for the selected observation (incidence angle of about 80° for cube C7, see Table 1). On the other hand, the spectra in the thermal range show consistency with the expected temperature regimes, with very cold ice clouds and the ocean hotter than liquid water clouds.

Ice is similarly widespread in other MAJIS and PRISMA data, so that some considerations on its distribution and correlations between its parameters can be made (Fig. 12). We compute the 1500 nm band asymmetry as a ratio of slopes, the first considered between 1415 and 1500 nm (left wing) and the second between 1500 and 1790 nm (right wing). The asymmetry correlates with the strength of the 1650 nm band (quantified as equivalent width, Fig. 12A and B), with higher values indicating increasingly crystalline ice (Mastrapa et al., 2008; Stephan et al., 2021a; Grundy and Schmitt, 1998). Different regimes of these two parameters map localized structures in MAJIS and PRISMA observations, as shown in Fig. 12C and D respectively where green and blue pixels refer to clusters contained within dashed ellipses sharing the same color in Fig. 12A and B. In MAJIS case, the blue cluster is characterized by an increasing 1650 nm equivalent width at constant 1500 nm band asymmetry. The green cluster, instead, shows a common trend of growth for the two parameters. On the other hand, the PRISMA ellipses identify well separated clusters of points within the two parameters' space.

It is interesting to note that the correlation between these clusters and those obtained from the K-means classification discussed in Sect. 3.2 (Fig. 12E–F) is not straightforward. For MAJIS, ice spectra with high reflectivity in the solar part of the spectrum (green in Fig. 12E) are mostly correlated with the blue cluster in Fig. 12C. This trend is not observed in PRISMA, where all K-means clusters are equally distributed over both the blue and green clusters shown in Fig. 12D, suggesting variable ice densities and grain sizes within the same regimes of crystallinity. This difference derives from the fact that, as explained in Sect. 3.2, for MAJIS the thermal wavelengths contribute to the K-means classification of the spectra, hence providing information also on the temperature of the ice (see also Sect. 4.2.3). This is verified by the trend of the 1500 nm band asymmetry with the radiance in the thermal part of the spectrum, shown for MAJIS cubes C16 and C17 in Fig. 13A–B: more crystalline ice (larger asymmetry) is correlated with lower radiances (i.e. temperatures) at thermal wavelengths. In particular, orbit C17 also shows a detached cluster in the distribution of the thermal radiance suggesting different regimes of temperature (hence different clouds' altitude). Finally, we show the correlation between the ice crystallinity and its temperature in Fig. 13C–D, where the intensity and wavelength of the Fresnel peak are compared to MAJIS thermal radiances. Consistently with previous studies (e.g. Stephan et al., 2021a), the intensity of Fresnel peak is higher when the temperature is low (Fig. 13C), indicating enhanced crystallinity (see also Poulet et al., 2026). The comparison in Fig. 13D shows two distinct regimes of the peak position, with the short wavelength cluster characterized by a larger spread of the thermal radiance (suggesting an enhanced temperature variability for a less crystalline ice, e.g. Stephan et al., 2021a).

We now discuss the altitudes of clouds derived with the different methods presented in Sect. 3.3.

The CO₂ and O₃ emissions introduced in Sect. 3.4 have been studied in all MAJIS cubes, deriving maps like those shown in the examples of Fig. 19 and Fig. 20. In Fig. 19, panels (A) and (B) show MAJIS cube C7 displayed at 3100 and 4512 nm, whose anti-correlation highlights the presence of the convective clouds discussed in Sect. 4.1, 4.2.2 and 4.2.4. Panels (C) and (D), instead, show the radiance of the peak of CO₂ emission at 4270 nm and the brightness temperature difference between the O₃ emission peak and its continuum (Sect. 3.4). It is evident how wavy patterns can be seen in the CO₂ map and are uncorrelated with the clouds beneath. No wave patterns are spotted from the O₃ emission, whose positive values (and hence the emission) are only detectable above the convective structures. This suggests that, while both phenomena are likely happening above the clouds' top, waves are generated at different altitudes with respect to those pertaining to the O₃ emission. However, the actual heights are not investigated here, since a rigorous retrieval accounting for non-LTE effects (required for the assessment of these high-altitude emissions) is beyond the scope of the paper.

Following the discussion in Sect. 3.4, in Fig. 20 we show the effect of the increased contrast that can be achieved through the spectral integration of the CO₂ emission (right panel), with respect to the single wavelength investigation (left panel). The integration reduces noise hence allowing enhanced accuracy in detecting the wave patterns. Indeed, if the radiance integrated in the band is considered, the detectable relative intensity drops from 1 % to about 0.5 %, which translates as an increased capability in characterizing the vertical structure of the waves.

The land/ocean-contrast detection method described in Sect. 3.1 has been applied to all MAJIS cubes, but only a few land features have been identified. The C1 and C2 cubes, expected to cover large land areas at nighttime, encountered very thick and extended storm systems that prevented any surface visibility. Hence, all observable land regions consist of small islands seen in twilight illumination, colder than the surrounding sea surface but barely observable at visible wavelengths. Besides the largest example (Fig. 7), other islands are found in the cube C5 (Fig. 21): Babuyan (region 3), Sabtang (region 4), and the very small Ibahos island (about 4 km × 2.5 km wide, region 5), all part of the Batanes archipelago. The nearby Dequey island, even smaller (∼0.7 km× 1 km), remains unresolved. With respect to the ocean, the brightness temperatures measured over land and cloud areas (Fig. 21b) are colder, with differences up to ∼6 and ∼8 K respectively. Even if fully located beyond the terminator (solar incidence angle ∼90.8°), a significant signal is detectable also at visible wavelengths, ascribable to light scattering in the upper illuminated portion of the atmospheric column, and to multiple scattering effects in the lower part.

The MAJIS sensitivity to temperature variations can be estimated from the signal fluctuations over cloud-free ocean regions. The resulting uncertainties in thermal brightness (at 4610 nm) vary between 0.5 and 1 K, which correspond to about 0.2 % and 0.4 % of the radiance at 293 K. This sensitivity appears sufficient to discriminate significant temperature variation not only between sea and land surfaces but also between different land regions. As an example, we show in Fig. 22 the variability of MAJIS brightness temperature inside the Babuyan island, which hosts a volcano of about 1 km in elevation (Babuyan Claro Volcano). Even if the spatial resolution is limited, a clear trend emerges with respect to the topographic altitude, suggesting that the MAJIS data are sensitive to the surface altimetric temperature change.

The detection of terrestrial ice clouds described in Sect. 4.1 represents the first spectral observations of water ice performed by MAJIS, and is therefore the first approach to establish the potential outcomes from observations of Jovian icy satellites, in particular for Callisto and Ganymede.

The investigation of ice properties possibly provides information on the differential evolution these bodies underwent in the Jovian system environment. For example, Callisto's surface is mainly covered by crystalline ice, while significant amorphous ice patches have been observed on Ganymede (e.g. Tosi et al., 2024; Bockelée-Morvan et al., 2024; Cartwright et al., 2024). These regions could indicate alteration through radiolysis induced by the impinging of charged particles on the ice (Khurana et al., 2007), hence providing information on the mechanisms connecting Jupiter's magnetic field lines and the moons' surfaces. Moreover, while Callisto is characterized by an overall low ice content on the surface (∼50 %) and presents a more ancient and stable scenario (Greeley et al., 2007), Ganymede's fresh ice patches are indicative of more frequent ice resurfacing and cryo-volcanism events (Ligier et al., 2019). Smaller ice crystals are observed at the poles, matching the distribution of the fresher ice deposits and hence acting as a tracer of geologic activity. In this view, the investigation of ice-related spectral parameters can be used to address many scientific goals of the JUICE mission (Stephan et al., 2021b; Poulet et al., 2024b).

Jupiter's atmosphere is thought to be dominated by the presence of three main cloud decks residing at different altitudes and mixed by convective processes and atmospheric circulation (Fletcher et al., 2023). From lower to higher heights these are respectively composed of a H₂O-NH₃ liquid solution, NH₄SH solid aggregates, and NH₃ ice crystals (Atreya et al., 1999). In particular, the NH₄SH and NH₃ clouds can be responsible for the chromatic differences in Jupiter's dark “belts” and bright “zones”. Above these structures, hazes composed of products of the photochemical disruption of CH₄ and NH₃ extend from the upper troposphere to the stratosphere (e.g. Sindoni et al., 2017; Biagiotti et al., 2025). Such cloud complexity is not present in Earth's atmosphere where water is the only condensible, aside from a variety of aerosols of different origin (e.g. maritime, volcanic, smog, stratospheric). Nevertheless, the study of EGA observations allows a first MAJIS data analysis devoted to disentangling the spectral information related to different sources, like gases, clouds and, in this case, also surfaces. In this manuscript we have investigated clouds under different points of view, including their detection, water vapour phase identification, vertical structure assessment, and microphysical properties estimation. All these techniques are applicable to Jupiter once adapted to the different composition and structure of the giant planet. For example, the RT modeling presented in Sect. 4.2.4 only dealt with the solar part of the spectrum, which would only allow the investigation of Jupiter's hazes and the NH₃ deck (e.g. the recent work of Biagiotti et al., 2025, on JUNO/Jiram data). The exploitation of the full MAJIS spectral range, including thermal wavelengths, is instead mandatory for characterizing the deeper NH₄SH (Grassi et al., 2021) and H₂O (Bjoraker et al., 2022) clouds, especially in “hot spot” regions.

The shadow technique for measuring cloud heights, commonly applied in planetary high-resolution imaging analysis, is also applicable to Jupiter (e.g. Orton et al., 2017). For instance, in observations acquired at the bottom of methane bands, Simon et al. (2015) were able to measure shadows 45 km long, revealing wavy structures less than 1 km in amplitude. In principle, MAJIS observations of Jupiter atmosphere will allow the application of this technique to limited cases, mostly near the terminator and in polar regions when observed from perijove. Maximum spatial resolutions of ∼120 km per pixel achievable in these conditions may enable detecting shadows related to vertical displacements of the order of 10 km.

The use of chemical atmospheric species as tracers for the atmospheric circulation, including wind measurements and wave detections, is widely applied to the investigation of both terrestrial (i.e. Hueso et al., 2008; Peralta et al., 2008) and giant planets (i.e. Müller-Wodarg et al., 2019; Grassi et al., 2020). A similar approach is valid for the upcoming MAJIS measurements at the Jovian system, whose upper atmospheric dynamical structure can be investigated through the monitoring of the distribution (in latitude and local time) of minor widespread species like H3+ and hydrocarbons deriving from the photolysis of methane (see Miller et al., 2020, for a thorough review) as demonstrated from both ground-based (see for example O'Donoghue et al., 2016) and space-based data analyses (e.g. Moriconi et al., 2020). MAJIS IR channel will allow to spectrally discriminate the CH₄ and H3+ contributions in the range 3000–4000 nm, where the two species present strong features (Castagnoli et al., 2025) identifiable within the fundamental 3300 nm CH₄ absorption band, similarly to the case of the 4300 nm CO₂ band in Earth's atmosphere (see Sect. 4.3). The study of CH₄ and H3+ (e.g. JWST data analysis, Melin et al., 2024) will give access to upper atmospheric layers which are hardly probed otherwise. Altitudes from about 200 km above the 1 bar level are typical of methane emission peak, while above 500 km the H3+ emission seems to dominate, as also shown by recent analyses of JIRAM-Juno data (Migliorini et al., 2023), where the two species have been spatially separated.