Pending the next "ocean worlds" missions: Callisto’s surface properties and composition from near-infrared telescopic data

Icy bodies are the most numerous and diverse bodies in the Solar System, but only a few have been visited yet by space probes. Some of these icy bodies, mostly giant planet satellites like Callisto, are called ocean worlds as they likely host most of the liquid water of our Solar System within sub-glacial oceans [Sotin & Tobie 2004]. The presence of liquid water under the icy crusts of these bodies raises questions relative to exobiology. In the coming years and decades, these ocean worlds will be the focus of major planetary science missions, firstly with JUICE (ESA, launch scheduled in April 2023) and Europa Clipper (NASA, launch scheduled in October 2024) which will be targeting the Galilean moons, then with still conceptual missions as we speak such as (i) Uranus Orbiter and Probe and (ii) Enceladus Orbilander. Despite not being the prime target of any of these missions, Callisto will be investigated by JUICE and Europa Clipper through multiple flybys.

One way to get information about Callisto’s sub-glacial ocean is to look at the moon’s surface, which may highlight markers of past or current activity and could lift the veil on its chemical composition. Spectroscopy is a powerful technique for such study, especially in the near-infrared domain where spectral signatures of H₂O-ices, silicates, salts are detectable. Consequently, in preparation for future infrared spectroscopic data, and specifically those of the infrared imaging spectrometer MAJIS of the JUICE mission, a ground-based campaign was led with an instrument sharing some similar and also complementary characteristics: SINFONI (SINgle Faint Object Near-infrared Investigation), which was mounted on the UT4 of the VLT at ESO. Here we present the results derived from the analysis and the modeling of four full-disk observations acquired with this instrument.

SINFONI combines an integral field spectrometer operating with different gratings with an adaptive optics module [Eisenhauer et al. 2003]. Our four observations were performed using the H+K grating, ranging from 1.46 µm to 2.41 µm, with a spectral resolution R = 1.500, but each spectrum acquired was automatically resampled in order to get a spectral sampling of 5 x 10^-4 µm. The campaign took place from January 2015 from March 2016, and the observations were acquired near the opposition to optimize the angular resolution with Callisto’s angular diameter close to 1.5 arcsec. As SINFONI’s field of view is 0.8 x 0.8 arcsec² divided into 64 x 64 pixels, one observation actually corresponds to a mosaic of ten overlapping acquisitions, as described in a paper about Ganymede using the exact same technique [Ligier et al. 2019]. All in all, after removing the high solar incident angle pixels we could not recover after applying photometric corrections (basically all pixels above latitude 60°, north and south), our dataset covers approximately 70% of Callisto’s entire surface with a spectral and spatial sampling (~40 km/pix) which allow to detect and map the spectral absorptions that may exist in this wavelength range of the Callisto’s spectrum. Spatial resolution is in par with that of the partial Callisto mosaic from the NIMS imaging spectrometer of the NASA Galileo probe.

The first results obtained concerns the physical properties of the moon’s surface directly derived from the reduction process and the photometric corrections; one obtains satisfactory results using the qualitative photometric model of Oren-Nayar, which is a generalization of the Lambertian model with the surface roughness (σ) as only additional parameter [Oren & Nayar 1994]. In the case of our study, the best results were obtained with  ranging between 17° and 19° whatever the observation. These values are much lower than the values in the 30° – 45° range provided by previous studies [Buratti 1991, Domingue & Verbiscer 1997], however these values were obtained through the modeling of the solar phase curve via the Hapke model [Hapke 1984, Hapke 1986] using a combination of telescopic observations and Voyager data; the major differences between the approaches make the respective results hard to compare (our data are < 10° emergence angle).

Our second result concerns the global shape and spectral signatures in the surface spectra: overall, they are much flatter and more distorted than those of Europa and Ganymede, especially the quite large 2-µm H₂O-ice band. To the exception of crystalline ice absorptions at ~1.50 µm, ~1.57 µm, ~1.65 µm and the 2-µm band, Callisto does not show other clear signatures (figure 1). However, very subtle signature(s), like the one at ~2.21 µm (figure 1), might actually be observed and are being investigated to confirm the detection(s). If confirmed it could suggest the presence of hydrated materials like salts or silicates. We will address this point during the meeting.

Figure 1: Comparison between Ganymede’s (blue) and Callisto’s (red) mean spectrum of the entire dataset.

And lastly our third main result comes from spectral modeling, where both linear and non-linear unmixing were performed. Unlike previous studies on Ganymede and Europa [Ligier et al. 2019, Ligier et al. 2016], the linear model shows that all of Callisto’s spectra are satisfactorily modeled without needing any hydrated salt and that the only two constituents would be (i) a darkening agent, with abundances always exceeding 80% even for the iciest spectra, and (ii) multiple grain sizes of the crystalline form of H₂O-ice, with an average abundance of 7% and exhibiting a strong latitudinal gradient for the smallest grains typical of the influence of the Jovian magnetospheric environment (figure 2). Such a low concentration overall of H₂O-ice challenged us about the legitimacy of the linear unmixing approach, so we decided to run the Shkuratov non-linear unmixing model [Shkuratov et al. 1999]. Tests are ongoing but very preliminary results show that the overall concentration of the crystalline ice should be much higher than the 7% coming from the linear model and hence much more representative of what should be the surface composition of an icy satellite. By the time of the meeting, we will be able to present the final results of the study.

Figure 2: Distribution of the smallest grains of crystalline ice.