Ray and Halo Impact Craters on Ganymede: Fingerprint for Decoding Ganymede's Crustal Structure

Abstract

Impact craters are a unique tool not only for inferring ages of planetary surfaces and examining geological processes, but also for exploring subsurface properties. We use ejecta blankets as proxies to obtain insights into the subsurface characteristics and the vertical stratification of Ganymede's icy crust. We investigated 36 prominent ray and halo craters using images acquired during the Voyager, Galileo, and Juno spacecraft missions. These craters exhibit diverse characteristics, including dark rays, bright rays, or their combination, in both continuous and discontinuous patterns as well as dark and bright halos. Dark halo craters (DHCs) have the smallest radial extents of their dark ejecta deposits, while dark ray craters (DRCs) have the largest. DRCs in dark terrain suggest a thickness of less than ∼2 km based on their excavation depths. DRCs and DHCs craters located in light terrain (LT) reveal significant heterogeneity in the uppermost portions of icy crust at various locations. DRCs and DHCs in the LT require the presence of at least one layer of dark material. This is the case if the LT is formed by tectonic rifting and graben formation. In contrast, if the LT is formed by tectonic spreading, bright halo and ray craters are expected to form.

Introduction

Ganymede, the largest moon in the solar system, is one of the most interesting celestial bodies in our Solar System due to its complex geological history and potential habitability. As a result, it is now the main target of ESA's current JUICE mission to the Jovian system (Grasset et al., 2013; Stephan et al., 2021). Among its intriguing surface features are a large spectrum of different impact crater morphologies as well as different ejecta materials (Schenk et al., 2004). Here we study impact craters that contain ejecta blankets with rays or halos composed of either bright or dark ice or non-ice material (Hibbitts, 2023; Schenk & McKinnon, 1991). They are located in both the dark and light terrains (LTs) on Ganymede, which have formed in different times due to different surface processes and tectonic resurfacing (Pappalardo et al., 2004). These processes, particularly the formation of the LT, are still not fully understood and reach from tectonic rifting of the ancient dark terrain to tectonic spreading, as modes of extension for creation of new icy crust (Pizzi et al., 2017). During rifting, extension occurs by graben formation and the downfaulted dark terrain is covered by LT. In contrast, if the LT is formed in a spreading mode, then an entirely new light crust is formed by pulling apart dark terrains. Since the emplacement of ejecta surrounding many of Ganymede's craters is potentially strongly sensitive to the vertical stratigraphy of the subsurface, we use the current understanding of the impact crater process (Kenkmann et al., 2014; Melosh, 1989) and here in particular, the excavation flow field of a hypervelocity impact, to deepen our understanding of the vertical stratification of the Ganymede's icy crust with implications for the formation processes of the LT.

Ray craters, which occur on most planetary bodies, are defined as impact craters with a prominent and distant ray system formed by radially to sub-radially oriented streaks of fine, either continuous or discontinuous, ejecta deposits extending outward from the crater rim. Rays may be created by high-speed debris ejected from the central crater close to the projectile (Melosh, 1989). On Ganymede, some of the rayed craters exhibit distinctly dark rays, and others have bright rays (Schenk et al., 2004; Stephan et al., 2021, 2024). A halo crater is defined as a crater with an ejecta deposit that is either darker or brighter than its surroundings. The ejecta deposits of a halo crater form a roughly circular to sub-circular area around the crater rim, often featuring a diffuse edge (Melosh, 1989).

Schenk and McKinnon (1985) conducted the first study on Ganymede's dark rays, floor and halo craters using Voyager imagery. In the region of Uruk Sulcus, they found that halo effects were only present in craters with diameters smaller than ∼12 km. They suggested that the dark material in this region could be buried ∼1 km below the grooved terrain. Later Schenk and McKinnon (1991) investigated the distribution of dark ray and dark floor craters on various terrains, and found no preference for either dark or LTs. They interpreted the dark material in these craters to be formed by remnants of the impactors, such as D-group asteroids or comets, rather than being a result of excavation of underlying stratigraphic units. Spectral investigations of dark ray craters (DRCs) on Ganymede using data from the Galileo Near Infrared Mapping Spectrometer (NIMS), which detected Ganymede's surface composition in the near-infrared wavelengths region between 0.7 and 5.2 μm (Stephan et al., 2020), have led to the assumption that the dark material of DRCs consists of varying amounts of non-ice hydrated endogenous material and less hydrated carbonaceous material (Hibbitts, 2023). The latter is discussed to be either a remnant of impactor material or primordial to Ganymede as it is for the neighboring moon Callisto (Hibbitts, 2023). Although, remnants of impactor material are discussed to possibly explain some of the analyzed dark ejecta, no dark crater material has been found to be entirely composed of carbonaceous compounds and thus dark ejecta materials are interpreted to be mostly the result of the excavation of endogenous crustal material. Hypervelocity impacts, however, are expected to be responsible for the tenuous dust cloud of Ganymede and the other Galilean satellites that have been detected with the dust detector onboard the Galileo spacecraft (Krüger et al., 2000, 2003). The directions, speeds and distribution of masses of the grains indicate that they come from Ganymede, and are consistent with an ejection process resulting from hypervelocity impacts of interplanetary dust onto Ganymede's surface. Even more, nearly all ejected refractory particles eventually fall back to the surface on ballistic trajectories and darken Ganymede's dark regolith (Krüger et al., 1999).

During the Galileo mission, in addition to NIMS observations, Ganymede was imaged by the Solid State Imaging (SSI) camera system (Belton et al., 1992). These images show Ganymede's craters in so far unprecedented resolution including some of Ganymede's unique dark ray and dark halo craters (DHCs). In addition, craters have been observed that were emplaced directly at the border between dark and bright terrain.

The idea for this study is based on ballistic ejecta emplacement models (Maxwell, 1977; Melosh, 1989; Oberbeck, 1977). The basic principle of excavation is that the innermost parts of a growing crater cavity that directly surround the penetrated projectile have experienced the highest shock, are launched first, and traveled fastest and form the top of the ejecta curtain and eventually the distal, discontinuous part of an ejecta blanket. Ejecta originating farther from the crater center are launched later, move more slowly, contribute to the base of the ejecta curtain and are deposited nearer to the crater rim and form the proximal ejecta blanket. With increasing radial range and ejecta velocity, more local surficial material is incorporated and mixed with the distal ejecta (Oberbeck, 1975). We apply these principles of excavation dynamics to obtain information about the crustal structure of Ganymede. A natural limit is given by the excavation depth, which is a measure of the impact energy and thus the crater size.

Previous Studies on Ray and Halo Craters on Other Bodies

On the Earth's Moon, rays appear mostly around fresh craters (Elliott et al., 2018). Distal rays were interpreted to be a mixture of primary crater ejecta and the ejecta of secondary craters (Hiesinger et al., 2012). Studying the bright rays of Tycho, Shoemaker (1965) found that the density of secondary craters increased closer to the midline of each ray, suggesting that rays are due to clusters of bright secondary craters. It was proposed that when the ejecta fragments are large enough, they can excavate bright material from beneath the mature space-weathered layer to create bright rays, while smaller ejecta fragments simply mix the dark mature lunar soil (Elliott et al., 2018). Robinson et al. (2015) who examined the ejecta blanket of a young Lunar crater, stated that the bright continuous ejecta blanket likely formed due to unweathered material excavated by the primary crater.

Craters with dark halos and rays, observed at large phase angles, can also result from surface roughness rather than compositional variance (Kaydash et al., 2014). Other findings revealed that lunar rays are bright due to their composition contrast with the surrounding terrain, the presence of immature material, or some combination of the two (Hawke et al., 2004).

Daubar et al. (2022) investigated more than 1,000 newly formed small impact craters on Mars containing halos, blast zones, and rays with crater sizes ranging from 58 m in diameter down to smaller than a meter. They found that light- and dual-toned blast zones indicate excavation of compositionally distinct material from different depth of the crater cavity, rather than representing projectile material or surficial material. They could also confirm experimental and numerical data that high porosity and loosely consolidated target material tends to form craters with little or no ejecta (Housen & Holsapple, 2012). Mars exhibits crater halos influenced by thicker dust layers and lower elevations, with halo size primarily controlled by impact energy (Bart et al., 2019).

Ceres show impact craters with diverse colored ejecta deposits, such as the dark material on the Dantu crater floor, due to compositional differences at subsurface layers, indicating a heterogeneous crust that was excavated during the impact (Williams et al., 2018). The various types of crater rays and ejecta materials on Vesta represent excavation from underlying layers (Krohn et al., 2014; Williams et al., 2014; Yingst et al., 2014). Callisto's DHCs reveal subsurface dark layer materials excavated during the impact process. In contrast, the craters Lofn and Heimdall feature bright ejecta facies with water ice-rich impact melt deposits and more non-ice components excavated from shallower depths (Greeley et al., 2000). Europa exhibits dark, red materials around craters, originating from depths of a few kilometers below the surface, with maximum excavation depths reaching up to 3.6 km (Moore et al., 2001). Heating and vapourization of ice during hypervelocity impacts darken Europa's impact crater ejecta, possibly due to inherent ice properties. Another interpretation suggests a localized dark subsurface layer, with craters and extrusive domes excavating dark material (Tomlinson & Hayne, 2022). Dione's ray craters reveal excavation through surface deposits to reach redder materials, characterized by higher IR/UV ratios (Schenk et al., 2011). On Rhea, a study using VIMS data indicates the presence of clean, water-rich ice below the weathered surface, starting at a depth of at least 4 km. This suggests that at least the uppermost crust consists of clean water ice (Stephan et al., 2012).

Data Basis and Methods

Voyager and Galileo Data

To identify ejecta blankets and map dark ray and halo craters across Ganymede's entire surface, we utilized a global mosaic formed by merging high-resolution images from Voyager 1, Voyager 2, Galileo, and Juno spacecraft missions (Kersten et al., 2022). Juno's wide-angle camera, JunoCam, captured images during its close approach to Ganymede on 7 June 2021, providing additional images in the region between longitudes 40°W and 25°E (1 km/px) with slightly better resolution (1 km/px) and better illumination compared to Voyager 1 + 2 and Galileo spacecraft (C. J. Hansen et al., 2022). The global mosaic incorporates images with spatial resolutions ranging from 100 m/px to 10 km/px (Kersten et al., 2021). The detailed geologic study of individual craters and their geologic context was performed based on the best resolved Voyager and Galileo images available.

Several of the studied craters were observed also by Galileo NIMS with spatial resolutions up to ∼2 km/px (Hibbitts, 2023; Stephan et al., 2008) allowing the combination with the camera images and to separately analyze the crater and its ejecta from the surrounding terrain material. We included NIMS-derived information about the varying abundance of water ice and dark non-ice material in order to verify that the mapped ejecta are not an effect of illumination conditions or surface texture such as roughness, but are indeed characterized by highly concentrated dark non-ice material differently from the crater's surroundings. We follow the approach presented in Stephan et al. (2020) by mapping the band depth (BD) of one of the major water ice absorptions at 1.5 or 2 μm to derive information of the relative abundance of water ice versus dark material. Stephan et al. (2020) also presents an easily applicable method to derive information on the water ice grains sizes by using the BD ratio (BDR) of two water ice absorptions such as the ones centered at 1.5 and 2 μm. BDRs have been found to be largely independent of the additional existence of most non-ice materials expected to exist on Ganymede (Stephan et al., 2020). Increasing BDRs, generally indicate smaller grain sizes. In comparison with water ice model spectra presented in G. B. Hansen (2009), a rough estimate of the grain size can be performed (Stephan et al., 2020). Generally, the surface ice grain sizes on Ganymede are highly affected by the surface temperature causing relatively small water ice grains in the cold polar regions and large ones close to the equator accompanied by the equalization of grain sizes between the craters and surrounding surface ice with time (Stephan et al., 2020). In case of fresh impact craters, however, differences in the water ice grain sizes between crater, ejecta and surrounding material should still be apparent. Stephan et al. (2020) also showed that significant amounts of hydrated material such as salts presented in McCord et al. (2001) would destroy the relationship between BDRs and particular water ice grain sizes. In such a case BDRs values are different (mostly >1) from what is expected for water ice in the studied regions. Therefore, we use the information about water ice grains sizes to verify that we study more or less recently emplaced craters and their ejecta material and not dark material concentrated by other surface processes such as sublimation effects and mass wasting and to identify any significant differences in the composition of the dark deposits.

Identification and Mapping of Ray and Halo Craters

All studied craters are supposed to be geologically relatively young and did not experience extensive weathering because of their pronounced crater morphology and their bright or dark ejecta, which distinctly stand out compared to the surroundings of the crater. The retention times of ray craters on Ganymede's LT are approximately 256 ± 40 to 409 ± 46 million years, while those on the dark terrain are ∼120 ± 29 to 236 ± 46 million years (Xu et al., 2017).

We conducted a visual inspection of the global mosaic of Ganymede images to identify ray and halo craters on both light and dark terrains. We traced and mapped the morphologic details of the craters including their inner crater features such as peak, pit, or dome, the crater rim as well as the extension of their halos or ray system, as precisely as possible. In order to determine the crater diameter as accurately as possible, we first calculated the area enclosed by the rim of the halo or ray craters. Then, we equated this area to that of a circle and derived the diameter from it. These tasks were carried out using QGIS 3.16.11. Mapping was performed in an equidistant cylindrical projection (Plate Carree projection) to ensure accurate diameter measurements, with each crater being centered for analysis. For areas poleward of 60°, the global mosaic was re-projected into a stereographic projection and centered on each crater. We considered only those craters, whose rays and halos were traceable enough to be accurately mapped and whose crater diameters could be measured. Additionally, we documented the geologic context of the mapped crater and whether these craters were located in light or dark terrain or at the boundary between these two terrain types.

Based on albedo differences in image data in combination with the geologic context, and differences in albedo and/or abundance of water ice, we identified 4 classes for ray and halo craters:

Bright ray craters (BRCs): BRCs are characterized by a bright, presumably, ice-rich crater, and extensive rays of bright icy material. Although, these types of craters make up the majority of impact craters on Ganymede, some of them were emplaced directly at the border between dark and LT and are extremely valuable to understand any differences in the subsurface composition of the dark and LT.

Dark ray craters: DRCs generally exhibit a bright, presumably ice rich crater as seen for the BRCs. In contrast, however, their ejecta deposits, which extend as rays over long distances are of low albedo and presumably composed of non-ice material (Stephan et al., 2024).

Bright and Dark Ray Craters (BDRCs): BDRCs are characterized by one half of their rays having a bright appearance, while the other half displays a dark appearance.

Dark Halo Craters: DHCs are characterized by double concentric zones of differently colored continuous ejecta material with circular to sub-circular diffused edges. These concentric zones consist of alternating dark and light ejecta materials around the crater rim.

To quantify the ejecta blanket radial extent, we calculate the mean ejecta blanket radius normalized to each of their crater radius for different crater types.

Determination of Excavation Depths

Hypervelocity impacts are extremely complicated geologic processes that have been studied by experiments (e.g., Housen & Holsapple, 2011; Kenkmann et al., 2018) and sophisticated numerical simulations (e.g., Collins et al., 2012). Senft and Stewart (2008) derived the equation-of-state and strength model parameters for H₂O and modeled impacts into an icy-layer underlain by a rocky target. Even more complexity arises when oblique impacts are considered (Pierazzo & Melosh, 2000). Here we use a simple but useful analytical approach to model the idealized excavation flow field of a vertical impact into a homogeneous target, the so-called Z-model (Maxwell, 1977). The streamlines of the excavation flow near the point of impact are directed downward and outward, away from the point of impact and lead to the opening of the transient cavity (Kenkmann et al., 2014; Figure 1). Sidewise streamlines radiate outward and gradually bend upward due to the reduced pressure gradient toward the surface, leave the crater, and form an ejecta curtain, and eventually an ejecta blanket. It is important to note that only the displaced material of the upper one-third of the cavity depth leaves the crater.

[IMAGE OMITTED. SEE PDF]

According to Maxwell (1977), the maximum depth of excavation (D_e) of the transient cavity diameter D_t depends on the Z-value: 1 $${D}_{e}=\left(1/2\,{D}_{t}\,\right)(Z-2){(Z-1)}^{(1-Z)/(Z-2)}$$

In rocky targets, Z = 3 leads to good results. The maximum excavation depth D_e is D_e = 1/8 D_t and the excavated material leaves the crater at an angle of 45°. However, in icy targets the excavation flow leads to steeper ejection trajectories (Senft & Stewart, 2008), which are better simulated by Z = 4 streamlines.

According to Melosh (1989), the maximum depth of excavation (D_e) of a crater is roughly 10% of the transient cavity diameter: 2 $${D}_{e}\approx 1/10\ast \,{D}_{t}$$ where, D_t is the parabolically shaped diameter of the transient crater cavity.

To determine the transient diameter (D_t) of a crater with known final crater diameter (D), we employ the equation by Schenk et al. (2004). This equation integrates the transient diameter (D_t), the simple-to-complex crater transition diameter (D_c), and the final crater diameter (D) for complex craters. Since all the craters we have mapped exceed 5 km in size and are considered complex craters, we can apply the Schenk et al. (2004) equation in our analysis. This equation is rooted in the crater scaling equation established by Schmidt and Housen (1987) within a gravity-scaling regime. Therefore, for complex craters (Schenk et al., 2004): 3 $$D={D}_{t}^{\varepsilon }\,\cdot \,{D}_{c}^{1-\varepsilon }$$

D_c = 2.5 km for Ganymede (Schenk et al., 2004), d and D_t are in kilometers, ε ∼ 1.13, which accounts for crater slumping (Schenk et al., 2004). Rearranging and substituting values in Equation 2 leads to: 4 $${D}_{t}=1.111\,{D}^{\,0.8849}$$

The radial distance of the ejecta from the crater center gives further hints to the original location and depth of the material prior to the impact (Figure 1). For example, proximal ejecta deposited close to the crater rim originates from a nearby position inside the crater and becomes involved in the excavation flow field only when the transient cavity reached almost its final size (Kenkmann et al., 2014). This lately launched ejecta is rather slow, its ballistic path short and originates usually from a near surface location (Oberbeck, 1977). In contrast, ejecta that forms a discontinuous distal blanket and is deposited multiple crater radii away from the source crater is ejected early in the cratering process, forms the upper part of the ejecta curtain, and stems from a central region of the crater, close to the point, where the projectile came to a halt. In analogy to explosion cratering this point is known as the equivalent depth-of-burst (Holsapple, 1980). In a simplified version, this depth-of-burst dob depends on the density contrast between the projectile ρ_p and the target ρ_t, as well as the diameter D_p of the projectile (Holsapple, 1980). 5 $$dob={D}_{p\,}\sqrt{\sfrac{{\rho }_{p}}{{\rho }_{t}}}$$

To calculate the diameter of a projectile as a function of crater diameter, scaling laws in the gravity regime are applied (Collins et al., 2005). Solving this for the projectile diameter yield: 6 $${D\,}_{p}^{-0.78}=1.161\,{D}_{t}^{\,-1}{\left(\sfrac{{\rho }_{p}}{{\rho }_{t}}\right)}^{\,\sfrac{1}{3}}{v}_{p}^{\ 0.44}\,{g\,}^{-0.22},$$ where v_p is the mean impact velocity, g, is the gravity of Ganymede. We used Ganymede's escape velocity of 2,740 ms⁻¹ as a conservative estimate of the impact velocity and a projectile density of 1,500 kgm⁻³. This is lower than the density of most meteorites but it is an estimate for the density of Kuiper belt objects based on Bierson and Nimmo (2019) and reflects a high ice content and porosity. With g = 1.428 ms⁻², and a target density of 1,000 kgm⁻³, Equation 5 is simplified to: 7 $${D}_{p}={\left(\sfrac{39.95}{{D}_{t}}\right)}^{\,-1.28}$$

For small craters up to 5 km diameter, the equivalent depth-of-burst calculation gives similar results as the excavation depth. For larger craters the dob exceeds 10% of the transient cavity size. The geometry of the excavated volume (Croft, 1980) is such that approximately 50% of the ejected volume is derived from the upper third of the excavation depth.

Results

Distribution of Ray and Halo Craters on Ganymede

We included 36 impact craters on Ganymede (Figure 2) ranging in size between 7 and 135 km diameter, where three of them categorized as DHCs s (Nergal, Khensu and Humbaba) while the remaining craters were classified as ray craters. These three DHCs exhibit a unique double-halo configuration, characterized by dark ejecta material surrounded by bright ejecta material. These halo craters have dark crater floors. Additionally, five of the ray craters are DRCs, with four located in dark terrain and 1 in LT. The ray crater Tammuz (DBRC) is a unique case, displaying half bright and dark ejecta. The 27 ray craters included in this study are BRCs were selected as representative of the majority of Ganymede's craters with respect to their size and location and geologic context. In detail, 20 are situated in the LT, eight in the dark terrain, and 8 were emplaced along the boundary between light and dark terrain.

[IMAGE OMITTED. SEE PDF]

Excavation Depths of Ray and Halo Craters on Ganymede

Crater parameters derived for the studied craters and discussed in this work are presented in Table 1. Figure 3 displays the maximum excavation depth versus the final crater diameter separated for (a), the LT, (b), the dark terrain, and (c) mixed terrain. The majority of rays and halo craters are situated in the LT (Figure 3a). The observed power law trend in the graph reflects Equation 4 and deviates slightly from a linear trend. The largest excavation depth, 8.5 km, was found for a 135 km final crater diameter, a BRC situated in LT. All craters with final diameter larger than 75 km excavate material from a maximum depth of 5 km have bright rays. In contrast, the smallest crater included in our study has a diameter of 7 km and excavates material from a maximum depth of ∼0.6 km.

Table 1 Parameters Derived for the Craters Selected for This Study Such as: Number of Crater in This Study (N), Latitude (Lat°) and Longitude (Lon°) Where the Crater Is Located, Crater Diameter (D), Transient Crater Diameter (D_t), Maximum Depth of Excavation (D_e), Depth of Excavation (D_e) Based on Z = 3, Depth of Excavation (D_e) Based on Z = 4, Class of Crater Type (Cc), Terrain Type (Tt), That Is, Dark (DT) or Light Terrain (LT), or Between Light and Dark Terrain (LT/DT), Minimum (R_min), Maximum (R_max) and Mean Ejecta Blanket Radius Normalized to the Crater Radius (R_A) and the Standard Deviation (S)

N	Crater name	Lat °/Lon °E	D (km)	Dt (km)	De (km)	De (Z = 3) (km)	De (Z = 4) (km)	CC	Tt	Rmin [km]	Rmax [km]	RA	S
1	Nergal	38.7/160.1	9	7.76	0.77	0.97	1.49	DRC	LT	11	17	3.11	0.94
2	Khensu	1.0/207.1	17	13.63	1.36	1.70	2.57	DRC	LT	19	29	2.82	0.83
3	Tammuz	13.9/129.1	51	36.04	3.60	4.51	6.94	BDRC	LT	124	486	11.96	10.04
4	Antum	5.5/141.1	15	12.02	1.20	1.50	2.31	DRC	DT	69	208	18.77	13.33
5	Kittu	0.5/25.2	15	12.20	1.22	1.52	2.35	DRC	LT	79	387	31.06	29.04
6	Tros	11.1/332.5	94	61.91	6.19	7.74	11.92	BRC	LT	254	785	11.05	7.99
7	Herschef	47.8/89.7	120	76.85	7.68	9.61	14.79	BRC	LT/DT	176	854	8.58	7.99
8	Tash-metum	−38.7/95.1	135	85.29	8.52	10.66	16.41	BRC	LT	305	1,351	12.26	10.96
9	n.n.	−12.6/165.6	46	32.89	3.28	4.11	6.33	BRC	LT	85	190	5.97	3.23
10	Amon	34.1/139.4	102	66.55	6.65	8.32	12.81	BRC	LT/DT	112	377	4.79	3.67
11	n.n.	20.9/191.5	67	45.88	4.58	5.74	8.83	BRC	DT	111	240	5.23	2.72
12	Nah-hunte	−17.8/274.5	47	33.52	3.35	4.19	6.45	BRC	LT	107	203	6.59	2.89
13	Cisti	−31.7/295.5	70	47.69	4.76	5.96	9.18	BRC	LT/DT	217	731	13.54	10.38
14	n.n.	−28.9/257.0	78	52.49	5.24	6.56	10.10	BRC	LT	151	363	6.58	3.84
15	n.n.	−10.8/254.2	83	55.45	5.54	6.93	10.67	BRC	LT	153	324	5.74	2.91
16	n.n.	−45.6/272.2	35	25.83	2.58	3.23	4.97	BRC	LT	129	269	11.37	5.66
17	n.n.	21.6/184.2	20	15.74	1.57	1.97	3.03	BRC	LT	40	96	6.80	3.96
18	n.n.	10.2/264.7	43	30.99	3.09	3.87	5.96	BRC	LT	113	281	9.16	5.53
19	Enkidu	−26.4/34.5	122	77.98	7.79	9.75	15.01	BRC	LT	179	920	9.00	8.59
20	n.n.	−37.9/328.7	37	27.13	2.71	3.39	5.22	BRC	DT	60	161	5.97	3.86
21	Min	29.2/358.5	31	23.19	2.31	2.90	4.46	BRC	LT	85	185	8.70	4.56
22	Shu	43.1/2.8	44	31.62	3.16	3.95	6.08	BRC	LT/DT	160	613	17.56	14.56
23	n.n.	23.9/6.5	21	16.43	1.64	2.05	3.16	BRC	DT	24	84	5.14	4.04
24	Ishkur	0.4/351.4	67	45.88	4.58	5.74	8.83	BRC	DT	150	280	6.41	2.74
25	Apophis	−7.6/83.6	57	39.76	3.97	4.97	7.65	BRC	LT/DT	85	293	6.63	5.16
26	n.n.	40.4/18.4	34	25.17	2.51	3.15	4.84	BRC	LT/DT	122	588	20.88	19.38
27	n.n.	−20.8/294.6	27	20.53	2.05	2.57	3.95	BRC	LT/DT	54	115	6.25	3.20
28	Osiris	−37.8/193.4	107	69.43	6.94	8.68	13.36	BRC	LT	460	1,826	21.36	18.05
29	Melkart	−9.7/173.9	103	67.13	6.71	8.39	12.92	BRC	LT/DT	194	545	7.17	4.82
30	Mir	−2.9/129.7	8	6.99	0.69	0.87	1.34	DRC	DT	41	352	49.12	54.98
31	Humbaba	−55.2/292.1	40	29.06	2.90	3.63	5.59	DRC	LT	80	120	5.00	1.41
32	n.n.	31.6/3.6	14	11.48	1.14	1.44	2.21	BRC	LT	22	33	3.92	1.11
33	n.n.	−65.4/134.1	29	21.87	2.18	2.73	4.21	BRC	LT	119	334	15.62	7.41
34	Andjeti	−52.6	45	32.26	3.22	4.03	6.21	BRC	LT	164	600	16.97	13.70
35	Enkidu (I)	−27.6/33.0	11	9.27	0.92	1.16	1.78	DRC	DT	11	31	3.81	2.57
36	Enkidu (II)	−25.9/33.7	7	6.21	0.62	0.78	1.20	DRC	DT	6	11	2.42	1.01

[IMAGE OMITTED. SEE PDF]

DRCs are typically found among the smaller craters. All DHCs are located on light grooved terrain. Nergal, the smallest DHC with a diameter of 9 km, excavates dark material from a maximum depth of approximately 0.77 km. In contrast, Humbaba, the largest DHC, excavates dark material from a maximum depth of about 3 km (Figure 3a). DRCs in dark terrain (Figure 3b), excavate dark material from a maximum depth of ∼1.2 km. Larger ray craters located at the borders of light and dark terrain are observed to be BRCs (Figure 3c).

Radial Ejecta Blanket Extent of Ray and Halo Craters on Ganymede

Figure 4 displays the mean ejecta blanket radius normalized to the crater radius (R_A) versus the standard deviation (S). The mean ejecta blanket radius normalized to the crater radius (R_A) provides a measure of the average extent of the ejecta blanket relative to the size of the crater. This normalized value allows for a comparison of ejecta blanket extents across craters of different sizes. R_A is calculated as follows: 8 $${R}_{A=}\frac{\left(\frac{{R}_{\min }}{{R}_{\text{cr}}}+\frac{{R}_{\max }}{{R}_{\text{cr}}}\,\right)}{2}$$ where, R_min is the minimum radial extent of the ejecta blanket, R_max is the maximum radial extent of the ejecta blanket, R_cr is the radius of the crater.

[IMAGE OMITTED. SEE PDF]

Standard deviation (S) considers the normalized differences between the maximum and minimum radial extents and the mean radial extent of the ejecta blanket, providing a measure of how much these extents deviate from the mean. It is calculated as follows: 9 $$S=\sqrt{{\left(\frac{{R}_{\max }}{{R}_{\text{cr}}}-{R}_{A}\right)}^{2}+{\left(\frac{{R}_{\min }}{{R}_{\text{cr}}}-{R}_{A}\right)}^{2}}$$

The resulting graph (Figure 4), calculated based on Equation 8 and Equation 9, shows that different crater types can be distinctly separated when plotted R_A versus S. DHCs, including Nergal (N) and Khensu (K), generally exhibit the smallest R_A values (<5) and standard deviations (∼1). Conversely, DRCs, including Mir (M), Antum (A), and Kittu (K) exhibit the largest R_A values (>15 km) and standard deviations (>10), possibly due to different material properties of the dark non-ice material compared to bright ice. The rays of BRCs exhibit intermediate values for R_A and S. Intriguingly, Tammuz (BDRC), whose ejecta are composed of both dark and bright portions, has an R_A value similar to that of other BRCs. The same accounts also for the special group of BRCs such as Enkidu (E) and Melkart (M) emplaced at the border between dark and LTs on Ganymede.

Individual Craters

Several craters have been observed during the Voyager and Galileo missions with sufficient resolution for analysis of their characteristics in detail. In the following, we present the results of the detailed geologic mapping of these craters, which include various dark ray and halo craters in addition to BRCs emplaced at the border between dark and LT.

Dark Ray Craters (DCRs)

Antum (∼5.5°N/∼141.1°E)

Antum, a ∼15 km diameter large DRC is positioned close to the equator of Ganymede's trailing hemisphere at ∼5.5°N/∼141.1°E (Figure 2). The maximum excavation depth reaches up to ∼1.2 km (Equation 2), 1.5 km (z = 3) or 2.31 km (z = 4) (Table 1). Antum was only imaged during the Voyager mission with ∼2 km/px, but observed at relatively high resolution by Galileo NIMS (2.1 km/px). Antum is entirely situated within the extensive dark terrain of Marius Regio. The main geological units observed are a crater with a distinct floor and rim, bright continuous ejecta, and extended dark rays (Figures 5a and 5b). The R_A value for the dark ray materials (discontinuous ejecta) is about 18.7 (Table 1). The crater interior and its outer rim appear predominantly bright. Due to low resolution, it is unclear whether a peak or pit is present. The dark rays represent the discontinuous ejecta and are presumably a very thin layer of dark material on top of the dark terrain of Marius Regio.

[IMAGE OMITTED. SEE PDF]

The pronounced morphology and albedo variations compared to the surroundings as seen in the camera images imply that Antum is a relatively fresh impact crater. This is supported by the BD map of the water ice absorption at 1.5 μm derived from the NIMS observation with the enrichment of water ice in the crater and dark non-ice material in the ray material (Figures 5c and 5d). The distribution of the concentrated dark non-ice material matches the dark rays seen in the camera image. BDRs in the Antum area indicate water ice grain sizes between 1 and 2 mm, which corresponds to the global grain size variations depending on the surface temperature on Ganymede as presented in Stephan et al. (2020). This, contrary to the visual albedo, implies that the water ice grain sizes of the crater as well as the ejecta material are already more or less adapted to the local surface temperature (Stephan et al., 2020). Slightly smaller grain sizes in the crater could likely result from a colder crater in comparison to larger grains in the area of the warmer darker regions (Pappalardo et al., 2004).

Mir (∼2.9°S/129.7°E)

Mir, a ∼8 km diameter DRC, is positioned on the trailing hemisphere at ∼2.9°S/129.7°E in the southwest of Antum (Figure 2). The maximum excavation depth of Mir could be estimated to be ∼0.7 km, 0.87 km or 1.34 km, depending on the method applied (Table 1). Our analysis utilized an available image resolution of ∼2 km/px and additional Galileo NIMS data acquired at 3.4 km/px. The main geological units identified are a bright crater with a distinct rim and floor and dark rays making up the discontinuous ejecta (Figures 6a and 6b). Similar to Antum, the low resolution prevents to identify any peak or pit located in the crater center. The dark rays cross a relatively narrow band of LT in the area and have R_A value of about 49.12 (Table 1).

[IMAGE OMITTED. SEE PDF]

Furthermore, the dark rays appear to be asymmetrically distributed and extend farther to the west than to the east, suggesting an oblique (30°–40°) impact from east. Also, two neighboring craters, with one superimposed onto the dark rays of Mir, might have contributed to the dark ejecta. The BD map of the water ice absorption at 1.5 μm derived from the NIMS data support that the ejecta concentrates downrange (west of the crater) with more non-ice material than the surrounding dark terrain, whereas the craters themselves are richer in water ice (Figures 6c and 6d). Similar to Antum, the BDR variations indicate water ice grains sizes between 1 and 2 mm, which correspond to the global variations in grain sizes on Ganymede related to the surface temperature, that is, relatively large grains close to the equator (Stephan et al., 2020). Only the dark ejecta exhibit slightly smaller water ice grain sizes. Following Hibbitts (2023) the dark ejecta of Mir is composed of a dark but hydrated non-ice material mixed with a less hydrated carbon-rich compound typically found on Callisto. Since for Antum, however, a similar mixture is assumed, it is difficult to explain the BDR variations related to an effect by the dark surface compound. Presumably, this can be explained by a very recent impact event with smaller grained water ice mixed with the dark material, with the recently emplaced ejecta material still being slightly colder than the older surroundings in the dark terrain of Marius Regio.

Kittu (0.5°N/25°E)

Kittu, a 15 km diameter DRC, is located on Ganymede's trailing hemisphere at ∼0.5°N/∼ 25°E within two distinct types of extended light grooved terrains (Mysia and Harpagia Sulci) (Figure 2). It is likely that the dark ejected material originated from a depth of ∼1.2, 1.52, or 2.35 km, depending on the applied method (Table 1). The R_A value for the dark ejecta blanket of Kittu is about 31 (Table 1). Kittu is one of the rare surface features on Ganymede that have been imaged at high resolution by Galileo SSI as well as NIMS. Therefore, we were able to use SSI image and NIMS data with resolutions of 145 and 279 and ∼3 km/px, respectively for our analysis.

The main geological units observed in our study comprise a broad central peak, a bright rim and floor, a continuous bright ejecta blanket extending about one crater radius beyond the crater rim, and discontinuous ejecta of extensive dark rays (Figures 7a–7d). Kittu displays a butterfly-shaped ejecta blanket comprised of dark ejecta material, suggesting a low-angle impact from east to west (Pierazzo & Melosh, 2000). A forbidden zone uprange, along with ejecta deposition cross-range and downrange, are characteristic features. Notably, the polygonal outline of the crater is governed by the orientation of grooves (Baby, Kenkmann, Stephan, & Wagner, 2024). It cannot be excluded that the emplacement of the crater itself within a band of grooves affected the configuration, distribution and orientation of the dark rays.

[IMAGE OMITTED. SEE PDF]

The NIMS derived BD map covers the Southern portion of Kittu and supports the dark non-ice material dominating the dark rays and an icier crater as well as icier surroundings (Figures 7e and 7f). Similar to Antum and Mir, the NIMS derived BDR map indicates water ice grain sizes that fit to the global trend with relatively large water ice grains expected close to the equator. Similar to Antum, the grain sizes are slightly larger, where dark ejecta cover the surface. Smaller grain sizes are found in the surrounding LT and within the crater. This could be explained by slightly different surface temperatures caused by the presumably warmer dark material (Pappalardo et al., 2004). This means that the water ice grain sizes of Kittu and its ejecta have already adapted to the temperature environment.

Half Bright and Dark Ray Crater (BDRC) Tammuz (∼13.9°N/∼129.1°E)

Tammuz is a ∼51 km diameter BDRC, located in the North of Antum and Mir at ∼13.9°N/∼129.1°E in Ganymede's trailing hemisphere (Figure 2). The maximum excavation depth is estimated to be ∼3.6, 4.51, or 6.94 km (Table 1). It exhibits an asymmetrical distribution of its ejecta. Half of the ejecta is dark, while the other half is bright, both radiating in opposite directions. The R_A value for the ejecta blanket of Tammuz is about 11.9 (Table 1). Like Antum and Mir, Tammuz was also imaged only during the Voyager mission with a resolution of 2 km/px, but observed by NIMS with 6.2 km/px.

The main geological units observed for Tammuz include: a bright crater floor and rim (constituting a major portion of the crater floor and rim), a dark crater floor and rim (constituting only a minor portion of the crater floor and rim), bright continuous ejecta, dark continuous ejecta, bright discontinuous ejecta, and dark discontinuous ejecta (Figure 8). The bright ejecta is predominantly located to the north, stretching from slightly west to slightly east of due North, while the dark ejecta is predominantly situated to the south, extending from slightly west to slightly east of due South. Due to the low resolution, it is challenging to precisely determine the exact position of the dual-colored ejecta. However, the NIMS derived BD map confirms the change in albedo related to the abundance in water ice/dark non-ice material with respect to the ejecta and the crater itself (Figure 8c) and between the Northern brighter and Southern darker ejecta. The NIMS derived BDR map (Figure 8d) shows a similar pattern as in case of Antum and Kittu. The estimated water ice grain sizes again fit to the global trend and are generally slightly smaller than observed for the other craters caused by location of Tammuz farther away from the equator. From the given data, it is not clear if Tammuz was emplaced within the dark terrain of Marius Regio or the LT of Tiamat Sulcus, which crosses the area north of Tammuz from Southeast to Northwest. Possibly, a piece of dark terrain related to Marius Regio reaches up to the southern crater rim of Tammuz, but is covered by its dark ejecta. This is also indicated in the global geologic map of Ganymede by Collins et al. (2013) favoring a heterogeneous target area, that is most likely composed of light ice in the northeastern half and dark ice in the southwestern half.

[IMAGE OMITTED. SEE PDF]

Dark Halo Craters (DHCs)

Khensu (∼1.04°N/207.1°E)

Khensu, a 17 km diameter DHC, features a central dark material region encircled by lighter materials. Positioned on the leading hemisphere at ∼1.04°N/∼207.1°E, it is situated close to the equator. Khensu lies within the light grooved terrain of Uruk Sulcus but relatively close to the border of the dark terrain of Galileo Regio in the North (Figure 2). The presence of double layer ejecta on a body like Ganymede indicates that an atmosphere is not required for the fluidization of ejecta materials during the impact event which are usually assumed to be the cause of halos (Boyce et al., 2010). The R_A value for the ejecta blanket of Khensu is about 2.8 (Table 1).For our analysis, one of the Galileo SSI images acquired at high resolution of 220 m/px is available. Unfortunately, no NIMS data exist for this unique feature. The excavation depth of Khensu has been estimated to be less than 1.36, 1.7, or 2.57 km (Table 1).

The main geological units related to Khensu are the peak, dark crater floor, bright crater rim, dark continuous ejecta including some even darker spots, and bright continuous ejecta (Figures 9a and 9b). Geomorphological analysis indicates the presence of a forbidden zone to the north of the crater, and we observed an off-centered positioning of the peak toward the northwestern side. Thus, an oblique impact from a northerly direction is inferred. The presence of the discontinuous ejecta is attributed to the influence of the topography, where ridges play a significant role in their formation, while other ejecta remain concealed within the grooves.

[IMAGE OMITTED. SEE PDF]

Nergal (∼38.8°N/∼160.1°E)

Nergal, a 9 km diameter DHC like Khensu, is positioned in the northern portion of Ganymede's trailing hemisphere at ∼38.8°N/∼160.1°E within the light grooved terrain of Byblus Sulcus (Figure 2). A high-resolution Galileo SSI imagery with 86 m/px is available. Similar to Khensu, Nergal was not observed by NIMS.

The four primary geological units identified in Nergal are the peak, rim and floor, a halo of dark material encircled by a halo of lighter material (Figures 9c and 9d). The peak exhibits a diameter of ∼2.5 km. Located just outside the crater's rim, within the dark ejecta region, is an adjacent crater measuring about 3 km in diameter. Both craters presumably representing a double impact (Baby et al., 2023). The dark ejecta layer is thinner compared to the bright ejecta. The R_A value for the ejecta blanket of Nergal is about 3.1 (Table 1).

From our calculations, the excavation depth of Nergal is estimated to be around 0.7, 0.97, or 1.49 km, depending on the applied method (Table 1). This suggests that both dark and bright materials originate from the near subsurface, with the dark material comprising the dark halo situating just beneath the LT surface. Consequently, the dark terrain is inferred to be significantly thinner, spanning only a few meters in thickness. The bright ejecta is observed to be concentrated solely on the impacted terrain, with no evidence of its presence on the neighboring dark terrain, furrows, or grooved terrain to the east. Additionally, a prominent ridge, likely broader and more elevated with an estimated thickness of ∼1 km is visible through both the dark and bright ejecta areas.

Bright Ray Craters (BRCs) Located at the Border Between Dark and Light Terrain

Enkidu (∼26.4°S/34.5°E)

Enkidu is a 122 km diameter large BRC and situated in the sub-Jovian hemisphere at ∼26.4°S/34.5°E (Figure 2). It shares the border between the dark terrain of Nicholson Regio and the LT of Harpagia Sulcus. Our analysis was conducted using the available image resolution of ∼2 km/pixel. Unfortunately, no NIMS data are available to study this unique crater.

The main units identified are as follows: a central bright dome (diameter of ∼31 km) of pancake shape almost filling a deeper nested crater, an outer flat floored crater that is terminated by the crater rim that forms a steep escarpment (Figure 10). The bright ejecta blanket expands ∼9 crater radii in the LT (Table 1). The rays that superpose the dark terrain are darker and less visible. The entire crater floor is composed of bright material, although the brightness gradually diminishes toward dark terrain side.

[IMAGE OMITTED. SEE PDF]

We also conducted mapping of three smaller craters superimposed on Enkidu's ejecta deposits. The nearest two neighboring craters feature a dark crater floor and dark ejecta, and have diameters up to ∼11 and 7 km implying excavation depths of ∼1, 1.16, or 1.78 km and ∼0.62, 0.78, or 1.20 km (Table 1). One is situated within Enkidu's discontinuous ejecta close to the border between dark and LT, with the majority of it located in the dark terrain. Its crater floor appears dark (Figures 10a and 10b). Presumably, the impact did not reach into the icy subsurface of Nicholson Regio. The second crater, located within Enkidu crater and close to its western rim, also features a dark floor and dark continuous ejecta (Figures 10a and 10b). Given the estimated excavation depth of 7.8, 9.75, or 15.01 km (Table 1) it is expected that Enkidu completely impacted into the icy subsurface of the dark terrain of Nicholson Regio excavating bright ejecta material. Possibly, dark material from Nicholson Regio were later transported in this portion of the crater during crater modification with the small crater superimposed on it. A third neighboring crater, with a diameter of ∼7 km, is entirely situated in the LT and does not show any sign of darker ejecta (Figures 10a and 10b). Its estimated excavation depth is ∼0.6 km, where during this impact event likely only bright material became excavated from the LT subsurface.

Melkart (∼9.7°S/173.9°E)

The 103 km diameter large BRC, Melkart, is located at ∼9.7°S/173.9°E. Like Enkidu, it is positioned at the border between the older dark terrain, Marius Regio, and the younger light grooved terrain situated between Tiamat Sulcus and Sippar Sulcus (Figure 2). Melkart is one of the rare areas on Ganymede that have been observed during the Galileo mission by the SSI camera system as well as NIMS at high resolution (Stephan et al., 2008). Our analysis was conducted on the available image resolution of less than 200 m/px. The associated NIMS observation exhibits a resolution of 3.3 km/px.

Melkart's extended ejecta rays are emplaced on dark and LT (Figures 11a and 11b). The off-centered location of the pit-dome suggests an impact direction from SSW to NNE (Lucchetti et al., 2023). The eastern border between the dark and LT is clearly visible crossing the crater right of the central dome. On the contrary, the western border between these two terrain types does not show up within the crater and is probably hidden by the extensive ejecta blanket of Melkart in this part. The R_A value for the ejecta blanket of Melkart is about 7.1 (Table 1).

[IMAGE OMITTED. SEE PDF]

The NIMS derived BD and BDR map indicate a slight change in abundance as well as water ice grain size between the portions of the crater located within the dark and LT (Figures 11c and 11d). The border, however, does not fit exactly the border between these two terrains. The darker or less icy material is only concentrated along the inner crater rim. The estimated water ice grain sizes correspond to the global variations with smaller water ice grains corresponding to the location of Melkart at higher latitudes compared to DRCs observed by NIMS (Stephan et al., 2020). Locally, slightly larger ice grains are associated with warmer darker regions. This fits to the pattern seen for the DRCs Antum, Tammuz and Kittu and thus support no differences in the thermal behavior appearing between these dark materials with a possible implication also for a similar composition of these dark materials.

The excavation depth of Melkart can be estimated to be ∼6.7, 8.39, or 12.92 km (Table 1). Thus, it is expected that Melkart impacted into the subsurface of dark and LT. Stephan et al. (2008) indicated that Melkart is not a very young crater (∼3.6–3.9 Ga based on LDM). Therefore, the dark material within the crater could have originated from the dark terrain of Marius Regio, either during the crater modification phase or through subsequent weathering processes.

The western border between the light and dark terrain is hidden below the extensive ejecta blanket of Melkart. Since no indication of this border can be seen inside the crater, it is assumed that the border is located outside the crater. However, similar processes resulting in the redeposition of dark material within the crater could also account for small amounts of dark material in the western portion of the crater.

Discussion

We use the simplified analytical approach of Maxwell's Z-model (Maxwell, 1977) to obtain information of a possible vertical stratigraphy of the subsurface. As a precautionary note, this model cannot take into account effects of oblique impacts with uneven ballistic trajectories in up range and downrange directions. It also neglects excavation complexities that arise from a layered target stratigraphy with uneven rheologies. Moreover, our approach does not take into account that the albedo of rays and halos is influenced by mixing with surficial material and ejecta from secondary craters (Elliott et al., 2018). The observed rays usually have the same albedo as the continuous distal ejecta, where secondary cratering should not play an important role. Furthermore, we observe that individual rays superpose dark and LTs and do not change their albedo. This suggests that their albedo is governed by composition of the excavated primary crater rather by the admixture of near-surface material from secondaries. From this we conclude that the effect of secondaries cannot be so strong that it obliterates the compositional signature. This observation goes along with Melosh (1980) who stated that the total amount of material that produces normal distal secondary craters is typically less than one percent of the excavated volume of the primary crater.

In the following, we discuss the implications of the observed ejecta pattern (Figure 12) with respect to the target stratification assuming that the target is composed of parallel layers. Basic principles for interpretation are (a) that distal ejecta is emanating from a central, deeper source and proximal ejecta is sourced from a shallower position near the edge of the transient crater and (b) and that an inverted (upside down) stratigraphy exists in the ejecta blanket with respect to the target stratification.

[IMAGE OMITTED. SEE PDF]

Formation Scenarios for Ray and Halo Crater Types on Ganymede

Figure 12a illustrates the stages involved in the formation of DRCs on dark terrain such as represented by Antum and Mir, where both of them have bright crater interiors (Sections 4.1.1 and 4.1.2). To explain dark non-ice dominated ejecta, a top stratigraphic unit of a dark ice is required to exist before the projectile impacts the surface. During the excavation stage, a transient cavity forms, affecting both the dark terrain and the bright ice layer underneath. As the entire ejecta blanket is dark the original layer should have a thickness corresponding to the depth of the excavation zone. The crater interior is largely bright because of a bright ice layer underneath the dark layer. The subsequent modification stage, caused the bright icy materials of the lower part of the transient cavity to move upward to form a central uplift. The measurements suggest that Antum excavated material from a maximum depth of ∼1.2, 1.50, or 2.31 km, while Mir reached a depth of ∼0.7, 0.87, or 1.34 km. Therefore, the thickness of the dark layer in both areas probably below ∼2 km. The admixture of surficial material to the ejecta by ballistic sedimentation (Oberbeck, 1975) and secondary cratering would not alter the albedo of the ejecta.

Figure 12b illustrates the formation of a DRC on LT, specifically as expected for Kittu (Section 4.1.3). Kittu formed in the LT and consequently forms an inner bright halo. The bright layer, however, is not thick so that the more distal ejecta is governed by dark ice that is excavated from a source region of dark material that is closer to the point of impact and deeper seated. The interior of the crater is bright because the dark layer is underlain by bright ice that forms large parts of the displaced zone. This material is uplifted upon crater collapse and central uplift formation developing the pronounced bright icy central peak. Both, the dark and the bright ejecta materials originate from a maximum depth of ∼1.2, 1.52, or 2.35 km below the surface for Kittu with the bright ice layer superposing the dark ice layer. Therefore, it can be inferred that the combined thickness of the two stratigraphic layers here, the light grooved and dark terrain materials, should be less than the depth mentioned above.

Figure 12c illustrates the stages of forming a DHC in Ganymede's LT such as observed for the two small craters Nergal and Khensu (Section 4.3). These craters exhibit distinct features, including bright outer ejecta, dark ejecta surrounding the crater rim, a dark crater floor, and a bright peak. This indicates that the subsurface in this area is extremely heterogeneous, with alternating layers of bright and dark ice. During the excavation stage, a transient cavity forms, affecting all of these layers. Initially, before the projectile impacts the surface, the top stratigraphic unit consists of a light grooved terrain. However, the bright ice layer should be so thin that the formation of a continuous ejecta blanket is suppressed. Instead, the dark inner halo is formed by the dark layer underneath. The dark layer may have a thickness of roughly half the maximum excavation depth of the specific crater, because the dark halo is surrounded by a halo of bright ejecta. This bright outer ejecta is formed due to a third, bright target layer considering that the outer ejecta is formed from a zone deeper sourced and closer to the point of impact. The two lowest layers 4 and 5 of the target are necessary to explain the dark crater interior and the bright tip of the central uplift. The transition from bright to dark layers is responsible for the dark material inside the crater. In the final modification stage, the bright icy materials from the lowest portions of the transient cavity move upward to form a central uplift. The alternating layers of bright ice and dark ice, have a combined depth of ∼1.36, 1.70, or 2.57 km for Khensu and ∼0.77, 0.97, or 1.49 km for Nergal.

Large impacts such as Melkart and Enkidu excavate deep into the icy crust and do not show a clear variation in ejecta brightness thus suggesting a simple crustal structure dominated by bright ice (Section 4.4). During subsequent processes such crater modification and later erosional processes due to sublimation and impact gardening some redistribution of dark material from the surrounding dark regions could be expected. Nevertheless, small craters such as the small dark craters superimposed within and outside of Enkidu's crater rim both featuring dark floors and ejecta, show very well the influence of the dark terrain material and resemble impacts such as Antum and Mir but with the differences that these craters do not excavate deep enough to reach the ice underneath. Therefore, thickness of the dark terrain of Marius Regio should not exceed 2 km as inferred from the excavation depth analysis of these craters.

If the target layers are not complete then the ejecta blanket should be somehow an image of it and is also not complete in one sector or the other. So, it is assumed that the impact resulting in Tammuz, exhibiting both bright and dark ejecta, it may indeed have impacted along the boundary between a substantial section of the LT (an unresolved portion of Tiamat Sulcus) and a smaller area of the dark terrain linked to Marius Regio, which is highly disrupted in this region (Section 4.2). The estimated excavation depth is ∼3.6, 4.51, or 6.94 km, allowing it to penetrate more deeply into the bright icy layer compared to Antum and Mir. However, higher-resolution imagery is required to definitively confirm the exact location and characterization of the crater and its ejecta pattern together with its geologic context.

The great variability of different halo and ray ejecta pattern reflects the heterogeneous structure of the ice crust of Ganymede. All pattern can be associated to a layered structure and the thickness of the variable layers can be roughly deduced. According to Senft and Stewart (2008), the ejection angles are much steeper in icy targets compared to homogeneous rocky targets.

For large ray craters exceeding 25 km in diameter, elliptic comets have proven less significant than previously assumed, while isotropic comets may hold a more pivotal role on Ganymede than previously believed (Xu et al., 2017). The distinct distribution of these craters across the two terrain types stems from preferential thermal sublimation on the dark terrain (Xu et al., 2017). Moreover, a considerable number of smaller craters are expected to exist on Ganymede, but current resolutions are inadequate in certain areas. Additionally, the distribution disparity between large and small ray craters (with diameters ranging from 10 to 25 km) indicates that the rays of smaller craters are more susceptible to erasure by surface modification processes, such as micrometeorite gardening (Xu et al., 2017).

Implications for the Structure and Formation Process of the Dark Terrain

Based on the furrows observed in the dark terrain of Galileo Regio, an estimated elastic thickness of ∼0.5 km is suggested for this region (Nimmo & Pappalardo, 2004). Studies by Murchie et al. (1990) indicate a global average thickness for dark terrain ranging from 3 to 8 km, while McKinnon and Parmentier (1986) have estimated a thickness of 5 km for dark terrain. Models presented by Golombek and Banerdt (1986) relate the width and spacing of furrows suggesting a lithosphere thickness of 5–10 km at the time of furrow formation. However, when examining the relationship between crater rim height and crater diameter in dark terrains such as Nicholson Regio, Marius Regio, and Galileo Regio, it has been found that the thickness of dark terrain in these regions is ∼1 km or less (Murchie et al., 1988). Even more, Prockter et al. (1998, 2000) proposed that formation of dark terrain material is thought to result from the concentration of admixed dark meteoritic material on the originally icy surface through processes such as sublimation, impact cratering, impact volatilization, and mass wasting.

Our analysis supports a relatively thin uppermost layer of dark material on top of an ice-rich substrate, because the depth of excavation for the observed small craters located in the dark terrain with dark non-ice ejecta but an icy crater such as Antum and Mir suggests the thickness of the dark layer not exceeding 2 km. This holds true, particularly within Marius Regio. But also, in other regions of the dark terrain, such as Nicholson Regio, the dark terrain thickness is estimated to be not more than ∼2 km thick based on the excavation depth measurements of the small craters superimposed on Enkidu. Therefore, our analysis supports the findings of Prockter et al. (1998) with the dark terrain composed of a thin surface layer of highly concentrated dark non icy material on top of an ice rich substrate.

Implications for the Light Terrain Formation

Large impact craters such as Melkart and Enkidu that impact into deeper portions of Ganymede's crust at the border between dark and LT imply that at these depths, that is, below of the dark surface layer in the dark terrain, the composition of the material in both terrains do not show significant compositional differences and indicate an ice-rich substrate for both terrains. Nevertheless, in some regions within the LT, such as the areas around DRC like Kittu, and DHCs like Nergal and Khensu, the presence of dark material in the subsurface is needed to explain the observed ejecta pattern and implies regional heterogeneities in the subsurface composition of the LT.

In regions of LT, such as where Kittu is present, the thickness of the dark layer in the subsurface remains less than or equal to ∼2.3 km and is similar to what could be estimated for the dark terrain of Marius and Nicholson Regio. Furthermore, in areas where Khensu is present, the excavated dark material originates from a maximum depth of ∼2.5 km, suggesting that this ejected material comes from three distinct stratigraphic layers: the upper LT, the intermediate dark terrain, and a lower bright ice layer. Similarly, in areas where Nergal is present, the excavated dark material originates from a maximum depth of ∼1.5 km, with the same three stratigraphic layers, indicating their limited thicknesses.

The inferred vertical stratification may also shed some light on the processes that formed the LT. Previous studies favor two major processes responsible for the LT: tectonic rifting and spreading as discussed in Pizzi et al. (2017) and the studies presented in Pappalardo et al. (2004), which differ in the point if the LT presents resurfaced dark terrain or newly formed crust by material from the subsurface. The sketch of Figure 13 illustrates the difference between spreading and rifting and the expected ejecta pattern when craters form in these LTs.

[IMAGE OMITTED. SEE PDF]

Existing dark material in the subsurface such as observed for Khensu, Nergal and Kittu favors at least partly resurfaced dark terrain in these areas. Possibly, downward movement of the dark terrain material due to downfaulting could explain the existence of dark material in the subsurface. This would imply that rifting plays a significant role in the formation of LTs on Ganymede in these locations (Figure 13b). The depth of downward faulting should roughly match the excavation depth, which corresponds to the depth from which ejecta material originates. For near-subsurface dark material, excavation depth measurements are down to 1 km and should align with the depth of downward faulting. Cameron et al. (2019) propose the possibility of Coulomb failure occurring at depths of approximately 1 km for high friction (µ_f = 0.6) cases and up to 2 km for low friction (µ_f = 0.2) cases in most fault zones. The extent of concentrated dark material in the subsurface should at least correspond to the extent of the developed crater diameter. The sandwiched portion of dark material in the subsurface should measure at least a few hundred meters in thickness in order to match the observed pattern of dark material within the crater and the ejecta (Figure 12). Particularly, Patel et al. (1999) described that the formation of the LT was started by means of extensional necking of the lithosphere causing fault systems of longer topographic wavelengths. This allows larger portions of dark terrain to be transported into the depths and then reworked by multiple shorter wavelengths formed by normal faulting of the brittle lithosphere.

It should be noted that the studied craters lie relatively close to the neighboring dark terrains of Galileo and Marius Regio, as in the case of Nergal and Khensu, and Nicholson Regio in the case of Kittu, where Kittu is surrounded by numerous pieces of dark terrain. The uppermost icy layer in these regions is expected to be only a few meters thick and thus thinner than the underlying dark terrain material. Therefore, it cannot be fully excluded that some initial overflooding of dark terrain by icy material in the early phases of grooved terrain formation took place, which could also explain this observation.

Finally, dark material re-excavated from the subsurface of a tectonically resurfaced region might slightly differ with respect to the composition from dark material excavated from the strongly weathered uppermost dark layer in the dark terrain and might help to explain the compositional differences in the dark material discussed in Hibbitts (2023).

Conclusion

We conducted a study of the ejecta blankets from 36 impact craters on Ganymede, ranging in size from 7 to 135 km in diameter. Our goal was to analyze the composition, color, and distribution of the ejecta blankets in order to decipher the stratigraphy of the icy subsurface in various light and dark terrains of Ganymede. The main conclusions of this study are:

• The ejecta of ice craters on Ganymede and the crater interiors are sensitive tools to probe the vertical stratification of the ice crust.

• The investigated craters show that the ice crust of Ganymede is laterally heterogeneous.

• DRCs and DHCs are distinguishable by their mean ejecta radial extent and standard deviations, with DRCs having the largest and DHCs the smallest values, while BRCs and BDRC fall in between.

• DRCs in the dark terrain allow to estimate the thickness of the dark terrain. The estimated thickness of dark terrain at Marius Regio and Nicholson region likely do not exceed a thickness of ∼2 km.

• In the LT where Kittu is located, the dark terrain material originated from a depth less than ∼2.3 km. The uppermost LT and near subsurface dark terrain should be only some meters thick.

• A very thin uppermost ice layer on top of a dark layer in the LT could imply the possibility of flooding of older dark terrain by an icy substrate.

• The DHCs Khensu and Nergal suggest that locally the ice crust is composed of multiple dark and light ice layers. This has important implications for the geological history of Ganymede. In regions where Khensu is present, the excavated dark material originates from depths less than ∼2.5 km. In areas with Nergal, the excavated dark material comes from depths less than ∼1.5 km. As a result, the topmost LT and the dark terrain underneath it is only a few hundred meters thick.

• The large DHC Humbaba (Table 1) suggests that dark terrain material can be excavated from a maximum depth of 3–5.6 km depth. Craters formed at geological boundaries indicate that these boundaries are not surficial features but extend into the crust. The BRCs Melkart and Enkidu, formed at these borders with large excavation depths, indicate that they penetrate deep enough to reach underlying bright ice, implying that the overlying dark terrain is relatively thin.

• The study documents that the albedo of halos and rays is a result of compositional differences that reflects heterogeneities of the source craters. Surficial admixture, ballistic sedimentation and secondary cratering during ejecta emplacement seem not to be capable to obliterate this signature.

Outlook

In this study, only a small number of medium and larger craters could be investigated as the availability of higher resolution data is limited. However, most of these craters are small and not covered with images offering sufficient spatial resolution to analyze them in detail. In order to verify our results and prove the role of secondary cratering and admixture of surface and near-surface substrate, particularly concerning the origin of dark ejecta deposits in the LT, highly resolved data covering more of Ganymede's surface and its unique features are needed. Better coverage of Ganymede by high-resolution images and spectral data, such as that presented by Hibbitts (2023), will enable the study of existing heterogeneities in the subsurface across Ganymede. This will also reveal more details about the processes responsible for the formation of dark and LT. The JUICE mission with its unique payload such as the Jovis Amorum ac Natorum Undique Scrutator (JANUS) camera (Della Corte et al., 2019; Palumbo et al., 2014) and Moons and Jupiter Imaging Spectrometer (MAJIS) spectrometer experiment (Langevin & Piccioni, 2017; Piccioni et al., 2019; Poulet et al., 2024) will enable to investigate these craters and more with an unprecedented resolution (Grasset et al., 2013; Stephan et al., 2021). The planned complete coverage of Ganymede's surface with a JANUS image resolution of 70 m during the GCO 5000 mission phase should make it possible to identify and define the distribution of these features and their association to major geologic terrain types. Particularly, the combination with color information given by the different filters of the JANUS camera in the visible and near-infrared light will help to avoid misinterpretations of different terrain types due to illumination conditions. Selected regions could be observed with a resolution of 7 m, when JUICE is orbiting the moon at relatively low altitude (GCO500 mission phase) and thus be analyzed in highest detail. Even more, joint observations of JANUS together with MAJIS will study the composition of the dark material and the surface ice properties. This could reveal differences in the chemical or physical properties of dark ejecta deposits. These deposits may originate from one of two sources. First, they could come from dark surface material that was excavated from strongly weathered dark terrain. Second, they might come from dark terrain that was resurfaced during the formation of LT and then excavated from beneath the LT. As mentioned above, since most of these craters are quite small, an exact observation pointing will be required for JANUS observations alone and joint observations with MAJIS. Even more, observations by the radar sounder RIME (Radar for Icy Moons Exploration) instrument (Bruzzone & Croci, 2019; Bruzzone et al., 2013) can verify subsurface characteristics particularly in the LT and topographic information by stereo imaging of JANUS and GALA (GAnymede Laser Altimeter) observations (Hussmann et al., 2013, 2019). The combination of these data can finally prove not only the compositional information about the dark material(s) existing on Ganymede's surface but also the processes responsible for the existence of these dark materials within relatively fresh crater and ejecta materials. Therefore, we would strongly support to push the planning of observing craters of the discussed types as regions of interests for future high-resolution observations.

Acknowledgments

This project is financed by the German Aerospace Center DLR, Grant 50QJ2403. N. R. B. also acknowledges the financial support of the DLR-DAAD PhD fellowship from the German Aerospace Center and the German Academic Exchange Service in the early phase of the project. T. K. and K. S. are co-investigators of JANUS and acknowledge the support by the JANUS team and PI Pasquale Palumbo of INAF-IAPS. We thank Kosuke Kurosawa and an anonymous reviewer for their comments, which greatly contributed to the improvement of this manuscript.

Open Access funding enabled and organized by Projekt DEAL.

Data Availability Statement

All the data used in this work can be found in a repository following the FAIR principles (Baby, Kenkmann, Stephan, Wagner, et al., 2024).

References

Baby, N. R., Kenkmann, T., Stephan, K., & Wagner, R. (2024). Polygonal impact craters on Ganymede. Meteoritics & Planetary Sciences, 59(3), 544–559. [DOI: https://dx.doi.org/10.1111/maps.14138]