Introduction

The link of Martian remote sensing data with laboratory analysis of Martian meteorites can provide an effective method to broaden the understanding of the geology of the red planet by improving the quality of interpretation of remote sensing data and also setting constraints in investigating possible past or present life on the planet. Thus the identification and analysis of source areas (i.e., craters) of certain Martian meteorites is important.

Previous studies that investigated source rocks of the Martian meteorites have only targeted large (>10 km) and rayed craters on the surface of the planet (Mouginis-Mark et al. 1992; Tornabene et al. 2006; Hargitai and Kereszturi 2015), such as the Zunil crater in Cerberus plains (McEwen et al. 2005). An elliptical crater in the Syrtis Major region was suggested by Harvey and Hamilton (2005), as a source for the Nakhlite/Chassigny meteorites based on analysis of mineralogical information and age correlation. The 55 km Mojave crater was also suggested to be the source of shergottites by Werner et al. (2014a, b), although there is still an ongoing debate regarding this crater as a candidate source. The connection however between Mojave crater and shergottites is problematic as the crust there seems to be much older than the shergottite meteorites. For the NWA 7533 Martian meteorite the 6.9 km diameter, ~5 Ma old Gratteri crater (Wittmann et al. 2015) was proposed as a source.

Mineralogical, petrological, and textural properties of nakhlites, as well as evidence provided by secondary alteration, show that all these meteorites came from the same source rock unit (Changela and Bridges 2011). This evidence strongly suggests that the bolide impact ejected the nakhlites from a differentiated thick basaltic lava flow or a shallow sub-volcanic intrusion (Treiman 1993; Lentz et al. 2001). The solidification of the basaltic parent material occurred at about 1.2–1.4 Ga ago (Ganapathy and Anders 1969; Nyquist et al. 2001; Park et al. 2009; Cassata et al. 2010; Korochantseva et al. 2011), thus the source crater should be located in terrains of this age (estimated by crater-counting on Mars). Moreover, the ejection event occurred about 11 Ma ago, thus, the nakhlite source crater has to be of this age. These constraints a larger frame in which detailed investigations are performed in this study.

The current work adds to past efforts by investigating all craters with diameters above 3 km at lava plains with the corresponding formation age around 1.2–1.4 Ga. We have decided to include craters <10 km in diameter because computer simulations (Head et al. 2002), have shown that bolide impacts that produced craters with diameters down to 3 km on homogeneous and layered Martian terrains could have launched ejecta that escaped Martian gravitational field.

Spectral Properties of Meteorites and the Martian Surface

Spectral comparison of remote based Martian surface data and laboratory based data of Martian meteorites also provide input for source crater identification. Corresponding results are listed below according to meteorites. Analyzing the general spectral properties of meteorites and Martian lava flows, the composition of basalts in the Tharsis region somewhat resembles those of SNC’s, especially shergottites, however no good match was found (Lang et al. 2009). The work of Ody et al. (2015) compared Los Angeles, Shergotty, QUE 94201, the lherzolitic shergottite ALH A77005, Nakhla, Chassigny and the orthopyroxenite ALH 84001 to the NIR data of the Mars Express OMEGA detector. The results suggest basaltic shergottites are similar in NIR spectral properties to early Hesperian volcanic provinces (Ody et al. 2015). Spectral resemblance between ALH 77005 and the Martian surface is poor, mainly its spectra similar to olivine-bearing regions like Nili Fossae and Noachian/early Hesperian terrains. For Chassigny similarities to spectra from the Nili Patera caldera was identified, while for ALH 84001 areal analogies exist at northeast of Syrtis Major and northwest of Hellas basin, regarding the low-calcium-pyroxene-rich composition and old age (but not for the modal mineralogy). Unfortunately no convincing spectral analogue was found for the Amazonian-aged Nakhla meteorite. It is generally expected that dust cover significantly influences the remotely acquired spectra (Ody et al. 2015) and makes possible identification difficult.

Non SNC member NWA 7034, 7533 and 7455 Martian brecciated meteorites showed spectral signatures consistent with a near surface Noachian aged terrain. Among them NWA 7533 do not show strong mafic mineral absorptions, but is consistent with a large part of the Martian highlands’ spectra (Beck et al. 2015). Besides classical spectral analysis, gamma-ray spectroscopy could also be used to compare meteorite and Martian surface composition. In the case of NWA 7475 the comparison to gamma-ray spectrometer results onboard Mars Odyssey shows that the Fe and Ti abundances are compatible with the composition of the estimated average southern highlands in general. At this area a possible source crater was also identified: the 6.9 km diameter Gratteri crater, aged about 5 Ma old (Wittmann et al. 2015).

A recent fall, the Tissint meteorite (Chennaoui et al. 2012), shows petrographical similarities to shergotites. Its powder measured with visible and near infrared spectrometer were compared to CRISM based spectra of the pyroxene and melt rich outcrop of Alga crater on the Martian surface (Mustard et al. 2013).

Characteristics of Nakhlites

Analysis of nakhlite meteorites suggests that the target area on the Martian surface requires a basaltic substrate, which is iron rich and aluminium poor (Treiman1993; Treiman 2005). Mineralogically clinopyroxene is the most abundant mineral in nakhlites (70–80 vol%), the second most common is olivine (9–12 vol%), while there are also feldspars and some refractory minerals in the mesostasis (8–11 vol%, or up to 20 vol% in NWA nakhlites, Treiman 2005). Their cumulate texture suggests that the parent rock of the nakhlites was formed by slow cooling below the surface (Treiman2005); however, the mesostasis is composed of fine-grained material which indicates faster cooling, similar to those close to the surface of a lava flow. The distribution of trace and minor elements in augites suggests that Nakhla, Governador Valadares, Lafayette, Y000593 (Mikouchi et al. 2003) and some other nakhlites (Corrigan et al. 2013) might be samples from different horizons of the same lithologic unit (Wadhwa and Crozaz 1995), although late magmatic and subsolidus diffusion might cause changes to this primary zoning (Harvey and McSween 1992). Similarly, the degree of alteration of the different nakhlites is interpreted to provide evidence for their origin from different horizons of the same parent rock (Changela and Bridges 2011).

Microphotographs of the major secondary mineral assemblages found in Nakhla are shown in Fig. 1 (Chatzitheodoridis and Turner 1990; Chatzitheodoridis et al. 2014). Alteration minerals include evaporitic assemblages such as carbonates, sulfates, and halides (Chatzitheodoridis and Turner 1990; Gooding et al. 1991; Bridges and Grady 1999, 2000; Chatzitheodoridis et al. 2014), hydrous minerals such as iron-rich silicate rust (probably smectite, Gooding et al. 1991) and clays (Changela and Bridges 2011; Chatzitheodoridis et al. 2014), and finally, hematite and marcasite (Chatzitheodoridis et al. 2014). All alteration is pre-terrestrial and formed by moderate to very low temperature olivine-water interaction, where water is either igneous in origin or produced by melting of ice from the permafrost (Karlsson et al. 1992; Chatzitheodoridis et al. 2014). Alteration minerals are important because they indicate that the rock of the nakhlites has undergone at least two alteration events. Alteration was initiated by hydrothermal fluids (about 913 Ma ago), which locally melted the parent rock and reset the age according from radiogenic isotopes (Cassata et al. 2010), and which might have formed rhoenite and Al-rich clinopyroxene rims in Nakhla (Chatzitheodoridis et al. 2014). The second impact ejected nakhlites at 11 Ma ago (Ganapathy and Anders 1969; Eugster et al. 2002).

Fig. 1
figure 1

Examples of secondary minerals in Nakhla made with the electron microscope from thin sections (backscattered images a, c, and d), or from pristine surfaces of Nakhla grains (secondary electron images b and e). a Sulphate crystal (anhydrite) in mesostasis. b A leaf-shaped halite, grown on a pristine clinopyroxene surface from a freshly opened, untreated Nakhla grain, most probably of Martian origin. c The first reported siderite in a Nakhla mesostasis (Chatzitheodoridis and Turner 1990). d Ovoid in Nakhla mesostasis, made of nanocrystalline clay (reproduced image from Chatzitheodoridis et al. 2014) and various minerals, such as cumulus clinopyroxene crystals (cpx) with Al-rich rims, olivine (Ol), a pyrrhotite grain altered to marcasite, and acicular apatite crystals in mesostasis glass. The inset image in (d) is the ovoid shown under the transmitted optical microscope. e A sulfide grain grown on a pristine clinopyroxene grain from an untreated, freshly opened Nakhla grain, therefore most probably of Martian origin

Full size image

In summary, it is expected that the source crater from which the nakhlites have been ejected was formed on a lava plain. Lithologies from vertical strata up to several tens of meters depth were extracted, which contain both slow-cooled subsurface cumulus rock units and fast-cooled near-surface rock units. Overprinted alterations might have been potentially formed by a previous impact that caused partial melting and initiated hydrothermal systems, before 11 Ma ago an impact finally ejected the meteorites. Water abundance in the area might have been too low to produce morphological surface manifestations at the lava flow, but it was enough in volume and time, to leave behind the strong alteration signatures that are observed in nakhlites.

Methods

We used remote sensing datasets from Mars to search for candidate source craters of the nakhlite meteorites. We surveyed all volcanic units dated to 2.1—0.3 Ga (Hartmann and Neukum 2001) with the upper limit was between 2.1 and 1.4 and the lower limit at 0.6–0.3 because of uncertainties. These areas were outlined according to the mapped units based on the Geological map of Mars Map 3292 of scale 1:20,000,000 made by USGS (Tanaka et al. 2014), with overlying the borders of the corresponding aged surface areas with THEMIS image maps. Altogether 17.3 million km2 area was surveyed. Those areas are indicated by a white outline in Fig. 6. The mapping procedure was to:

  1. 1.

    select areas that are covered with lavas 2.1–0.3 Ga old,

  2. 2.

    survey those areas both on daytime and night-time THEMIS mosaics looking for the rayed craters (Christensen et al. 2003, 2004) with detailed analysed the candidate craters using THEMIS and CTX (Malin et al. 2007) data. THEMIS dataset was selected as a main source of data in this study because it enables the firm identification of fresh craters with diameters down to the km size. The comparison of the daytime and nighttime IR images of Mars clearly shows the albedo contrast between the ejecta and the surrounding area (Tornabene et al. 2006). Additionally, THEMIS data provide almost continuous coverage of the planet, although—as it turned out during this analysis—not all probable recent craters showed rays both in infrared and optical images (Piatek et al. 2014). CTX dataset was too detailed for the analysis of relatively large craters and the survey of large area. This step was done using the JMARS software (Gorelick 2003),

Basic Constraints

Based on crystallization age constraints of nakhlite meteorites (1.24 Ga ±0.01 by Gale et al. 1975 or 1.28 Ga ±0.06 by Nakamura et al. 1977) and taking into account the uncertainty of the crater density based surface age estimation (Hartmann and Neukum 2001; Hartmann 2005) we used the extended time interval to 2.1–0.3 Ga for the selection of the surveyed surface area. (However, based on isotopic dating of hydrothermal alteration produced minerals in the meteorites, the area where the nakhlites were ejected has undergone at least two alteration events.) Although lava flows of Amazonian age also exist at the Hesperia volcanic province in the southern hemisphere of Mars, the youngest lava flow that is present there is about 2.2 Ga (Lehmann et al. 2012) which is significantly older than the age of nakhlites crystallization. Another small area of young lava flows exists in the Syrtis Major region with an age younger than 1.4 Ga (Jodlowski et al. 2013). These areas were not included in this study.

Constraints for the crater size are influenced by the mechanical strength of the targeted terrains; i.e., a porous terrain requires larger craters. Based on model computations of impact process and particle ejection, the source craters of the Martian meteorites have to be larger than 3 km in diameter (Head et al. 2002), while earlier studies suggested that 12 km is a minimal crater diameter (Melosh 1985). Specifically for Nakhla, some models calculate a minimum of 15 km diameter (Warren 1994). In this survey of rayed craters on young Amazonian lava plains we have included craters from 0.5 km up to 100 km in diameter, to see how our findings fit to the expected crater production rate, and such comparison is more relevant if a wider range of crater size was taken into account.

The third requirement is the meteorite ejection age (source crater age) of about 11 Ma. Young craters are usually identified based on their pristine appearance and existence of easily erodible ejecta, i.e., features such as rays (Hawke et al. 2004), and containing few or no overprinting of small craters (Boyce and Garbeil 2007).

Context for the Potential Nakhla Target Crater

From the ejection age of Nakhla, the target crater should have formed at around 11 Ma ago, and the area where it is located should form around 1.2 Ga. It is therefore worth searching areas with a spatial density of craters indicating such an age. Figure 2 shows a crater size-frequency distribution diagram with isochrones (after Hartmann and Neukum 2001) to visualize the expected areal density of potential source crater. The candidate craters should show a population of superimposed craters following the 11 Ma. The spatial density of km sized craters (horizontal axis) should be about 10−6–10−7craters/km2. This confirms that within reasonable limits our methodology is valid. However, a few larger craters launched most ejecta (Gladman 1997), thus the best candidate craters should be the larger ones.

Fig. 2
figure 2

Crater size frequency distribution diagram displaying the expected areal occurrence of identified rayed craters (Table 1) craters (square in the centre) with threshold limit (the 3 km) to launch projectiles on interplanetary orbit, and other parameters (indicated as an arrow points to the right at the top of the diagram)

Full size image

Results

This section outlines the basic constrains of the survey, e.g., where the survey was realized and what type of craters we searched for. The identified rayed craters are listed in Table 1. Based on this, general and specific cases, including the best candidates, will be presented in the discussion.

Table 1 List of rayed craters identified within lava surfaces dated to (2.1–0.3Ga) in the Elysium and Tharsis regions. The appearance on daytime and nighttime THEMIS images was characterized by numbers between 0 and 3 (0:no rays or halo, 1:weak rays or weak halo, 2:>4 distinct rays, 3:>10 separated rays)
Full size table

The characteristic appearance of fresh craters targeted this study are shown in Fig. 3. A recently formed large crater is Mojave crater (327.0E 78.5 N), its age is estimated to be from 2 to 10 Ma old (age estimation was based on the craters’ spatial density on its ejecta; Werner et al. 2014a, b Fig. 3a, d). This 55–60 km diameter crater retained its ray system (Werner et al. 2014a, b). Several smaller craters in this area appeared very recently, i.e. during the time were images of Mars were recorded (Dauber et al. 2013). These very recent craters are not indicated in Fig. 3 as we only look for larger, km scale craters. Zunil crater (143.5E 7.4 N; Fig. 3b, e, McEwen et al. 2005) is also a relatively young (1.5–2.7 Ma) and large (10 km diameter) crater with a ray system, although less prominent than that of the Mojave crater (Preblich et al. 2007).

Fig. 3
figure 3

Example THEMIS IR daytime (top) and nighttime (bottom) images of three recently formed craters: (a, d) Mojave crater (58 km); (b, e) Zunil crater (10.4 km); and (c, f) an ‘unnamed’ very young small crater (about 2 km). Note that the young ejecta and rays cannot be observed in every case both on daytime and nighttime images. The craters sought in this work are larger than the ‘unnamed’ one of this figure. For comparison, all images are extracted from JMARS with the same pixel resolution. (THEMIS images are from the Planetary Data System)

Full size image

Craters that are formed during the acquisition of images from the Martian terrain can provide a good tool to estimate the impact rate, especially when it is combined with information on the erasing rates of the dark rays (Geissler et al. 2010). A good example can be found at the coordinates 246.434E, 16.969 (Fig. 3c, f), of which a detailed view is on the HiRISE PSP_002183_1970 image (Morgan and Murchie 2013). It shows radial rays only in the daytime IR image of the THEMIS detector (Fig. 3c). As in many cases, also in this example fresh impact craters are not erased by global dust storms for several years.

Ray Identification and Usage

Two important questions and issues regarding the rayed craters on Mars are their retention age, and the identification of crater rays. While obvious ray systems for km sized craters are relatively rare in the visual part of the spectrum (Tornabene 2005), using THEMIS images and careful inspection, a reasonable number of them could be identified especially in the small size range (McEwen et al. 2005). However there is an ongoing debate on which features should be classified as rays: the case of the largest ones is mostly evident, but it is more difficult to classify the smaller and less pronounced ones, especially as erosional agents make them fading with time. At certain locations rays could show specific appearance and regionally elevated concentration on infrared images like at the Cerberus region where many small (100–200 m) bright-rayed craters are observed (McEwen et al. 2005) beside the large ones (Malin et al. 2006). In this work we searched for candidate source craters that impact ejected the nakhlite meteorites, and listed craters including those with the less pronounced ray systems. The reason for this strategy is that most probably at certain regions rays are fading away faster (see the Discussion for the inhomogeneous spatial distribution of rayed craters).

In this work we add an effort to identify the features that appear as several radially elongated albedo stripes beyond the ejecta blanket (Hartmann et al. 2010). We noted great differences between certain ray systems (see Fig. 4). Although these central craters do not exhibit obvious rays, the rays are still observable (probably somewhat older than their more obvious counterparts), thus they also need to be considered during the analysis. Such candidate craters with less obvious ray system should require more detailed analysis with higher resolution images – that could be a next step after this research.

Fig. 4
figure 4

Two examples for the observable, although not obvious crater rays for two pedestal craters on optical (a, a1, a2) and infrared (b, b1, b2) THEMIS images. Inserts a1, a2, b1, b2 show magnified versions of locations indicated by boxes in a and b insets. Small arrows mark the radial streaks that form ray-like features occasionally to several crater radii from the central crater, the longer dashed arrows point to the direction of the central crater (which are outside the inset images’ area)

Full size image

Based on image analysis performed in this work, daytime IR images show the fresh ray system more frequently than the nighttime IR images. This is a good indicator of the young age, using the albedo contrast of the ejecta rays to reveal subsurface material, with the high albedo contrast both in the optical and the infrared region (Morgan and Murchie 2013). There is also a possible age and climate related connection that explains the relatively young ejection age of Martian meteorites. Based on Fritz et al. (2007) model, surface regions not covered with ice were especially favoured for impact ejection of the past ∼ 5 Myr after the last major glaciations on Mars (Fritz et al. 2007; Schon and Head 2012). A fraction of the impacted body’s energy is released in this upper icy unit. As impact ejects part of the topmost unit of the surface layer to interplanetary orbit, and if m thick ice-dust layer covers it, the fraction of the ejected part from this unit because its fragility and volatile content probably cannot recovered as meteorites on the Earth or other planetary body, decreasing the probability of having meteorites from ice covered area on Mars.

Candidate Craters

Table 1 describes the characteristics of those rayed craters that were identified during this survey. Here craters below 3 km diameter are also indicated to provide possibility to analyze their size frequency distribution in a wider range and fit to the model based expectations. The best candidates are larger than 3 km, are sharp-edged craters displaying obvious ray systems, both in the daytime and the nighttime images. Craters smaller than 3 km are indicated here, despite they are not likely candidates. Their distribution and characteristics however could be used as indicators for the general properties of recent impact scars on Mars.

A selection of interesting, but less “regular” craters is displayed in Fig. 5a–c. They show unusual “rays” as overprinted elongated topographic anomalies in addition to structures of different albedo. These “rays” are not real ejecta rays, but topographic structures that probably formed by selective erosion of material surrounding the crater, structures more resistant to erosion is shielding the material underneath and leaves behind raised platforms. The cemented/hardened nature of these rays can be seen in the nighttime images, where ejecta rays are almost invisible (except the craters on Fig. 5b, c). Such structures characteristics of older craters, and thus are not relevant candidates, thus were not included in the table.

Fig. 5
figure 5

Daytime (1, 4 columns) and nighttime (2, 5 columns) THEMIS image pairs of craters, and their morphological outline (3, 6 columns). These craters present characteristic features considered during this work. (a–c) Craters with ray like elongated parts of the ejecta blanket, formed by old remnants left behind selective erosion (see the text for details). (d–f) Craters with faint rays beyond a thick proximal (d) and fluidized ejecta blanket (e, f). (g–h) Examples of more abundant ray systems with various morphologies (for even higher resolution versions see also Fig. 4). (THEMIS images are from the Planetary Data System)

Full size image

Few pedestal craters also show weak rays (Fig. 5e, f) extending beyond the edge of the fluidized ejecta blanket. These craters are important targets for further analysis as they might hint to ice stored in the subsurface that produced the fluidized ejecta blanket.

The spatial distribution of the 53 identified candidate craters is visualized in Fig. 6. During the survey, 23 rayed craters were identified at the Elysium area (surveyed an area covering 3,505,573 km2) with areal density of 6 × 10−6 craters/km2, and 30 craters at the Tharsis region (at 13,857,576 km2 area) with 2 × 10−6 craters/km2.

Fig. 6
figure 6

Spatial distribution of identified rayed craters on a MOLA shaded topographic map in Mercator projection. The gray lines shows the edges of the analysed young lava plains at Elysium (gray line) and Tharsis (black line) while the white dots show the locations of candidate craters

Full size image

Discussion

Analysing the sizes of the identified young candidate craters, a roughly exponential increase from smaller to larger craters is present, with more craters of smaller diameters (Fig. 7). The identified craters are 0.5–15 km in diameter. The theoretical size-distribution model curve for 11 Ma aged craters (Hartmann and Neukum (2001) is also indicated with grey colour (grey solid line), calculating with spatial crater density (grey logarithmic axis at left) for the total surveyed 17.3 million km2. The observed size frequency distribution is in good agreement with the expectations regarding the trend – although the craters are a bit smaller in number, possibly because of some process that removed or less preserved craters at certain locations.

Fig. 7
figure 7

Size frequency distribution of the identified craters grouped by 2 km diameter bins

Full size image

Most of these young structures (left part of the diagram in Fig. 7) fall below the crater size required to launch substantial mass that escapes Mars, but these smaller craters suggesting that the selection of young craters used in this work was relatively effective. The diagram also shows the theoretical size-frequency distribution of small craters formed in the last 11 million years using data from Hartmann and Neukum (2001).

In theory the craters should be distributed homogeneously, but at the southern and western part of the Tharsis area only few candidates were identified. The lack of young candidate craters in these areas might be connected to the topography (elevation and/or slopes). For example, the rayed craters of the Elysium region are located below −1200 m topographic height; in the Tharsis region they are mostly below −1000 m (except of two cases that are at the elevation of +2200 m). Rayed craters exist on airless bodies too, thus the rare atmosphere need not necessary produce absence of rays on Mars; however, the difference in air density might have an effect on ray formation or retention on Mars, and it is also possible that strong slope winds are present at the steep slopes of Tharsis (Magalhaes and Gierasch 1982) and could have erased them quickly. But other possibilities also exist, thus more detailed work is required for the exact analysis of this issue that should be the topic of another work.

The Best Candidates

From the list of all rayed impact craters identified on roughly Middle Amazonian aged volcanic units (see the outline of the surveyed areas in Fig. 8), we have selected craters that seem to be the best candidate sources crater of the nakhla meteorites (Table 2). These craters show a ray system both on daytime and nighttime THEMIS images, and they have diameters above 3 km. The largest rayed craters on the lava terrains of appropriate age are shown in Fig. 7 on daytime (left) and nighttime (right) images. A critical point here is the firm identification of ray systems, as they show quite a wide range of appearances. For example the ray system in the case of the a1-a2 crater was less evident (faint and scattered rays with low spatial density), and in the case of the d1-d2 crater the rays show only a moderate density and one of them shows an unusual curved shape.

Fig. 8
figure 8

Image mosaic of the best candidate craters on daytime (a1, b1, c1...) and nighttime (a2, b2, c2...) THEMIS images. Arrows mark those rays, which are difficult to identify otherwise

Full size image
Table 2 Characteristics of the best candidate craters listed according to decreasing size order
Full size table

All of the candidate craters are located at relatively smooth lava plains, without signs of significant vertical lava mixing (like large tilted, chaotic lava blocks or small cones), suggesting rock units originally crystallized at the bottom or top of the lava flow could not have brought closer to each other, as it is observed in the nakhlites’ mineral pattern; although, small scale vertical mixing could easily take place during the horizontal movement of the lava flow without observable surface manifestation. In three cases (see them in Table 2) several cases, the candidate craters show fluidized ejecta blankets and valleys, which are suggestive of subsurface water or ice —it is however not possible to know whether this H2O was present there during the cooling of lava, or came later as ice precipitation on the surface.

An important consideration is also the depth of excavation during a meteor impact, which increases from 0.25 km to about 1 km when the diameter of the crater increases from 1 to 10 km (Tornabene et al. 2013). Numerical simulations suggest that for material that reaches the Martian escape velocity the depth of origin is ~10 % of the impactor’s diameter (Artemieva and Ivanov 2004).

The best candidates for source craters are listed in Table 2. It is difficult to select one ideal crater among them for this purpose, as the larger ones show quite poor and faint rays around them, although signatures for the presence of subsurface H2O exist there, which is compatible with the water related alterations in the nakhlite meteorites. Craters with more pronounced ray systems are smaller, around the threshold limit of 3 km diameter, with moderate chance to launch rocks to heliocentric orbit. The survey in this work points to several questions where more focused research is required to consider potential source craters of nakhlites, i.e. the evaluation of ray systems with different level of appearance, and the ranking of effects and their consequences that erase ray systems.

During this work several difficulties emerged to identify rayed craters, especially if they are heavily eroded and/or relatively old. Bright or dark halos can occasionally look somewhat similar to the proximal part of coalescing rays — their separation from real rays requires careful analysis. Halos might occasionally form by the wind activity (forming by small patches that might look like a halo, close to the limit of the resolution, but not radial elongated streaks). They can also form by the exhumation of different materials that are not visible on the surface, and the search for elongated rays requires special care. The quality of images significantly influences the investigation, and it is getting more difficult towards higher latitudes where data are scarce. In some cases, older ray-bearing craters might have been misidentified as fresh ones, since these rays are not only albedo features but also the result of selective erosion that enhances the system of radial ridges (i.e. Fig. 5a, b, c) – but careful analysis helps here much.

Cases, where large ray system does not exists, while some radial albedo features are still present beyond the edge of the ejecta are difficult to interpret. These might be older and degraded systems of formerly existing well-developed rays. In these cases the age constrain might not work well. The inhomogeneous areal distribution of the identified craters suggests that on Mars erosion acts in different intensity at surfaces of different altitude, thus producing eroded ray systems with different appearance. This makes the definition and use of age constrains uncertainty, and more focused analysis is necessary.

Conclusion

We have surveyed an area of 17.3 million km2 of about 1.2 Ga old Martian lava terrains, an age that is corresponding to the crystallization age of the nakhlites, to search for craters with rays that are probably young, similar to the ejection age of the nakhlites about 11 Ma. Altogether, 53 rayed craters within late amazoninan aged lava flows were identified. However, most of them are too small (between 1 and 3 km diameter) to match all required criteria. The methodology of using the associated ray system as an indicator for a young age seems to work well, because the measured areal density of craters around 10−6 km−2 agrees will with the number of craters predicted by recent bombardment rate models. A profound uncertainty of the analyses of the remote sensing data is the absence of about 1/3 of potential candidates, which might be located on terrains with a high topographic elevation or on surfaces of steep slopes – but the exact role of such terrains on the appearance of crater rays require more work beyond the scope of this paper. These regions display an apparent deficit of rayed craters, suggesting that the ray retention could be influenced by slope winds (strong katabatic winds with daily cycle, caused by the high daily temperature and pressure change especially along slopes of large topographic undulation (Kite et al. 2013) above the generally low thermal inertia Martian surface). A detailed analysis of this possibility is beyond the scope of this paper.

From the 53 rayed craters on corresponding age lava terrains that were identified in this study, only about a dozen can be considered as actual candidates. They are all above 3 km in diameter what means they have sufficient size to launch a meteorite on the heliocentric orbit, with Earth as their final destination. Roughly half of the best candidates are located close to possible fluvial valleys, while only three of them show a fluidized ejecta blanket. The latter implies the presence of subsurface water ice, but in these cases the rays are poorly developed, and probably they are older craters, not satisfying the constrain for a young age. Here the poor, fragmented ray system requires improved interpretation for age estimation. Water is a strong indicator for a nakhlite parent rock, since all nakhlites are hydrothermally altered, or contain secondary minerals that imply hydrous activity. We already have certain geochemical, mineralogical, and petrological information from the nakhlites that could be matched with existing remote sensing data but possibilities will improve in the future with improved and more detailed remote studies. More detailed studies could, especially focusing on chemical and mineralogical information, similar to those mentioned in section 3.1 (“Basic constrains”) could help to narrow the list of candidate craters. An effort should also be made to locate those candidate craters that are not identified with the current methodology due to the lack of (probably eroded) ray systems. This effort however requires a better understanding, which features shall we classify as a ray system and how do they form. The reason is that ray systems cannot be easily classified because variable methods of erosion influenced their appearance to different degrees and on different terrains.