Mini-RF S-band Radar Characterization of a Lunar South Pole–crossing Tycho Ray: Implications for Sampling Strategies

One of the youngest features on the Moon is Tycho, an 85 km diameter impact crater with a vast ray system that spans much of the lunar nearside. As such, it serves as an important stratigraphic marker for the Moon. One of Tycho’s longest rays crosses the South Pole, where it intersects several candidate landing sites for NASA’s Artemis III mission, which intends to return new lunar samples. Identification of ray-related effects are thus important to understand the provenance of collected material. To help contextualize sampling strategies, here we characterize the South Pole–crossing Tycho ray using monostatic S-band radar observations from the Lunar Reconnaissance Orbiter’s Miniature Radio Frequency instrument. We found that the ray is a ∼15 km wide radar-bright feature extending at least ∼1600 km from Tycho. Polarimetric analysis revealed that the measured radar backscatter is consistent with a terrain enhanced in centimeter-to-decimeter-scale scatterers. Moreover, we found that the abundance of these scatterers likely decreases with distance from the primary crater, suggesting there may be less Tycho-disturbed material, in particular, poleward of 85°S, where the candidate landing sites are located. Nevertheless, we identified craters along the ray and, importantly, within the Haworth candidate landing site that exhibit secondary crater characteristics, such as radar-bright, asymmetric ejecta deposits. We showed, based on solar illumination and topographic slopes, that the likely Tycho-related secondaries within Haworth are accessible by landed missions. Exploration of this site may thus directly sample Tycho-disturbed material, including a nearby permanently shadowed region, providing new insights into lunar surface processes.


Introduction
Crater rays are optically bright, narrow, filament-like features that extend radially from some fresh impact craters on various bodies in the solar system.On the Moon, rays are classified by the likely cause of their optical brightness as (1) compositional rays, which are bright due to differences in the albedo of ray-forming material relative to the background terrain; (2) immature rays, which are bright due to deposition of fresh, unweathered material; or (3) a combination of these (Hawke et al. 2004).Radar-bright rays, on the other hand, are attributed to an enhancement in wavelength-scale complexity within the radar penetration depth along the path of the ray (Campbell et al. 1992;Harmon et al. 2007;Neish et al. 2013).These physical and compositional properties are believed to result from the emplacement of primary ejecta along the ray path, some of which impact with sufficient energy to form secondary craters.This process induces local mixing and the deposition of secondary, ejected, fragmented material (Oberbeck 1971(Oberbeck , 1975;;Pieters et al. 1985;Elliott et al. 2018).
A lunar crater with a prominent and extensive ray system is the 85 km diameter impact crater Tycho (43°.3S, 348°.8E).Its rays may extend out to ∼2200 km (Elliott et al. 2018) and cover an area of ∼560,000 km 2 (Dundas & McEwen 2007).Effects from the Tycho-forming impact are also evident at the antipode, where multiwavelength analysis suggests impact melt ponding and infilling of local topography from Tycho ejecta (Carter et al. 2012;Robinson et al. 2016;Bandfield et al. 2017).Additionally, studies of the Apollo 17 landing site in the Taurus-Littrow valley (20°N, 31°E) revealed evidence of Tycho-disturbed material (e.g., landslide from South Massif) and secondaries (e.g., Central Cluster; Arvidson et al. 1976;Lucchitta 1977).Cosmic-ray exposure ages and isotopic measurements of Apollo 17 samples from these regions suggest a formation age of ∼100 Ma (Arvidson et al. 1976;Drozd et al. 1977;Zellner et al. 2009).However, the validity of these ages is questioned, as the direct association of the sample's origins with the Tycho impact remains contentious (Stöffler & Ryder 2001;Robbins 2014).Nevertheless, due to its extensive ray system and sample-based inferred age, Tycho is a critical stratigraphic marker for the Moon; thus, characterizing its secondaries and rays is important to further our understanding of lunar chronology.
Although secondary craters are typically identified by their morphology, orientation, and proximity to a primary crater, identifying distal secondaries can be more challenging.Generally, the impact velocity of distant secondaries is high enough that the resulting craters are deeper and more symmetric, complicating their identification (e.g., Watters et al. 2017).However, radar observations, which are sensitive to wavelength-scale complexity, can reveal additional characteristics of secondary cratering.For example, using Arecibo S-band (12.6 cm, 2380 MHz) radar observations, Wells et al. (2010) identified peculiar features among some small, 2 km diameter craters within the floors of the Newton (76°.5S,342°.6E) and Newton A (80°. 1S, 339°.2E)craters, ∼1060 km or ∼25 crater radii away from Tycho.In Figure 1, for reference, we show an Arecibo S-band radar image highlighting the increased backscatter from the Newton crater group (Campbell et al. 2010).In their work, Wells et al. (2010) found that these small craters are associated with asymmetric ejecta deposits heightened in wavelength-scale roughness, as indicated by their S-band radar circular polarization ratio (CPR).These ejecta have tail-like geometries extending downrange from Tycho crater.Crater counts of these features demonstrated that they follow a steep power-law cumulative frequency slope, a characteristic often associated with secondary craters (Hirata & Nakamura 2006;Wells et al. 2010;Watters et al. 2017;Bierhaus et al. 2018).Consequently, these craters were interpreted as Tycho secondaries.Notably, similar radar properties were also observed in secondaries within a Tycho ray in Mare Nectaris (15°.2S, 34°.6E) using X-band (3 cm, 10 GHz) radar observations collected at the Haystack Observatory (Campbell et al. 1992).The enhanced wavelength-scale roughness within the extended ejecta has been attributed to a debris surge resulting from ray formation and/or secondary cratering (Campbell et al. 1992;Martin-Wells et al. 2017).
Subsequent analyses of Clementine 750 nm and color-ratio maps, as well as reflectance measured by the Lunar Reconnaissance Orbiter's (LRO) Lunar Orbiter Laser Altimeter (LOLA), revealed a feature along the Newton crater group that extends over the South Pole and originates at Tycho (Dundas & McEwen 2007;Denevi & Robinson 2020;Bernhardt et al. 2022).This terrain also exhibits increased roughness at baselines of <0.4 km, as determined by analysis of data from LRO's LOLA, attributed to an increased density of small craters (Barker et al. 2023).These observations of a polecrossing Tycho ray overprinting the Newton crater group lend support to the radar-enabled identification of secondary craters.Furthermore, detailed analysis of the extended, tail-like CPR features associated with such secondaries suggested a potential component of primary impact melt (Martin- Wells et al. 2017).Thus, samples from these regions would provide valuable opportunities to directly age the Tycho impact event (Bernhardt et al. 2022).
In light of NASA's Artemis III mission (NASA 2020), which aims to send astronauts to locations near the Tycho ray and return new lunar samples, here we utilized radar observations from the Miniature Radio Frequency (Mini-RF) instrument on board LRO to characterize this South Polecrossing ray.Radar observations can reveal wavelength-scale variation in regolith properties (e.g., scatterer size-frequency distribution and morphology), provide insights into subsurface structure, and constrain bulk density and composition (i.e., dielectric permittivity).Characterization of such properties can provide new insights into ray-forming processes on the Moon.Moreover, due to observing limitations of ground-based facilities (e.g., high incidence angles that affect the measured CPR), orbital radar assets can help reveal small-scale features relevant for crewed exploration.Thus, we used high-resolution, S-band Mini-RF images along the ray, focusing particularly on an Artemis III Candidate Science Program landing site, to identify Tycho-related secondary cratering in order to inform sample return strategies.

Mini-RF S-band Radar Observations
Mini-RF is a hybrid-polarimetric, dual-frequency, sidelooking, synthetic aperture radar (SAR) that can operate in either baseline mode (0.15 km resolution) or zoom mode (0.03 km resolution).It is designed to record the full polarization state of received backscattered echoes in the S (12.6 cm, 2380 MHz) and C (4.2 cm, 7140 MHz) bands, though for historical reasons it has been referred to as the X band (Raney 2007;Raney et al. 2011).During observations collected before 2010 December, the instrument operated monostatically by transmitting a left-circularly polarized wave and receiving reflections off of the lunar surface in orthogonal linear polarizations (i.e., horizontal, H, and vertical, V ).After 2010, the instrument's transmitter failed.Since then, Mini-RF has been used in a bistatic configuration, where the receiver and transmitter are not colocated, by using ground-based radar telescopes such as the Arecibo Observatory and the Goldstone Solar System Radar as the transmitter.In this work, though, we made use of monostatic S-band observations only.
From the received complex backscattered echoes, the Stokes vector, which describes the polarization state of electromagnetic radiation (Stokes 1851), can be derived following where E is the complex voltage; the angle brackets denote spatial averaging; Re and Im denote the real and imaginary value of the complex cross-product amplitude, respectively; and the negative sign in S 4 follows the backscattering alignment coordinate convention (Raney et al. 2011).Here, S S S S is the total received backscattered power, where the inequality arises due to depolarized (i.e., randomly or unpolarized) echoes; S 2 describes the linearly polarized power; S 3 is the linearly polarized power at an orientation of 45 • ; and S 4 is the power received in the circular polarization.In essence, the Stokes parameters S 2 , S 3 , and S 4 are Cartesian points on a sphere of radius mS 1 , where m is the degree of polarization, and conversely, (1 − m) is the degree of depolarization.The so-called Poincaré sphere is further parameterized by χ, the ellipticity parameter, and ψ, the linear polarization angle, which are dependent on the Stokes parameters following The Stokes parameters and Poincaré variables can be used for detailed polarimetric analysis of Mini-RF data.Indeed, Raney et al. (2012) proposed a technique that decomposes imaging radar data into three components: (1) odd bounce scattering, which is associated with quasi-specular reflections from radarfacing facets, Bragg scattering, and single scattering from wavelength-scale features; (2) even bounce scattering, which is associated with scattering from angular geometries (e.g., dihedral scattering), irregularly shaped wavelength-scale scatterers, and echoes from ice; and (3) depolarized scattering, which is associated with diffuse scattering, multiple scattering events, such as inside a radar quasi-transparent medium, and scattering from short-scale surface roughness and long-scale slopes.This decomposition technique makes use of two of the three Poincaré variables, m and χ, and hence is named m-χ decomposition.Each of the scattering components can be viewed simultaneously with a false-color red, green, and blue (RGB) image, where Here, red indicates even, green depolarized, and blue odd scattering, and Although there are other decomposition methods commonly used in terrestrial SAR applications (e.g., Cloude & Pottier 1997;Raney 2007), m-χ decomposition has been widely adopted for analysis of Mini-RF data (e.g., Cahill et al. 2014), as well as for analysis of ground-based radar observations of near-Earth asteroids (Hickson et al. 2021).
Typically, though, ground-based radar observations are analyzed in terms of their individual derived products, in particular the backscattered power in the opposite-circular (OC) and same-circular (SC) polarizations as transmitted, and CPR = SC/OC.For clarity, we note that in this work "depolarized" scattering refers to (1 − m) and not SC, as in some other planetary radar works (e.g., Hagfors 1967;Campbell et al. 1992).Both SC and OC can be derived from the Stokes parameters as At low local incidence angles, OC backscatter primarily results from specular, single-scattering events from large-scale topography, while at high incidence angles, the diffuse component of the backscatter from small-scale topography increases (Ostro 1993;Black 2002).On the other hand, SC backscatter primarily results from diffuse scattering at all incidence angles.Additionally, both polarizations are affected by regolith properties, such as dielectric permittivity and the size distribution of scatterers.Thus, CPR can be complex to interpret, particularly in scenes with substantial topography causing changes in the local incidence angle (e.g., Jawin et al. 2014).Holding the local incidence angle constant (e.g., on flat surfaces), CPR is a first-order approximation of wavelength-scale complexity (e.g., Campbell et al. 1993;Campbell & Campbell 2006), where terrains with high CPR are considered rough (e.g., Fa & Cai 2013).High CPR can also arise due to scattering from ice via the coherent backscatter opposition effect (e.g., Hapke 1990;Harmon et al. 1994;Stacy et al. 1997;Rivera-Valentín et al. 2022;Hofgartner & Hand 2023).
Virkki & Bhiravarasu (2019) suggested analysis of SC and OC separately as a method to untangle some of the ambiguity in interpreting CPR.In their work, they showed that the slope and intercept of a linear least-squares fit (LSF) to SC and OC data, i.e., OC = SCb + a, where b is the slope and a is the intercept, are related to the abundance and/or morphology of wavelength-scale scatterers and the medium's bulk dielectric permittivity, respectively.This is because the intercept, a, occurs when specular reflections dominate (i.e., when SC = 0), and so the radar brightness is well approximated by the Fresnel reflection coefficient, which is dependent on the dielectric permittivity.Furthermore, it can be shown that in this framework the slope is related to the intercept for the linear equation of the form CPR −1 = b + aSC −1 .Thus, b is inversely proportional to CPR in the case where SC −1 = 0, or, in other words, in the diffusive scattering regime.This derived CPR has been termed the diffusive circular polarization ratio in asteroid radar astronomy and numerical modeling of radar backscatter (e.g., Magri et al. 2001), and here we define it as CPR diff .The method by Virkki & Bhiravarasu (2019) was applied for Sband radar observations of Mercury's north polar region and was demonstrated to improve characterization of the potential ice deposits (Rivera-Valentín et al. 2022).
Mini-RF S-band measurements of the polarization state of received echoes and transmission capabilities have been calibrated separately with help from the Arecibo and Green Bank radio telescopes, as well as with self-calibration procedures, as described by Raney et al. (2011).Calibration issues, though, have been identified with some collects that are most apparent in S 2 measurements.Additionally, a cross-track gradient in backscatter intensity has been observed in monostatic Mini-RF collects that reflects imperfect correction of the antenna beam pattern (e.g., Spudis et al. 2013;Carter et al. 2017;Fassett et al. 2024).This results in a lower signalto-noise ratio at the edges of images.Such artifacts are accounted for here when interpreting Mini-RF monostatic radar products.

Ray Characterization
Ray characterization was conducted using a South Pole controlled polar mosaic of the Stokes parameters.The procedure to mosaic multiple Mini-RF collects into east-and west-looking polar stereographic maps is described by Kirk et al. (2010), and the first analysis of these mosaics to search for evidence of water ice in the lunar South Pole was provided by Spudis et al. (2013).In Figure 2, we show a polar stereographic mosaic of S 1 in dB scale (i.e., S 10 log 10 1 ).As in Spudis et al. (2013), the mosaic averages both east-and westlooking observations, but here it excludes regions where data existed only in one look direction; however, we ensured that all interpretations continued to be valid within the individual lookdirection-dependent mosaics.The presented map shows values poleward of 70°S.After averaging both look directions, to ensure conservation of energy, we masked all pixels where S S S S 1 2 2 3 2 4 2 < + + , since this would be nonphysical, as well as locations where CPR > 10, since such values are spuriously high even for enhanced dihedral scattering (Campbell 2012).These likely artifacts were 0.1% of the pixels.To improve visual inspection of radar-bright features, the color mapping was clipped to range from the image median to 1.5× the interquartile range above the upper quartile (Q 3 ) of S 1 .Thus, the color mapping follows a box plot from the median to the upper whisker.In Appendix Figure 10, for completeness, we also show an unclipped S 1 mosaic in the typical grayscale color map.Although the polar mosaic combined zoom mode collects, here we made use of a map product resampled to a resolution of 0.24 km pixel −1 to reduce speckle noise.
As seen in the high-resolution LRO Narrow Angle Camera (NAC) image, the largest of these craters has a diameter of 0.9 km.It is expected that the largest secondary crater produced by Tycho would be 1.4 km in diameter at a distance of ∼1540 km, according to the model by Singer et al. (2020).Thus, the identified secondary crater cluster fits expectations for Tycho.It is notable that because the ray is seen to go over the pole, it likely also transects the Connecting Ridge Extension (89°.0S,258°.8E) and Connecting Ridge (89°.5S,225°.2E)Artemis III candidate landing sites.However, because Mini-RF is a side-looking SAR and LRO is a polar orbiter, gaps in coverage result within a couple of degrees of latitude around the pole.These two candidate landing sites fall within this central coverage gap.
The radar-bright ray agrees well with the location of a previously identified enhanced LOLA reflectance feature originating at Tycho crater (Denevi & Robinson 2020;Bernhardt et al. 2022).However, in their work using LOLA reflectance, Bernhardt et al. (2022) found that the ray has a total length of ∼2040 km, which exceeds the length of the radarbright feature.Radar-bright rays emanating from De Forest crater (76°.9S, 196°.7E),though, may interfere with our length measurement.As seen in Figure 2, De Forest's rays overprint regions equatorward from Wiechert J crater.The remaining radar signature from Tycho's ray equatorward from this point may have been erased.Alternatively, radar-bright extended ejecta deposits may become difficult to identify at great distances from their source craters.Indeed, we found that backscatter from the ray is not consistently high throughout its extent, becoming tenuous poleward of 85°S, contrasting the nearly consistent LOLA reflectance brightness over the pole (Denevi & Robinson 2020;Bernhardt et al. 2022).As such, our extent measurement is likely a lower bound estimate.
To further quantify the detection of the ray and improve identification of the fine-scale variation within it, we reduced Figure 2 into a plot showing radar power as a function of distance from the center of the ray.In Figure 4, we show the lower (Q 1 ) and upper (Q 3 ) quartiles for S 1 within ±100 km from the center of the ray, focusing on the region equatorward from the Haworth candidate landing site, where the ray is most readily detectable.To reduce noise in the plotted values, we applied a Gaussian-weighted moving average filter with a length of ∼1 km.Additionally, the curves are compared to the median, Q1, and Q 3 of the total backscattered power from the South Pole to identify significant variation.
In Figure 4, the ray can be seen as a broad peak in total backscatter power, rising above the median S 1 of the South Pole, aligned about the center of the ray.In particular, between ±10.5 km, the ray's Q 3 values are above the South Pole's Q 3 values, indicative of its radar-bright nature and suggesting a width of ∼21 km.However, we note that the ray does not directly follow a great circle, as some curvature is observed, potentially due to interactions with local topography.The average of 20 spatially separate measurements of the ray's width along its extent suggest a width of 15 ± 3 km, in agreement with albedo-based measurements (Bernhardt et al. 2022).Furthermore, the well-defined peak of the values suggests an emplacement environment where ray-forming material was concentrated along a radial line from Tycho with diminishing ray-forming material away from the center.
To constrain the potential physical properties leading to the observed total power backscatter, in Figure 5, we show the results of the polarimetric decomposition of the ray as intensity of the scattering components as a function of distance from the ray center, as well as the CPR.The depolarized backscatter intensity across the ray follows the same broad peaklike structure as for the measured S 1 , where both quartile values are above the polar median.The ray is also characterized by enhanced odd bounce backscatter, though to a lesser extent.While odd bounce scattering Q 3 values are above the polar median and respective quartile, Q 1 values only approach the polar median but are above the respective quartile.Odd bounce backscatter from the central region of the ray is also muted, compared to depolarized intensity, and overall follows more closely a uniform distribution about the center of the ray.In contrast, the variation of even bounce scattering and CPR is within the variation experienced across the pole.Furthermore, the profiles do not follow a distinguishing shape across the ray.Indeed, using a Student's t-test, we found that backscatter from the ray terrain is distinguishable from the overall south polar values at the 95% confidence level by its depolarized and odd bounce scattering, but the measured even bounce scattering and CPR are not discerning features.These results further demonstrate the importance of using multiple radar metrics to identify and characterize features.
Our results show that the ray is defined as a radar-bright feature in the S band whose return is dominated by depolarized and odd bounce backscatter; however, the feature does not result in enhanced CPR.Considering scatterers above the Rayleigh size limit in the S band (i.e., 2 cm), depolarization of radar waves can arise from scattering through a quasitransparent regolith column due to multiple internal reflections and single scattering from nonspherical particles  .Controlled polar mosaic of Mini-RF S-band S 1 (i.e., total backscattered power) images in dB scale for areas poleward of 70°S.Color mapping is clipped from the scene median to 1.5× the interquartile range above the upper quartile (Q 3 ) such that black represents either areas with no data or areas below the median, shades of blue represent terrain with backscatter close to the scene median, and shades of yellow represent terrain near or above Q 3 .Magenta arrows point out the location of the studied Tycho ray, and white arrows point out the location of a radar-bright ray from De Forest crater.Regions noted in the text are annotated as A, Newton crater; B, Weichert J crater; and C, De Forest crater.The Haworth candidate landing site is noted with a white square.Image resolution is 0.24 km pixel −1 .East longitude is annotated for reference.study of various particle shapes that a medium comprised of particles below the Rayleigh size limit would also result in only odd bounce scattering, whereas larger, decimeter-scale particles result in increased even bounce intensity.Based on these modeling results and our observations, we infer that the ray is likely characterized by a size distribution of scatterers mostly in the centimeter-to-decimeter range and is generally deficient in large-scale boulders.This scattering medium would not exhibit enhanced CPR and would offer limited opportunities for even bounce scattering, although such scattering can still occur, such as through biparticle scattering (i.e., scattering between nonspherical particles).
It is important to note, though, that the results presented in Figures 4 and 5 are averages over large areas on the Moon, and as such, diverse geologic features are impacting the overall radar return.For example, the increased number of small craters within the ray, as suggested by enhanced roughness at baselines of <0.4 km (Barker et al. 2023), likely result in opportunities for dihedral scattering from the interface between the crater wall and floor.Boulder fields associated with fresh   impact ejecta would also offer opportunities for enhanced depolarized and even bounce backscatter.Furthermore, volume scattering from subsurface structures, such as from buried fragmental, ray-forming material, likely also contributes to the received echoes.These diverse features may on average over a large area result in the observed polarimetric response.
Nevertheless, if the ray terrain has more centimeter-todecimeter-sized scatterers, it would be consistent with a model where ray formation included a significant debris surge component.In such a model, ejecta released by many small, secondary craters impacting nearly simultaneously during ray formation interact with each other resulting in a debris surge moving downrange through the secondary crater field.This process may create distant features from the primary impact, such as herringbone structures (Melosh 1989), which were observed by Martin-Wells et al. (2017) within the ray.The debris surge would also transport wavelength-scale scatterers along the ray path, a process that would then interact with secondary craters and produce tail-like, radar-bright ejecta deposits that point downrange from Tycho (Campbell et al. 1992;Martin-Wells et al. 2017).In this scenario, the peaklike distribution of depolarized backscatter intensity observed in Figure 5 could be the result of debris deposition primarily concentrated along the center of the ray.Indeed, in their work, Martin-Wells et al. (2017) found evidence of smooth regions within the ray, as well as infilled small craters.These geologic observations lend support to the interpretation of significant debris deposition associated with this Tycho ray.Alternatively, the peaklike distribution of both S 1 and depolarized backscatter could be the result of secondary cratering (Neish et al. 2013).For example, Shoemaker (1965) found that the crater density increases closer to the middle of some of Tycho's rays.The more likely scenario, though, is a combination of both mechanisms.
As noted earlier for Figure 2, the ray becomes tenuous poleward of 85°S.To investigate the cause of this decrease in backscatter power, we leveraged the methods of Virkki & Bhiravarasu (2019) to study the potential variation in wavelength-scale complexity.Here we discretized the ray into 15 km × 15 km areas and calculated SC, OC, and CPR from the Stokes vectors for every square element.Using the methods described in York et al. (2004), we fit a linear LSF to the SC and OC backscatter for every square element assuming an uncertainty on the intensity of 2% for each circular polarization, following the cited uncertainty for Mini-RF measurements in Cahill et al. (2014).The inverse of the slope of an LSF is related to the CPR, specifically to the portion of roughness due to the size-frequency distribution and/or morphology of wavelength-scale scatterers, which we define here as CPR diff .The resulting values are plotted in Figure 6(a) as a function of distance with respect to Tycho radii.For comparison, in Figures 6(b) and (c), the same analysis was conducted on either side of the ray with respect to distance from the top of the map.Additionally, the analysis was repeated for CPR in Figures 6(d) and (e), where we show the mean and standard deviation of CPR over the square elements.
We tested for significant changes in CPR diff and CPR along the ray via a two-tailed model utility test for the LSF slope at 95% confidence.For the diffuse CPR of the ray, we found that a zero slope can be ruled out, and so CPR diff likely changes with distance from Tycho.The change in CPR diff is 9.9 0.2 10 4 ( ) - ´km −1 , while the change on either side is 2.4 0.1 10 4 ( ) - ´km −1 .Thus, the rate of change of CPR diff over the ray is ∼5× higher than nonray terrain.Furthermore, values of CPR diff skew higher than for nonray terrain, reaching 0.86 compared to a maximum of 0.65.The median CPR diff for nonray terrain is 0.47.On average, CPR diff over the ray is at or below this value at a distance of 30 Tycho radii (i.e., 1300 km) from Tycho, which occurs near 85°S.This agrees with our observed transition to the radartenuous ray in Figure 2.
In contrast, repeating the analysis for the CPR, we found that a zero slope cannot be ruled out, and thus there is no significant evidence suggesting a change in overall roughness with distance from Tycho using this traditional metric.Along the ray, CPR decreases with distance as 1.1 0.9 10 4 ( ) - ´km −1 , while nonray terrain is 0.6 0.5 10 4 ( ) - ´km −1 .Thus, following the model utility test to 95% confidence, CPR likely does not change along the ray and is not distinguishable from nonray terrain.This demonstrates the additional insight afforded from the polarimetric analysis method proposed by Virkki & Bhiravarasu (2019).
These results suggest that the brightest segment of the ray is characterized by a scatterer size-frequency distribution skewing to a few wavelengths (i.e., 12-60 cm) in size.Inferred scatterer sizes agree with earlier multiwavelength radar observations of several of Tycho's rays, which found that they were readily identifiable in the X band (3.0 cm, 10 GHz) but not in the P band (70 cm, 430 MHz), indicative of a terrain enhanced in 1-50 cm scatterers (Campbell et al. 1992;Neish et al. 2013).Indeed, although not noted in their work, X-band radar observations from Haystack Observatory reported by Zisk et al. (1974) show this South Pole-crossing Tycho ray as a feature bright in SC backscatter.Overall, these results provide evidence in support of our interpretation that radar backscatter intensity decreases with distance from Tycho due to less centimeter-to-decimeter-sized scatterers, which may have been deposited as part of a debris surge component to ray formation.

Implications for Landed Exploration
The observation of enhanced LOLA reflectance over the South Pole-crossing Tycho ray, as well as its radar brightness, already provides important context for sampling from regions overprinted by this ray (Bernhardt et al. 2022).As we have shown, though, there may be a decreased component in Tycho impact-related debris deposition poleward of 85°S, where the candidate Artemis III landing sites are located.However, the radar-enabled observation of small, secondary craters with taillike ejecta deposits that point downrange from Tycho provides additional opportunities.In particular, detailed analysis of some of these features within this South Pole-crossing Tycho ray suggested a potential component of impact melt by geomorphological observations, such as smooth terrains with cracks (Martin- Wells et al. 2017).The melt could be a mixture sourced from primary and secondary cratering, though due to energy scaling relationships it is likely predominantly primary melt (i.e., derived from the Tycho-forming impact), if present.From a lunar exploration perspective, sampling of such material would provide an opportunity to more directly age the Tycho impact event, furthering our understanding of the lunar bombardment history.Here, we further characterized local-scale features via m-χ decomposition and in particular searched for such secondary signatures within Artemis III candidate landing sites.
In Figure 7, we show an m-χ decomposition RGB image over the Newton crater group.This figure is formed by combining two Mini-RF zoom mode (0.03 km resolution) images that were orthorectified following the methods described in Fassett et al. (2024).Additionally, we applied a 2σ stretch to the color map of each RGB decomposition component to avoid biasing by outliers.We found that the taillike geometries enhanced in CPR are also enhanced in depolarized backscatter and even bounce scattering, resulting in the yellow color.Thus, the m-χ decomposition suggests that the depolarized backscatter intensity is likely primarily due to short-scale roughness, given the enhanced even bounce backscatter.The observed enhanced CPR of the secondaryassociated ejecta is then likely due to deposition of wavelengthscale blocky and/or rough material, in agreement with Martin- Wells et al. (2017), who used high-resolution LRO NAC images over these regions for their interpretations.
In light of the Artemis III mission (NASA 2020) and the Mini-RF monostatic S-band coverage, we investigated the Haworth candidate landing site.In Figure 8, we show an m-χ decomposition RGB image over the region using two orthorectified, zoom mode Mini-RF collects.Several small craters within this candidate landing site have tail-like ejecta features pointing downrange from Tycho that are enhanced in depolarized scattering and even bounce backscatter, similar to the features observed in the Newton crater group.In particular, we point out one secondary of diameter 0.4 km (86°.6S,331°.9E) with an elongated ejecta tail extending out ∼48 radii from the crater rim.As observed in Figures 8(b) and (c), the asymmetric ejecta is most readily identifiable in the m-χ image compared to the CPR map.Furthermore, the region around this crater includes several other smaller craters with tail-like ejecta enhanced in depolarized scattering and even bounce backscatter.Given their similarities to secondaries likely formed by Tycho primary ejecta across the ray, these features are likely Tycho-related, making them ideal targets for sample collection.
In terms of landed exploration, we studied three conditions that are relevant for the Artemis III mission: (1) areas illuminated a large fraction of the year (>40%; Mazarico et al. 2011), (2) terrains with small slopes of <5°that would not likely pose a hazard for landing (Barker et al. 2023), and (3) locations of permanently shadowed regions (Mazarico et al. 2011).Results are shown in Figure 9.The small crater noted in Figure 8 is identified as a permanently shadowed region (PSR), increasing its interest for exploration.Additionally, there are two regions  near the identified extended ejecta deposits from the noted crater that meet both illumination and slope criteria.These regions are both <2 km from the radar-identified extended ejecta deposits and could thus serve as landing sites that would permit access to them.Regolith samples from this region, and in particular from the highlighted secondary crater, would thus likely reveal a radiometric age spike near 100 Ma, providing further validation of Tycho's age.On the other hand, if no cluster of ages in the regolith samples is observed, then either the ability to trace distal deposits to their source craters with radar requires additional constraints or the generally accepted age for Tycho requires revisiting.

Conclusions
Here we reported on Mini-RF S-band polarimetric radar characterization of a South Pole-crossing ray emanating from Tycho crater.The ray was previously identified as an enhanced LOLA reflectance feature that extends at least past de Gerlache and Shackleton craters (Denevi & Robinson 2020;Bernhardt et al. 2022), as well as an area of increased roughness at baselines of <0.4 km due to an increased number of small craters (Barker et al. 2023).The ray terrain has been suggested as an area of interest for future landed missions due to opportunities to directly sample Tycho-related material (Bernhardt et al. 2022).Particularly, here we explored the Haworth candidate landing site for areas that would improve opportunities to sample material related to the formation of Tycho crater.
Using a controlled polar mosaic of the Stokes parameters over the lunar South Pole, we found that the Tycho ray is a ∼15 km wide, radar-bright feature that extends at least ∼1600 km from the primary impact site.Although the width of the radar-bright portion of the ray is consistent with visible albedo-based measurements, the length is much shorter.This may be due to overprinting by radar-bright rays emanating from De Forest crater (76°.9S,196°.7E) or decreased wavelength-scale ray material at distance.
To characterize the physical properties that may lead to the observed radar backscatter, we conducted a polarimetric analysis.We found that the radar backscatter from the ray is defined by depolarized and odd bounce scattering but is not distinguishable by even bounce scattering or CPR.The measured polarimetric response from the ray may suggest that the terrain is characterized by centimeter-to-decimeter-scale scatterers and generally deficient in large-scale boulders.This would be consistent with a model where ray formation included a significant debris surge component, which would serve to transport wavelength-scale scatterers along the ray.Indeed, Martin- Wells et al. (2017) found evidence of smooth regions within the ray, as well as infilled small craters, which lends support to this interpretation.Furthermore, by investigating the radar backscatter as a function of distance from the center of the ray, we found that the total backscattered power and the depolarization backscatter intensity follow a broadly peaked distribution, with the strongest reflections arising from the center of the ray.This may be suggestive of an emplacement environment where debris deposition and/or secondary cratering was concentrated along a radial line from Tycho with diminishing ray-forming material away from the center.
We also found that radar backscatter from the ray generally decreases poleward of 85°S, where it becomes tenuous.To further investigate this possibility, we studied the variation of wavelength-scale complexity along the ray by leveraging methods proposed by Virkki & Bhiravarasu (2019).Our results indicated, at the 2σ level, that the diffusive CPR decreases with distance from Tycho and remains above background values down to 85°S.In contrast, after lateral averaging, CPR shows some evidence for a weak negative correlation with distance from Tycho; however, this was only detected at the 1σ level and is therefore not robust.These measurements are consistent with a decrease in the abundance of centimeter-to-decimeterscale scatterers with distance from Tycho.
Although there may be a decrease in Tycho-related debris poleward of 85°S, where the candidate Artemis III landing sites are located, there are other opportunities at the candidate landing sites to improve the potential to collect samples directly related to the Tycho impact event.Following earlier descriptions of radar-enabled identification of secondaries (Campbell et al. 1992;Wells et al. 2010), we found several small, 2 km craters that have radar-bright asymmetric, taillike ejecta extending downrange from Tycho along the ray.In particular, using radar decomposition false-color images, we found that the tail-like ejecta are enhanced in depolarized and even bounce scattering.This suggests that the secondaryassociated ejecta is characterized by blocky and/or rough material, which is supported by detailed analysis of LRO NAC images (Martin- Wells et al. 2017).As such, in light of the Artemis III mission (NASA 2020), we investigated the Haworth candidate landing site, where Mini-RF radar data suggests the Tycho ray overprinted.We found that within this region, several small (0.3 km) craters have elongated ejecta tails extending downrange from Tycho.We showed that the illumination conditions and regional slopes of nearby terrain would allow access to the ejecta deposits by crewed missions.Furthermore, we found that some of the craters of interest are themselves PSRs, increasing their relevance for exploration.Thus, these locations offer excellent sample return opportunities that meet multiple science priorities.
Our work provides further support for the radar-enabled identification of secondary craters (Campbell et al. 1992;Wells et al. 2010).We found that such craters are characterized by extended ejecta with tail-like morphologies heightened in CPR, depolarized, and even bounce backscatter.Additionally, we showed that distal secondaries are more readily identifiable by depolarized intensity rather than CPR.These results further demonstrate the importance of using multiple radar metrics to identify and characterize lunar features.
Figure 10.Controlled polar mosaic of Mini-RF S-band S 1 (i.e., total backscattered power) images in dB scale for areas poleward of 70°S in grayscale.Here, the grayscale map ranges from 1.5× the interquartile range below the lower quartile to 1.5× the interquartile range above the upper quartile and as such follows a box plot from the lower and upper whiskers.This removes the effects of outliers.

Figure 1 .
Figure 1.Arecibo S-band SC radar backscatter in dB scale.Radar backscatter intensity increases from cool colors to warm colors.The magenta arrows point to the Tycho ray, and the letter A denotes the location of Newton crater.The Artemis III Haworth candidate landing site is noted with a white square.Image resolution is 0.08 km pixel −1 .

Figure 2
Figure2.Controlled polar mosaic of Mini-RF S-band S 1 (i.e., total backscattered power) images in dB scale for areas poleward of 70°S.Color mapping is clipped from the scene median to 1.5× the interquartile range above the upper quartile (Q 3 ) such that black represents either areas with no data or areas below the median, shades of blue represent terrain with backscatter close to the scene median, and shades of yellow represent terrain near or above Q 3 .Magenta arrows point out the location of the studied Tycho ray, and white arrows point out the location of a radar-bright ray from De Forest crater.Regions noted in the text are annotated as A, Newton crater; B, Weichert J crater; and C, De Forest crater.The Haworth candidate landing site is noted with a white square.Image resolution is 0.24 km pixel −1 .East longitude is annotated for reference.

Figure 3 .
Figure 3. Map of total backscatter power (i.e., S 1 ) in dB scale around the region of Wiechert J crater, annotated as B following Figure 2. The color map scales from 1.5× the interquartile range below the lower quartile (blue) to 1.5× the interquartile range above the upper quartile (yellow) and as such follows a box plot from the lower and upper whiskers.The radar image is draped over a terrain hillshade model (Barker et al. 2023).Magenta arrows point out craters that exhibit asymmetric, extended ejecta deposits that point downrange from Tycho.The image ranges from 88°S, 205°E (top left) to 84°S, 159°E (bottom right).A high-resolution LRO NAC image for the secondary craters noted by the magenta box is shown on the right.

Figure 4 .
Figure 4. Power as a function of distance from the ray center produced by collapsing the S 1 controlled polar mosaic between 254°E and 74°E and calculating the median (black line) and interquartile range (gray area) as a function of distance from the ray center.The shaded gray area between the quartiles represents the most likely values across the ray for that given distance from the center.For comparison, the mosaic-wide upper quartile (top dashed magenta line), median (solid magenta line), and lower quartile (bottom dashed magenta line) are shown.

Figure 5 .
Figure 5. Intensity of each m-χ decomposition component for (a) red, even bounce; (b) green, depolarized; and (c) blue, odd bounce scattering, as well as (d) CPR, as a function of distance from the center of the ray.Following Figure 4, the black line is the median, and the gray area represents the interquartile range.For comparison, the South Pole upper quartile (top dashed line), median (solid line), and lower quartile (bottom dashed line) values for the studied radar parameters are shown.The mosaic-wide quartile values follow the color coding for each m-χ component, with CPR represented by orange.

Figure 6 .
Figure 6.Diffusive CPR (left) and CPR (right) as a function of ((a) and (d)) distance along the ray in terms of Tycho crater radii (42.5 km), as well as distance from north to south for the ((b) and (e)) left and ((c) and (f)) right sides of the ray.Values were calculated over 15 km × 15 km square elements, shown as filled points with corresponding uncertainty for CPR diff (purple) and CPR (orange).Some error bars are smaller than the point size.The 2σ prediction interval for the LSF line, shown in black, is the shaded gray area.

Figure 7 .
Figure 7. Map of m-χ decomposition over the Newton crater group, annotated following Figure 2, where A denotes Newton crater.The image is formed by combining two orthorectified Mini-RF zoom mode (0.03 km resolution) collects.Magenta arrows note examples of craters with extended ejecta deposits.The image ranges from 77°S, 343°E (top) and 81°S, 338°E.North is up.

Figure 8 .
Figure 8.(a) Map of m-χ decomposition over the Haworth candidate landing site region formed by combining two orthorectified Mini-RF zoom mode (0.03 km resolution) collects.The white outline notes the boundaries of the Haworth candidate landing site.The image ranges from 86°S, 331°E (top left) to 88°S, 354°E (bottom right).North is up.Zoom-in of the region noted by the magenta box for (b) m-χ and (c) CPR.Magenta arrows point to the in-text referenced crater's (86°.6S, 331°.9E) extended ejecta deposit.To reduce speckle noise, more obvious in zoomed-in images, a Gaussian smoothing filter was applied to the m-χ while the CPR image was box averaged with a 5 × 5 filter.

Figure 9 .
Figure 9. Zoom-in of a portion of the Haworth candidate landing site (region is to the right of the white line, which indicates a portion of the bounding box) noting illumination and slopes relevant for human exploration of the surface.Areas in yellow correspond to regions that are illuminated a large fraction of the year (>40%), areas in dark turquoise indicate where slopes are less than 5°, and green areas are where both conditions are met.Regions in dark blue indicate areas in permanent shadow (Mazarico et al. 2011).The background image is a hillshade map derived from LOLA data (Barker et al. 2023).The magenta arrow indicates the corresponding crater noted in Figure 8. North is up.