Lunar Boulder Fields as Indicators of Recent Tectonic Activity

Wrinkle ridges are the predominant tectonic structure on the nearside lunar maria. Although lunar wrinkle ridge formation began as early as ∼3.9–4.0 Ga, recent investigations have identified wrinkle ridges in the lunar maria that were tectonically active as recently as the Copernican period of lunar geologic history. Some of those geologically young wrinkle ridges were identified by the presence of dense fields of meter-scale boulders on their scarps and topographic crests. Other investigations have identified recently active lunar wrinkle ridges that lack the ubiquitous presence of boulder fields, thereby rendering the presence of boulder fields ambiguous in the search for ongoing tectonic activity on the Moon. Here we assess boulder populations associated with 1116 wrinkle ridge segments on the lunar maria that are inferred to be recently active (<1.5 Ga) based on their crisp morphologies and crosscutting relationships with small impact craters. We utilize data from the Lunar Reconnaissance Orbiter Mini-RF and Diviner Lunar Radiometer Experiment instruments to assess surface rock populations across these recently active structures. Our results indicate that, where present, meter-scale boulder fields are likely indicators of fault-slip-induced ground acceleration given the short lifespan of lunar surface boulders. However, elevated boulder populations are not observed on all recently active ridges analyzed here. This latter observation supports the notion that wrinkle ridge boulder fields are a nonunique indicator of recent tectonic activity. Furthermore, the spatial distribution of those boulder fields indicates that variable mare protolith properties may play a role in boulder field formation.


Introduction
Meter-scale boulders at the lunar surface break down on geologically short timescales (∼150-300 Ma) due to macroscopic space weathering processes (e.g., Hörz et al. 1975;Basilevsky et al. 2013;Ghent et al. 2014;Molaro et al. 2017;Vanga et al. 2022).Given their short lifetimes on the lunar surface, the presence and distribution of meter-scale boulders can serve as a useful indicator of recent geologic processes.One such recently active process is tectonism.The lunar crust has undergone structural deformation from numerous stress mechanisms at regional and global scales throughout lunar history (e.g., Schultz 1976;Melosh 1978;Watters 1988Watters , 2022)).Geologically recent lunar tectonism and crustal deformation is expressed by small-scale, morphologically undegraded tectonic surface features such as lobate scarps (e.g., Watters et al. 2010Watters et al. , 2015;;Banks et al. 2012;Clark et al. 2017;van der Bogert et al. 2018), graben (Watters et al. 2012;French et al. 2015;Clark et al. 2022), and wrinkle ridges (e.g., Lu et al. 2019;Williams et al. 2019;Nypaver & Thomson 2022a;Frueh et al. 2023) and is supported by shallow moonquakes (e.g., Nakamura et al. 1982) recorded by the Apollo Passive Lunar Seismic experiment and their possible association with lobate scarps (Watters et al. 2019).Such structures have generally been interpreted as the result of a combination of stresses from solid body tides, orbital recession, and global contraction acting on the lunar lithosphere (Watters et al. 2019) in the last ∼1.0Ga of lunar history (e.g., Watters et al. 2010;van der Bogert et al. 2018).
The presence of meter-scale boulder fields on the steeply sloping scarps of lunar wrinkle ridges has been cited by past work as evidence of recent slip along the wrinkle-ridgeforming faults (e.g., French et al. 2019;Valantinas & Schultz 2020).In those investigations, the ground acceleration (seismic shaking) that results from fault slip events was hypothesized to cause the preferential transport of fine-grained material on steeply sloping scarps, with larger, meter-scale boulders being exhumed and retained on the slope (Figure 1).The short lifetime of boulders at the lunar surface indicates that such seismic acceleration has occurred in the Copernican period (<1.1 Ga) of lunar history (e.g., Basilevsky et al. 2013;Ghent et al. 2014).Using this logic, past work has identified numerous wrinkle ridges on the lunar nearside mare that were inferred to be potentially recently active due to the presence of boulders on their bounding scarps (French et al. 2019).Moreover, that work also used boulder dimensions to infer regolith properties and mare lava flow thickness.Another study separately identified a population of wrinkle ridges in the lunar mare inferred to be recently active based on their elevated boulder populations (Valantinas & Schultz 2020).The wrinkle ridges identified in that study-termed the Active Nearside Tectonic System-were attributed to antipodal stresses related to the South Pole-Aitken basin.In a more recent study, 1116 wrinkle ridge segments in the lunar nearside mare were interpreted as recently active, based on their undegraded morphologies and crosscutting relationships with decameterscale primary impact craters (Nypaver & Thomson 2022a, hereafter NT2022).That work noted that, while some of the recently active ridges exhibited dense boulder fields on their bounding scarps (Figure 1(A)), many of the ridges appeared to be morphologically smooth and boulder-free at the resolution scale of the available data, but they did not investigate that observation further.Hence, the utility of boulder fields as a ubiquitous and conclusive direct indicator of recent tectonic activity on the lunar surface remains unclear.
The work presented here is the first comparison of recently active lunar tectonic landforms and enhanced boulder populations where both observations are made independently of each other.As such, our work represents an unbiased comparison of recent lunar tectonic activity and enhanced mass wasting on the lunar surface.The goal of the work presented here is to assess the relationship between boulder fields and recent tectonic activity on the lunar nearside mare.We estimate the relative boulder density along all recently active wrinkle ridge segments proposed in the NT2022 database using newly derived data products from the Miniature Radio-Frequency (Mini-RF) and Diviner Lunar Radiometer Experiment (Diviner) instruments on board the Lunar Reconnaissance Orbiter (LRO).Through this analysis, we present a better understanding of the degree to which the Moon, specifically its nearside mare regions, is recently or currently tectonically active.The work presented here also quantifies the degree of surface alteration and mass wasting caused by fault slip and moonquake events on a regional and semitemporal scale.Such a hazard assessment will be useful for the safety of future lunar surface exploration initiatives (e.g., Wootton & Jablonski 2021).

The NT2022 Wrinkle Ridge Database
The wrinkle ridges measured in this investigation were first documented and mapped in NT2022.For a full explanation of those features and the methods used in their mapping, we recommend a thorough review of that work.The majority of the 1116 wrinkle ridges in the NT2022 database are morphologically small-measuring <1.0 km in width and <10 km in length -and are typically concentrated in discrete clusters of ∼5-10 ridge segments that are oriented roughly parallel to each other along strike.Some of those wrinkle ridges identified in NT2022 are <100 m wide and are not visible at the resolution scale of the Lunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC) Orthophoto base map.Those ridges were found to be recently active based on their crisp morphologies and crosscutting relationships with decameter-scale primary impact craters-neither of which persist on the lunar surface for prolonged periods (10 1 -10 2 Ma) due to space weathering processes, regolith overturn, and distal ejecta mantling (e.g., Soderblom 1970;Fassett & Thomson 2014;Speyerer et al. 2016;Fassett et al. 2022Fassett et al. , 2024)).Although not heavily emphasized in their work, the recently active wrinkle ridges identified in NT2022 consistently occur in close spatial proximity to small (∼5-20 m wide) graben that appear to be kinematically linked to ridge formation (Figure A1).Similar structures have been cited as evidence for formation or reactivation in the larger structures with which they are spatially associated (e.g., French et al. 2015;Clark et al. 2022).Recent diffusion modeling predicts a topographic lifetime of 118.4 Ma for a 20 m wide crater at the lunar surface (Fassett et al. 2022).Hence, while ∼1.5 Ga is presented as an upper limit for the wrinkle ridge activity in NT2022, it is likely that many, if not all, of the structures identified and mapped in that work are younger than the small surface features that they interact with (i.e., ∼<120 Ma).Importantly, the NT2022 wrinkle ridges also include a small percentage of larger (∼1-2 km wide) wrinkle ridges that were also inferred to be recently active based on the aforementioned geomorphic attributes and geologic relationships with small lunar surface features.Those larger ridges are likely to involve older faults that were reactivated in a similarly recent time frame (i.e., ∼<120 Ma).The stresses responsible for the formation or reactivation of the recently active ridges from NT2022 appear to be consistent with those of lobate scarps globally (e.g., Watters et al. 2019).This is to say that, at present, recently active wrinkle ridges on the lunar maria appear to be formed or reactivated by global compressional stresses that are a combination of orbital recession, solid body tides, and global contraction, although further work is necessary to validate that hypothesis (e.g., Watters et al. 2019;NT2022).Though enhanced boulder populations were qualitatively observed in association with the NT2022 wrinkle ridges in some cases, the presence of boulder fields on the scarps of those ridges was not used as a diagnostic indicator of recent tectonic activity in that work.In NT2022, the relationship between boulder fields and ridges was also not quantified, which we do here using Mini-RF δCPR and Diviner rock abundance data sets.

Mini-RF δCPR
The radar data set used to characterize wrinkle ridge rock populations in this work is derived from data collected by the LRO Mini-RF instrument.Mini-RF is a hybrid dual-polarimetric synthetic aperture radar instrument that was originally designed to transmit S-band (∼12.6 cm) and X-band (∼4.2 cm) radar signals in a left circular manner of polarization and receive both the horizontal (H) and vertical (V) orthogonal components of that signal (e.g., Raney et al. 2010).In the work presented here, we exclusively use S-band Mini-RF monostatic data products.The derived H and V components of the received radar signal allow for the calculation of the Stokes parameters and derived radar metrics such as the circular polarization ratio (CPR; e.g., Campbell 2012).The CPR is defined as the ratio of the radar signal returned to the instrument in the same sense (SC) of transmitted polarization over that signal returned in the opposite sense (OC) of circular polarization (e.g., Campbell et al. 2009;Campbell 2012).The OC component of the returned radar signal represents the amount of the incident radar signal that has undergone a single scattering event (i.e., specular scattering) at the lunar surface, whereas the SC component of the signal represents the amount of incident signal that has undergone double-bounce scattering events off of low-loss reflectors (i.e., rocks) at the lunar surface and in the subsurface down to a depth that is ∼10× the radar wavelength, which, in the case of the S band, would be a depth of ∼1.2 m.
The CPR serves as a useful indicator of wavelength-scale rocks at the lunar surface and shallow subsurface.One caveat to the use of CPR as a metric of surface rock populations is the dependence of CPR on local surface topography.Given the reflected nature of radar signals, an instrument-facing slope will reflect a disproportionately large amount of the incident signal back to the radar receiver in the OC sense of circular polarization.That enhanced specular reflection on radar-facing slopes leads to a suppressed CPR (lower than what would be obtained for the same surface/subsurface regolith properties if not on the radar-facing slope).To account for the steep slopes commonly associated with lunar wrinkle ridges, we utilize the δCPR data set (Fassett et al. 2024).The δCPR was derived by modeling the surface response of radar signals with a variable incidence angle and then normalizing the S-band monostatic Mini-RF CPR data set to the expected radar response from a 49°incidence angle.The result of that normalization method was a δCPR data set that empirically mitigates most of the effects of variable local incidence angles-and, therefore, local slope-on the data.The δCPR data set also has improved geospatial control over the prior Mini-RF S-band monostatic data set (i.e., Cahill et al. 2014).For a full description of the δCPR derivation, we refer the reader to Fassett et al. (2024).

Diviner Rock Abundance
In addition to radar data, we use the Diviner rock abundance data sets to assess wrinkle ridge surface rock populations in the work presented here (Bandfield et al. 2011;Powell et al. 2023).Diviner rock abundance is derived using a two-component thermal model of rocks and regolith at the lunar surface that represents the observed nighttime anisothermality over the lunar surface (Bandfield et al. 2011).That data set was originally derived by determining the modeled rock/regolith combination at the lunar surface that best fits the observed lunar surface radiance from channels 6 to 8 of Diviner (Paige et al. 2010).The result of that model fit is a data set that represents the percentage of the lunar surface that is covered by rocks >∼1 m in size.Recent work has improved the Diviner rock abundance data set by correcting for a known pointing error in Divinerʼs elevation actuator, making use of the accrued Diviner data volume to improve effective field-of-view sampling, interpolating radiance values to a consistent local time, accounting for the effects of reflected and emitted radiation from surrounding terrain, and accurately modeling shadows due to nearby local topography (Powell et al. 2023).The result of these improvements is an updated rock abundance data set with a higher effective resolution that provides a better representation of surface rock populations on small, steeply sloping features such as the wrinkle ridges investigated here.

Data Collection
We characterized recently active wrinkle ridges in each of the aforementioned data sets by first building 600 m wide polygon buffers over each wrinkle ridge in the NT2022 database using the buffer tool in ArcGIS Pro 3.0.Those buffers extended to 300 m on either side of the wrinkle ridge crests that were mapped as line features in NT2022.Given that the majority of measured wrinkle ridges ranged in width from ∼0.03 to 0.9 km, the ridge buffer distance was chosen to accurately capture the RA and δCPR of each ridge without including excessive background data values or missing any boulder-bearing slopes altogether.For those wider, potentially reactivated ridges in the NT2022 database, the corresponding feature polyline was drawn on the vergent side of the wrinkle ridge at the top of the ridge scarp, thus ensuring that the overlying buffers would sample the area most likely to exhibit dense boulder fields.We extracted the median RA and δCPR values within each buffer using the ArcGIS zonal statistics tool.By default, the zonal statistics tool establishes a sampling grid that matches the input data pixel scale (∼237 m pixel −1 for Diviner, 90 m pixel −1 for δCPR) so that only those pixels with an exact center point covered by the buffer are sampled.However, that center point sampling may lead to an increased sampling of background terrain in a scenario where small ridges are well within a buffer.Similarly, rock abundance pixels that are just on the edge of a buffer may be excluded from the median.As a solution to this sampling issue, we reduced the zonal stats sampling grid to 10 m pixel −1 for the work presented here.This artificial up-sampling of the δCPR and rock abundance data under the 600 m wide wrinkle ridge buffers can be done in the environments settings of the zonal statistics tool.8Hence, instead of the default sampling of a δCPR or RA pixel, the sampling method used here is weighted by the amount of the pixel covered by the buffer.This sampling method remains imperfect in that it fails to completely account for the variable widths of the wrinkle ridges sampled here.Although most of the NT2022 wrinkle ridges were similar in width, some of those ridges were narrower than the standard 600 m buffer used for the calculation of RA and δCPR medians.In those instances, more background terrain is incorporated into the corresponding median data value, thereby potentially decreasing the resulting median RA or δCPR median.Future work would benefit from improved handling of wrinkle ridge width variations and/or a better understanding of how feature width may affect the corresponding median RA or δCPR values.The effects of variable wrinkle ridge widths on our data are discussed further in Section 4 of this manuscript.

Results
Of the 1116 wrinkle ridges sampled in RA data, 169 ridges (15.1%) exhibited RA median values above 0.01, 605 ridges (54.2%) exhibited RA median values greater than 0.005, and 1086 ridges (97.3%) exhibited RA median values greater than 0.002 (Figure 2(B)).Past work has indicated that the background intercrater RA for the nearside lunar maria is ∼0.004 ± 0.001 (Vanga et al. 2022), though young, fresh impact crater ejecta deposits may display RA median values of ∼0.01 or higher (e.g., Bandfield et al. 2011;Ghent et al. 2014).The wrinkle ridges with the highest relative RA are located in Mare Serenitatis, Mare Frigoris, Mare Cognitum, and Mare Humorum (Figure 2(A)).Small wrinkle ridges in Mare Imbrium and central Mare Procellarum exhibit consistently low RA median values relative to other ridges in surrounding maria.However, low RA ridges exist in all major mare basins.In some cases, low RA ridges occur in close spatial proximity to the highest RA ridges documented in this work.In addition to median RA values, we also collected 95th percentile RA values for all recently active ridges (similar to the methods used by Ghent et al. 2014).Those wrinkle ridge 95th percentile values show similar regional trends to the median RA values and a nonnormal (right-skewed), unimodal distribution (Figure 3).
Due to the reduced data coverage of Mini-RF relative to the RA data set, we only sampled 419 small wrinkle ridges (37.2% of the total NT2022 database) in the Mini-RF δCPR radar data.Of those ridges sampled, 29 ridges (6.9%) exhibited δCPR median values higher than 0.5 compared to an observed ).Many of those regions that contain ridges with relatively elevated δCPR median values also appear to contain ridges with elevated RA values.Similar to the largescale regional distribution of wrinkle ridge median RA, Mare Humorum, Mare Serenitatis, Mare Cognitum, and Mare Frigoris contain ridges with elevated δCPR median values (Figure 4(A)).In contrast to maria with similar wrinkle ridge δCPR and RA values, differences between the δCPR and RA wrinkle ridge maps presented here include a cluster of ridges in NE Mare Imbrium (2.8566237°W 45.4833852°N) that have high δCPR and low RA and a small, isolated cluster of high δCPR/low RA ridges in central Mare Procellarum (56.7421026°W 14.7672486°N).

Discussion
In this work, we use multiple remote sensing data sets to measure the surface and subsurface rock populations associated with 1116 small wrinkle ridge segments on the lunar surface that were inferred by NT2022 to be recently active based on their morphologies and stratigraphic relationships with small impact craters.Past work has identified young impact ejecta deposits as some of the rockiest terrains on the lunar surface where RA values commonly exceed 0.01-0.02(e.g., Bandfield et al. 2011;Ghent et al. 2014).In contrast to fresh ejecta deposits, the background RA of the lunar nearside mare commonly ranges from ∼0.003 to 0.005 (e.g., Cahill et al. 2014;Vanga et al. 2022).Our analysis indicates that many of the boulder fields sampled along wrinkle ridge scarps exhibit RA values that are equivalent to those associated with geologically young (i.e., Copernican) ejecta deposits (i.e., >0.01).However, we consider any RA value associated with a wrinkle ridge to be elevated if the corresponding RA median is greater than that of the background mare terrain (i.e., >0.005).Our work demonstrates that the highest RA pixel values (i.e., >0.01) observed here are indicative of dense boulder fields on steeply sloping wrinkle ridge slopes (Figures 5 and 6), whereas the lowest RA ridges (i.e., <0.005) in our results exhibit few or no boulders at their surface (Figure 7).We can validate the sensitivities of RA data to meter-scale rocks at the lunar surface using LROC Narrow Angle Camera (NAC) data over wrinkle ridges in Mare Cognitum (Figures 5 and 6) and Mare Imbrium (Figure 7).Those local analyses indicate that the dense boulder fields with visibly higher optical albedos in LROC NAC data are spatially coincident with the higher RA pixels.In contrast to the high RA values over the wrinkle ridges (Figures 1, 5, and  6), the surrounding terrain exhibits low RA values and surface rock populations.Furthermore, elevated RA pixel values associated with a narrow, <50 m wide ridge (Figure 5) appear to refute the previous assertion that small-scale, recently active ridges in the lunar maria would not be discernable in the Diviner RA data set (i.e., NT2022).Our results demonstrate that when dense boulder fields are present on their scarps, even the smallest wrinkle ridges in the NT2022 database are visible in the Diviner RA data sets.
Past investigations of CPR variability have indicated that the freshest impact ejecta and melt deposits at the lunar surface exhibit CPR and δCPR values of ∼0.6-1.0 (e.g., Neish et al. 2013;Jawin et al. 2014;Fassett et al. 2024), whereas background mare CPR values are typically ∼0.3-0.5 (e.g.,  A2) demonstrates that agreement between individual ridge RA and δCPR medians, such as this one, is uncommon.Cahill et al. 2014;Nypaver et al. 2021;Fassett et al. 2024).Hence, we refer to any wrinkle ridge δCPR values >0.5 as "high" relative to the surrounding terrain, and we infer those high δCPR ridges to exhibit increased surface and subsurface rock populations.In general, we find good regional agreement between the high RA and high δCPR ridges.For example, wrinkle ridges with elevated RA and δCPR values are present in Mare Humorum, Frigoris, and Cognitum.However, aside from several isolated ridges (Figure 6), a direct comparison between wrinkle ridge RA and δCPR reveals considerable discrepancies and no clear correlation between the median RA and median δCPR of individual ridges (Figure A2).Several explanations exist for the observed RA/δCPR discrepancies.First, surface and subsurface rock populations may be affecting the δCPR median values to differing degrees.For example, a ridge with elevated subsurface rock populations but relatively few surface rocks may exhibit a low RA median but a high δCPR median.Second, the highly variable loss properties of the maria (e.g., Campbell et al. 2014) may influence the contribution of subsurface rock populations to the δCPR signature from location to location.Alternatively, the discrepancy may be due to artifacts within the Mini-RF monostatic data set or the differing statistical distributions of the RA and δCPR data sets.A known across-track gradient within some Mini-RF monostatic data swaths may result in some wrinkle ridge buffers-particularly those associated with N-S striking ridges-sampling artificially elevated δCPR pixels and exhibiting an elevated δCPR median.Given the consistent correlation between RA and high-albedo boulder fields observed in NAC data (Figures 5 and 6), the observed RA/δCPR discrepancy indicates that δCPR is potentially a nonunique identifier of wrinkle ridge boulder fields on a local scale and may be more well suited for larger-scale, regional analyses of surface and subsurface rock populations.These ambiguities surrounding δCPR are not inconsistent with past work that investigated impact ejecta breakdown using Mini-RF S-band CPR data and Diviner RA data (Ghent et al. 2016).That work determined that, while an ejecta breakdown trend was decipherable using Diviner RA data, a similar trend could not be established using Mini-RF CPR data for those same impact ejecta deposits due to the prolonged retention of subsurface rock populations.In light of these uncertainties surrounding δCPR, we focus our discussion and interpretations on the RA results described in Section 3. Future work comparing Diviner rock abundance data with other radar polarimetric parameters, such as a closer examination of radar scattering behavior as observed with the m-χ decomposition or degree of linear polarization, may yield additional insight into the δCPR variability observed here and the sensitivity to CPR data sets to surface rock populations (e.g., Rivera-Valentín et al. 2024).
As discussed in Section 2.4, the standard 600 m buffer sampling method is an inherent limitation of our work.One alternative method would be to manually map the topographic and structural boundaries of each wrinkle ridge in the NT2022 database and then use those mapped boundaries to bin the corresponding data values.However, doing so properly at a consistent mapping scale would require a significant volume of LROC NAC or other high-resolution orthophoto data covering all of the ridges in the NT2022 database with a consistent solar incidence angle.Such a task would require significant time components corresponding to image processing and mapping and is beyond the scope of the work presented here.However, a direct comparison of RA values collected over a sample set of 50 ridges with varying widths and median RA values in Mare Humorum indicates that there is good agreement between the results derived from both data collection methods (i.e., standardized buffer and mapped boundary; Figures 8 and  A3).Hence, our method of standardized wrinkle ridge buffer data collection provides a good approximation of the RA and δCPR associated with each ridge in the NT2022 database.
Based on the wrinkle ridge RA values derived here, 47.2% of wrinkle ridges interpreted as recently active (NT2022) have moderate or low boulder populations.This interpretation is supported by the histogram distribution of wrinkle ridge RA values relative to the documented RA associated with blocky ejecta deposits and rock-free surfaces (Bandfield et al. 2011;Ghent et al. 2016;Vanga et al. 2022).For example, young, Copernican-age impact ejecta blankets can exhibit RA medians >0.01 (Ghent et al. 2014(Ghent et al. , 2016;;Nypaver et al. 2021), whereas the background RA of the nearside maria is estimated to be <0.005(Bandfield et al. 2011;Vanga et al. 2022).Those low RA, recently active wrinkle ridges are unlikely to be devoid of rocks entirely, but the dense boulder fields that have been documented on some wrinkle ridge scarp faces (Figures 1, 5, and 6;French et al. 2019) are not ubiquitous at the resolution scale of the remote sensing data.The observed lack of wrinkle ridge boulder fields is also supported by the presence of low Mini-RF δCPR median values representing meter-scale rock fragments at the lunar surface and in the subsurface down to a depth of ∼1.2 m (∼10× the S-band wavelength).However, the lack of a correlation between wrinkle ridge RA and δCPR median values indicates that further work is necessary to validate the utility of δCPR as a metric of surface rock content The agreement between the results of these two sampling strategies indicates that a standardized buffer is a suitable approximation for true wrinkle ridge widths.A visual representation of these two data collection methods is presented in supplemental information (Figure A3). on the small-scale features such as those investigated here.The aforementioned variability in wrinkle ridge δCPR and RA data supports the notion that recent tectonic activity and the resulting coseismic ground shaking does not consistently result in the formation of dense boulder fields on small lunar wrinkle ridges.Such a hypothesis could only be fully supported if dense boulder fields were found along all recently active wrinkle ridge scarps and crests.Where present, dense boulder fields may indicate recent seismicity from reactivated ridge faults; but, since not all recently active wrinkle ridges exhibit dense boulder fields, there are likely additional conditions controlling the presence of rock fields, some plausibly unrelated to tectonic activity.Thus, boulder fields along wrinkle ridges are only a partial indicator of recent tectonic activity and/or seismicity, and any documentation of recently active tectonic features based solely on the presence of boulder fields is likely to be incomplete.However, the criteria set forth as a definitive indicator of recent tectonic activity in NT2022 may be incomplete as well.That study inferred the wrinkle ridges' crosscutting relationships with small lunar impact craters and crisp morphologies to indicate recent activity.However, numerous wrinkle ridges on the lunar maria exhibit boulder fields but are topographically muted with no visible boundaries between the ridge scarps and flat-lying background terrain (Figure 9).Hence, it appears that wrinkle ridge morphology and crosscutting relationships with other surface features may also be nonunique indicators of recent tectonic activity in maria.One possible scenario is that the presence of a crisp scarp boundary and crosscut impact craters indicates recent ridge formation, whereas the presence of boulder fields alone indicates recent reactivation of the fault associated with the wrinkle ridge in question.This hypothesis remains speculative and still does not explain the lack of ubiquitous boulder field presence on what may be the recently formed wrinkle ridges assessed here-i.e., some other geologic mechanism or attribute must play a role in the spatial distribution of boulder fields on the nearside lunar maria.Alternatively, those boulder fields located on topographically muted ridges may have been exposed by seismic activity, such as impact-induced seismicity, that is unrelated to slip along the ridge-forming fault.
One potential boulder field formation mechanism may be regolith draining (e.g., Miyamoto et al. 2007).Any deformation of the mare basalts-or regolith in the case of small, decameterscale wrinkle ridges-is likely to result in the creation of void space along the associated fault zone given the high porosity and pulverized nature of the lunar regolith (e.g., Hapke & Sato 2016;Wyrick & Buczkowski 2022).As the upper mare basalts and regolith buckle under compressional stresses to form wrinkle ridges, it is expected to result in layer parallel extension due to flexural bending (Watters 1988;Watters & Johnson 2010).The layer parallel extension that coincides with ridge formation likely opens subsurface void spaces into which surface regolith may drain and leave behind only the larger rock fragments and boulders at the surface (a form of granular convection or "Brazil nut effect" in which small particles can more readily fill in void space left under large particles during vertical motion, causing net exhumation of large grains to the surface).In addition to testing this regolith drainage hypothesis, future work focused on establishing a positive correlation between wrinkle ridge slope and surface boulder density may yield new insights into lunar boulder field formation.Further analog or laboratory investigations of fault structures in loosely compacted alluvium or other sediments are also likely to improve our understanding of how lunar boulder fields form and evolve over time.
The nonuniform spatial distribution of high δCPR and RA wrinkle ridges in the nearside lunar maria also supports the notion of a regional substrate and or protolith control on regolith development and surface rock populations (e.g., Head & Wilson 2020).Recent work has indicated that some mare surfaces, such as Mare Humorum, S. Mare Procellarum, and the features contained therein, are rockier than they are expected to be given their geologic age (e.g., Cahill et al. 2014;Vanga et al. 2022;Chertok et al. 2023;Elder et al. 2023).It was postulated that those surfaces were underlain by a mare basalt that was more resistant to macroscopic space weathering processes or more conducive to the production of larger rocks (e.g., Chertok et al. 2023).When those more resistant rock layers are fractured and exhumed by impacts and/or structural uplift, the exposed rocks survive under lunar surface weathering conditions for prolonged periods of time.In the work presented here, we also observe increased populations of high RA wrinkle ridges in Mare Humorum and in parts of Mare Procellarum.Additionally, we note that Mare Cognitum and Mare Frigoris exhibit enhanced populations of high RA wrinkle ridges.Such regional variations in wrinkle ridge rock populations support the preexisting hypothesis of a substrate control (i.e., mare basalt thickness, basalt layering, mechanical strength, etc.) on surface rock populations and, more specifically, wrinkle ridge boulder populations (French et al. 2019).

Conclusions
In the work presented here, we systematically assess the surface rock populations associated with ∼1116 small, recently active wrinkle ridges using δCPR and RA data from the LRO Mini-RF and Diviner instruments, respectively (Powell et al. 2023;Fassett et al. 2024).The geologic sensitivities and slope corrections associated with both of those derived data products make them ideal data sets for measuring the presence of dense boulder fields on the slopes of lunar wrinkle ridges.Based on the results of our investigation, we put forth the following conclusions.
1.While the majority (54.2%) of recently active wrinkle ridges measured in this work exhibit elevated rock populations according to their respective Diviner rock abundance values, a portion (47.8%) of our wrinkle ridge sample set exhibits rock populations that are roughly equivalent to or indistinguishable from the background terrain of the nearside lunar maria.2. Given the short (<1.1 Ga) lifetime of meter-scale boulders on the lunar surface (e.g., Basilevsky et al. 2013;Ghent et al. 2016), those wrinkle ridges that do exhibit enhanced boulder populations have likely undergone some amount of seismic activity/ground acceleration due to slip along the associated fault structure.However, the lack of ubiquitous boulder fields across our sample set of recently active wrinkle ridges indicates that the presence of boulder fields is likely an incomplete indicator of recent tectonic activity.3. The nonrandom distribution of elevated δCPR and RA values regionally across the nearside lunar maria supports the previously established hypothesis that rock populations on the nearside lunar maria are likely to be at least partially controlled by variable geologic properties of the preexisting bedrock.

Figure 1 .
Figure 1.(A) LROC NAC image (NAC M1197583152; 1.2 m pixel −1 ) of a wrinkle ridge in East Mare Serenitatis showing boulder fields on the east-facing scarp (yellow arrows).The blue inset box shows the density of meter-scale boulders present over one small area of the wrinkle ridge scarp.(B) LOLA/Kaguya ∼60 m pixel −1 slope map over the wrinkle ridge from Figure 1(A).

Figure 2 .
Figure 2. (A) NT2022 recently active wrinkle ridges colorized by median rock abundance (RA) value overlaid onto the LROC WAC global mosaic (∼100 m pixel −1 ).The highest 5% RA wrinkle ridges are primarily located in S. Mare Cognitum, Mare Procellarum, Mare Humorum, and Mare Serenitatis.Ridges with the lowest 5% RA are concentrated in Mare Imbrium and central Mare Procellarum.(B) The overall distribution of wrinkle ridge RA median values for every ridge in the NT2022 database.

Figure 3 .
Figure 3. NT2022 recently active wrinkle ridges colorized by rock abundance (RA) 95th percentile value overlaid onto the LROC WAC global mosaic (∼100 m pixel −1 ).(B) The overall distribution of wrinkle ridge RA 95th percentile values for every ridge in the NT2022 database.

Figure 4 .
Figure 4. (A) NT2022 recently active wrinkle ridges colorized by median δCPR value overlaid onto the LROC WAC global mosaic (∼100 m pixel −1 ).Due to the limited coverage of the Mini-RF data set, only 419 of the 1116 total wrinkle ridges from the NT2022 database are measured in δCPR data.(B) The overall distribution of wrinkle ridge δCPR median values for every sampled ridge in the NT2022 database.

Figure 5 .
Figure 5.A recently active wrinkle ridge from the NT2022 wrinkle ridge database in Mare Cognitum (19.89°W 12.46°S) shown in LROC NAC image data (A; NAC M1175531291) and overlain by LRO Diviner RA data (B; Powell et al. 2023).This wrinkle ridge segment ranges from ∼30 to 100 m in width and exhibits multiple high-albedo boulder fields along strike (C; red inset).Diviner RA pixels remain elevated along the ridge as a result of the exposed boulder fields.

Figure 6 .
Figure 6.NE-SW striking wrinkle ridge from the NT2022 database located in Mare Serenitatis (4.009°W 45.997°N) shown in Diviner RA data (A; Powell et al. 2023), δCPR (B; Fassett et al. 2024), and LROC NAC data (C; LROC NAC M1304401329RE).Boulder fields on this ridge exhibit visibly high albedo in LROC NAC data (C) with corresponding elevated δCPR and RA pixel values.A direct comparison of RA and δCPR median values for 419 wrinkle ridges with δCPR coverage (i.e., Figure A2) demonstrates that agreement between individual ridge RA and δCPR medians, such as this one, is uncommon.

Figure 7 .
Figure 7. E-W striking wrinkle ridge cluster from the NT2022 database located in NE Mare Imbrium (4.009°W 45.997°N) shown in LROC NAC data (A; LROC NAC M1274280167RE) and Diviner RA data (B; Powell et al. 2023).Similar to other wrinkle ridges in Mare Imbrium, these structures do not exhibit visible boulder populations on their scarps or crests in LROC NAC data (C; red inset) even though they appear to be recently active based on the presence of meter-scale graben (D; yellow inset).Similarly, the Diviner RA pixels associated with each ridge are not elevated relative to the background terrain.

Figure 8 .
Figure 8.A comparison of wrinkle ridge median RA (A), 95th percentile RA (B), and median δCPR values (C) collected over a sample set of recently active wrinkle ridges in SE Mare Humorum (n = 51 ridges in RA, n = 24 ridges in δCPR).The agreement between the results of these two sampling strategies indicates that a standardized buffer is a suitable approximation for true wrinkle ridge widths.A visual representation of these two data collection methods is presented in supplemental information (FigureA3).

Figure 9 .
Figure 9.An example of a topographically muted lunar wrinkle ridge with increased boulder populations (11.12°S, 37.04°W).Unlike those ridges documented in NT2022, this ridge does not exhibit a crisp boundary between the wrinkle ridge scarps and the surrounding terrain or any crosscutting relationships with small surrounding craters.Hence, this ridge was not included in the NT2022 recently active wrinkle ridge database, but the presence of boulder fields on the ridge scarps and crest may indicate potential recent tectonic activity (NAC M147225042RC, M147225042LC).

Figure A3 .
Figure A3.Mapped wrinkle ridge boundaries overlain onto the 600 m standardized ridge buffers for the subset of recently active ridges in Mare Humorum used to derive the quantitative relationships in Figure7.Most of the ridges sampled here are contained within the standard buffers, but several exceed their boundaries.For those wider ridges, the buffers were drawn to include the vergent side of the ridge, where the majority of the boulder fields should be located if present.