Heterogeneity of the Noachian Crust of Mars Using CRISM Multispectral Mapping Data

We used Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) multispectral mapping data to assess heterogeneity of primary and secondary mineral compositions of Noachian‐aged regions of Martian crust. Multispectral data corroborate interpretations from CRISM targeted observations and Observatoire pour la Mineralogie, l’Eau, les Glaces et l’Activité global mapping of large horizontal differences in vertical crustal structures, and degree and grade of alteration to secondary minerals. CRISM multispectral data analysis conducted at a combination of high spatial resolution and coverage not available in other data sets also reveals previously unrecognized exposures. At one extreme, basaltic crustal material is minimally altered, mostly to smectite clay; at the other, Al‐enriched alteration products are present, alteration is widespread, and superposed surface materials rich in salts and precipitates, suggesting multiple episodes of alteration. The revealed vertical structure of primary mineralogy is consistent with that inferred in previous studies. Controls on the extent and metamorphic grade of secondary mineral assemblages are proximity to heat sources including large impact basins and inferred magmatic bodies.


Introduction
Observations of the Noachian crust of Mars by the Thermal Emission Spectrometer on Mars Global Surveyor (Christensen et al., 2001), the Thermal Emission Imaging System (THEMIS) on Mars Odyssey (Christensen et al., 2004), the Observatoire pour la Mineralogie, l'Eau, les Glaces et l'Activité (OMEGA) onboard Mars Express (MEx) , and targeted observations by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) on Mars Reconnaissance Orbiter (MRO) (Murchie et al., 2007) have provided a first-order understanding of the primary and secondary mineralogies present in the Martian crust. The highest spatial and spectral resolution at the visible and short-wave infrared wavelengths that are most sensitive to secondary minerals comes from CRISM targeted observations (544 wavelengths at 0.4-3.9 μm, 18-36 m/pixel). Targeted observations cover ∼2%-3% of Mars' surface, often targeted at locations where aqueous alteration was known previously from THEMIS or OMEGA (Murchie et al., 2007;Murchie, Seelos, et al., 2009); sampling of the surface is thus biased and non-uniform. Aggregated TES, THEMIS, OMEGA, and CRISM data indicate a predominantly basaltic Noachian crustal composition, whose primary mineralogies dominate the spectral signature of the crust in almost all areas Christensen et al., 2000;Murchie et al., 2019;Mustard et al., 2005;Ody et al., 2012;Poulet et al., 2007Poulet et al., , 2009Riu et al., 2019;Ruff & Christensen, 2002). In CRISM and OMEGA data, low-Ca pyroxene (LCP) dominates the spectral signature of southern highlands units (Mustard et al., 2005) that formed during the pre-Noachian through mid-Noachian periods. Rocks dominated spectrally by high-Ca pyroxene (HCP) form younger volcanic units that are late Noachian to Amazonian in age, indicating a shift in igneous rock composition in the late Noachian around ∼3.5 Ga (Mustard et al., 2005;. Coherent outcrops dominated by olivine have a range of ages, and are found in multiple settings (Ody et al., 2013;Poulet et al., 2007): in the rims and ejecta of Argyre, Hellas, and Isidis and in central peaks of large craters (Skok et al., 2012); as outcrops in the lower walls of Valles Marineris (Viviano-Beck et al., 2017) Abstract We used Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) multispectral mapping data to assess heterogeneity of primary and secondary mineral compositions of Noachian-aged regions of Martian crust. Multispectral data corroborate interpretations from CRISM targeted observations and Observatoire pour la Mineralogie, l'Eau, les Glaces et l'Activité global mapping of large horizontal differences in vertical crustal structures, and degree and grade of alteration to secondary minerals. CRISM multispectral data analysis conducted at a combination of high spatial resolution and coverage not available in other data sets also reveals previously unrecognized exposures. At one extreme, basaltic crustal material is minimally altered, mostly to smectite clay; at the other, Al-enriched alteration products are present, alteration is widespread, and superposed surface materials rich in salts and precipitates, suggesting multiple episodes of alteration. The revealed vertical structure of primary mineralogy is consistent with that inferred in previous studies. Controls on the extent and metamorphic grade of secondary mineral assemblages are proximity to heat sources including large impact basins and inferred magmatic bodies. emplaced at least in part as dikes (Flahaut et al., 2011; in some intercrater plains (McBeck et al., 2014;Ody et al., 2013;Rogers & Fergason, 2011); and as buttes near Hellas (Phillips et al., 2022).
A wide range of secondary minerals is found in different geologic settings, in approximate order of decreasing age Ehlmann & Edwards, 2014;Murchie, Mustard, et al., 2009;Murchie et al., 2019): (a) hydrated and hydroxylated silicates and silica glass, some of which have been exhumed from depth with some formed by the impact process associated with craters and basins in the Noachian highlands crust (following paragraph); (b) carbonates of differing composition and stratigraphic position formed during the pre-Noachian to late Noachian periods (Brown et al., 2010;Edwards & Ehlmann, 2015;Ehlmann et al., 2008;Viviano et al., 2013;Viviano-Beck et al., 2017;Wray et al., 2016); (c) two classes of lacustrine deposits: fan-shaped deposits of detrital sediments and silica within hundreds of craters and other topographic lows at the mouths inflowing valley networks , Goudge et al., 2017Pan et al., 2021), and deposits bearing acid sulfates and kaolinite in deep craters that lack inflowing channels, providing evidence for river-and groundwater-fed lakes, respectively Wray et al., 2011); (d) a shallow layer of compositionally stratified phyllosilicate with Fe/Mg-rich clay typically overlain by Al-rich clay, which is itself discontinuously overlain by silica, sulfates, and/or amorphous material, occurring in scattered highland locations widely interpreted as remnants of a pedogenic layer (Bishop & Rampe, 2016;Bishop et al., 2008;Le Deit et al., 2012;Loizeau et al., 2007;McKeown et al., 2009;Wray et al., 2008Wray et al., , 2009; (e) scattered, eroded remnants of late Noachian and younger sulfate-rich layered deposits superposed unconformably on older eroded surfaces (Arvidson et al., 2005; see review by Murchie et al., 2019); (f) silica and other alteration products associated with volcanic constructs (Ackiss et al., 2018;Skok et al., 2010); and (g) thin beds of silica and acid sulfates within and adjacent to Valles Marineris (e.g., Milliken et al., 2008).
Hydrated and hydroxylated silicates exhumed from depth in the Noachian crust are the most widespread secondary mineral assemblage that formed under aqueous conditions Ehlmann & Edwards, 2014). The most common minerals are Fe/Mg-smectite clays, which dominate at depths of origin from 0 to 5 km, and chlorites, which dominate at greater depths to ≥7 km (Sun & Milliken, 2015). Preservation of smectite at 0.3-5 km depth without alteration to chlorite or illite may have resulted from the limited availability of liquid water (Tosca & Knoll, 2009). In some places, other secondary minerals indicate higher-temperature metamorphism through zeolite to sub-greenschist (prehnite-pumpellyite) facies (Ehlmann et al., 2011;McSween et al., 2015). Zeolite, serpentine, and prehnite indicate temperatures ranging from ambient to 400°C (Ehlmann et al., , 2011. There is evidence for both magmatic and impact heating having supported low-grade metamorphic alteration. Evidence for enhanced heating in the crust includes the occurrence of minerals that form at higher temperatures nearby dike-intruded tectonic fractures (Viviano-Beck et al., 2017), alteration assemblages that require enhanced geothermal gradients (McSween et al., 2015), and formation of silica-bearing secondary mineral assemblages in close association with volcanic constructs (Ackiss et al., 2018;Skok et al., 2010). Other evidence supporting a major role of impact heating in the alteration of the Noachian crust includes the association of silica, prehnite, and serpentine with the Isidis and Hellas impact basins (e.g., Ehlmann et al., 2009Ehlmann et al., , 2010 as well as with highlands impact craters (Sun & Milliken, 2015Tornabene et al., 2013). Alteration at significant depth is discontinuous, occurring over domains that are not traceable for hundreds or even several tens of kilometers, consistent with alteration by localized impact heating (Ding et al., 2015;Viviano-Beck et al., 2017) and/or limited by a discontinuous supply of liquid water (Sun & Milliken, 2015. Global surveys using both high spatial resolution CRISM targeted observations and OMEGA global mapping at lower spatial resolution suggest that the alteration of Noachian crust is regionally heterogeneous Riu et al., 2022).
Here we utilize a complementary part of the CRISM data set distinct from targeted observations to investigate heterogeneity in composition and alteration of the Martian crust: multispectral mapping data (Murchie et al., 2007;Murchie, Seelos, et al., 2009;Seelos et al., 2022). These data fill a gap in resolution and coverage between CRISM targeted observations and OMEGA global mapping: ∼86% areal coverage in 72 bands clustered at the wavelengths of key absorptions, with data binned spatially to ∼180 m/pixel which is several times better spatial resolution than OMEGA but lacking fully hyperspectral sampling. Challenges to the interpretability of CRISM multispectral data compared to targeted observations are (a) fewer wavelengths, which degrade separation of closely related Fe/Mg-or Al-rich hydroxylated silicates, and (b) lower signal-to-noise ratio (SNR) at ≥3 μm, which challenge the identification of chlorides and carbonates. Hence, the focus of our analysis is both reproducing results from earlier data, and improved coverage and resolution supporting discoveries of new mineral exposures. We have analyzed the distribution of chlorides separately using targeted observations in a companion study (Beck et al., 2019), and found their ages and spatial distribution to be consistent with results reported by Osterloo et al. (2008Osterloo et al. ( , 2010 and Leask and Ehlmann (2022).

Study Regions
We used multispectral mapping data to ascertain spatial variations in composition and alteration of the Noachian crust by determining the mineralogy and mineral stratigraphy of the following three study areas reported by  to span a range of alteration environments: a relatively high spatial density of alteration (Terra Sirenum), a moderate spatial density of alteration (Tyrrhena Terra), and a relatively low spatial density (Margaritifer Terra). These areas have been the subject of numerous previous studies that used CRISM targeted observations. Margaritifer Terra consists of highland plains, chaos terrain, outflow channels, and sedimentary deposits indicative of massive flooding. Its surface is superposed by widely scattered, mid-to late-Noachian examples of possible pedogenic horizons (Buczkowski et al., 2021;Le Deit et al., 2012;Wilson et al., 2018). It also contains portions of the Uzboi-Ladon-Morava valley outflow system (including Ladon, Eberswalde, and Holden craters), formed near the Noachian-Hesperian boundary and modified during the Hesperian to early Amazonian (e.g., Grant, 1987). The spectral similarity of clay minerals within fan and associated deposits in these craters suggests these local lows acted as sinks during sedimentary transport (Milliken & Bish, 2010). Mineralogy suggests transport and redeposition of materials sourced from altered Noachian crust in the drainage basin (Poulet et al., 2014).
The more diverse and higher-grade metamorphic alteration in Tyrrhena Terra has been inferred, based on the spatial distribution of alteration, to have been driven at least in part by heating from the Hellas-and Isidis basin-forming impacts (Loizeau et al., 2012;Viviano & Phillips, 2019). Unlike Margaritifer Terra, Tyrrhena Terra also contains a thick near-surface layer of overlapping ejecta from the two large basins. The northern Hellas rim also reveals the most extensive exposure of feldspathic material identified thus far on Mars, in blocks of pre-Noachian crust uplifted by the Hellas basin-forming impact (Phillips et al., 2022). These materials are identified by a signature of Fe-bearing plagioclase using CRISM data Phillips et al., 2022;Wray et al., 2013).
Terra Sirenum lies southwest of Tharsis, and is unique in its exposure of diverse hydrated mineral deposits, with significant occurrences of chlorides, sulfates including gypsum, alunite, and jarosite, Al-and Fe/Mg-phyllosilicates, and hydrated silica (Ehlmann & Edwards, 2014;Osterloo et al., 2008Osterloo et al., , 2010Wray et al., 2011). Sulfates in this region appear restricted to interiors of large, deep craters, suggesting paleolakes fed by upwelling groundwater (Wray et al., 2011), whereas locations of chloride deposits are at higher elevation consistent with deposition from ponded surface runoff (Leask & Ehlmann, 2022). A large graben system that extends radially from Tharsis cross-cuts circular mounds that have been interpreted as mud volcanoes sourced from groundwater (Hemmi & Miyamoto, 2017).

Methods
From west to east the locations of the study areas are: (Figure 1, row 1) 17.5°S to 32.5°S, 155°W to 170°W, in Terra Sirenum northeast of Eridania basin and southwest of Arsia Mons, called "Terra Sirenum" hereafter; (Figure 1, row 2) 22.5°S to 32.5°S, 30°W to 40°W, including Holden crater and the Uzboi Vallis, hereafter "Margaritifer Terra"; and (Figure 1, row 3) 12.5°S to 27.5°S, 60°E to 80°E, a region just north of Hellas basin, hereafter "Tyrrhena Terra". For each study area, multiple ∼5° × 5° "tiles" of CRISM multispectral mapping data (72 wavelengths, 180 m/pixel; Seelos et al., 2022) were mosaicked. Terra Sirenum is covered by 9 tiles, Margaritifer by 4 tiles, and Tyrrhena Terra by 12 tiles. In each case, strips of mapping data were custom-processed and mosaicked using an advanced prototype of the pipeline used to generate the final version of tiled, multispectral mapping data that is being delivered to the Planetary Data System (PDS; Seelos et al., 2022). Compared to previously released tiled multispectral mapping data, noise is strongly remediated and inter-strip residuals from atmospheric variations are minimized.
We utilized thematic CRISM browse product color composites comprised of summary parameters (i.e., spectral indices) with related mineralogic significance (Viviano-Beck et al., 2014), custom image analysis, ArcGIS interfaces, and additional data described below to classify and map mineral diversity within the study areas. Mapped regions of interest (ROIs) consist of spatially contiguous regions having a consistent spectral signature judged as representing the underlying geologic unit, distinct from the signature of surrounding widespread dust and sand that blankets most surfaces. The ROIs, shown in Figures 1-3 (panels b, e, g), occur within the intersection of high-quality multispectral data included in the mosaics (e.g., low IR detector temperature, low atmospheric opacity; Seelos et al., 2022) with relatively low surface dust coverage as indicated by the TES dust cover index indicate new locations where kaolinite has been identified compared to previous studies (e.g., Wray et al., 2011) as enabled by the coverage of coverage and spatial resolution in multispectral mapping data. (Ruff & Christensen, 2002). Warmer tones in the TES dust cover index map indicate dustier surfaces and correspond to parts of the study areas where variations in surface composition are either not detectable and/or likely related to eolian sediment ( Figure S1, panels a, c, e in Supporting Information S1). Relative ages of deposits in the ROIs were assigned based on co-registered geologic mapping based on the dominant unit in each ROI (Tanaka et al., 2014; Figure 1, panels a, c, f). Custom RGB CRISM browse products (examples shown in Figure 1d and Figure S1 in Supporting Information S1) were constructed to visualize compositional variations using summary parameters as described by Viviano-Beck et al. (2014).
Analysis of the maps was conducted by a single user (AWB) and reviewed by a second user (CEV) to achieve consistency in mineral identification and mapping across the regions, and was consistently performed at 1:500,000 scale. The identified mineralogies were grouped into two broad categories, (a) primary mineral phases (LCP; HCP; olivine; and plagioclase), and (b) secondary alteration phases, including those with Fe/Mg-OH bonds (chlorite, smectite clay-nontronite or saponite), phases with Al-OH bonds (montmorillonite, kaolinite, or spectrally similar phases, e.g., alunite), hydrated silica (Si-OH bonds), and polyhydrated sulfate. The multispectral data do not easily discriminate polyhydrated sulfate phases from each other or from hydrated chlorine salts or most zeolites, and in general, occurrences of those minerals could be present within the polyhydrated sulfate group. However, polyhydrated sulfate phases were only identified in the Terra Sirenum region, in Columbus crater where hydrated sulfates have been previously reported in analyses of CRISM targeted observations Wray et al., 2011); thus, we use the "polyhydrated sulfate" classification. Spectra from CRISM targeted observations that overlapped mineral identifications in the study areas were extracted to confirm the accurate identification of secondary phases in the multispectral data ( Figure 2). Identification of LCP, HCP, olivine, and plagioclase using CRISM multispectral mapping data has also been demonstrated in previous publications (e.g., Phillips et al., 2022). Extracted spectra from Phillips et al. (2022; supplementary data set G50341) were also used to validate mapping in the Tyrrhena Terra region.
Relative abundances of mineral exposures are reported in this work as pie charts showing the proportion of area identified in that mineral category (i.e., Primary or Secondary) within a given age range (early Noachian-massif, early Noachian-highlands, middle Noachian, late Noachian, and Amazonian-impact unit). For example: % Kaolinite = kaolinite area area in age range and study area classif ied as secondary minerals We note variation in total study area size and in the geographic area in each age range, warranting caution when interpreting differences in abundances of phases between regions and units of different ages. Percentages related to exposure of primary and secondary minerals in each study area and in the total area are reported by age group for comparison of type and degree of alteration. Figure 3 shows relative surface exposure of spectrally-dominant primary and secondary minerals in units of each age range in each of the three study areas. The Magaritifer Terra study area, which from targeted observations shows the least evidence for alteration of the three areas , exhibits characteristics typical of Martian highlands crust reported in this region by studies cited above including Quantin et al. (2012), Le Deit et al. (2012), and Viviano-Beck et al. (2017). Older, mid-Noachian materials have primary mineralogy dominated spectrally by LCP. Younger, late Noachian materials are dominated spectrally by HCP. The mid-and late-Noachian highlands expose alteration to secondary mineralogy dominated by Fe/Mg-phyllosilicates. The low total coverage by secondary minerals (3%) corroborates that this is the least altered and diverse in secondary mineralogy of the three study areas. This same conclusion is supported by global mapping of hydrated silicate abundance using OMEGA data (Riu et al., 2022). Significant phyllosilicate exposures are observed in the walls of Nirgal Vallis, consistent with findings from Wilson et al. (2018) and Buczkowski et al. (2021), although our mapping of the altered wall unit extends further west yielding improved coverage and determination of extent of the deposit. These exposures are consistent with a shallow, pedogenic horizon of alteration in this region (Buczkowski et al., 2021;Le Deit et al., 2012). Fe/Mg-phyllosilicates are also observed in the ejecta materials of small craters in the mid-and late-Noachian highland terrain, and in the rim and ejecta of Amazonian/Hesperian-aged Holden crater.

Results
The Tyrrhena Terra study area exhibits greater mineralogic diversity and a higher degree of alteration (6%) than the Margaritifer Terra study area, consistent with results reported from studies of CRISM targeted and Figure 2. Sample corroborating Compact Reconnaissance Imaging Spectrometer for Mars targeted-observation spectra of mineral phases mapped in the study areas, ratioed to in-column spectrally-bland materials. Alphanumeric identifiers are observation IDs of observations from which the spectra are extracted (see Table S1 in Supporting Information S1). Laboratory spectra of analog materials plotted below each endmember (thick lines) are from Viviano-Beck et al. (2014); identifiers represent spectra archived in the Planetary Data System Geosciences Node. mapping observations of that region (e.g., Loizeau et al., 2012;Phillips et al., 2022;Viviano & Phillips, 2019) and from OMEGA global mapping (Riu et al., 2022). The temporal evolution in the spectrally dominant pyroxene, from LCP in the early-Noachian to HCP in the late-Noachian in Tyrrhena Terra, corresponds to CRISM targeted  and OMEGA global (Mustard et al., 2005) observations. However, compared to Margaritifer, olivine is more abundant in units of all ages including underlying basin ejecta (early-Noachian massif unit) and superposed younger plains. Plagioclase-rich material occurs predominantly in the massif unit, suggesting pre-Noachian formation in material that pre-dates the Hellas impact basin, which is consistent with interpretations by Phillips et al. (2022). Finally, crustal materials are more altered in Tyrrhena than Margaritifer Terra: a larger fraction of the exposed substrate exhibits a spectral signature of secondary minerals (6% vs. 3%, respectively). Enhanced hydrothermal and/or metamorphic conditions are also evidenced by a greater proportion of chlorite in mapped early to mid-Noachian and Amazonian/Hesperian materials. Hydrated silica was identified but only in an Amazonian/Hesperian impact, suggesting it may be a product of impact alteration.
Terra Sirenum is distinct from the two other study areas, both in its primary and secondary mineralogies and in the stratigraphic position of secondary mineralization. Primary mineral composition is more olivine-rich than in either Margaritifer or Tyrrhena, with olivine exposed in crater floor materials. A larger fraction (30%) of crustal materials is altered to secondary minerals than in either of the two other study areas, and exposure of significant Al-enriched kaolinite-bearing units at the pixel scale of mapping data is unique to Noachian materials of this study area, implying a higher water-to-rock ratio that leached more soluble Fe and Mg cations. These findings are consistent with analyses of CRISM targeted observations  and global mapping from OMEGA data (Riu et al., 2022). Several new locations of alteration to kaolinite have been identified on account of improved coverage at medium resolution (e.g., Wray et al., 2011), in intracrater plains materials (Figure 1b,   Figure 3. Relative abundances of primary and secondary minerals in units of each age range in each of the three study areas. The percentages above and to the right of each pie chart indicate the relative proportions of mapped primary and secondary minerals within a given age group within the given study area (e.g., 86% of mapped mineral exposures in the Terra Sirenum Early Noachian units are primary minerals). white arrows). These additional locations coincide with predicted locations of groundwater discharge from hydrological simulations (Andrews-Hanna et al., 2008 as shown in Wray et al. (2011). Sulfates at the pixel scale of mapping data are restricted to floors of deep craters; occurrence of both kaolinite and sulfates in those locations is consistent with previous interpretations that they may have formed in groundwater-fed paleolakes (Leask et al., 2019;Wray et al., 2011). In addition, alteration appears to have persisted later into the Noachian period than in the other study areas (80% of exposures in the late-Noachian are secondary minerals).

Discussion
Broadly, the analysis of multispectral mapping data presented here corroborates previous inferences about each study area, the differences between them, and the overall evolution of the Martian crust that have been made based on CRISM targeted observations and OMEGA global mapping. Overall, the combination of high coverage and spatial resolution of the multispectral mapping data reveal new insights into the broad-scale composition and alteration of the crust: (a) evolution of pyroxene composition over the Noachian period from low-Ca-dominated to high-Ca-dominated (e.g., Margaritifer Terra and Tyrrhena Terra) matches that inferred from global OMEGA coverage (Mustard et al., 2007;Poulet et al., 2009) and from CRISM observations of the intact pre-Noachian to Amazonian stratigraphic section present in the Tharsis plateau Quantin et al., 2012;Viviano-Beck et al., 2017); (b) the stronger spectral signature of olivine in Tyrrhena and Sirenum compared to Margaritifer is consistent with OMEGA mapping of elevated olivine in Tyrrhena Terra (Ody et al., 2012); (c) the increased fraction of secondary minerals from Margaritifer to Tyrrhena to Sirenum corroborates the trend identified from CRISM targeted observations having sparser coverage (e.g., , indicating that observational bias in CRISM targeted observations is a tractable issue. Variations in primary mineralogy between the three study areas support a major role of impact basins in configuring crustal primary mineralogy. The lower crust has been shown in studies of rocks exhumed by impacts to be enriched in olivine and LCP (Buczkowski et al., 2010;Phillips et al., 2022;Quantin et al., 2012;Skok et al., 2012), as well as plagioclase (Phillips et al., 2022). The enhanced olivine and plagioclase concentrations in the Tyrrhena Terra early-Noachian massif unit is consistent with excavation of olivine-and plagioclase-rich lower crustal materials by the Hellas impact. We infer the olivine-rich composition of Noachian plains in Terra Sirenum to correspond with previously identified olivine enrichment in inter-crater plains (McBeck et al., 2014;Ody et al., 2013;Rogers & Fergason, 2011).
Our results also suggest a role for both basin-forming impacts and endogenic heating in driving alteration to secondary minerals. The region between the Isidis and Hellas impact basins accumulated intense impact heat during the early Noachian period (Viviano & Phillips, 2019). The high concentration and enhanced metamorphic grade of hydrated minerals in this region  is consistent with a major role of impact heating driving alteration. In contrast, the Terra Sirenum region may be strongly altered due to increased heat flow due to long-lived volcanism, similar to but larger in scale than the enhanced alteration observed in materials exhumed along the Claritas Rise on the eastern flank of the Tharsis plateau (Viviano-Beck et al., 2017). Michalski et al. (2017) noted that the extent of phyllosilicates and other secondary minerals formed during the first stage of alteration at the site of the nearby Eridania paleolake was far larger than expected for a sedimentary deposit (based on comparison with other Martian paleolakes), and that the rock units bearing these minerals were massive in appearance (lacking bedding) and pervasively veined. They proposed that alteration occurred in a subaqueous hydrothermal environment in which water circulated through fractured rock, analogous to terrestrial mid-ocean ridges. Our finding of a high abundance of kaolinite in these rocks supports leaching of the protolith in such an environment. We also note that other lake deposits in this region, which are rich in acid sulfates, have been proposed to be driven by discharge of warm groundwater Wray et al., 2011).

Conclusions
CRISM multispectral mapping data fill a resolution and spatial coverage gap between CRISM targeted observations and OMEGA global mapping. Multispectral mapping data results presented here corroborate the regional and vertical variations in primary mineral abundances in the Noachian crust indicated by CRISM targeted and OMEGA global data. They confirm interpretations from targeted observations of large horizontal differences in degree of alteration to secondary minerals, and reveal new exposures that increase overall understanding of evolution of the Martian crust. Fractions and grades of secondary mineralization vary dramatically from one region to another. Important controls on the abundance and metamorphic grade of secondary mineral assemblages appear to include proximity to large impact basins and to magmatic bodies that supplied endogenic heat.

Data Availability Statement
Data sets for this research are available at the PDS Geoscience node, and described in the in-text data citation reference: Seelos et al. (2022). Data are available at the following URL: https://pds-geosciences.wustl.edu/, and labeled as CRISM derived data: Multispectral Reduced Data Record (MRDR). A repository including ArcGIS shapefiles with the mapping results from this work is available at the following URL: https://lib.jhuapl.edu/ papers/heterogeneity-of-the-noachian-crust-of-mars-using-/.