A Spectral and Morphometric Analysis of a Small Lunar Dome Complex Blanketed by Ejecta in Euclides-J Region

Abstract

This study focuses on a relatively unexplored area located some 33.4 geodetic km away from the central floor of Euclides, in the NE direction, precisely at −6.62° Latitude and −29.89° Longitude. Land-based CCD observations of this area showed a small group of dome-like structures situated on an elevated basaltic platform. One hypothesis is that these features represent a series of mounds made up of a collection of undifferentiated crater materials or ejecta as described by USGS (Geologic Atlas of the Moon. Department of Interior, US Geological Survey—I-458, 1965). This hypothesis is tested against an alternative one in that the area shows the presence of domes that are partly covered by these ejecta. The present spectrometric results place the basalt within the chosen AOI subset in the low-Ti category (1–5 wt% TiO₂) and a FeO abundance wt% of between 12 and 14. Spectrometrically, similar dome features have been observed by other investigators near the Milichus/Tobias Mayer regions. A possible classification for this feature would be DW/1a/6h/8j. Details of the cluster analysis used for raster classification is presented as an effective approach to discriminate between different classes of the lunar surface. This approach was used to map the lunar soil geology over the selected showing better lunar geological information than the current mapping information available by USGS. This same approach can be used to effectively map out other regions of the lunar surface.

1 Introduction

Mare Cognitum (which is the area chosen in 1964 by the Ranger program to gather the first high resolution imaging of the Moon), is located in the second ring of Oceanus Procellarum. To the northwest of the mare is the Montes Riphaeus mountain range which date back some 4 billion years, which are in turn the likely remains of a very large crater rim that was not completely buried by the lava flood.

To the west of M. Raphaeus is the prominent Copernican era crater Euclides (Lat: −7.40085°, Long: −29.56347°), which is a 12 km circular bowl-shaped crater with steep-slopes surrounded with bright ejecta and with a height/width ratio of 0.055 (using GLD 100 elevation data).

Euclides crater and its surrounding area are an interesting, yet somewhat overlooked lunar region. A number of low lying domes have been documented, notably dome ‘470’ which lies to the East of Euclides crater, some 21 geodetic km away.¹ Lunar domes are circular features of varying diameter and have a gentle slope rising in elevation a few hundred metres to the midpoint. These are structures of volcanic origin that at the time of eruption, are typically viscous, possibly silica-rich lava, erupting from localized vents having low effusion rates. Domes and associated volcanic features yield information about the volcanic history of the Moon (Stopar et al. 2009). Some of the domes may also show a small craterlet or vent at the peak.

Terrestrial observations of domes and related features (such as cones)² represents a particular challenge in that they are usually difficult to identify due to their low height, thus requiring particularly low sun illumination angles in order to be detected and studied.

Visual observation techniques of such volcanic features can obviously be supplemented by the acquisition of CCD images, which can undergo subsequent processing and geometric transformations to produce their 3-dimensional models, thus facilitating their identification even with modest amateur telescopic equipment. In addition to visual and photographic low sun-angle observations, spectrometry in the ultraviolet and visible electromagnetic spectrum can provide evidence of their mineralogical features against a known background of lunar surface, which according to Papike et al. (1991) is generally composed mainly of plagioclase, pyroxene, olivine, silica, ilmenite (Iron Titanium Oxide), mare and highland glass. In the case of the Euclides region, Clementine-derived geologic information points to a general presence of iron-rich, high titanium soils. Superimposed on and admixed with these materials include ejecta from young craters (USGS 1965).

Spectrometric observations are one of the most elaborate optical remote sensing techniques used by many lunar investigators, and has been carried out as a quantitative tool for the understanding of the chemical and mineralogy examination of the lunar surface (Wohler et al. 2011a, b). This is made possible because UVVIS electromagnetic radiances are associated with silicate absorption bands.

Before the provision of the Clementine satellite data (which among the wide-ranging sensors included a UV/Visible Camera, a Near Infrared Camera, a Long Wavelength Infrared Camera, and a High Resolution Camera), the mineral composition of the lunar surface was principally obtained using Earth-based telescopes having a high spectra resolution (allowing minerals to be identified with high accuracy) at the expense of low spatial resolution (2–10 km). Moreover, the availability of an orbiting satellite meant that lunar geological studies could now focus on the mapping of lunar mineral distribution on a global scale global. In 1994, the Clementine UVVIS camera acquired global multispectral images with pixel resolution of 100–300 m, for five spectral channels (Eliason et al. 1999) at 415, 750, 900, 950, 1000 nm wavelengths. For the first time, the acquisition of such data enabled the mapping of the global lunar mineral distribution (Simpson et al. 1995) on a global scale.

Calibration algorithms for Clementine data were derived primarily by Lucey (2004a) and Lawrence et al. (2005) based on Apollo lunar samples. For these samples the chemical determination was used and their albedo values were found at the sampling sites against Clementine images having a ground pixel resolution of 100 m. These algorithms enabled the calculation of percentage weight of FeO and TiO₂ from the radiances collected at the 5 wavebands (Shkuratov et al. 2011) and such estimates allowed the distinction among different types of lunar surface materials. Lunar remote sensing algorithms are now being used by investigators to retrieve mineral distributions (Cahill et al. 2009) where colour ratios [R₄₁₅/R₇₅₀] and [R₉₅₀/R₇₅₀] are used as diagnostic absorption bands in reflectance spectra to allow the identification of mafic minerals.

2 Aim of the Study

The Euclides region is surrounded by several domes (Fig. 1). A cursory investigation using LROC data shows that ‘Euclides 3’ could possibly be a complex dome system residing some 15 km to the east of Euclides crater (Kapral and Garfinkle 2005).

Fig. 1

Lunar map of the Euclides region and its environ showing the location of domes reported by several lunar observers, in need of further verification. (Source BAA). The dashed area represent the region of interest used in this study (South is at the top)

Of particular interest is an unexplored area residing some 33.4 geodetic km away from the central floor of Euclides, in the NE direction, precisely at −6.62° Latitude and −29.89° Longitude. In this area lies a small group of dome-like structures situated on an elevated basaltic platform. This topographic feature was detected by telescopic observation and CCD imagery carried out by the author on May 30th 2012 at 19:56UT (Fig. 2). One hypothesis is that these features represent a series of mounds made up of a collection of undifferentiated crater materials or ejecta as described by USGS (1965). According to USGS (1965), (Geological Atlas of the Moon—I-458; LAC-76; Geology of the moon in the M. Raphaeus region), the area of interest (AOI) consists of predominantly crushed rock and great blocks derived mainly from the region Mare Imbrium. This layer forms a coating probably ranging from a few meters to about 500 meters in thickness, and has a heterogeneous composition.

Fig. 2

CCD image showing an overview of Mare Cognitum. The white frame indicates the area of interest with the inset magnified to show further detail. Image taken on May 30th 2012 at 19:56UT with a 200 mm SCT and DMK21AU04 camera at f/10 using a red filter (Znith Observatory Malta; http://znith-observatory.blogspot.com). North is at top

This hypothesis is tested against an alternative one in that the area shows the presence of domes that are partly covered by these ejecta.

3 Method and Measurement

3.1 Geographic Setting

A subset covering a cartographic width of approximately 10.74 geodetic km and situated between Euclides-J and Euclides was selected and analysed (Fig. 3). This subset of the AOI indicated in Fig. 1 was chosen on the basis of the terrain and related shadows that were highly suggestive of the presence of a dome-like feature as recognised by the CCD image (inset, Fig. 1). The LROC NAC high resolution image (0.5 m/pixel) laid over the WAC Global Morphological basemap (100 m/pixel) is available from http://bit.ly/1INJ6Nn.

Fig. 3

This subset area—http://bit.ly/1zXZepg—lies within the lunar coordinates shown at the top right hand corner, and falling within the area of interest—http://bit.ly/1INJ6Nn. North is at top

3.2 Telescopic CCD Observation

For the CCD observation (Fig. 2), the local solar altitude and the sun’s selenographic colongitude were calculated using the LTVT software package (Mosher and Bondo 2006) which requires a calibration of the image by identifying the precise selenographic coordinates of some landmarks on the image. This calibration was performed based on a standard list of control points. This calculation was performed so as to obtain predictions of the same solar illumination conditions (Fig. 4).

Fig. 4

The header information below shows the selenographic information at the time of CCD image capture (North is at top). Sub-solar Pt = 58.321E/0.096S Sub-Earth Pt = 5.825 W/5.890 N Center = 29.566°W/7.402°S Zoom = 40.000 Vertical axis: central meridian

The morphometry of the dome-like feature residing in the AOI subset, including slope, height and diameter was calculated and further refined using GLD100 (Global Lunar DTM 100 m topographic model) elevation model (NAC-derived DTMs are included, where available) produced by Scholten et al. (2012).

3.3 Spectrometry

The UVVIS spectrum based on Clementine imagery at wavelengths of 415, 750, 900 and 950 nm were downloaded from the LPI Clementine Mapping Project (http://www.lpi.usra.edu/lunar/tools/clementine/) for an area that covered the AOI subset: South = −10; North = 10; West = −40 and East = −20 to produce a simple cylindrical output map projection at a projected scale of 303 pixel per degree. The reflectance values were derived based on the calibrated and normalised Clementine UVVIS reflectance data as provided by Eliasson et al. (1999). The extracted Clementine UVVIS data were reprocessed in terms of R₄₁₅/R₇₅₀ and R₉₅₀/R₇₅₀ colour ratios for the AOI.

The R₄₁₅/R₇₅₀ colour ratio was used as a measure for the TiO₂ content of mature basaltic soils (Charette et al. 1974). However, one has to keep in mind that for many lunar regions the relation between the R₄₁₅/R₇₅₀ colour ratio and TiO₂ content displays a significant scatter (Gillis and Lucey 2005). The R₉₅₀/R₇₅₀ colour ratio can be used to assess the strength of the mafic absorption band, representing a measure of the FeO content of the soil (Lucey et al. 1998). Contour analysis was used to assist in the identification of the spatial distribution of the colour ratios.

FeO and TiO₂ abundance (wt%) data were also downloaded from the LPI Clementine Mapping Project. The algorithms used to produce these data are according to Lucey et al. (1996). Contour analysis was also applied to denote the spatial distribution of FeO and TiO₂ abundance (wt%).

The scatter between these two spectrometric information was thus analysed over the subset AOI in order to determine the natural scatter over the area for subsequent accuracy assessment.

It is interesting to note that Blewett et al. (1997) found excellent linear correlations between spectral Fe and Ti parameters from Clementine mission and the average FeO and TiO₂ contents of soils sampled at the Apollo 15, 16 and 17 landings sites. Blewett et al. (1997) based their investigations on the method used by Lucey et al. (1996) to derive FeO abundance from reflectance spectrometry modelling. More detailed analysis are included in Blewett et al. (1997).

A series of co-located spectrometric values were extracted at different mineralogical sites as defined by USGS (1965) from the TiO₂ and FeO abundance, R₄₁₅/R₇₅₀ and R₉₅₀/R₇₅₀ raster images within the larger Euclides region and analysed. These regions included the maria region (M), ejecta deposits (E), enclosed ejecta (EnE) as well as crater walls (CW). This was carried out in order to better define the characteristic spectrometric response from various areas in the Clementine data representing the main mineralogical classes within the Euclides region.

3.4 Image Processing

Unsupervised classification (or cluster analysis) was used as the statistical method to aggregate digital number pixels into surface mineralogical categories (or classes) that have similar spectral character. The methodological concept here is that values within a given lunar soil cover type should have similar measurements (i.e. have similar DN levels), whereas data in different classes should be statistically well separated (Lillesand and Kiefer 1994). In doing so, image processing can determine and map out spectrally separable classes. A range of clustering algorithms may be used to produce these natural clusters of data. The processor needs to provide some minimum information to these algorithms, such as the number and statistical separation between the different clusters, etc. These parameters will determine the sensitivity of the algorithm to cluster very close, yet distinct groups of pixels with similar spectral characteristics (Gupta 2013).

Clustering is done using one of these statistical routines where classes of pixels are created based on their shared spectral signatures. During this process, clusters are split and/or merged until further clustering does not improve the presence of the variation in the scene. Successful clustering is achieved by a series of iterations until the variability that resides “between clusters” reaches a maximum when compared to that “within clusters”.

The Iterative Self-Organizing Data Analysis Technique (ISODATA clustering algorithm) method was used for this study (De Sanctis et al. 2010; Allender and Stepinski 2014). The ISODATA algorithm merged clusters together based on a separation distance in multispectral feature space that was defined by a maximum standard deviation of 2 and a minimum distance between class means of 1. The maximum number of classes was set to 20. This method used a large number of iterations through the dataset until the specified results were obtained. Thus, UVVIS Clementine image based on the reflectance in the 750 nm band was classified to derive 20 spectrally different classes based on the individual relationship of every collocated pixel value in the original albedo image. Similarly, the FeO abundance (wt%) raster data was classified to derive congruent spectral classes based on the degree of FeO abundance as originally calibrated using the algorithm developed by Lucey et al. (2000). The resulting classified map represented the major terrain geology according to the FeO abundance (wt%).

Unsupervised classification can also be considered as a highly sensitive technique that is efficient in identifying the main clusters from the FeO abundance (wt%) image. Clustering techniques might also provide equally good and possibly more accurate classification results, but were not explored in this study.

It is important to note that the identity of the final classes can only be ascertained by comparing the final maps with in situ or accurate secondary sources of information (although their presence remains doubtful) in order to determine the absolute typology of each of the spectral classes. Once this is made available then this classification approach provides a complementary rapid means of producing high-quality lunar soil cover data.

4 Results and Discussion

4.1 The Geological Setting from Spectrometric Results

The lunar stratigraphy of the AOI can be described by the superposition and chemistry of the most recent crater materials. The impact which formed Euclides crater, for example, has excavated and spread underlying crustal material. This ejecta material has of course different chemical composition from that of the surface basaltic layer. Spectrometric information collected from the points marked as M and MC (Fig. 5) suggest a slightly different material than that detected over the dome region. This difference is most probably due to crustal ejecta from neighbouring Euclides, giving it a closer range of spectrometric values as the ones observed at E, EnC and CW. Results suggest the presence of ejecta overlaying the dome structure that might be altering the spectral parameters of its original chemical composition.

Fig. 5

Classified albedo map based on the reflectance at 750 nm over the Euclides region [Mare Cognitum (MC), Montes Raphaeus (MR), Euclides and Euclides-J]. The resolution of the data is 303 pixels per degree. A number of sampling points were selected to derive mineralogical characteristics as shown in Table 1

The scatteplot between TiO₂ abundance (wt%) and R₄₁₅/R₇₅₀ over the subset AOI is shown in Fig. 6. A rather strong positive linear correlation coefficient of 0.6, with a linear regression model of R₄₁₅/R_{750 =} [−21.56 + 35.26.TiO₂].

Fig. 6

Scatterplot between TiO₂ abundance (wt%) and [R₄₁₅/R₇₅₀] over the subset AOI. The linear regression model of R₄₁₅/R_{750 =} [−21.56 + 35.26. TiO₂]; R = 0.6; SE: 0.003

The Clementine UVVIS spectral data covering the small AOI subset, which included a single dome feature in the Euclides-J region (Fig. 7a–d), revealed a low average value of the UVVIS colour ratio of R₄₁₅/R₇₅₀ = 0.61, indicating a low TiO₂ content, and an average weak mafic absorption with R₉₅₀/R₇₅₀ = 1.04, suggesting a high soil maturity (Table 1). Using the Lucey et al. (2000) algorithm, the FeO abundance over the AOI subset showed a value of around 13 wt% while TiO₂ abundance showed a value of around 2.6 wt%.

Fig. 7

The spatial distribution over the subset AOI of a [R₄₁₅/R₇₅₀]radiance ratio b [R₉₅₀/R₇₅₀]radiance ratio c FeO abundance (wt%) and d TiO₂ abundance (wt%). Colour range is dependent on each of the respective DN values. (Color figure online)

Table 1 Spectrometric results derived over the Euclides region derived from the Clementine UVVIS reflectance, FeO abundance wt% and TiO₂ abundance wt% data

Spectrometrically, similar dome features have been observed by other investigators near the Milichus/Tobias Mayer regions (Table 2) which according to Wohler and Lena (2011) result from both effusive and intrusive volcanic processes. Yet, overlaying these domes, USGS (1965) described the presence of “freshly exposed crushed rock derived in part from primary rayed craters…” (USGS Geological Atlas of the Moon: I-515; LAC 58).

Table 2 Comparative spectrometric results showing similar attributes between the dome structure under investigation in Euclides-J region and other domes as documented by Wohler and Lena (2011)

Based on these spectrometric results, the Clementine dataset was used to detect new features that could highlight the potential location of previously undetected domical structures within the Euclides region. Table 3 shows an additional feature in the Euclides region (http://bit.ly/1ISL1QQ) that was detected after the raster data were filtered to only show FeO abundance wt% is between 13 and 14 and TiO₂ abundance wt% of between 2 and 3.

Table 3 Detection of another domical structure in the Euclides region when FeO abundance wt% is between 13 and 14 and TiO₂ abundance wt% is between 2 and 3

4.2 Lunar Soil Mapping of the Euclides Region

Figure 8 a shows a classified map derived after cluster analysis on the basis of the main R₉₅₀ albedo spectrometric classes of the Euclides region, which covers both the AOI and its subset. When compared to the corresponding albedo image (using the UVVIS 950 nm band—Fig. 8b), it can be noticed that the AOI subset where the feature is found, forms part of the distinctive light-green coloured class which spectrometrically is different from either the red or orange coloured classes, both of which are statistically closer to the ejecta seen next to the walls of Euclides crater. This means that the light green colour has a higher basaltic composition, and therefore lies somewhere in between the two extreme classes, i.e. between the basaltic, maria surface (dark blue) with little ejecta coverage, and the exposed crustal material situated within and around Euclides crater itself (orange and red coloured classes).

Fig. 8

a Shows part of the classified map (subset AOI boxed) based on R₉₅₀ reflectance, b shows raw R₉₅₀ reflectance. The colour scheme used for (a) is dependent on the clusters identified by the unsupervised classification. (Color figure online)

Moreover, the present digital image classification of the Euclides region on the basis of FeO abundance (wt%) has produced a map that is similar to, but is spatially richer (in terms of mineralogical distribution based on FeO abundance) than the published geological map made available by the United States Geological Survey (1965–1969). The classification of the lunar surface into discrete classes based on the statistical distribution of FeO abundance (wt%) seems to provide an improved map product (Fig. 9). This new spatial map shows that the AOI subset is definitively within the 11.5–13.2 wt% region, roughly equivalent to the “If = ejecta blanket” type as defined by USGS (1965). This study confirms that the “If” region is in fact made up of four statistically distinct classes in terms of FeO abundance (wt%), One must keep in mind that this categorisation is based purely on spectral and statistical information. The resulting classification maps should be checked using ground truth information and field validation surveys whenever possible.

Fig. 9

a Additional geological classes as detected by the Iterative Self-Organizing Data Analysis Technique (ISODATA algorithm) method. Insets (black outlines) show detection of additional categories compared to USGS (1965), and b lunar soil map published by USGS (1965) over the Euclides region

4.3 Morphometric Properties

The morphometry of the most prominent dome feature identified within the AOI subset, which caused an identifiable shadow cast from the terrestrial CCD observation (inset, Fig. 1), were derived using GLD100 data from LROC database.

Results show that the dome feature reaches up to 112 m, with an approximate flank slope of 4.9°. The dome volume V was computed by integrating the data provided in Fig. 10 assuming a circular feature of 2 km in diameter. A rough quantitative measure for the shape of the dome is equivalent to 0.3 form factor (Lena et al. 2009), which is equivalent to a conical shape. The volume amounts to 0.13 km³. A digital elevation map of the area including this feature is shown in Fig. 11 below. A possible dome classification for this feature, based on the one used by Westfall (1964) would be DW/1a/6h/8j.

Fig. 10

Cross section of the dome based on data derived from GLD100 data from LROC database

Fig. 11

The dome-shaped structure found within the AOI subset has a diameter of 2 km and height of 112 m. Source: GLD100 data from LROC database http://target.lroc.asu.edu/qm3d/o2w_3d_451478048_46_11_18_82_0/

In morphometric terms, a similar dome structure is located on the floor of the crater Stevinus (Lat: −32.75200°, Lon: 53.71112°) which is some 2.3 km in diameter. Using GLD100 data, the dome has a height of 103 m (NAC frame M113603383L)—http://bit.ly/1oL4uYD.

5 Conclusion

The rapid growth of image acquisition technology is leading to a large amount of planetary remotely sensed images now available for astronomy research in both professional and amateur domains. The use of such imagery for terrain and soil classification is instrumental to study the geologic evolution of moons and planets. This study focuses on lunar image classification to understand lunar mineralogy using images collected by the Clementine mission.

Lunar geologic classification is particularly challenging because the shapes and colours of each type of terrain depend on the spatial location, while the brightness and shadow caused by the angle of sunlight can significantly change the appearance of lunar features.

Meanwhile, studies of lunar domes and related volcanic features continue to provide us with clues into the interior and evolution of the Moon as well as the composition of the lunar crust. The methodology presented in this study has the potential to provide new insight into the distribution of volcanic features, in otherwise overlooked regions of the lunar surface. Finding these small features using traditional terrestrial telescopic techniques, coupled with new online research tools such as LROC Act React Quick Map demonstrate the ability to amalgamate with the latest technological developments that bring within reach planetary information to whoever is interested to discover and further understand the lunar geological surface.

This study makes use of an automatic image classification system with the following conclusions:

• 1. This study is strongly suggestive of the presence of volcanic structures close to Euclides-J crater, and thus adds to the already identified similar features in the Euclides region, as reported by other lunar observers.

• 2. Present results place the basalt within the AOI subset in the low-Ti category (1–5 wt%.TiO₂). Multispectral Clementine data indicate that the AOI subset has an FeO abundance (wt%) between 12 and 14. The low-Ti category points at relatively old basalts compared to more recent ones that tend to become progressively Ti-richer (i.e. from 2–3 to 4–5 wt% TiO₂).

• 3. A rather strong positive linear correlation coefficient exists between the TiO₂ abundance (wt%) and [R₄₁₅/R₇₅₀] over the subset AOI (R = 0.6) with a regression model of R₄₁₅/R_{750 =} [−21.56 + 35.26*TiO₂].

• 4. The AOI subset is definitively within the 11.5–13.2 wt% FeO which is geographically resides over the “If = ejecta blanket” soil type as defined by USGS (1965). This “If” region is in fact made up of four classes in terms of FeO abundance. Spectrometrically, similar dome features have been observed by other investigators near the Milichus/Tobias Mayer regions. These domes have been as most likely resulting from volcanic processes.

• 5. A possible classification for this domical feature would be DW/1a/6h/8j. In morphometric terms, a similar dome structure is observed from LROC data located on the floor of the crater Stevinus (Lat: −32.75200°, Lon: 53.71112°) and is some 2.3 km in diameter.

• 6. Using the restricted range of Fe- and Ti-content observed over DW/1a/6h/8j, a possible, so far unlisted domical feature is also present at Lat: −5.53534°, Lon: −29.66767°.

• 7. Image classification detail is presented which is considered to be effective to discriminate between different classes of varying FeO abundance. This approach can be used to effectively map out other regions of the Moon.

• 8. Higher resolution mapping of the lunar soil geology is presented which is more spatially informative than currently available lunar geological mapping information.