Lunar domes in the Doppelmayer region: Spectrophotometry, morphometry, rheology, and eruption conditions

Transcription

1 Lunar domes in the Doppelmayer region: Spectrophotometry, morphometry, rheology, and eruption conditions Raffaello Lena Via Cartesio 144, sc. D, 00137, Rome, Italy Christian Wöhler DaimlerChrysler Research and Technology, Machine Perception P. O. Box 2360, D Ulm, Germany Jim Phillips 101 Bull Street, Charleston, SC 29401, USA Michael Wirths RR#3 Perth, Ontario, Canada K7H 3C5 Maria Teresa Bregante Via Antica Romana Occ. 13, 16039, Sestri Levante (Genova), Italy Geologic Lunar Research (GLR) group 1

2 Abstract In this study we examine a lunar volcanic region near Doppelmayer, composed of two domical structures previously not studied in detail and a well-known lunar pyroclastic deposit. Dome 1 is situated at selenographic coordinates W and S, dome 2 at W and S. We perform a spectrophotometric study of representative locations of the volcanic region based on Clementine UVVIS data. Relying on ground-based high-resolution CCD imagery, we furthermore examine the morphometric characteristics of the two domes, making use of a combined photoclinometry and shape from shading technique. Based on a rheologic model, we examine the physical conditions under which the domes were formed. Dome 1 is spectrally atypically red for mare domes and shows a very weak mafic absorption, implying a low TiO 2 and FeO content. The overall spectral signature corresponds to that of a mixture between mare and highland soils. Dome 1 was formed of lava with a relatively high viscosity value of 10 7 Pa s, situated between the ranges of values typically observed for mare domes and for the Gruithuisen and Mairan highland domes, respectively, erupting at moderate rates over a long period of time. Dome 1 is larger, steeper, and more voluminous than typical mare domes but has a much lower flank slope than the Gruithuisen and Mairan highland domes. It is an exemplar of a rare type of unusually steep and voluminous mare domes, similar to the well-known mare domes Hortensius 5 and 6 and Herodotus ω. We discuss the relevance of vertical (assimilation of crustal material) vs. lateral (distribution of material across mare-highland boundaries by random impacts) mixing mechanisms as being responsible for the observed spectral appearance of dome 1. The thermal conditions in the lunar interior did not favour the assimilation of crustal wallrock into the ascending magma. Due to the fact that dome 1 is located right on the boundary between hummocky terrain and a mare pond, lateral mixing of mare and highland soils is a much more natural explanation for the observed spectral signature. For dome 2, we find that it is a typical effusive mare dome, given its spectral and morphometric properties and inferred rheologic parameters. An estimation of the dimensions of the feeder dikes of the two domes reveals that the dike which formed dome 1 was five times as broad as the one which formed dome 2, while the dike lengths only differ by about one third. The dike dimensions suggest that their source regions were located below the lunar crust. Moon; volcanism; image processing; spectrophotometry; geological pro- Keywords: cesses 2

3 1 Introduction The Humorum impact basin is of Nectarian age and was formed 3.9 billion years ago (Wilhelms, 1987). The basin is filled with basaltic lava. Loading by lava effusion has caused a subsidence of the central part of the basin. The effects of this crustal settlement are visible at the surface in the shape of long wrinkle ridges, which are due to compressive forces, in the eastern part of Mare Humorum, and in graben due to crustal extension, especially visible at the western borders of Mare Humorum (Wilhelms, 1987). A section of Consolidated Lunar Atlas image F19 (Kuiper et al., 1967) providing an overview about the Mare Humorum region is shown in Fig. 1. Different lithological units, included in the USGS lunar geologic map I-495, are apparent in Mare Humorum (cf. Titley (1967) and references therein). Clementine multispectral data show that the highlands around the Humorum basin are feldspathic, but somewhat more mafic (4 8 wt.% FeO) than the deposits of other basins, while the mare deposits of central Humorum show an FeO content between 16 and 20 wt.% (Bussey et al., 1997). Humorum basalts have been mapped as four distinct units, Ipm1 through Ipm4, in the USGS lunar geologic map I-495. Ipm units denote flows of volcanic material or pyroclastic material or both. Pieters et al. (1975) describe three distinct basaltic units. The basalts covering the western third and the southeastern part of Mare Humorum are spectrally red, implying a low TiO 2 content. The western unit has a lower albedo than the southeastern unit. Spectrally bluer basalt units situated in the basin centre and displaying a higher TiO 2 content are similar to others that occur in Oceanus Procellarum but are not contiguous with them. Pieters et al. (1975) suggest that these basalts do not originate in Mare Humorum itself but may have entered the basin through the northeastern entrance. The crater Doppelmayer has a diameter of 64 km and is located on the southwestern edge of Mare Humorum. To the northeast of the rim of Doppelmayer crater the nearly submerged crater Puiseux is situated. Very dark material partially covering its floor, also known as the Doppelmayer Formation, has been interpreted as fragmental volcanic ejecta or flows, or both, from vents or fissures along the edge of Humorum basin (Titley, 1967). The dark material of the Doppelmayer Formation (Eid unit in USGS map I- 495) is prominent in the Clementine 750 nm albedo image and coincides with units of high FeO content within Mare Humorum found by Bussey et al. (1997). Moreover, these two Eid units, located just to the west and north of Doppelmayer, have been spectrally characterised as lunar pyroclastic deposits (LPDs) by Gaddis et al. (2003). They assign the spectrally red appearance of the LPDs to iron-rich glassy deposits, like those found in the Aristarchus-Harbinger region. Hackwill et al. (2006) further subdivide the three main units found by Pieters et al. (1975) into 109 units differing in FeO and TiO 2 content and crater density. While a uniform age of about 3.2 Ga is reported for the Mare Humorum basalts by Hiesinger et al. (2003), individual ages are determined by Hackwill et al. (2006) for 33 major units based on crater counts and isotopically dated Apollo samples. The highest age is found for the unit in the basin centre, which supports the suggestion that the central unit sank due to the load of the basalt layer, causing the lithosphere to bend and create 3

4 dikes through which lava ascended and generated the basaltic units near the basin rim. Furthermore, it is shown that the TiO 2 content of the basaltic units is correlated with their FeO content. Later episodes of lava effusion may have occurred in the region around Doppelmayer. In this study we perform a detailed examination of a domical structure south of the rim of Doppelmayer, first reported by Wood (2005), and a previously unreported further domical structure located southwest of it. These two domical structures lie in a complex region, mapped as a basaltic Ipm unit but also consisting of hummocky material in USGS map I-495. In this study, based on high-resolution telescopic CCD observations carried out under oblique illumination conditions, we explore the region around the domical structures in more detail. We examine their morphometric characteristics by making use of a combined photoclinometry and shape from shading approach (Horn, 1989; Wöhler and Hafezi, 2005; Lena et al., 2006; Wöhler et al., 2006). The obtained values are used to derive information about the physical parameters of dome formation (lava viscosity, effusion rate, duration of the effusion process, magma rise speed, dike dimensions), employing the rheologic model by Wilson and Head (2003). We provide a geological interpretation of our spectrophotometric, morphometric, and rheologic modelling results, comparing them to the corresponding parameters observed for typical lunar mare and highland domes. 2 Observations 2.1 Telescopic CCD imagery Fig. 2 displays our CCD images of the region around Doppelmayer. They were taken with telescopes of apertures between 130 and 450 mm. Details about the UT date and time of each image and the telescope and CCD camera used are given in the figure caption. Both utilised camera types have an image size of pixels and a pixel size of 5.6 µm. Each image was generated by stacking several hundreds of video frames. For this purpose we made use of the Registax and Giotto software packages, employing a cross-correlation technique similar to the one described by Baumgardner et al. (2000). In that work, however, digitized analog video tapes were processed, while we directly acquired digital video frames. The scale of the images is between 200 and 500 m per pixel on the lunar surface. Due to atmospheric seeing, however, the effective resolution (corresponding to the width of the point spread function) is not much better than 1 km. All images are oriented with north to the top and west to the left. 2.2 Lunar Orbiter and Clementine imagery, soil composition Fig. 3 displays Lunar Orbiter frame IV-143-H1, distinctly revealing the two previously mentioned domical structures. From our telescopic CCD images (Fig. 2) and using Lunar Aeronautical Chart (LAC) #93, we determined their selenographic positions to W and S for dome 1 and W and S for dome 2 (cf. Table 1). We found evidence for the assumption that both domical structures are effusive volcanic constructs, which is justified in more detail in Section

5 The Clementine UVVIS multispectral image data were obtained at five wavelengths: 415, 750, 900, 950, and 1000 nm. Fig. 4b reports reflectance values derived for the two domes and several further geologic units, relying on the calibrated and normalised Clementine UVVIS reflectance data as provided by Eliason et al. (1999). Prior to the Clementine mission, the characterisation of lunar soil types was performed by means of reflectance spectra measured with earth-based telescopes (Adams and McCord, 1970; McCord et al., 1972; McCord and Adams, 1973). A characterisation of spectral features attributable to titanium in lunar soils is provided by Burns et al. (1976). The extracted Clementine UVVIS data were examined in terms of 750 nm reflectance ( albedo ) and the R 415 /R 750 and R 950 /R 750 colour ratios. Albedo at 750 nm is an indicator of variations in soil composition, maturity, particle size, and viewing geometry. The R 415 /R 750 colour ratio essentially is a measure for the TiO 2 content of mature basaltic soils, where high R 415 /R 750 ratios correspond to high TiO 2 content and vice versa (Charette et al., 1974). Recent work by Gillis and Lucey (2005), however, relying on TiO 2 abundance data obtained with the Lunar Prospector neutron spectrometer, indicates that other effects such as ilmenite grain size or FeO content may contribute to the UV/VIS ratio. Hence, accurate estimates of the absolute TiO 2 weight percent values from spectral properties require supporting information beyond the UV/VIS spectral ratio, such as reflectance at 2.7 µm and radar backscatter (Gillis et al., 2005). Although TiO 2 content is monotonously increasing with R 415 /R 750 ratio, the correlation is only moderate and the data display a strong scatter. Gillis and Lucey (2005) establish two linear trends. A first trend with a higher slope is apparent for TiO 2 contents of more than 2 wt.% found e. g. in the Mare Tranquillitatis region with R 415 /R 750 larger than 0.62, while a distinct second trend valid for smaller R 415 /R 750 ratios displays a lower slope and is represented e. g. by several types of soils in Oceanus Procellarum. The R 950 /R 750 colour ratio is related to the strength of the mafic absorption band, representing a measure for the FeO content of the soil, and is also sensitive to the optical maturity of mare and highland materials (Lucey et al., 1998). Fig. 4a shows the locations in the Clementine 750 nm albedo image at which the spectra were obtained. The reflectance spectra for the two domical structures examined in this study and further geologic units including the dark and smooth terrain west of dome 1, the nearby hummocky terrain located just to its south-east, and the LPD northeast of Doppelmayer are shown in Fig. 4b. The sample area amounts to 2 2 km 2. Both domes are spectrally red with their low R 415 /R 750 ratio of 0.58, indicating a low TiO 2 content of less than 2 wt.% according to Gillis and Lucey (2005). The high R 950 /R 750 ratio suggests that they consist of mature material (Fig. 5). Dome 2 is spectrally not distinguishable from the mare-like surface into which it merges to the north, while dome 1 has a spectrum that is intermediate in reflectance between the dark and smooth mare unit to the west and that of the hummocky terrain to its south-east, which is of higher reflectance and shows a typical highland signature. The material of dome 1 appears to be a 1:1 mixture of the corresponding soils (Fig. 4b). In the Clementine 750 nm albedo image shown in Fig. 4a, the outline of dome 1 is indicated by a white circle. The location labelled dark smooth terrain with its mare-like spectral appearance is situated right at the bottom of the western flank of 5

6 dome 1. Hence, dome 1 is situated on a mare-highland boundary. The easternmost part of the crescent-shaped mare pond west of dome 1 forms the lower part of the western dome flank. This slope is clearly visible in Fig. 2d, appearing bright due to the sunset illumination from western direction. The observed intermediate spectral signature of most of the surface of dome 1 may be due to vertical mixing effects, e. g. assimilation of crustal wallrock into ascending basaltic magma (Grove, 2000), or to lateral mixing caused by random impacts (Li and Mustard, 2000, 2005). We will discuss the relevance of these possible mixing mechanisms in detail in Section 5. Two regions mapped as Eid units in the USGS map I-495 are characterised as LPDs by Gaddis et al. (2003). A large LPD with an area of 2628 km 2 is located inside Doppelmayer, centred at 40.5 W and 28.1 S. A further LPD measuring 1472 km 2 is situated at a larger distance north-west of Doppelmayer at 44.4 W and 26.6 S. Clementine UVVIS data have been used to characterise LPDs and their composition (Gaddis et al., 2003; Lena et al., 2006). The spectral properties likely represent a complex combination of the degree of crystallinity (e. g. ilmenite content) and of the iron-titanium content of LPDs. In Fig. 5a and 5b, mare soils are represented by dark grey regions and highland soils by light grey regions, according to Gaddis et al. (2003). In Fig. 5c these fields are omitted since the regions representing mare and highland soils strongly overlap. In the R 750 vs. R 415 /R 750 diagram (Fig. 5a), the LPD in the northeastern part of Doppelmayer is located near the margin of the mare field towards lower values of R 750 (Gaddis et al., 2003). We obtain R 415 /R 750 and R 950 /R 750 ratios of and 1.027, respectively, which are somewhat different from the values and reported by Gaddis et al. (2003). The LPD displays spectral variations across its surface, hence we presumably measured its spectrum at a slightly different location with a possibly different sample area. In the R 750 vs. R 950 /R 750 diagram (Fig. 5b), this LPD is darker and shows a stronger mafic absorption than the two domes and the associated dark smooth mare-like terrain, which can be explained by an iron-rich glassy soil component (Gaddis et al., 2003). Similar properties are found for several other large LPDs situated near Aristarchus and Harbinger, inferred to have a lower TiO 2 content in the Fe 2+ bearing orange glass spheres than the Taurus-Littrow, Sinus Aestuum, Vaporum, and Rima Bode LPDs, which are spectrally very blue and are known or inferred to have a significant component of high-tio 2 materials in the form of ilmenite-rich black beads (Gaddis et al., 2003). Weitz and Head (1999) explain the differences between orange and black bead LPDs by variations in cooling time in a fire fountain probably resulting in quenched, crystallized, and composite droplets. The spectral behaviour of LPDs is analysed in terms of a radiative transfer model by Wilcox et al. (2006), who find low TiO 2 contents and FeO contents of wt.% for three LPDs on the Aristarchus Plateau, in Mare Humorum, and near Sulpicius Gallus. 6

7 3 Morphologic and morphometric properties 3.1 Morphology of the domes The structure situated south of the rim of Doppelmayer at W and S, described as dome 1 in Table 1, appears to be smooth with a shallow and elongated crater on the summit (Fig. 3). The location of this elongated crater pit on a typical dome relief (cf. Fig. 2) is suggestive of its volcanic origin, even if it might also be, as an alternative explanation, a degraded impact or secondary crater. However, our interpretation that it is of volcanic origin is based on the observation that in the Lunar Orbiter image shown in Fig. 3, the summit crater is elongated and appears rimless and without a sharp outline. Hence, it looks different from nearby degraded small craters of impact origin. Furthermore, in the Lunar Orbiter image its rim does not cast a black shadow at a solar altitude of 16, under which the image was acquired. We will see below that its depth is significantly different from the D/5 ratio typical of small fresh impact craters of similar diameter (Pike, 1974; Wood and Andersson, 1978). In contrast, impact craters in Fig. 3 appear to have significantly steeper inner walls than the suspected vent since they are filled with black shadows, even when their appearance is not fresh. As found in Section 2.2, the lower part of the western flank of dome 1 consists of mare material, which contradicts the possible interpretation that this feature is merely an elevated deposit of hummocky material. It rather seems that a significant component of highland material has been entrained into the surface of a domical structure consisting of mare basalt, situated near a mare-highland boundary (cf. Section 5). These findings support our interpretation that the structure denoted as dome 1 with its elongated crater pit is of volcanic origin. In Fig. 2a the dome diameter amounts to D = 16.8 ± 0.3 km. The vent has a diameter of D c = 3.3 ± 0.3 km. Its depth d c was estimated by measuring the length of the shadow cast by its rim, which yields d c = 128 ± 30 m. The depth derived from the shadow length measurement must be considered as a lower limit because the shadow is not cast right into the middle of the vent but slightly off-centre on its inner wall. In Fig. 2b, the dome diameter was determined to D = 16.4±0.2 km, the vent diameter to D c = 3.2±0.2 km, and the vent depth to d c = 145 ± 20 m, again based on shadow length measurement. In the Lunar Orbiter image shown in Fig. 3, the vent diameter amounts to 3.3 ± 0.3 km, which is consistent with the values derived from Figs. 2a and 2b. According to the empirical relation between vent diameter and dome base diameter established by Head and Gifford (1980) for effusive mare domes, the expected value for the vent diameter corresponds to 3.2 km, given the dome diameter of D = 16.8 km, which is in very good agreement with our measured values of D c. Dome 2, located at W and S and situated adjacent to a non-volcanic mountain, has a diameter of 12.6 ± 0.3 km according to Fig. 2a and of 12.0 ± 0.3 km according to Fig. 2b. These values are consistent with the diameter of 12.6 ± 0.3 km measured in the Lunar Orbiter image shown in Fig. 3. In Fig. 6 an enlarged section of the Lunar Orbiter image is shown where the dome appears to have a smooth surface with an outflow channel or chain of vents (feature A) and linear rilles (features B and C). 7

8 Narrow rilles crossing the dome are also visible in the Clementine 750 nm albedo image shown in Fig. 7. A discussion of the possible mode of formation of dome 2 is given in Section Image-based 3D reconstruction of the domes For an in-depth morphometric and rheologic analysis of the domes, we performed a reconstruction of their three-dimensional shape based on the available image data, relying on a combined photoclinometry and shape from shading method. Such techniques take into account the geometric configuration of camera, light source, and the surface normal, as well as the reflectance properties of the surface to be reconstructed. We will only give a short summary of our photometric 3D reconstruction approach since it has been described in detail by Lena et al. (2006) and by Wöhler et al. (2006). A good description of the reflectance properties of the surface is the physically motivated photometric model by Hapke (1993) which is based on the theory of radiative transfer. It is difficult, however, to use the Hapke model directly for 3D reconstruction purposes. Therefore, in many astrogeological applications the comparably simple, empirical Lunar-Lambert law is used (McEwen, 1991). It fits the true reflectance behaviour of many planetary surfaces equally well as the Hapke model, making use of a single phase angle dependent parameter which has been tabulated by McEwen (1991) for planetary surfaces with a wide range of regolith properties. In a first step we follow a photoclinometric approach, which consists of computing height profiles along image rows, since the problem of directly determining two surface gradients per pixel (in east-west direction and in north-south direction) from a single intensity measurement is an underdetermined ( ill-posed ) problem. For lunar domes, the surface slopes are small, and the illumination is highly oblique. The scene is illuminated nearly exactly from the east or the west. It can be shown that the Lunar- Lambert reflectance then shows a very weak dependence on the surface gradient in north-south direction, which we set to zero. This approximation is exact for crosssections in east-west direction through the summit of a feature, while it is otherwise a reasonable approximation. We furthermore impose a zero average surface slope over the region of interest. Unter these assumptions, the surface gradient in east-west direction can be computed for each pixel based on the measured pixel intensity, respectively. For each image row, a height profile is then obtained by integration of the surface gradients. The result of photoclinometry is refined in a second step by means of a variational approach described in detail by Horn (1989). Applications to the scenario of lunar surface reconstruction are described by Wöhler and Hafezi (2005). The algorithm minimizes the mean square deviation between the observed pixel intensities and the modelled reflectances under the constraint that the surface gradients are not independent but must form an integrable vector field. To increase reconstruction accuracy, the width of the point spread function is computed based on the appearance of the boundaries of shadows in the image (Lena et al., 2006) and incorporated into the optimisation scheme (Joshi and Chaudhuri, 2004). The dome volume V was computed by integrating the reconstructed 3D profile over an area corresponding to a circular region of diameter D around the dome summit. For 8

9 the Gruithuisen and Mairan highland domes, which will be regarded for comparison in Section 5, Wilson and Head (2003) assume a parabolic shape with V = (π/2)(d/2) 2 h. Lunar Orbiter images are not suitable for 3D reconstruction of lunar surface parts based on photometric methods since the relation between incident flux and pixel greyvalue is nonlinear and unknown. The reason is that the images were acquired on a photographic film scanned on board the spacecraft. The same problem arises for the high-resolution orbital images taken with hand-held and aerial cameras from the Apollo command modules. Moreover, solar altitudes are between about 20 and 30 for Lunar Orbiter images, and nearly all Clementine images were acquired at low phase angles, corresponding to steep illumination angles in the equatorial regions. Hence, illumination is not sufficiently oblique to apply photoclinometric methods to shallow features like lunar domes. For generating digital elevation maps we thus have to rely on telescopic CCD images. The utilised CCD cameras perform an internal adjustment of the gamma value γ, leading to a pixel intensity I proportional to F γ with F as the incident flux. We calibrated the gamma scale of the camera control software by evaluating flatfield frames of different intensities acquired through different neutral density filters with known transmission coefficients, fitting a characteristic curve of the form I = af γ to the measured flatfield intensities. Three main sources of error may be relevant for the described 3D reconstruction scheme. The phase-angle dependent parameter of the reflectance function is not exactly known and may show variations over the surface for different terrain types. Furthermore, the width of the point spread function is uncertain by a value of about ±0.5 pixels. Both effects, however, change the elevation values by only a few metres in our examples. In contrast, the uncertainty of the γ value of the CCD cameras of about 0.05 results in a relative standard error of the dome height of 10 percent, which is independent of the height value itself. Based on experiments, we found that the uncertainties in dome diameter (cf. Section 3.1) and dome height lead to a standard error of the edifice volume of less than 20 percent. The resulting digital evelation maps for dome 1 and dome 2 are shown in Fig. 8. For dome 1, we obtain a height of 410±40 m, a flank slope of 2.8 ±0.3, and an edifice volume of 34 ± 7 km 3. The flank slope is an average value since the profile of dome 1 is somewhat asymmetric, with the eastern flank being steeper than the western flank. Dome 2 is shallower and less voluminous, displaying a height of 160 ± 20 m, a flank slope of 1.15 ± 0.15, and an edifice volume of 2.8 ± 0.6 km 3. 4 Physical parameters of dome formation 4.1 Classification, mode of formation Lunar domes are formed either by outpouring of magma from a central vent or by a subsurface accumulation of magma that causes an up-doming of the bedrock layers, creating a smooth, gently sloping positive relief (Head and Gifford, 1980; Basaltic Volcanism Study Project, 1981). Domes representing volcanic sources are smoothsurfaced and usually have a summit crater pit. Most vents related to domes appear to 9

10 be associated with surrounding lava plains of known volcanic origin or in association with pyroclastic deposits. The extrusive origin of lunar domes and their similarity to terrestrial features like small shield volcanoes have been described in the literature (cf. e. g. Head and Gifford, 1980). The presence of a summit crater pit argues against an intrusive or laccolithic origin for the majority of these features. The issue of intrusive lunar domes and how they are possibly related to equivalent terrestrial features is not yet well understood. They are probably formed by subsurface intrusions, similar to laccoliths on Earth, where magma has flowed under a surface of solidified lava and lifted it up (Head and Gifford, 1980; Basaltic Volcanism Study Project, 1981). Effusive lunar domes probably formed during the terminal phase of a volcanic eruption. Initially, lunar lavas were very fluid due to their high temperature. Thus, they were able to form extended basaltic mare plains. Over time, the temperature of the erupting lavas became lower, flow rate decreased, and crystallisation occurred. This changed the characteristics of the lava such that it began to pile up around the effusion vent and formed a dome (Cattermole, 1996; Mursky, 1996). Head and Gifford (1980) define seven classes of lunar domes. Classes 1 3 refer to volcanic features resembling terrestrial shield volcanoes. Class 1 domes have diameters between 5.5 and 15 km and circular to elliptic outlines, they exhibit slopes smaller than 5, and they have rimless summit crater pits. However, this diameter range is not based on a statistical analysis but rather an estimate. Class 2 domes differ from those of class 1 essentially by their pancake-like shape. Class 3 domes are similar to but lower than those of classes 1 and 2. While class 4 denotes domes associated with mare ridges and arches of possible tectonic origin, class 5 describes domes originating from lava mantling of previously existing highland terrain. Class 6 includes domes with higher albedo than mare material and comparably steep slopes, such as the Gruithuisen highland domes. Class 7 describes complex mare domes with irregular outline and topography such as Arago α and β and many of the domes in the Marius Hills region. Wöhler et al. (2006) introduce a classification scheme for effusive mare domes which is complementary to the one by Head and Gifford (1980) in that it bases the distinction between their classes 1 3 on spectral and morphometric quantities rather than a rough estimation of 3D shape. In this scheme, class A denotes spectrally blue domes (high R 415 /R 750 ratio) of low slope and volume, while domes of moderate R 415 /R 750 ratio and moderate diameter are assigned to class B. Steep domes of high edifice volume are denoted by subclass B 1, shallow domes of low edifice volume by subclass B 2. Class C represents large, shallow, high-volume edifices either of low to moderate (subclass C 1 ) or high (subclass C 2 ) R 415 /R 750 ratio. Class D is made up by large and complex edifices like Arago α and β and corresponds to class 7 of the scheme by Head and Gifford (1980). The diameter of dome 1 is slightly above the (not too strictly defined) range of class 1 by Head and Gifford (1980) and at the same time not steep enough for their highland dome class 6. In the scheme by Wöhler et al. (2006), dome 1 with its spectrally red soil, large diameter, and high edifice volume fits into class C 1 with a tendency towards B 1 due to its relatively steep flank slope. Dome 2 at W and S may be categorised as class 5 using the Head and Gifford (1980) scheme and as a typical C 1 dome in the scheme by Wöhler et al. (2006). The lava forming this feature appears to be draped on or around a portion of 10

11 nearby highland terrain. Figs. 6 and 7 show the presence of several rilles crossing the surface of dome 2. Rilles that begin in craters have two common means of formation. Some represent sinuous rilles, where lava from a source poured out onto the surface (Wilhelms, 1987). These extend from the source by some distance out into the mare surface. Other rilles represent some kind of collapse, commonly due to updoming of the surface, generating extension and subsequent down-dropping of the surface between parallel faults (Masursky et al., 1978; Wilhelms, 1987). These rilles generally contain straight segments and may be rather short. Tensional stresses are most consistent with laccolith formation but in the case of dome 2 the rille should then end where the dome merges with the mare surface. Some rilles in Figs. 6 and 7, however, extend into the surrounding mare surface and show bends in their trace. Several rilles appear to begin in oblong craters of sizes close to the resolution limit. According to the mechanisms suggested by Wilson and Head (1996) and Petrycki and Wilson (1999), narrow linear rilles and graben can be explained as the surface manifestations of dikes. A detailed geophysical model of the involved processes that lead to the formation of linear graben due to the near-surface stresses generated by emplacement of dikes is suggested by Wilson and Head (2002), who apply their model to graben systems on Mars. The model is extended to explain the emplacement of giant dikes of around 1000 km length by Scott et al. (2002). For the regarded Martian dikes, the model by Wilson and Head (2002) yields a depth of the source region of several tens of kilometres, with the dike top situated at shallow depths of a few hundred metres. Lunar graben-forming dikes have widths of the order of hundreds of metres and lengths of typically several tens of kilometres but sometimes more than 100 km (Wilson and Head, 1996; Jackson et al., 1997; Petrycki and Wilson, 1999). Head et al. (1997) argue that depending on the depth to dike top, eruptions may occur in the form of degassing, forming rimless crater pits along the graben as observed e. g. for Rima Hyginus, or as lava effusion, which formed the cones Isis and Osiris situated along a graben in Mare Serenitatis. In the eruption scenario, the depth to dike top amounts to only a few tens of metres, and the dike penetrates the surface in places. More extensive effusion processes may lead to the formation of domes (Wilson and Head, 2003). In Fig. 6, two linear rilles are apparent, marked as B and C, which extend to considerable distances from dome 2. The linear feature A rather appears to be an outflow channel or a chain of vents. A possible interpretation for dome 2 is that dikes ascended to shallow depth below the surface. The associated stress fields generated the linear rille structures B and C according to the mechanisms suggested by Wilson and Head (2002). Additionally, one of the dikes gained surface access at some points such that an extensive lava effusion could occur. According to this scenario, we interprete dome 2 as an effusive structure. 4.2 Rheologic modelling We assume that the two examined domes were formed by extrusion of magma onto a flat plane spreading in all directions from the vent, in contrast to lava flows resulting from lava extrusion onto an inclined surface. Wilson and Head (2003) provide a quantitative treatment of such dome-forming eruptions, which we will follow in our study. This 11

12 model estimates the yield strength τ, i. e. the pressure or stress that must be exceeded for the lava to flow (Wolff and Sumner, 2000), the plastic viscosity η, yielding a measure for the fluidity of the erupted lava, the effusion rate E, i. e. the lava volume erupted per second, and the duration of the effusion process. The modelling results may be of limited accuracy for dome 2 since the employed model assumes lava spreading out radially from a vent, while during the formation of dome 2 the lava was restricted at the southern side by hummocky deposits. Hence, we regard the modelling results for dome 2 merely as order of magnitude estimates. In the model by Wilson and Head (2003), the magma is treated as a Bingham plastic with a yield strength of τ = h2 ρ g D/2 (1) (Wilson and Head, 2003). The plastic viscosity is estimated by the empirical relation η(τ) = τ 2.4, (2) where τ is expressed in Pa and η in Pa s. In Eq. (1), ρ denotes the lava density, for which Wilson and Head (2003) apply a value of 2000 kg/m 3, g = 1.63 m/s 2 the acceleration due to gravity, h the height of the dome, and D its diameter. Assuming a higher density ρ will increase the viscosity by a constant factor. For a high magma density of ρ = 2800 kg m 3, this factor amounts to 2.2, compared to the values obtained with ρ = 2000 kg m 3, which is not too significant when regarding the broad range of viscosities inferred for lunar mare and highland domes (Wöhler et al., 2006). Hence, we compute the lava viscosities for ρ = 2000 kg m 3, but for the subsequent determination of dike dimensions we will consider the effects of possibly higher lava density and thus viscosity values. It is assumed that the advance of the front of a lava flow unit is limited by cooling once a critical depth of penetration of the cooled boundary layer into the flow is reached. A corresponding estimate of the lava effusion rate E is then obtained by E = /2 300 κ (D/2) / h for a dome with a parabolic cross-section (Wilson and Head, 2003). Here, κ 10 6 m 2 s 1 denotes the thermal diffusivity of the lava. The duration T of the lava effusion process amounts to T = V/E, (4) with the edifice volume V determined according to Section 3.2. Eqs. (1) (4) are valid for domes that formed from a single flow unit (monogenetic volcanoes). Otherwise, the computed values for τ, η, and E are upper limits to the respective true values. For the two domes examined in this study, we obtained comparable effusion rates of 121 m 3 /s for dome 1 and 173 m 3 /s for dome 2. They formed out of lava with viscosities of Pa s and Pa s over periods of time of 8.9 and 0.5 years, respectively (Table 2). Hence, dome 1 and 2 are quite different with respect to both their morphometric properties and the conditions under which they formed. 12 (3)

13 In the scenario of lava effusion through dikes, an important parameter is the magma rise speed U at which the dike propagates. According to Wilson and Head (2003), the value of U depends on the pressure in the dike, the buoyancy and the viscosity of the magma, and the dimensions of the dike. It is found by balancing the driving vertical pressure gradient dp/dz against the wall friction, taking into account the need to overcome the yield strength τ: U = W 2 [ dp 12η dz 2τ ]. (5) W Wilson and Head (1996) state that for eruption of magma on the surface an excess pressure at the dike source of 28 MPa is required for a crust of 64 km thickness, corresponding to a driving pressure gradient of 328 Pa m 1. For the Gruithuisen and Mairan highland domes, Wilson and Head (2003) assume no excess pressure but a pressure gradient dp/dz = g(ρ c ρ) which is due to the positive buoyancy of the magma, with ρ c as the density of the crustal material and ρ as the magma density, leading to a value of dp/dz = 1300 Pa m 1. It is realistic to assume a positive buoyancy in the case of the highland domes since they were formed from non-basaltic magma which may be less dense than the crustal material. For the case of basaltic magma reaching the bottom of the lunar crust, Wilson and Head (1996) point out that the magma is positively buoyant in the mantle and negatively buoyant in the crust. Without an excess pressure in the source region, the magma was able to rise above the base of the crust to a distance where the negative buoyancy component from the magma above the crust-mantle boundary compensated the positive buoyancy from the magma below the boundary. Since, however, the magma could not rise to the surface unless it had an untypically low density, an excess pressure arising due to the melting of the magma is assumed, leading to an additional force that drives the magma to the surface. However, Wieczorek et al. (2001) argue that also basaltic magmas may have been positively buoyant in the lower lunar crust. They show that since the lunar crust becomes more mafic with depth, basaltic magma should be less dense than the material of the lower crust. Hence, in places where the upper anorthositic crust was removed by an impact event, basaltic magma could have been driven to the surface by its positive buoyancy alone. This assumption is supported by the observation that mare basalts are present where geophysical models of crustal thickness predict the upper crust to be absent. For modelling the feeder dike dimensions of the Doppelmayer domes we will adopt the minimum value of dp/dz = 328 Pa m 1 necessary for basaltic magma to erupt at the surface (Wilson and Head, 1996). We will examine the implications of possibly higher values of dp/dz. The magma rise speed U, the dike geometry defined by the width W and the length L, and the eruption rate E are related by E = UW L. (6) However, L and W are not independent. A detailed model for the dependence between lava viscosity and the ratio L/W is suggested by Rubin (1993), regarding a pressurised dike propagating in a linear viscoelastic medium. In this model, the first parameter 13

14 governing dike dimensions is p 0 /G, representing a measure for the elastic response of the host rock, where p 0 is the magma pressure at the dike entrance and G the elastic stiffness of the host rock. The second parameter is η/η r, where η is the magma viscosity and η r the host rock viscosity. Rubin (1993) points out that the value of p 0 /G is situated between 10 4 and Wilson and Head (2003), who also apply this model, assume η r = Pa s for estimating the dike properties of the Gruithuisen and Mairan highland domes and obtain L/W = 200 for the lava viscosities around Pa s they infer, which corresponds to setting p 0 /G = With these parameters for p 0 /G and η r, we obtain L/W = 1300 for dome 1 and L/W = 4470 for dome 2. Inserting Eq. (6) into Eq. (5) then yields a relation for the dike width W according to W 4 = 12ηE (L/W ) [dp/dz (2τ/W )]. (7) We solved Eq. (7) numerically for W with the nested intervals method. For dome 1, we obtained a low magma rise speed of m s 1, a dike width of 127 m, and a dike length of 167 km. The less viscous lava of dome 2 ascended at a higher speed of m s 1, and its feeder dike is narrower (W = 27 m) than the dike that formed dome 1 but of comparable length (L = 121 km). The influence of the driving pressure gradient dp/dz is such that increasing dp/dz by a factor of two yields for both domes a magma rise speed U which is larger by a factor of four, while dike width W and length L decrease by a factor of two. 5 Discussion The region around Doppelmayer shows many traces of ancient volcanic activity. The described two lunar domes near Doppelmayer do not appear to be connected with any of the two LPDs in southern Mare Humorum described by Gaddis et al. (2003), mapped as Eid units in USGS map I-495. Both domes are spectrally red (low R 415 /R 750 ratio), indicating a low to moderate TiO 2 content, and consist of mature material as shown by their high R 950 /R 750 ratios. Dome 1 is a large, comparably steep (flank slope nearly 3 ), and voluminous edifice. Dome 2 is slightly smaller than dome 1, significantly shallower with a flank slope of just above 1, and less voluminous by an order of magnitude. Spectrally, it is not distinguishable from the mare-like surface to its north. The rheologic model by Wilson and Head (2003) yields comparable effusion rates for the two domes. However, lava viscosity was higher by a factor of 30 for dome 1 than for dome 2, and the duration of the effusion process was nearly 20 times as long. Hence, although the two domes are located close to each other, their spectral, morphometric, and effusion properties indicate significantly different eruption conditions. Dome 1 has been formed by a much broader dike than dome 2. In contrast, the dike lengths are not strongly different. If we assume that the vertical extension of a dike is similar to its length (cf. Jackson et al. (1997) and references therein), we conclude that the magmas which formed the Doppelmayer domes originate from well below the lunar crust, for which a total thickness of 50 km is given by Wieczorek et al. (2006) for the region in which the domes are located. 14

15 5.1 Dome 1: Lateral vs. vertical mixing mechanisms, comparison to mare and highland domes The UVVIS spectrum of dome 1 is intermediate in reflectance between the dark and smooth mare pond to its west and that of the nearby hummocky terrain having a higher reflectance, suggesting that it appears to consist of a 1:1 mixture of the corresponding mare and hummocky terrain soils (Fig. 4b). Such a mixing effect between mare and highland soil has also been found by Lena et al. (2006) for the material of the LPD partially covering the lunar dome in the south of Petavius. However, the lowest 3 4 km of the western flank of dome 1, corresponding to the easternmost part of the crescent-shaped mare pond (Fig. 4a), display a mare-like spectral appearance. The highland component inferred from the UVVIS spectrum (Figs. 4b and 5) may have been incorporated into the dome material either during the lava effusion process or later by lateral mixing with impact ejecta. A vertical mixing mechanism that could explain the high lava viscosity and in turn the morphometric properties of dome 1 is assimilation of crustal wallrock into the up-welling basaltic magma during the eruption process (Grove, 2000). In this scenario, basaltic magma from the mantle melts parts of the crustal wallrock through which it rises upwards, such that the molten wallrock mixes with the magma and rises to the surface. Hence, the fact that the material of dome 1 spectrally appears intermediate between basaltic lava and highland material could be explained as the result of the mixing between up-welling lava and surrounding crustal wallrock. The high lava viscosity of 10 7 Pa s inferred from the morphometric properties of dome 1 may be partially due to the low metal content of the soil (Melendrez et al., 1994; Williams et al., 2000), where the very high R 950 /R 750 ratio implies a low FeO content (Lucey et al., 1998), while the low R 415 /R 750 ratio suggests a low TiO 2 content (Gillis and Lucey, 2005). However, although lava viscosity is known to increase with decreasing metal content, the analysis by Wöhler et al. (2006) reveals that other domes with similar spectral properties were formed from lavas of much lower viscosity. In contrast, heat transfer from the up-welling magma into the wallrock implied by the assimilation scenario might explain the inferred high viscosity as a result of low magma temperature (Spera, 2000). However, Warren (1985) states that the magma has to be considerably superheated to be able to assimilate a significant fraction of crustal material. Hess (1994) shows that 120 C of superheat is a realistic upper bound under lunar conditions. The crustal material has to be brought to its solidus temperature before assimilation becomes possible. According to Hess (1994), even in the early history of the Moon the temperature of the lower crust was not higher than 800 C, while the solidus temperature of the crust corresponds to at least 1000 C, i. e. about 200 C above the crustal temperature. Hess (1994) demonstrates that even under these favourable circumstances all available superheat is expended already for increasing the crustal temperature to the lowest reasonable melting temperature under realistic circumstances, however, crustal solidus temperatures are likely above 1200 C. Assuming that the maximum temperature of the wallrock at the border of the dike is equal to the average of the initial magma 15

16 and wallrock temperatures 1, Hess (1994) argues that the crustal temperature will not exceed 1150 C, rendering the assimilation of a significant fraction of crustal wallrock very unlikely. As an alternative mechanism to explain significant highland components in mare soils and vice versa, lateral mixing due to random impacts of small bodies is suggested by Li et al. (1997) and modelled in more detail by Li and Mustard (2000). They infer the relative fraction of mare and highland soil along mare-highland contacts based on spectral mixture modelling of Clementine UVVIS data and introduce a so-called anomalous diffusion model that fits well the observed relative abundances at distances of up to 10 km from the boundary. The symmetric shapes of the fraction profiles perpendicular to the boundary indicate that the efficiency of vertical mixing by impact cratering, leading to contamination of mare soil by highland material excavated from below the mare basalt, is negligible, compared to lateral mixing. In a later work (Li and Mustard, 2003), mapping of mare and highland abundances is refined based on multiple end-member spectral mixture analysis. Li and Mustard (2005) provide a detailed model for the ejecta thickness resulting from impact cratering by bodies covering a broad range of sizes. Their study demonstrates that lateral mixing turns out to be efficient enough to distribute a fraction of 20 30% of exotic components even over distances larger than 100 km. Accordingly, the much more natural explanation for the observed spectral appearance of dome 1 is lateral mixing. Dome 1 is situated right on the boundary between the mare pond to its west and the hummocky terrain south of Doppelmayer. As expected from the model by Li and Mustard (2000), the fraction of highland material is lowest for the westernmost part of the dome surface, which is most distant from the mare-highland boundary. At this point it is illustrative to compare dome 1 to non-mare (highland) and mare domes. Wood and Head (1975) describe various Red Spots displaying a relatively high albedo and a strong absorption in the near UV. Their spectral signatures are different from those of mare basalts but also of the highland areas surrounding them. Among the Red Spots, the highland domes Gruithuisen γ and δ and the nearby Northwest Dome, situated at the northwestern border of Mare Imbrium, and the three Mairan domes, nearby located at the eastern border of Sinus Roris, are described as volcanic edifices by Head and McCord (1978). Head et al. (1978) discuss their formation from highly silicic non-mare lavas. Stratigraphic relations indicate that they were formed after the Imbrium impact but before the basin was flooded by mare basalts. Detailed mapping of the distribution of non-mare volcanic material is provided by Chevrel et al. 1 In the case of stationary magma (U = 0), the wallrock obtains the average of the initial magma and wallrock temperatures independent of the dike width W (Carrigan et al., 1992) for a limited period of time and then cools down slowly. The assumption of this maximum temperature value should still be a good approximation for the magmas that formed the Doppelmayer domes, ascending at very low speeds of 10 5 m s 1. However, for velocities so high that the magma reaches the surface long before it has lost a relevant amount of its heat to the wallrock, Carrigan et al. (1992) show based on numerical modelling that the magma temperature in the dike is largely uniform and hardly deviates from the initial temperature over large distances from the dike source, while the wallrock temperature at the border of the dike is close to the magma temperature. This effect is demonstrated for magma velocities of 0.1 m s 1 and higher in a dike of 3 m width. 16

17 (1999) based on Clementine UVVIS data. Wilson and Head (2003) quantitatively determine the rheologic properties of the lavas from which they formed and estimate the dimensions of their feeder dikes, utilising the model we have adopted for our study (cf. Section 4.2). Hawke et al. (2003) demonstrate that a further Red Spot, Hansteen α situated at the southwestern border of Oceanus Procellarum, is most likely of volcanic origin and was also formed by extrusion of viscous non-mare lavas. The Gruithuisen domes are known to have large diameters of up to 20 km, heights of more than 1000 m, steep flank slopes between 7 and 15, and very high edifice volumes of several hundred km 3. Gruithuisen δ is composed of a north-western and a south-eastern partial dome. Weitz et al. (1999) show that steeper domes represent the result of cooler, more viscous lavas with high crystalline content, possibly at the final stages of the eruption. According to Chevrel et al. (1999), the Gruithuisen domes have been formed by viscous lava of very low TiO 2 content (corresponding to a very low R 415 /R 750 ratio) and more silicic composition, presumably prior to the lava flooding of the Imbrium basin. Wilson and Head (2003) show that while the eruption processes that formed Gruithuisen δ and the Northwest Dome occurred over more than 20 years at low effusion rates between 6 and 50 m 3 s 1, the effusion rate was 119 m 3 s 1 for Gruithuisen γ over a period of 38 years. The lava that formed the Gruithuisen highland domes had viscosities between 10 8 and 10 9 Pa s. All three Gruithuisen domes are characterised by very low R 415 /R 750 ratios (Fig. 5). Compared to the Gruithuisen domes, the highland domes near Mairan located in northern Oceanus Procellarum are somewhat smaller and shallower. According to the morphometric data provided by Wilson and Head (2003), Mairan T is the largest and steepest of them with a diameter of 13 km, a flank slope of 7.9, and an edifice volume of about 60 km 3, while the other two domes, Mairan middle and south, have diameters around 10 km, flank slopes around 6, and edifice volumes between 20 and 30 km 3 (Table 1). The rheologic model yields lava viscosities between and Pa s, effusion rates around 50 m 3 s 1, and durations of the effusion process of 42, 18, and 13 years for Mairan T, middle, and south. The average magma rise speed for the Gruithuisen and Mairan highland domes amounts to 10 5 m s 1 for a magma density of 2000 kg m 3 assumed by Wilson and Head (2003). The inferred dike widths are between 80 and 200 m while the dike lengths range from 16 to 48 km (Table 2). These dike dimensions are given by Wilson and Head (2003), who assume a rather low lava density of 2000 kg m 3 (cf. Section 4.2). Setting the magma density to a higher but realistic value of 2400 kg m 3 reduces the driving pressure gradient by a factor of two and in turn increases the dike width and length by a factor of two. One must be careful when interpreting the flank slope values of the highland domes, since according to Wilson and Head (2003), the Gruithuisen domes and Mairan T have been formed during at least two distinct subsequent eruption phases, respectively, a process that may build up steeper edifices. Hence, the rheologic properties given in Table 2 for Gruithuisen δ and NW are reported for the individual layers, respectively, as determined by Wilson and Head (2003). In contrast, like most mare domes the Doppelmayer domes show no traces of multiple lava flows and are thus presumably monogenetic. Doppelmayer dome 1 is intermediate in diameter between the Mairan and 17

18 Gruithuisen domes. While its flank slope is far below the range of values covered by the highland domes, its edifice is of comparable volume. The estimated value for the lava viscosity of Doppelmayer dome 1 is about an order of magnitude lower than that of the Gruithuisen and Mairan domes, its effusion rate is similar to that of Gruithuisen γ, and the duration of the effusion process is similar to that of Mairan south (cf. Table 2). Compared to the Mairan domes, dome 1 displays a comparable R 415 /R 750 ratio and a slightly lower 750 nm albedo (Fig. 5). The Gruithuisen domes are spectrally redder than dome 1. In the R 750 vs. R 415 /R 750 and the R 750 vs. R 950 /R 750 diagrams, dome 1 is located rather close to the Gruithuisen and Mairan highland domes, being situated between mare and highland soils. In the R 950 /R 750 vs. R 415 /R 750 diagram, mare and highland soils overlap (Gaddis et al., 2003), and dome 1 is located near the Gruithuisen and Mairan domes but at the same time close to dome 2. For a set of 38 effusive lunar mare domes, Wöhler et al. (2006) report the morphometric properties (obtained using the photoclinometry and shape from shading approach outlined in Section 3.2) and corresponding rheologic parameters according to the model by Wilson and Head (2003). For the effusive mare domes examined in that study, steeper flank slopes of more than 2 only occur for smaller dome diameters below 13 km, while all examined mare domes with diameters larger than 13 km have flank slopes shallower than 2. With its lava viscosity of 10 7 Pa s, dome 1 is in the uppermost range of viscosities determined for mare domes. Its effusion rate lies well within the range determined for mare domes, while the estimated duration of the effusion process is atypically long. In the classification scheme for mare domes introduced by Wöhler et al. (2006), dome 1 belongs to class C 1 with a tendency towards B 1 due to its relatively steep flank slope. A lunar mare dome that comes close to dome 1 in its morphometric and rheologic properties is a member of the well-studied dome suite north of the crater Hortensius in Mare Insularum, termed Hortensius 6 by Head and Gifford (1980) and located at W and 7.82 N. Using the approach outlined in Section 3.2 and the high-resolution image of the Hortensius region shown in Fig. 9, we found that it has a base diameter of 12.5 km, a flank slope of 3.6, and an edifice volume of 32 km 3. Note that with its steep flank slope and large edifice volume, Hortensius 6 is not a typical representative of lunar mare domes (Wöhler et al., 2006). Its rheologic parameters are η = Pa s for the lava viscosity, E = 70 m 3 s 1 for the effusion rate, and T = 14.6 years for the duration of the effusion process. Assuming the same driving pressure gradient of 328 Pa m 1 as for dome 1, we found that the feeder dike of Hortensius 6 has a width of 157 m and a length of 160 km. Hortensius 6 is spectrally slightly bluer than Doppelmayer dome 1, implying a higher TiO 2 content, and has a somewhat lower 750 nm albedo (Fig. 5). It is spectrally not distinguishable from the mare surface surrounding it. In the case of Hortensius 6 consisting of typical mafic high-feo and moderate-tio 2 mare material, factors apart from lava composition, e. g. low eruption temperature with resulting high crystallinity of the lava, must have led to the high lava viscosity. Dome 1 presumably formed under similar conditions as Hortensius 6 from lava of comparably high viscosity, intermediate between the viscosities typically observed for mare domes and for highland domes, respectively, erupting at moderate rates over longer periods of time than typical of mare 18

19 domes. Interestingly, the dike widths inferred for dome 1 and also for Hortensius 6 are of the same order as the values inferred for the dikes that are supposed to have formed Rima Parry V and Rima Hyginus by intrusion to shallow depth below the surface (Wilson and Head, 1996; Jackson et al., 1997). Possibly these domes formed from dikes filled with relatively cool, highly viscous magma nearly saturated with crystals which gained surface access at some points (in contrast to the graben-forming dikes), leading to dome-forming effusion of lava. Given the large diameter of dome 1, its flank slope is atypically steep for mare domes but lower than the values observed for highland domes. The only further similarly steep and voluminous mare domes known to date, with flank slopes steeper than 2.5 and edifice volumes above 20 km 3, are Hortensius 5, situated immediately west of Hortensius 6 (cf. Fig. 9), and Herodotus ω, located about 90 km south of the crater Herodotus in western Oceanus Procellarum (Wöhler et al., 2006). Presumably, Doppelmayer dome 1 is a further exemplar of this rare type of mare domes. 5.2 Dome 2: Comparison to other mare domes Dome 2 is a typical effusive mare dome, given its spectral and morphometric properties (cf. Fig. 5 and Table 1). It belongs to class C 1 according to the scheme by Wöhler et al. (2006). It consists of mare material partially draped over the hummocky deposit to its south. Possibly, it is a source vent of the mare-like region to its north. A comparable mare dome with similar spectral and morphometric properties, examined by Head and Gifford (1980) and termed Tobias Mayer 1 in their catalogue, is located at W and N in the dome field between the craters Milichius and Tobias Mayer. Employing the 3D reconstruction approach outlined in Section 3.2 based on the image of the Milichius region shown in Fig. 9, we determined a base diameter of 13.4 km, a flank slope of 0.9, and an edifice volume of 8.2 km 3 for Tobias Mayer 1 (Table 1). Dome 2 and Tobias Mayer 1 show spectral signatures characteristic for lunar mare soils (Fig. 5). The LPD displays a somewhat stronger mafic absorption than the domes, which can be explained by an iron-rich glassy soil component (Gaddis et al., 2003). As a note of interest, Murase and McBirney (1970) experimentally found very low viscosity values of 1 Pa s for the lunar lavas that flooded the large mare basins. According to the mare-like spectral signature obtained for dome 2, the much higher viscosity of the lava that formed it ( 10 5 Pa s) was probably not caused by its composition but likely originated from a low eruption temperature (presumably higher, though, than that of the lava forming dome 1) and thus high crystallinity of the lava. According to the dike intrusion mechanisms described by Wilson and Head (2002), the narrow linear rilles near dome 2 (Fig. 6) were probably formed by dikes that ascended to shallow depths below the surface. One of the dikes, which initially remained subsurface, gained surface access at some points due to failure of the overlying rocks. As a consequence, an extensive effusion of relatively cool and viscous lava occurred, leading to the formation of dome 2 and the generation of the associated outflow channel or chain of effusive vents (cf. feature A in Fig. 6). 19

20 6 Summary and conclusion In this study we have examined two effusive lunar domes near Doppelmayer, displaying markedly different spectral, morphometric, and rheologic properties. Dome 1 is interpreted to be an effusive volcanic structure, due to the presence of a shallow rimless crater pit on its summit. Its spectral signature indicates a highland component in the dome material. Dome 1 is atypically large, steep, and voluminous, compared to lunar mare domes. It was formed of lava with a relatively high viscosity of 10 7 Pa s, which is between the ranges of values typically observed for lunar mare domes and highland domes, respectively, erupting at moderate rates over a long period of time of nine years. With respect to the determined spectral and morphometric properties and the inferred rheologic parameters, we conclude that dome 1 is an exemplar of a rare type of mare domes with intermediate to large diameters, flank slopes steeper than 2.5, and edifice volumes larger than 20 km 3. Only three further mare domes of this kind are known to date. Our discussion of lateral vs. vertical mixing mechanisms has led to the conclusion that the thermal conditions in the lunar interior did not favour the assimilation of crustal wallrock into the ascending magma. Due to the fact that dome 1 is located right on the boundary between hummocky terrain and a mare pond, lateral mixing of highland into mare soil is a much more natural explanation for the observed spectral signature of dome 1, taking into account the high efficiency of lateral transport mechanisms as described by Li and Mustard (2000). Dome 1 likely formed from basaltic lava, the spectral signature of which is still apparent at the bottom of its western flank. Supposedly, the main reason for the inferred high viscosity of the lava that formed dome 1 is a low eruption temperature and thus a high crystallinity of the lava. Dome 2 is a typical effusive mare dome, given its mare-like composition, shallow flank slope, and low edifice volume. It formed from fluid lavas of a viscosity of 10 5 Pa s that erupted at a high rate over a short period of time of only half a year. The rilles and tensional fractures observed on the surface of dome 2 indicate that it was formed by a dike that initially remained subsurface but gained access to the surface at localised positions, resulting in extensive lava effusion. While we inferred that dome 1 was formed by a considerably wider feeder dike than dome 2 (127 vs. 27 m), the dike lengths are comparable and amount to 167 and 121 km, respectively. If we assume that the vertical extension of a dike is similar to its length, the magma reservoirs of both domes were situated below the lunar crust. Acknowlegdements: We are grateful to S. Hanmer for his critical reading of the manuscript. We extend our thanks to two anonymous referees and Dr. A. Basilevsky, whose comments and suggestions improved this paper. References [1] Adams, J. B., McCord, T. B., Remote sensing of lunar surface mineralogy: implication from visible and near infrared reflectivity of Apollo 11 samples. Proc. 20

21 Apollo 11 Lunar Sci. Conf., [2] Basaltic Volcanism Study Project, Basaltic Volcanism on the Terrestrial Planets. Pergamon Press, Inc., New York. [3] Baumgardner, J., Mendillo, M., Wilson, J. K., A digital high definition imaging system for spectral studies of extended planetary atmospheres, 1. Initial result in white light showing features on the hemisphere of Mercury unimaged by Mariner 10. Astronomical Journal 119, [4] Burns, R. G., Parkin, K. M., Loeffler, B. M., Leung, I. S., Abu-Eid, R. M., Further characterization of spectral features attributable to titanium on the moon. Lunar Planet. Sci. VII, [5] Bussey, B. J., Spudis, P. D., Hawke, B. R., Lucey, P. G., Peterson, C., Taylor, G. J., Humorum basin geology from Clementine data. Lunar Planet. Sci. XXVIII, abstract #1294. [6] Carrigan, C. R., Schubert, G., Eichelberger, J. C., Thermal and dynamical regimes of single- and two-phase magmatic flow in dikes. J. Geophys. Res. 97(B12), [7] Cattermole, P., Planetary Volcanism. 2nd ed., John Wiley and Sons Ltd., Chichester. [8] Charette, M. P., McCord, T. B., Pieters, C. M., Adams, J. B., Application of remote spectral reflectance measurements to lunar geology classification and determination of titanium content of lunar soils. J. Geophys. Res. 79, [9] Chevrel, S. D., Pinet, P. C., Head, J. W., Gruithuisen domes region: A candidate for an extended nonmare volcanism unit on the Moon. J. Geophys. Res. 104(E7), [10] Eliason, E., Isbell, C., Lee, E., Becker, T., Gaddis, L., McEwen, A., Robinson, M., Mission to the Moon: the Clementine UVVIS global mosaic. PDS Volumes USA NASA PDS CL [11] Gaddis, L. R., Staid, M. I., Tyburczy, J. A., Hawke B. R., Petro, N. E., Compositional analyses of lunar pyroclastic deposits. Icarus 161(2), [12] Gillis, J. J., Lucey, P. G., Evidence that UVVIS ratio is not a simple linear function of TiO2 content for lunar mare basalts. Lunar Planet. Sci. XXXVI, abstract #2252. [13] Gillis, J. J., Lucey, P. G., Campbell, B. A., Hawke, B. R., Clementine 2.7 µm data and 70-cm earth-based radar data provide additional constraints for UVVIS-based estimates of TiO 2 content for lunar mare basalts. Lunar Planet. Sci. XXXVI, abstract #

22 [14] Grove, T. L., Origin of magmas. In: Encyclopedia of Volcanoes, H. Sigurdsson (ed.), Academic Press, San Diego, USA. [15] Hackwill, T., Guest, J., Spudis, P., Stratigraphy and evolution of basalts in Mare Humorum and southeastern Procellarum. Meteoritics and Planetary Science 41(3), [16] Hapke, B., Theory of reflectance and emittance spectroscopy. Cambridge University Press, Cambridge, UK. [17] Hawke, B. R., Lawrence, D. J., Blewett, D. T., Lucey, P. G., Smith, G. A., Spudis, P. D., Taylor, G. J., Hansteen Alpha: A volcanic construct in the lunar highlands. J. Geophys. Res. 108(E7), [18] Head, J. W., Gifford, A., Lunar mare domes: classification and modes of origin. The Moon and Planets 22, [19] Head, J. W., Hess, P., McCord, T. B., Geologic characteristics of lunar highland volcanic domes (Gruithuisen and Mairan region) and possible eruption conditions. Lunar Planet. Sci. IX, [20] Head, J. W., McCord, T. B., Imbrian-age highland volcanism on the moon The Gruithuisen and Mairan domes. Science 199, [21] Head, J. W., Wilson, L., Anderson, K. A., Lin, R. P., Lunar linear rilles, models of dike emplacement and associated magnetization features. Lunar Planet. Sci. XXVIII, abstract #1242. [22] Hess, P., Petrogenesis of lunar troctolites. J. Geophys. Res. 99(E9), [23] Hiesinger, H., Head, J. W., Wolf, U., Jaumann, R., Neukum, G., Ages and stratigraphy of mare basalts in Oceanus Procellarum, Mare Nubium, Mare Cognitum, and Mare Insularum. J. Geophys. Res. 108(E7), [24] Horn, B. K. P., Height and Gradient from Shading. MIT technical report 1105A. [25] Jackson, P. A., Wilson, L., Head, J. W., The use of magnetic signatures in identifying shallow intrusions on the moon. Lunar Planet. Sci. XXVIII, abstract #1429. [26] Joshi, M. V., Chaudhuri, S., Photometric Stereo Under Blurred Observations. Proc. 17th Int. Conf. on Pattern Recognition 3, , Cambridge, UK. [27] Kuiper, G. P., Whitaker, E., Strom, R., Fountain, J., Larson, S., Consolidated Lunar Atlas. Lunar and Planetary Laboratory, University of Arizona. Digital version edited by E. Douglass, 2003, LPI Contribution no. 1114, 22

23 [28] Lena, R., Wöhler, C., Bregante, M. T., Fattinnanzi, C., A combined morphometric and spectrophotometric study of the complex lunar volcanic region in the south of Petavius. Journal of the Royal Astronomical Society of Canada 100(1), [29] Li, L., Mustard, J. F., He, G., Compositional gradients across mare-highland contancts: The importance of lateral mixing. Lunar Planet. Sci. XXVIII, abstract #1724. [30] Li, L., Mustard, J. F., Compositional gradients across mare-highland contacts: Importance and geological implication of lateral transport. J. Geophys. Res. 105(E8), [31] Li, L., Mustard, J. F., Highland contamination in lunar mare soils: Improved mapping with multiple end-member spectral mixture analysis (MESMA). J. Geophys. Res. 108(E6), [32] Li, L., Mustard, J. F., On lateral mixing efficiency of lunar regolith. J. Geophys. Res. 110(E11), E [33] Lucey, P. G., Blewett, D. T., Hawke, B. R Mapping the FeO and TiO2 content of the lunar surface with multispectral imagery. J. Geophys. Res. 103(E2), [34] Masursky, H., Colton, G., El-Baz, F., Apollo Over the Moon. NASA SP-362, GPO, Washington. [35] McCord, T. B., Adams, J. B., Progress in optical analysis of lunar surface composition. The Moon 7, [36] McCord, T. B., Charette, M. P., Johnson, T. V., Lebofsky, L. A., Pieters, C., Adams, J. B., Lunar spectral types. J. Geophys. Res. 77, [37] McEwen, A. S., Photometric Functions for Photoclinometry and Other Applications. Icarus 92, [38] Melendrez, D. E., Johnson, J. R., Larson, S. M., Singer, R. B., Remote sensing of potential lunar resources. 2. High spatial resolution mapping of spectral reflectance ratios and implications for nearside mare TiO 2 content. J. Geophys. Res. 99(E3), [39] Murase, T., McBirney, A. R., Viscosity of lunar lavas. Science 167, [40] Mursky, G., Introduction to Planetary Volcanism. Prentice Hall, Upper Saddle River. [41] Petrycki, J. A., Wilson, L., Volcanic Features and Age Relationships Associated with Lunar Graben. Lunar Planet. Sci. XXX, abstract #

24 [42] Pieters, C., Head, J. W., McCord, T. B., Adams, J. B., Zisk, S., Geochemical and geological units of Mare Humorum: Definition using remote sensing and lunar sample information. Lunar Sci. Conf. VI, [43] Pike, R. J., Depth/Diameter relations of fresh lunar craters: Revision from Spacecraft Data. Geophys. Res. Lett. 1, [44] Rubin, A. S., Dikes vs. diapirs in viscoelastic rock. Earth and Planet. Sci. Lett. 199, [45] Scott, E. D., Wilson, L., Head, J. W., Emplacement of giant radial dikes in the northern Tharsis region of Mars. J. Geophys. Res. 107(E4), [46] Spera, F. J., Physical Properties of Magma. In: Sigurdsson, H. (ed.), Encyclopedia of Volcanoes, Academic Press, San Diego, USA. [47] Titley, S. R., USGS Map I-495. [48] Warren, P., Genesis of the geochemical bimodality of prostine nonmare rocks. Lunar Planet. Sci. XVI, [49] Weitz, C. M, Head, J. W., Spectral properties of the Marius hills volcanic complex and implication for the formation of lunar domes and cones. J. Geophys. Res. 104(E8), [50] Wieczorek, M. A., Zuber, M. T., Phillips, R. J., The role of magma buoyancy on the eruption of lunar basalts. Earth and Planet. Sci. Lett. 185, [51] Wieczorek, M. A., and 15 coauthors, The Constitution and Structure of the Lunar Interior. Rev. Mineralogy and Geochemistry 60, [52] Wilcox, B. B., Lucey, P. G., Hawke, B. R., Radiative transfer modeling of compositions of lunar pyroclastic deposits. Lunar Planet. Sci. XXXVII, abstract #1490. [53] Wilhelms, D., The geologic history of the Moon. USGS Prof. Paper [54] Williams, D. A., Fagents, S. A, Greeley, R., A reassessment of the emplacement and erosional potential of turbulent, low-viscosity lavas on the Moon. Lunar Planet. Sci. XXXI, abstract #1102. [55] Wilson, L., Head, J. W., Lunar Linear Rilles as Surface Manifestations of Dikes: Theoretical Considerations. Lunar Planet. Sci. XXVII, [56] Wilson, L., Head, J. W., Tharsis-radial graben systems as the surface manifestations of plume-related dike intrusion complexes: Models and implications. J. Geophys. Res. 107(E8), [57] Wilson, L., Head, J. W., Lunar Gruithuisen and Mairan domes: Rheology and mode of emplacement. J. Geophys. Res. 108(E2),

25 [58] Wöhler, C., Hafezi, K., A general framework for three-dimensional surface reconstruction by self-consistent fusion of shading and shadow features. Pattern Recognition 38(7), [59] Wöhler, C., Lena, R., Lazzarotti, P., Phillips, J., Wirths, M., Pujic, Z., A combined spectrophotometric and morphometric study of the lunar mare dome fields near Cauchy, Arago, Hortensius, and Milichius. Icarus 183, [60] Wolff, J. A., Sumner, J. M., Lava fountains and their products. In: Encyclopedia of Volcanoes, H. Sigurdsson (ed.), Academic Press, San Diego, USA. [61] Wood, C. A., Lunar Photo of the Day, March 23, [62] Wood, C. A., Andersson, L., New morphometric data for fresh lunar craters. Lunar Planet. Sci. IX, [63] Wood, C. A., Head, J. W., Geologic setting and provenance of spectrally distinct pre-mare material of possible volcanic origin. Conf. on Origins of Mare Basalts and their Implications for Lunar Evolution, abstract #234,

26 Feature Long. [ ] Lat. [ ] D [km] h [m] slope [ ] V [km 3 ] Dome ± ± ± ± 7 Dome ± ± ± ± 0.6 Gruithuisen γ Gruithuisen δ Gruithuisen NW Mairan T Mairan middle Mairan south Tobias Mayer ± ± ± ± 2.0 Hortensius ± ± ± ± 6 Table 1: Morphometric properties of the domes near Doppelmayer, the Gruithuisen and Mairan highland domes, and the mare domes Tobias Mayer 1 and Hortensius 6. The values for the diameters and heights of the highland domes were adopted from Wilson and Head (2003), and their volumes were computed under the assumption of a parabolic shape. Feature η [Pa s] E [m 3 /s] T [years] U [m s 1 ] W [m] L [km] Dome Dome 2 (estimated) Gruithuisen γ Gruithuisen δ NW partial dome , , Gruithuisen δ SE partial dome , , 19.3 Gruithuisen NW , , Mairan T Mairan middle Mairan south Tobias Mayer Hortensius Table 2: Rheologic parameters and dimensions of feeder dikes for the domes near Doppelmayer, the Gruithuisen and Mairan highland domes, and the mare domes Tobias Mayer 1 and Hortensius 6. The values for the highland domes were adopted from Wilson and Head (2003). For Gruithuisen δ and Gruithuisen NW, the rheologic values are given for the upper and lower flow layer, respectively. To estimate the magma rise speed U and feeder dike dimensions W and L, the driving pressure gradient was set to 328 Pa m 1 for the Doppelmayer domes and the mare domes regarded for comparison and to 1300 Pa m 1 for the Gruithuisen and Mairan highland domes. 26

27 Figure 1: Section of Consolidated Lunar Atlas image F19 (Kuiper et al., 1967; digital version edited by E. Douglass, 2003), showing the Mare Humorum region. The scale bar indicates the undistorted image scale as a 100 km reference distance (in telescopic images, scale is direction-dependent due to perspective distortion). The two domes are marked by circles and labelled 1 and 2. 27

28 Figure 2: Telescopic CCD images of the region around Doppelmayer. (a) November 13, 2005, 00:40 UT, 200 mm APO TMB refractor, Atik CCD camera. (b) March 22, 2005, 00:59 UT, 450 mm reflector, Atik CCD camera, 685 nm IR pass filter. (c) September 15, 2005, 01:43 UT, 200 mm APO TMB refractor, Atik CCD camera. (d) August 30, 2005, 03:51 UT, 200 mm Newtonian, Philips ToUCam CCD camera. (e) November 12, 2005, 20:25 UT, 130 mm APO TMB refractor, Philips ToUCam CCD camera. In all images, north is to the top and west to the left. Scale bars indicate undistorted image scale. The crater Doppelmayer is labelled as D. The line pairs mark the two domes south of Doppelmayer. Images (a) (c) and (e) were acquired under local sunrise illumination, image (d) under local sunset illumination. 28

29 Figure 3: Lunar Orbiter high-resolution image IV-143-H1 of the region around Doppelmayer. North is to the top and west to the left. The two domes are indicated by horizontal line pairs. 29

30 Figure 4: (a) Clementine 750 nm image of the region south-west of Doppelmayer. The LPD near and inside Doppelmayer, the two domes, the dark smooth terrain west of dome 1, and the hummocky terrain south-east of it are indicated. The outline of dome 1 is drawn as a circle. (b) Clementine UVVIS spectra of the indicated locations. The spectra of dome 2 (circles, dashed line) and the dark smooth terrain (crosses, dashed line) are very similar. 30

31 Figure 5: Spectral diagrams of the locations indicated in Fig. 4, the highland domes Gruithuisen γ, δ, and NW, the highland domes Mairan T, middle, and south, and the mare domes Tobias Mayer 1 and Hortensius 6. (a) 750 nm albedo vs. R 415 /R 750 colour ratio. (b) 750 nm albedo vs. R 950 /R 750 colour ratio. (c) R 950 /R 750 vs. R 415 /R 750 colour ratio. In (a) and (b), mare soils are represented by the dark grey regions and highland soils by the light grey regions, according to Gaddis et al. (2003). 31

32 Figure 6: Enlarged section of the Lunar Orbiter image shown in Fig. 3, centred around dome 2. North is to the top and west to the left. Feature A, marked by horizontal parallel lines, appears to be an outflow channel or chain of vents. Features B and C, denoted by vertical parallel lines, are linear rilles (cf. Section 4.1). The dotted line and the two long solid lines mark the outline of dome 2. Figure 7: Clementine 750 nm image of the region around dome 2. North is to the top and west to the left. Three linear features are indicated, the northwestern one of which corresponds to feature B in Fig. 6, while the other two are not apparent in Fig

33 Figure 8: (a) Digital elevation map of dome 1, viewed from south-eastern direction. (b) Digital elevation map of dome 2, viewed from north-eastern direction. Figure 9: Top: Lunar mare dome Tobias Mayer 1 (marked by horizontal line pair). Image taken on December 22, 2004, at 02:37 UT, using a 200 mm f/9 APO TMB refractor and an Atik CCD camera. Bottom: Lunar mare dome Hortensius 6 (marked by horizontal line pair). The dome Hortensius 5 is situated to the immediate left of Hortensius 6. Image taken on February 19, 2005, at 02:28 UT, using a 450 mm reflector and an Atik CCD camera. In both images, north is to the top and west to the left. Scale bars indicate the undistorted image scale. 33