Lobate debris aprons, known to be geomorphic landform indicators of the presence of ground ice, are of special interest for future missions devoted to the research of water on Mars. Lobate debris aprons in fretted terrains of Deuteronilus and Protonilus Mensae (35°–50°N) show typical convex shapes interpreted to be the result of viscous deformation. At the scale of Mars Orbiter Camera (MOC) high-resolution images the surface of these debris aprons shows complex patterns with small pits and buttes. These patterns can be explained by the mantling of dust, the accumulation of interstitial ice, and the subsequent removal of ice by sublimation. The sublimation of the ground ice is especially initiated and accelerated by subsurface heterogeneities like fractures. Theoretical quantification of sublimation rates therefore minimizes sublimation, which is not a homogeneous process, at least over the landforms studied. Crater counts show that the sublimation occurred in the last tens of millions of years up to the recent past. In the point of view of future searching of subsurface ice, only surface layers are submitted to sublimation favoring the conservation of ground ice in deeper layers since the formation of the landform. The geophysical survey of lobate debris aprons would give interesting insights into the subsurface distribution of ice and its seasonal variations, especially in order to measure current sublimation of ground ice.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Water on Mars is a major issue for the future exploration of this planet. New high-resolution images of the Mars Orbiter Camera (MOC) show unexpected landforms related to liquid water or ground ice [Baker, 2001] like recent gullies [Malin and Edgett, 2000] and small-scale polygons [Seibert and Kargel, 2001]. Nevertheless, many landforms are difficult to interpret because their scale is very different from landforms visible on Viking images. This is especially the case at the surface of landforms observed in the fretted terrains named lobate debris aprons and lineated valley fills. Icy debris aprons of Deuteronilus Mensae are young landforms whose evolution can give an insight to the recent Martian climate like glaciers do on Earth. This study focuses on the complex patterns of lobate debris aprons and lineated valley fills using new MOC images. At the scale of Viking high-resolution pictures, previous authors noticed mainly the presence of lineations due to deformation of ground ice [Squyres, 1978], but also possible aeolian etching [Zimbelman et al., 1989] and depressions possibly related to sublimation of ice [Squyres, 1989]. Based on the geomorphic analysis of high-resolution images and the chronology given by impact craters, the present study shows that the sublimation of ground ice helped by subsurface fractures explains most of the textures observed over lobate debris aprons. In the last section, we show that the future subsurface data of lobate debris aprons could be of great interest for the calibration of instruments and the understanding of the processes of ice cycle between atmosphere and cryosphere on Mars.
2. Geological Context and Topography of Lobate Debris Aprons
 Fretted terrains containing lobate debris aprons are located at the northern highland margin in the regions of Deuteronilus and Protonilus Mensae which are dissected into flat-floored valleys and angular mesas (Figure 1). These mesas are bordered by 1 to 2 km high scarps surrounded by debris aprons up to 800 m thick and 30 km long [e.g., Lucchita, 1984]. Lineated valley fills consist of the same kind of features but confined inside large valleys [Squyres, 1978]. Both landforms have been interpreted as the result of the creep of ground ice especially because of their lobate and convex shape [Squyres, 1978]. This interpretation is supported by the latitudinal distribution of these landforms between 35°–55° in both hemispheres [Squyres and Carr, 1986]. Indeed, these latitudes correspond to the locations where the ground ice is supposed to be stable at few meters deep under the subsurface [Fanale et al., 1986]. The age of these landforms is still poorly constrained but the scarcely cratered surface gives an age in the Late Amazonian period, significantly younger than the surrounding plains dated of the Late Hesperian [Squyres, 1978]. The analysis of recent Mars Orbiter Laser Altimeter (MOLA) topographic profiles of lobate debris aprons confirms their formation by viscous deformation of ice mixed with rocks similar to terrestrial rock glaciers [Mangold and Allemand, 2001]. Indeed, the convex topography of lobate debris aprons fits models using plastic deformation (Figures 1b and 1c), i.e. uniform basal stress at plastic equilibrium, a kind of model used for terrestrial ice caps [Paterson, 1994]. Terrestrial glaciers and rock glaciers usually show transverse ridges, fractures or longitudinal lineations due to the surface expression of viscous internal deformation [e.g., Crown et al., 1992]. Such lineations have been observed over lobate debris aprons at the Viking scale [Squyres, 1989].
3. Geomorphic Analysis of Lobate Debris Aprons Texture at MOC Scale
3.1. Description and Interpretation of MOC Images
 MOC images presented in this study are located in Deuteronilus Mensae and in Protonilus Mensae. Lobate debris aprons are analyzed over their central part (Figure 1d) where the slope is low, less than 2° as deduced from MOLA profiles. We classify the surface of lobate debris aprons in 3 geomorphic units [see also Mangold et al., 2000a]: (1) uniform smooth unit with no or few pits, (2) very rough unit with network of pits and/or small buttes and (3) unit with lineations or residual buttes.
 The smooth material of unit 1 on which we observe no lineations like on Viking images may not correspond to the material of the debris aprons itself but to a mantling. The unexpected smoothness of unit 1 may thus be explained by accumulation of aeolian dust deposits over the previous surface of lobate debris aprons. The observation on several MOC images of the continuity of this unit through aprons front into nearby plains is in favor to such interpretation.
 Many images show a complex surface texture (Figure 2) that corresponds to pitted surface at the ten-meter scale (Figure 2a) or to a dense network of small buttes (Figures 2b and 2c). We interpret unit 2 to be composed by two end-members of the same morphologic process. Unit 2a corresponds to densely pitted surfaces (Figure 2a) dissecting the smooth surface of unit 1. Unit 2b is composed by a lot of small buttes. After the surface was intensively pitted, pits probably become coalescent thus isolating small buttes. The fact that the top of buttes is at the same altitude as the smooth unit 1 (Figures 2a and 2b) is the main evidence of such continuity. Figure 3 shows one of the highest resolution of MOC pictures centered over the surface of a lineated valley fill. Buttes are about 10 to 30 m across with a maximum elevation of about 10 m measured using the incidence angle of the sun above horizon (i = 14°).
 Unit 3 is observed in topographic depressions relative to unit 1. This unit is covered by fresh dust deposits (Figure 2b), sometimes associated with remaining small buttes or by lineations (Figure 4, bottom). This unit likely corresponds to a completely dissected surface. Where lineations are present, the surface may correspond to outcrops of the actual lobate debris apron under surficial deposits. These interpretations lead to a scenario of formation of the observed patterns (Figure 5). This scenario implies that the smooth mantling of unit 1 is dissected into pits and small buttes that progressively collapse forming the unit 3.
 Pit-and-butte textures are also often oriented along a preferential direction (Figures 2b–2d). In Figure 4, the border of unit 1 follows the orientation given by lineations in unit 3, even when these lineations are strongly bent. In this image, lineations are bent probably because these patterns result of the collision of two lobate debris aprons that induces irregular internal deformation with fractures and shear zones. The relation between dissection and underlying deformational features can only be explained if the network of fractures influences the process of dissection. This does not prove that the deformation process was still active at the time of the dissection, but that the existence of this pattern is important in the development of the texture. On the Earth, the flow of glaciers usually creates longitudinal lineations parallel to the flow and orthogonal fractures perpendicular to it [Paterson, 1994]. These fractures are sometimes widened by effects of rain, ice melting and surface run-off making similar rectangular patterns that come from the widening of fractures. Deformational features of the glacier could thus explain both the rectilinear orientations of the pit-and-butte texture observed in Figures 2 and 3 and the curved pattern observed in Figure 4.
3.2. Explanation of the Phenomenon: Sublimation of Ice Initiated by Fractures
 Lobate debris aprons are located in the range of latitudes between 35°–50° where ice is supposed to be stable in the subsurface [Squyres and Carr, 1986]. However, the average annual temperatures likely exceed the frost point of several degrees, so the ground first meters can be desiccated by sublimation. This limit shifts in latitude with variations of temperatures set off by seasonal variations as well as obliquity and eccentricity variations for long term modifications [Mellon and Jakosky, 1993]. According to numerical simulations, at the latitudes considered, down to about 10 m of soils could have been affected by sublimation depending on physical parameters like grain size or porosity [Fanale et al., 1986]. This thickness is roughly consistent with that of the dissected terrain. In such case, dust deposits over debris, if not the uppermost debris, were enriched in ice before losing it by sublimation. The ice enrichment could be possible by pumping deeper ground ice inside shallow layers or by periodic frost deposits inside dust porosity [Clifford, 1993; Mellon and Jakosky, 1993].
 Sublimation is usually a gradual and slow process. In the case of lobate debris aprons, buried fractures could favor the sublimation (Figure 6). Indeed, ice grains sublimate in vapor molecules which cross the porosity of the ground before reaching the atmosphere. In a granular medium, the sublimation rate decreases progressively because water vapor molecules need to cross more thickness of material since the solid grains are not removed. On the contrary, the sublimation is faster in a fractured medium because fractures bring ice grains more quickly in contact with the atmosphere. Moreover, when one ice grain sublimates from a fracture, nearby solid grains become unstable and may be detached. In such removal process, next ice grains are quickly put in contact with the atmosphere and can sublimate more rapidly than on a flat surface. There is therefore a rapid regression of the vertical fracture or small cliff and a subsequent widening of the pits (Figure 6). Such process can explain why the vertical dissection is the dominant process observed on the MOC images. Carr  proposed that the upper surface of buttes is more consolidated to explain their mesa-like morphology. This assumption is possible but not necessary in the process described in Figure 6. The thickness of the dissected materials may correspond to the thickness of dust over the aprons but it may depend also on the depth and size of fractures. On the other hand, sublimation rates of ground ice are usually calculated theoretically using diffusion laws for porous medium [e.g., Fanale et al., 1986], but the process presented in Figure 6 does not allow to calculate any steady state rates because the sublimation does not occur homogeneously due to the fractured medium. Such models would only give minimum thicknesses of desiccated terrains.
3.3. Role of the Wind
 No features indicating active aeolian erosion are observed over debris aprons. Indeed, wind erosion produces yardangs, i.e. small hills elongated in the direction of the wind. No preferential orientations of small buttes of unit 2 have been observed except those related to structural patterns which can display various orientations inconsistent with the regular wind-related orientation of yardangs (Figure 4). Nevertheless, removed solid grains can be raised in the atmosphere by wind during their fall (Figure 6e). No strong erosion threshold needs to be reached by wind as if dust grains were on a flat ground. Thus, all the dust grains removed from fractures or small cliffs will not accumulate in the pits even if dusty surfaces of unit 3 may be explained by the persistence of loose dust. So, wind action can contribute to the dissection process by transporting desiccated dust raised by sublimation but wind is not a major process.
3.4. Role of the Viscous Deformation of Ground Ice
 Very few craters observed at MOC scale are elongated by the deformation. This implies that the viscous deformation of the lobate debris aprons is probably very slow or stopped at the present time [Carr, 2001]. The process of dissection helped by fractures does not need active deformation, but it can be helped by a preexisting network of fractures due to previous deformation phases. The penetration of fractures inside mantling material is possible by either subsisting movements inside the glacier or a propagation of fractures inside upper layers due to the load of the mantle during compaction like it occurs for glacier during the compaction of snow. Geomorphic modifications due to ice creep are thus not needed in the process of dissection described in Figure 6.
 On the other hand, some geomorphic features indicate possible recent deformations due to ice creep. Small buttes display round shapes sometimes exhibiting a central depression that suggests a beginning of collapse (Figure 3). Ten meters of rocks produce a shear stress usually not sufficient for ice to creep at mean annual temperatures of −70°C [e.g., Duval et al., 1983], but higher temperatures may have occurred in a recent past. Fresh flow marks that could correspond to activity of ice creep are also observed over the 20°–30° dipping walls of fretted terrains [Carr, 2001; Milliken et al., 2002]. These flow marks are recent because they apparently cover the smooth unit 1 at the foot of hillslopes. Carr  and Malin and Edgett  noted that they are more developed on poleward-facing slopes. Such dissymmetry could be related to periods of high obliquity like those proposed to explain recent gullies [Costard et al., 2002]. The occurrence of creep of ground ice contemporaneously to the sublimation is thus possible. In this case, recent up-warmings may have induced these ice creep explaining that it mainly affect uppermost layers, but it would have modified conditions of sublimation too.
3.5. Chronology of the Processes Using Degraded Impact Craters
 Most craters over lobate debris aprons are degraded by the same process of dissection which creates the pitted texture at MOC scale (Figure 2). Few craters are fresh without apparent degradations (Figure 7a) and some of these fresh craters can show a beginning of degradation (Figure 7b). Degraded craters have strong modifications in comparison to fresh craters (Figures 7c–7e). The interior of several of them presents a form of an “oyster shell” (Figures 7d and 7e). This shape could result of the material melted during the impact which becomes more resistant. “Ghost craters” are craters in a very late stage of degradation (Figure 7f). Craters are counted and classified over a total surface of 329 km2 corresponding to the lobate debris aprons observed on the 12 MOC images of Figure 1. In order to improve the statistical sampling, craters are summed over the whole surface of the 12 images. Diameters are measured by √2 bin increments ([1–1.4],[1.4–2],[2–2.8] etc.) and craters are classified in fresh, degraded and ghost craters, respectively class A, B and C (Figure 8).
 The incremental plot (Figure 9) is particularly adapted to show turndowns and variations of populations that could be interpreted in terms of obliteration, erosion or deposition [e.g., Hartmann, 1971]. Isochrons are plotted according to new models of Martian cratering (W. K. Hartmann, personal communication, 2002). These isochrons correspond to the density of craters that should be observed on a surface not modified since its formation. In addition to the absolute age, the major fact of such distribution is the shift of the crater population from ideal isochrons. The model takes in account the production function of the Moon which fixes the slope of craters smaller than 1 km. The fresh craters of the Arsia Mons caldera follow this slope with a good accuracy [Hartmann and Neukum, 2001]. A shift from this slope can therefore be interpreted in terms of surface modification. Indeed, the turndown from isochrons displayed by crater populations is characteristic of a continuous and progressive process rather than a unique and quick process at the geological timescale (Figure 10). Assuming the process of resurfacing to be steady state, this means that small crater lifetimes would be roughly proportional to their size, with small craters erased more rapidly than larger ones [Hartmann, 1971; Hartmann and Neukum, 2001]. In Figure 9, fresh craters (class A) are plotted separately from the total number of craters (class A + B + C). The diagram shows that distribution of the total number of craters does not follow isochrons except for craters larger than 175 m. According to Figure 10, the craters larger than 175 m give an age of about 300 My that may correspond to the approximate age of the lobate aprons itself. Moreover, the turndown of the total number of craters of about 300 Ma to less than 10 Ma is thus the signature of a continuous process during the last tens of My, that is possibly still going on.
 In agreement with these conclusions, craters located in the same area often show various degradations like in Figure 2d. If the process was unique with a short duration, craters of the same size and located in the same area should show the same degradation. The histogram of Figure 8 gives also an argument against a short duration. The total number of degraded craters (class B + C) is much higher than the number of fresh craters (Figure 8): 90% of craters are degraded. Comparing fresh craters (class A) to degraded (class B), we observe that the relative proportion of fresh craters increases with the decrease of their size: from 5 versus 37 in the class [44–63 m], it shifts to 12 versus 13 at [22–30 m]. Moreover, “ghost” craters are more frequent at larger sizes of more than 88 m. So, in the diameter range of 88–125 m, or 63–88 m, the Figure 8 shows the presence of fresh, degraded and ghost craters, i.e. craters at different stages of degradation. They can therefore not be explained by a single and recent event of degradation otherwise they would have displayed the same morphology.
 On the other hand, most fresh craters are smaller than 63 m. They indicate that the process of dissection was going on recently. Indeed, the total number of craters shows an age of less than 10 My (last isochron crossed by the total number of craters). Nevertheless, small fresh craters follow an isochron at about 1 My. This is given by two points with a lot of uncertainty, so it must be taken cautiously, but these two points correspond to a significant number of 20 fresh craters. If it is true, this could correspond to the end of the activity of the dissection by sublimation. We thus could conclude from the distribution of craters that (1) they have been degraded during a continuous degradation period over tens of million years and (2) that this period possibly stopped about 1 My ago. This last conclusion is not satisfying because there should be no reason for a process running for such a long time to be inactive just at the time we are looking at. Nevertheless, a long duration does not mean that the process was continuous but that it was not a single episode. Indeed, the repetition of episodes could have the same consequence, for example by quasi-cyclic phases. This last case can not be distinguished from a continuous degradation with that method because its consequence would be exactly similar to what is observed in a continuous process. Such interpretation may be consistent with a process controlled by the sublimation of ice which should vary accordingly to variations of climate related to astronomic cycles. Alternatively, the sublimation rate in the last million years was not sufficient to degrade the youngest craters, but it did not stopped completely.
3.6. Landforms Due to Sublimation on Mars and Other Planets
 In summary, we interpret the texture of lobate debris aprons as the result of the sublimation of ground ice inside dust and surface deposits with a possible contribution of wind and viscous ice creep. On Earth, the result of ground ice removal from the permafrost is called thermokarst. It usually forms lakes and depressions filled by water because it is related to the thaw of ground ice. On Mars, no ice melting is needed to explain these features, so the removal of ground ice by direct sublimation could be named “dry thermokarst” by analogy.
 Landforms due to the sublimation or the melting of ground ice have been detected at the scale of Viking images especially in areas of outflow channels and km large craters [Carr and Schaber, 1977; Lucchita, 1985; Costard, 1987; Costard and Baker, 2001]. However, these landforms are related to ancient episodes of thermokarst which possibly involved liquid water at one stage of the process like on the Earth. At MOC scale, lobate debris aprons of the southern hemisphere, especially in the region East of Hellas, display textures similar to those of Deuteronilus aprons that could be interpreted by same process of sublimation [Pierce and Crown, 2001]. Outside lobate aprons, many images show the dissection of surface layers [Mustard and Cooper, 2000; Mangold et al., 2000b] which are dissected randomly without organization along preferential lineations, except few ones dissected along straight fractures (Figure 11). It is likely that the formation of dissected units over lobate debris aprons and lineated valley fills may have the same origin than dissected units at the 10 meters scale involving sublimation of ice in many locations on the planets.
 The dissected terrains in smooth plains are distributed in regions of latitudes 30° to 60° [Mustard et al., 2001]. These authors assume cycles of frost deposits and desiccation over timescales defined by astronomical parameters like obliquity. The chronology that we deduced from craters counts is consistent with Mustard et al.  from the point of view of the range of the possible cycles, but the absolute ages obtained on the surface of debris aprons are significantly older than on desiccated plains where small craters are almost absent. The surface of lobate aprons may thus be covered by older more compacted dust in which ice could be more resistant to sublimation than young dust with fresh incorporation of interstitial ice. The difference in shape between dissected terrain over debris aprons and in smooth dust deposits over plains could then be the result of the differences in the initiation of the dissection. Indeed, fractures are not necessary to initiate the process in smooth plains due to reduced compaction and cementation. Nevertheless, other processes may occurred on lobate debris aprons to explain why the surrounding plains are less modified by sublimation. Does the ice may have been cold-trapped over lobate debris aprons especially [Carr, 2001]? Some authors also postulate the presence of groove marks of removed ice glaciers [Hamlin et al., 2000].
 Finally, ice sublimation is also involved to explain textures observed at the surface of Callisto and maybe other Galilean satellites [Moore et al., 1999; Greeley et al., 2000]. Indeed, Callisto is strongly cratered at the scale of large craters but the surface shows the lack of small craters, the strong degradation of large craters and the occurrence of chaotic regions with residual buttes and dust deposition. Such landscape is very similar to what is observed on Mars over lobate aprons at a smaller scale. The compared study of this process over different bodies should also give a feedback to the understanding of the process. For example, no winds can raise dust in the atmosphere on Callisto on the contrary to Mars.
4. Lobate Debris Aprons as Potential Targets for Geophysical Exploration
 Future geophysical exploration of Mars will be devoted to the search of water in its liquid or icy phase. New data from Mars Odyssei GRS show that the first meter of polar regions is filled by interstitial ice up to 50% in volume [Boynton et al., 2002]. Such proportion could also be possible for the ice present in the upper layers of the lobate debris aprons. The Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) experiment on board the Mars Express ESA orbiter and the Shallow Radar sounding experiment (SHARAD) planned for 2005 on board the NASA Mars Reconnaissance Orbiter (MRO) would have the possibility to analyze landforms like lobate debris aprons. MARSIS will be mainly operating at a frequency around 2 MHz [Picardi et al., 1999], with possible penetration depths of few hundreds of meters and vertical resolution around a hundred meter in basaltic soils [Heggy et al., 2001]. This resolution is adapted to the thickness of lobate debris aprons and a spatial resolution of 20 km is just sufficient to analyze lobate debris aprons of 20–40 km large. SHARAD will operate with a central sounding frequency around 20 MHz [Beaty et al., 2001] and a vertical resolution of about 10 m adapted to identify heterogeneities in the first layers of aprons and a spatial resolution of 300 m that fits aprons length. In the three next paragraphs we focus (1) on the specific geometry of the lobate debris aprons that would be interesting to solve problems of calibration and interpretation for future radar sounding experiments in the 2–20 MHz band, (2) the internal structure of debris aprons and (3) on problems related to sublimation and water vapor transport inside the debris aprons.
 Debris aprons were initially formed by the progressive accumulation of debris from scarps and ice from atmosphere [Squyres, 1978] and/or mass wasting of ice-rich material of surrounding mesas [Lucchita, 1984; Mangold and Allemand, 2001]. No large troughs or collapse of materials indicate that the dissection observed at MOC scale removed ground ice deeper than the 10 m observed. Assuming ice is still present deeper in the subsurface, the specific geometry of lobate debris aprons is interesting for the calibration of the instrument and the interpretation of data (Figure 12). The viscous deformation of a ice + rock mixture may occur if a sufficient proportion of ice is present, more than 28% of ice in volume according to Mangold et al. , and the distribution of ground ice with lots of fractures is different from surrounding terrains that are likely horizontally layered and undeformed. Consequently, SHARAD backscattered echo from those structures should show brighter on radar echo than their surrounding geological environment because this instrument is very affected by the segregations of ice and the presence of fractures [Beaty et al., 2001]. Lobate debris aprons are thus ice-rich landforms clearly identifiable from other landforms by subsurface radar. For the MARSIS experiment sounding such a structure would be of great interest to evaluate the radar attenuation in the Martian ground ice, as the thickness of the lobate debris can be easily defined from the MOLA topography, and by assuming that the dielectric contrast (due to different content of ice) between the debris apron and the underlying material is sufficient to produce a backscattered echo. We would thus have a realistic estimation of the dielectric constant of the exposed part of the ground ice. This would be of a great help to the inversion of future radar data and to reduce ambiguities about the dielectric description of the Martian soil. There may be very few landforms on Mars with a known subsurface geometry that could lead to such study. For example, lavas are also flow features but the underlying surface corresponds to previous episode of lava deposition. In the same way, the subsurface geometry of sedimentary layers are also not known. Only ice caps could display equivalent geometry but ice caps are ice deposits with dust and not ground ice inside deposits.
 On the other hand, high-resolution radar data such as provided by SHARAD could also be used to understand the fine structure of the rock glacier like basal deformation, ice lenses, and undeformed mantle, like it is done for terrestrial rock glaciers [e.g., Barsch, 1996]. These data would be useful to understand the whole evolution of the lobate debris aprons. For example, lobate debris aprons may have formed slowly at the current temperature until be stopped by a plastic threshold [Squyres, 1978], but transient warmer temperatures could also provided higher strain rates needed for the ground ice to creep. Both hypotheses can explain that lobate debris aprons seem inactive at the present time. A better understanding of the structure of lobate debris aprons could thus give an insight into the climate of the Amazonian period.
 In previous sections, we concluded that the textures over lobate debris aprons are related to the sublimation of interstitial ice helped by fractures. Because fractures inside aprons may produce a vertical flux of water vapor higher than usual porous medium, lobate aprons are interesting features to analyze the freezing and sublimation of water molecules. The study of the ice sublimation by seasonal survey of lobate debris aprons is possible because the dielectric constant may be known from the analysis at long wavelength. Indeed, seasonal observations may show variations in the attenuation in the radar echo due to ice sublimation and vertical water vapor transport across the frozen medium. This few amount of water vapor could increase the dielectric constant of the soil leading thus to an increasing in the signal attenuation and a significant limitation in the penetration depth of the 2 MHz radar wave. Such phenomena could be very important to consider for deep water sounding using 2 MHz sounding radar. Thus seasonal observations would help us to quantify the effect of water vapor transport for a 2 MHz sounder. The MARSIS experiment can not deduced the spatial organization of the process of sublimation over lobate aprons due to their small size but instruments with better spatial resolution would be useful to map spatial variations of sublimation in order to quantify sublimation over the different geomorphic units.
 Finally, lobate debris aprons are interesting for many aspects of the future subsurface sounding by radar. Similar discussion of the utility of the lobate debris as a calibration target for the planed radar orbital sounder could be also valuable for the possible use of Time Domain Electromagnetic Method (TDEM) to uniquely identify water in the Martian sub-surface [Grimm, 2002]. Since TDEM is a diffusive method it is more sensitive to the water vapor vertical transport compared to the low-frequency sounding radar. Hence an aircraft-based TDEM on Mars should be able to monitor the change in the electrical properties of rock ice mixtures due to sublimation processes. Thus coupling data from low-frequency sounding radar and TDEM sounding for the lobate debris would increase significantly our knowledge of the Martian geo-electrical conditions as a function of the water transport through the Martian regolith.
 The analysis of MOC data over lobate debris aprons shows that their surface has been submitted to recent geomorphic processes. We highlight some points in the following:
This study agrees with recent works showing that debris aprons have been mantled and that sublimation of near-surface ice from that dust mantle is a major process of modification of their surface at 10 meters scale [Carr, 2001; Malin and Edgett, 2001].
This study especially shows that this process of sublimation is significantly guided by heterogeneities like subsurface fractures. Numerical models measuring the progressive sublimation of ice from the ground can thus not be used to estimate rates of sublimation.
The gradual modification of craters deduced from their distribution shows that the process of dissection occurred over the last tens of millions years. Variations in many episodes rather than a pure steady state process are likely.
Due to the modification of the surface by deposits and desiccation, no evidences can be obtained about the duration of the setting of lobate debris aprons, except that they likely exist for several hundreds of million years.
Ground ice present in the underlying material is likely still present inside aprons at the present time because it is trapped under ice-rich dust deposits which are only partially dissected by sublimation.
The subsurface sounding of lobate debris aprons would be interesting to calibrate instruments because of the specificity of these landforms: high ice content, well-defined geometry, numerous fractures without horizontal layers.
The temporal and spatial survey of lobate debris aprons could provide information about the process of sublimation into a medium that is different from the theoretical homogeneous porous layer.
 The author acknowledges the use of Mars Orbiter Camera images processed by Malin Space Science Systems that are available at http://www.msss.com/moc_gallery/. The author acknowledges two anonymous reviewers and F. Costard, D. Crown, E. Heggy, J.-P. Peulvast, and T. Pierce for helpful discussions.