In this report we show evidence of widespread ripple migration over the stoss side of dark barchan dunes in Nili Patera on Mars. The measured average migration of ∼1.7 meters in less than 4 terrestrial months clearly indicates that active sand saltation is occurring in the study area. In addition, we document widespread changes in the dune base-ground surface contact and in the slip face structures, showing that not only the ripples, but the whole dunes are actually migrating in the present-day atmospheric setting. These results provide unequivocal evidence of recent aeolian activity and suggest that other dunes and ripples on Mars may also be active.
 High-resolution images from orbiters and in situ analysis of the Martian surface by rovers clearly show that sand deflation and erosion currently take place on Mars in present-day atmospheric conditions [Bourke et al., 2008; Bridges et al., 2007; M. Chojnacki et al., Orbital observations of contemporary dune activity in Endeavour Crater, Meridiani Planum, Mars, submitted to Journal of Geophysical Research, 2010]. Conversely, direct evidence of bedform migration are scarce, and limited to small advancement of sand ripples (2 cm in 5 Martian sols) detected by Spirit in Gusev crater [Sullivan et al., 2008]. Also previous studies conducted with lower resolution orbital images have failed to detect evidence of aeolian bedform migrations [Zimbelman, 2000; Malin and Edgett, 2001; Schatz et al., 2006]. This apparent lack of movement has led some authors to propose that Martian aeolian bedforms are relict features formed in the past when atmospheric densities or wind speeds were higher [Merrison et al., 2007] leaving unresolved the question of the present stability of the Martian dunes in the present-day wind regime [Breed et al., 1979].
 Wind ripples represent the smallest features in the hierarchical system of aeolian bedforms consisting of ripples, dunes and draas [Wilson, 1972; McKee, 1979]. Their movement is related to the processes of saltation and reptation [Anderson, 1987], and their characteristics (wavelength and amplitude) are modulated by the grain size, wind speed, and roughness height [Bagnold, 1941; Sharp, 1963; Pelletier, 2009].
 On Earth, where winds above the threshold for sand movement frequently occur at the surface, patterns of wind ripples can rapidly change in response to a variation in the wind direction, providing an instantaneous indication of the local sand transport [Lancaster, 1995]. Conversely, on present-day Mars, wind with sufficient energy to saltate sand rarely blows at the surface so the transport capacity of the wind is the main factor controlling the sediment state of a dune field. For this reason, wind ripples, rather than sand dunes, are the features that we would expect to see migrate during highly energetic wind events. Because sand should be more easily entrained by the wind than dust [Greeley et al., 1992], the lack of evidence of bedform migration on Mars compared to the high frequencies of dust storms, represents a long-standing paradox [Sullivan et al., 2008]. However, such paucity of evidence of sand movement could be only a matter of image resolution. The High Resolution Imaging Science Experiment (HiRISE) on the Mars Reconnaissance Orbiter (MRO), with a pixel scale up to 25 cm [McEwen et al., 2007], has the potential to show bedform migration and sand movement which, if found, would demonstrate that ergs are active under current atmospheric densities and winds. To test this hypothesis we analyzed two HiRISE images acquired along the edge of a dark erg in Nili Patera (Table S1 of the auxiliary material), the northern caldera of the Syrtis Major shield volcano. The ∼20 km2 region where the two images overlap represents our study area (Figure 1). The Nili Patera erg cover an area of ∼530 km2 and consists of large dunes up to 50 meters in height [Hayward et al., 2006]. These dunes consist of low albedo sand that, like dark sand elsewhere on Mars, is likely basaltic in origin [Bandfield, 2002; Tirsch et al., 2008]. Most of the dunes are barchan (horn to horn distance up to 100 meters) and barchanoid ridges with sharp brinks (Figures 1b–1d). Because these kind of dunes are the morphologies that we would expect to see migrate rather than grow vertically like reversing and star dunes, the Nili Patera dune field represents an ideal site for the detection of signs of recent dune activity.
 Two overlapping HiRISE images (orbits PSP_004339_1890 and PSP_005684_1890) were processed in a cylindrical projection on the Mars IAU spheroid using the United States Geological Survey-ISIS (Integrated Software for Imagers and Spectrometers) software. HiRISE images were acquired on 30 June and 13 October 2007 at Ls = 267.5 and Ls = 330.0 ° representing the state t0 and t1 respectively (Table S1). Because the image's orbital parameters are different, two kinds of distortion could influence our analysis and the calculation of ripple displacement in particular: 1) Geometric distortions: these arise because of the difference in the emission angle (the spacecraft look angle relative to zenith), at 0.4° and 4.4° respectively. However, because such divergence is quite small, parallax distortions are negligible and the images align well one with the other. 2) Radiometric distortions arise from changes in lighting geometry between the two images. The seasonal differences cause a divergence of the solar incidence angle (the angle the sun makes with the surface relative to zenith). This causes variations in the shadows that appear more prominent in winter images, causing albedo and contrast variations between the two images. We normalize these variations using the “cosi” function in ISIS. This program applies a basic photometric normalization by dividing each pixel of the images by the cosine of the incidence angle. In addition we adjusted the images in ArcGIS (brightness and contrast of one image has been adjusted to make it as similar as possible to the other). The study images were acquired at 2:51 PM and 2:12 PM – local Mars time. This means that for both images light comes from the bottom left (the sub solar azimuth is similar, Table S1) and that difference in the direction of the shadows of the study ripples is minimal. To calculate the ripple displacement we assumed a global migration of less than one crest wavelength (migration of more than one crest wavelength was judged as unlikely given the <4 month time spacing and the fact that the rates derived for less the one crest wavelength are already significant). To obtain a reasonable calculation of the displacement we selected the area shown in Figures 3a–3c (see Figure 1c for caption). We chose this zone because: 1) the horn of the selected barchan directly overlies the brighter fractured bedrock. Control points are taken on the bedrock, so the alignment of the overlapping images is ideal in this area. However, an error of ∼1–2 pixels (0.3–0.6 meters) close to the control points has been calculated in ArcGIS and affects our measurement. 2) The selected area (the extended horn of a barchan) is flatter than the rest of the dune, so errors caused by topography are minimized. 3) The crests of the ripples can be followed along their entire length because they are oriented nearly perpendicular to sub-solar azimuth, facilitating the mapping process. In this area the ripple crests have been manually mapped (Figure 3a). The assumption that ripples here moved less than one wavelength implies that we can recognize the same ripple in the two images. In this way, we can correlate the crests in the two images and quantify the minimum displacement along the entire crest length.
 Three kinds of modification are detected on the study dunes: (1) changes in the ripple pattern, (2) changes in the dune edges, and (3) changes in the slip face structures (Figure 1b).
 (1) Examples of changes in the ripple pattern are shown in Figure 2. Figures 2a (left), 2b (left), and 2c (left) represent the state t0 (30 June 2007) while Figures 2a (right), 2b (right), and 2c (right) represent the state t1 (13 October 2007) (Table S1). Ripple crests are outlined in yellow and blue. Such modifications are recognizable by means of: a) the modification of Y junctions at crest terminations (Figure 2a) caused by the migration of defects (ends of crest lines) [Werner, 1995; Werner and Kocurek, 1999]. This process causes the displacement of junctions like in Figure 2a where the junction (Y) is displaced toward the southwest. These kinds of modifications are easily recognizable comparing the two images and represent strong evidence for ripple pattern modification (Figure S1). b) The movement of the ripple crests with respect to the dune brink. We can easily observe how the pattern changes taking the brink of the dune (in pink) as reference and looking at how ripples r1 and r2 (in blue) have moved closer to the brink at t1 (Figure 2b). Part of r1 disappears beyond the brink itself (Figure S2). Changes in the position of Y junctions are also visible but not highlighted in this image. c) The displacement of a ripple with respect to the underlying bedrock (Figure 2c). In this image we highlight the displacement of the ripple r3 (in blue) toward the west. Such movement is detectable by taking the fracture on the bedrock (in white) as reference. In addition, the edge of the dune, outlined in green in Figure 2c, displays significant changes. These modifications in the dune contour will be discussed later.
 In the described examples the ripples moved an average of 1.7 meters toward the WSW in less than 4 Earth months. The direction of the migration is best viewed in the area showed in Figures 3a and 3b (Figure S3). In this flatter area ripple crests (∼4 meters spaced on average) are continuous and can be mapped along their entire length (Figure 3a), providing a clear record of migration toward the southwest. For this reason we chose this zone to calculate crest displacement (see Methods section). In Figure 3b we highlight the non-uniform migration rate of these large ripples with some areas migrating more than 5 meters. Similar situations have been reported from computer simulation and field observations of terrestrial ripple fields [Yizhaq et al., 2004; Anderson and McDonald, 1990] where the crest terminations were observed to migrate faster. Taking into consideration the present Martian atmospheric conditions, and that the measured displacement could represents a minimum (see method section), the computed average migration of ∼1.7 meters in less than 4 terrestrial month is surprisingly high and represents the highest rate of bedforms migration ever measured on Mars.
 (2) Changes in the dune edges are most easily identified where the sediment cover/thickness is low at the dune/bedrock interface (Figure 1b) and are shown in Figures 3c and 3d (Figure S4). These changes appear as consistent modifications in the albedo, presumably caused by the removal of the dark sand that saltates downwind, leaving the brighter underlying bedrock exposed. There is no evidence for significant amounts of dust in this area that could otherwise account for such albedo variations. The presence of these kinds of changes has never been reported on the Martian dark dunes and provides further indication of sand saltation at this site.
 (3) Changes in the slip face structures are common on the study area (Figure 1b) and are shown in Figure 4. They appear as rectilinear streaks (in black in Figure 4a) affecting the dune slip faces (shaded in white in Figure 4a) that change their pattern between t0 and t1. Similar features have been observed elsewhere on Mars [Fenton, 2006]. We interpret these changes as new grainflows events [Hunter, 1977] occurring between t0 and t1 caused by sand grains which move along the stoss side of the dune and avalanche downslope over the slip face. The sand moving along the slip face locally covers a ripple pattern on the steep dune slopes (white arrows in Figure 4b). These features are different from the flow structures observed by other authors on the slopes of the dark dunes [Gardin et al., 2010; Reiss et al., 2010] and from the dust slope streaks described in other region of Mars [Sullivan et al., 2001]. It is unlikely that slope streaks could form over a dust-free surface like the study dunes which are frequently cleaned from dust fallout by saltating sand.
 Changes in the ripple pattern are widespread in the study area suggesting that the wind event(s) causing ripple movement was strong enough to cause a regional migration of the ripples. Sand movement on the Martian surface has been proposed to be limited by the pervasive induration of the regolith and the low frequency of wind events having sufficient energy to saltate sand [Sullivan et al., 2001]. The migration of the large ripples over the Nili dunes suggests that, like the El Dorado ripples imaged by Spirit, they are not crusted or indurated by dust. This implies that saltation events causing ripple migration on the study dunes are frequent enough to prevent the formation of a thick stabilizing crust. The lack of prominent erosional features like fractures or cohesive mass wasting on dune slopes further suggests that these dunes are not cemented.
 The changes in the grainflow scars over the study dunes suggest that not only do the ripples migrate, but that also the whole dunes are migrating downwind. The position of these streaks (always on the SW dune slopes, the slip face) suggests that they are created by the same easterly winds causing the migration of the ripples.
 This study shows the first evidence of widespread ripple migration detected from orbital data that, together with the changes in the dune edges visible at the dune/bedrock interface, and the occurrence of new grainflow events over the dune slip faces, indicates that the study dunes are active in present-day atmospheric conditions. The documented aeolian activity at this site is driven by high shear stress winds from the ENE blowing between the Martian late autumn and winter. Most of the study dunes present signs of recent activity suggesting that the Nili Patera erg is a fully active erg. These dunes are currently being monitored by the HiRISE instrument to better constrain the rate of migration of Martian aeolian bedforms.
 The author would like to thank Gaetano Di Achille and Randy Kirk for useful advices and comments. This research has been supported by the Agenzia Spaziale Italiana.