Detecting Bedform Migration on Mars: A Review of Current Results and Plans for Sub-Pixel Detection Techniques (Invited)

Mars is a planet rich in aeolian bedforms, with dunes, ripples, and ergs seen in orbital and surface images. However, the low density of the Martian atmosphere results in threshold friction wind speeds required for saltation that are about an order of magnitude greater than on Earth. Measurements by landers and predictions from global circulation models indicate that winds above threshold occur about 2-3 orders of magnitude less frequently on Mars than in typical deserts on Earth. Prior to HiRISE, no bedform migration was detected from orbit in more than 3 decades of observation. However, changes in the patterns of downslope streaks on the slipfaces of dunes in Rabe crater were interpreted as grain flow associated with dune migration at rates of 1-2 cm/year [1]. The shrinkage and disappearance of ice-cored dome dunes in the North Polar Erg [2] indicated an active saltation-induced process of sand transport and probably abrasion of ice-cemented sand, perhaps assisted by sublimation of interstitial ice, or a combination of the two. However, most other areas of the planet appear static. We recently examined 26 sets of HiRISE images acquired under nearly identical lighting conditions and separated in time by one Mars year or more. Of these, only one image set, in the North Polar Erg, showed changes of dune-surface contact and ripples, with the lee-surface contacts advancing 2-5 m/year. In addition, recent analysis of dunes ripples in Nili Patera show advance rates of ~1.7 m in 4 terrestrial months [3]. It therefore appears that HiRISE, with higher resolution and greater SNR than previous imaging systems, and now with nearly two Mars-years of observations, is capable of detecting some bedform changes. The implication that some bedforms on Mars are migrating provides important insight on the near-surface wind activity and surface physical state. However, the fact that 25 areas examined showed no changes in bedform position, indicates that either no migration is taking place or rates are below HiRISE resolution of about 1 m between images. Because a full spectrum of rates is likely, the ability to detect sub-pixel changes is desired. The COSI-Corr sub-pixel change detection methodology has been successfully applied to terrestrial satellite images to measure seismic fault offsets, ice-flow, landsliding, and dune migration [4]. Since the studies mentioned above, we have begun using this same technique with HiRISE images. Requirements include dejittered images, precise orthorectification using a digital elevation model derived from stereo pairs, and change detection image sets acquired under similar lighting conditions. With this technique, it is theoretically possible to measure bedform migration down to ~3 cm, vastly expanding the data that up until now could only be obtained from visual comparisons. (1) Fenton, L.K. (2006) GRL, 33, doi:10.1029/2006GL027133. (2) Bourke, M. C., et al. (2008), Geomorphology, 94, 247-255. (3) Silvestro, S. et al. (2010), in press at GRL. (4) Avouac, J.P et al. (2006), EPSL, 249, 514-528.