 TIMESCALES FOR CRATER DEGRADATION AND BURIAL IN THE PHOENIX LANDING REGION.  E.Z. Noe Dobrea1,2 and C.R. Stoker2, D. C. Berman1, M. Krčo3, A.F. Davila2,4, C.P. McKay2.  1Planetary Science Institute, Tucson, AZ, 85719 (eldar@psi.edu); 2NASA Ames Research Center, Mountain View, CA; 3Cornell University, Ithaca, NY; 4SETI Institute, Mountain View, CA.  Introduction:  The ice-rich regolith of the northern martian plains is thought to act as a water reservoir. However, much remains to be understood with respect its origin, abundance, and depth. In a recent study [1, 2], we performed a photogeologic investigation of the morphology of small (<1 km) craters near the Phoenix landing site to gain insight into the distribution and depth of ground ice in the region. Particular questions of interest to the study were: 1) What are the geological processes and timescales responsible for crater loss in the region? 2) What are the emplacement mechanisms of the subsurface ice? 3) What is the thickness of the ice-cemented soil layer? Geology: The Phoenix landing site is located at approximately 68°N latitude on the floor of Green Valley, a roughly 50 km wide trough located in the martian northern plains. The basement material of Green Valley belongs to the Vastitas Borealis Unit [3], which is thought to underlie most of the northern plains [4]. Overlying this unit within Green Valley are deposits of the Scandia unit [3], which exhibit regularly spaced boulder piles [5]. Superposed on this is the ejecta blanket of the 600 Myr old Heimdal crater, a 10 km diameter double layer rampart crater [4]. This ejecta blanket has a low rock abundance [6,7], and appears deflated or even removed in places [8]. Most of the ground ice exposed by the Phoenix lander is thought to have been emplaced by vapordeposition, whereas a smaller amount may have originated as the result of ice segregation caused by thin film migration and the formation of ice structures akin to ice lenses [5]. However, analysis of epithermal neutron data [9] suggests that the regional abundance of water ice in the top 1 meter of regolith at these latitudes can be as high as 60% wg., which exceeds the expected pore-filling capacity of the soil and is inconsistent with a formation solely by vapor diffusion. Procedure:  We used HiRISE and CTX images over a 4000 km2 area containing the Phoenix landing ellipse to identify and classify craters on the basis of size, degree of degradation, and presence of ejecta. Degree of degradation was defined by the presence of a bowl, a rim, ejecta blocks, radial and concentric fractures, and modification of the interior by the formation of patterns or rock piles.  Ejecta blocks were identified by an increase in spatial density of boulders with proximity to the crater. Boulders were arranged in an arc around the crater's edge or in straight lines radiating away from the crater. Boulders found around craters were not considered ejecta if their spatial density relative to that of boulders in the surrounding plains did not increase with proximity to the crater.  Timing. The present-day cratering rate on Mars for small impacts has been modeled [10,11], and subsequently constrained by observations with the Mars Orbiter Camera [12] and HiRISE [13]. For any given crater size, crater morphology can be used to estimate its degree of modification relative to craters of comparable size.  Given a model size-frequency distribution, it is possible to derive a relative chronology for craters at different stages of degradation, and from that establish a rough estimate on the timing of geological processes in this region. We considered the timescales for ejecta burial and for loss of relief. Results and Discussion:  Rate of Ejecta burial - Ejecta blocks were typically not detected for craters smaller than 200 m nor for the most degraded (i.e. older craters). The absence of boulders in larger, degraded craters is interpreted as loss due to burial, although the same cannot be assumed in the case of smaller craters. We estimated the timescale of ejecta burial by identifying all craters larger than 250 m that still exhibit discernible ejecta and plotting their size-frequency distribution (SFD) against the SFD in [11] - Fig. 1).  We selected 250 m as the minimum size because it falls within the ejecta transition range and it is consistent with the binning adopted by [11] in their model.   The size-frequency distribution for all craters larger than 500 m is consistent with a surface age of approximately 3 Gy.  A rolloff towards small craters indicates progressive loss of small craters, which is typical of ice-rich surfaces.  We also note a difference in the SFD of craters greater than 250 m and of those that still have ejecta blocks, which translates to a difference in age of approximately ~ 1 Gy.  These findings suggest very low net depositional rates, with ejecta being buried over time scales of ~1 Gy. Rate of Loss of relief -We also estimated the rate of crater degradation by counting all craters whose bowls exhibited visually discernible relief in HiRISE images. The relatively flat appearance of the resulting crater SFD suggests that an equilibrium state exists (Fig. 1). Loss of relief occurs faster for smaller craters (<<200 m), which are being removed in timescales of thousands of years. Craters in the 200 m range loose discernible bowl relief over time scales of a few million years. Given the generally low relief of terrain in this region (sharp shadows are absent), visual determination of bowl relief was subjective. Slopes were typically shallow in small craters so shadows could not be used to estimate crater depths, and the presence of varying degrees of frost in most images made the use of photoclinometry impractical. With these limitations in mind we erred conservatively by possibly overinterpreting the presence of the bowl, which arguably resulted in a potential overestimate for the timescales for loss of relief.    Figure 1.  Measured SFD compared to model SFD [10] for (Top) craters exhibiting ejecta blocks (red dots) and for all craters (black dots) and (Bottom) craters exhibiting evidence for a bowl.  The observation of the geologically short timescale for crater survival in this region is intriguing. Small craters were observed to be completely filled with frost in winter images. In springtime and summer images, they still contained a lag of material on their floors. This deposit was likely a combination of ice and dust, where the dust may have acted as a nucleation site for ice in the wintertime, resulting in its precipitation and subsequent retention in the crater's bowl. Alternatively, dust and frost may have blown throughout the winter into the bowls of small craters, from which it is difficult to remove. Mid-size and larger craters (>200 m) also lose relief and volume over time, although the process responsible for this may be different. In particular, we note the development of multiple concentric fractures in the craters' interiors. One potential process for loss of relief is the same as that for small craters: seasonal infilling. However, seasonal lag deposits inside larger craters were not apparent in the images. Additionally, we observed ejecta boulders within the bowl area for many craters exhibiting no discernible relief. It is difficult to invoke a mechanism whereby the bowl gets filled in without burying the boulders. We posit instead that craters larger than approximately 200 meters may be losing relief due to viscous relaxation. This process would be consistent with the formation of concentric fracture systems within the craters' bowls and with the presence of ejecta boulders within craters that exhibit no apparent relief.  Preliminary models testing this hypothesis suggest that 200 m craters in this region can in fact relax over the time scales proposed here [14]. A particularly startling discovery is the fact that most craters smaller than 200-350 meters also lack ejecta blocks, suggesting that there is an incompetent layer of poorly consolidated fine-grain material approximately 40-70 meters deep. This incompetent layer extends beyond the extended ejecta blanket of Heimdall crater, and likely predates the impact. This finding is consistent with the shallow reflector identified in SHARAD data of the region [15]. Based on the geographic location of this unit and the relatively rapid relaxation of craters greater than 200 m in diameter, we suggest that this incompetent layer consists dominantly of ice cemented soil. This finding may imply that the high abundance of water ice extends to greater depths than those that can be probed by Mars Odyssey. How such relatively high abundances of water ice may have been emplaced within a 40-70 meter thick layer is not clear, and additional work is needed to understand the true nature and extent of this layer. References: [1] Noe Dobrea et al. (2016) Icarus, submitted.  [2] Noe Dobrea et al. (2015) Workshop on Issues in Crater Studies ... #9035. [3] Tanaka et al. (2005) Geologic map northern plains Mars.  [4] Heet et al. (2009) JGR 114.  [5] Mellon et al. (2009) JGR 119, 1936-1949.  [6] Golombek, et al (2008), JGR 113.  [7] Bonfiglio, et al. (2011) J. Spcrft Rckts 48, 784-797.  [8] Arvidson et al. (2008) JGR 113, E00A03.  [9] Maurice et al. (2011) JGR 116.  [10] Neukum,  In: Chronol and Evol. of Mars, 87-104. [11] Hartmann (2005) Icarus 174: 294-320.  [12] Malin et al. (2006) Science 314, 1573.  [13] Daubar et al. (2013) Icarus 225, 506-516.  [14] Dombard and Noe Dobrea (2016) LPS XLVII, ibid. [15] Putzig et al. (2014) JGR 119. 
