Volume 124, Issue 11 p. 2852-2862
Research Article
Free Access

High-Resolution Thermal Environment of Recurring Slope Lineae in Palikir Crater, Mars, and Its Implications for Volatiles

N. Schorghofer

Corresponding Author

N. Schorghofer

Planetary Science Institute, Tucson, AZ, USA

Correspondence to: N. Schorghofer,

[email protected]

Search for more papers by this author
J.S. Levy

J.S. Levy

Department of Geology, Colgate University, Hamilton, NY, USA

Search for more papers by this author
T.A. Goudge

T.A. Goudge

Jackson School of Geosciences, University of Texas at Austin, Austin, TX, USA

Search for more papers by this author
First published: 21 October 2019
Citations: 10

Abstract

A thermophysical model for rough terrain is developed that is capable of processing spatial domains of megapixel size. This computational advance makes it possible to characterize thermal environments on Mars at unprecedented scale and at a resolution of 1 m per pixel. The model is applied to Palikir Crater, Mars, where many recurring slope lineae (RSL) are located, often in bedrock alcoves. In areas with RSL, subsurface water ice is not stable, that is, any subsurface ice is lost to the atmosphere in the long term. On large portions of the craters walls, water frost accumulates continuously for up to hundreds of sols each Mars year, but no relation is found between the location of RSL and seasonal water frost accumulation. Some RSL do not have access to even 1 m2 of water frost, at any time of the year. Where water frost is present, it stops accumulating in early southern spring at the latest, long before major RSL activity. Based on the model results, neither CO2 frost, perennial subsurface ice, nor seasonal water frost patches (>1 m2) are connected with RSL activity.

Key Points

  • Thermal model developed for rugged topography on Mars can process megapixel domains
  • Subsurface ice is unstable at locations with Recurring Slope Lineae (RSL)
  • Water frost accumulates seasonally but is unrelated to RSL locations and activity period

Plain Language Summary

The possibility of liquid water on present-day Mars has long been of great interest, in part because water is a universal requirement for life. A feature called recurring slope lineae (RSL), dark narrow streaks that form on slopes primarily in the warm season, has been widely considered evidence for liquid water on Mars. In this study, we model temperature at high spatial resolution, along with the evolution of frost and ice. A highly resolved model can reproduce the stark temperature contrasts between pole-facing and equator-facing slopes. A new model has been developed that is capable of predicting temperatures and shadows in rugged terrain, such as alcoves on crater walls. Model results for Palikir Crater, in the southern midlatitudes of Mars, reveal that nearly all RSL form outside of areas with carbon dioxide frost and subsurface water ice. Water frost accumulates seasonally but is unrelated to RSL locations and activity period.

1 Introduction

On Mars, surface temperature can vary greatly over a small lateral distance due to the influence of small-scale topography and the tenuous atmosphere. Several types of active surface features are located in rough terrain: gullies, recurring slope lineae (RSL), and slope streaks. RSL are narrow, relatively dark-toned, Martian surface features that form on steep slopes (McEwen et al., 2011). They typically appear during the warm season on Mars and, over the past decade, have been suggested as evidence for surficial liquid water or aqueous solutions, but dry mechanisms have also been proposed and are consistent with many of the observed RSL properties (Dundas et al., 2017; McEwen, 2018; Stillman, 2018; Vincendon et al., 2019). A water-related formation process would require a source of H2O, such as subsurface ice or seasonal frost. RSL have been observed from 53°S to 42°N (Stillman, 2018), mostly outside the regions with contiguous shallow perennial subsurface ice. The abundance of equatorial RSL, where shallow ice is unlikely, suggests that RSL are not related to perennial ground ice.

Seasonal CO2 and H2O frost, common in the polar regions, have been identified at isolated locations at low latitudes and midlatitudes (Carrozzo et al., 2009; Piqueux et al., 2016; Schorghofer & Edgett, 2006; Vincendon, Forget, & Mustard, 2010; Vincendon, Mustard, et al., 2010). The accumulation and loss of volatiles are temperature-sensitive processes, so temperature modeling is key to quantitatively assessing the role of volatiles. Here we use three-dimensional surface energy balance calculations with topographic models that consist of millions of pixels to model the surface temperature with a spatial resolution as fine as 1 m per pixel. Model temperatures are then used to determine whether there is any relation of RSL locations with seasonal CO2 frost, subsurface water ice stability, or seasonal water frost.

Our study site is Palikir Crater (42°S, 202°E, 16-km diameter), which has been studied extensively with respect to RSL activity (Levy, 2012; McEwen et al., 2011; Ojha et al., 2014; Stillman & Grimm, 2018). The RSL occur on the west facing but not on the east-facing crater wall. The site has many gully alcoves and present-day gully activity (Dundas et al., 2012).

2 Models and Methods

2.1 DEMs

The input for our thermal model is high-resolution stereo topography of the Martian surface. We used the National Aeronautics and Space Administration Ames Stereo Pipeline (ASP; Shean et al., 2016; Beyer et al., 2018) to create Digital Elevation Models (DEMs) from stereo-pair images collected by the High Resolution Imaging Science Experiment (HiRISE; McEwen et al., 2007) and Context Camera (CTX; Malin et al., 2007) instruments, both onboard the Mars Reconnaissance Orbiter. We produced one CTX DEM at 18 m per pixel from Images D09_030825_1378 and D10_031102_1378 (image resolution ~ 5 m per pixel) and one HiRISE DEM at 1 m per pixel from Images PSP_005943_1380 and ESP_011428_1380 (image resolution ~ 0.25 m per pixel). During processing, we used ASP functionality to tie the CTX point cloud to Mars Orbiter Laser Altimeter (Smith et al., 2001) point shot data, following which we tied the HiRISE point cloud to the tied CTX point cloud, in an effort to improve topographic consistency of the output DEMs and to reduce systematic slope errors. Finally, to provide a smooth topographic surface for our thermal model, we used the ASP hole-filling algorithm to fill all holes larger than ~250 m for HiRISE during DEM production. These and other DEMs have been archived and are available for public download (Goudge, 2019).

2.2 Thermal Model for Megapixel Domains

We have developed a computationally efficient thermal model for rugged topography. It incorporates incidence on a sloped surface, terrain shadowing, subsurface conduction, and, in an approximate way, scattering of short- and long-wavelength irradiance between surface elements. The steepest slopes occur at the finest resolution and horizons from far cast shadows, and hence, a thermal model needs to consider DEMs with large domains at high spatial resolution.

The model incorporates shadowing from near and distant horizons (terrain shadowing), using multigrid acceleration. A multigrid method uses a coarser grid at large distance from the point of interest, which results in dramatically faster numerical processing. The model also incorporates one-dimensional subsurface conduction and is run for at least 5 Mars years at time steps of ~30 min (50 steps per sol). A one-dimensional heat equation solver for the subsurface is run for each pixel in the domain. The model domain is three seasonal thermal skin depths thick and has vertically uniform subsurface properties. The role of absorption, scattering, and emission in the atmosphere is incorporated as in the classic model by Kieffer et al. (1977) but modified to account for the obstructed sky view.

The reemission of infrared light from one surface element to another (self-heating) and scattering of visible light from one surface element to another are incorporated in a simplified way. In this approximation, a reference thermal model is run for a horizontal and unobstructed surface, and this is taken to be the temperature of all land surfaces within field of view. This approach of using a horizontal reference surface has been previously used for planar slopes by Aharonson and Schorghofer (2006) and Schorghofer and Edgett (2006), while here the topography is three-dimensional. An advantage of this approximation technique is that once the horizon elevations have been determined, portions of the domain can be run independently from one another, which leads to nearly perfect scalability on a parallel computer cluster.

Because the subsurface thermal model is semi-implicit and the horizons calculations are multigrid accelerated, it is, for the first time, possible to model domains of the size of megapixels, with diurnally resolved temperatures over multiple Mars years, and still incorporate nonlocal effects, such as horizons. This computational advance makes the present study possible. Technical details of the model are described as part of the User Guide that accompanies the online code archive (Schorghofer, 2019). All model results presented here were obtained on a single multicore CPU (central processing unit).

2.3 Input Parameters for Palikir Crater

Table 1 lists input parameters for the thermal model. The Thermal Emission Spectrometer (TES) onboard Mars Global Surveyor (1997–2006) provided global albedo and surface temperature measurements (Christensen et al., 2001). The TES-derived albedo of the Palikir Crater area is 0.12, but the albedo may vary within the 3-km footprint.

Table 1. Default Input Parameters for Thermal Model at Palikir Crater
Parameters Values
Latitude and longitude 41.6°N 157.7°W
Frost point of CO2 145 K
Frost point of H2O 200 K
Infrared emissivity 0.98
Frost-free albedo 0.12
CO2 frost albedo 0.65
Atmospheric long-wavelength irradiance 4% of noontime insolation
Atmospheric short-wavelength irradiance 1% of solar constant

TES temperatures are used to determine the thermal inertia at a coarse resolution. (Thermal inertia, I, is a combination of near-surface thermal properties that determines the amplitude of the surface temperature variation.) The TES data tool (http://tes.asu.edu/data_tool/) was used to obtain atmospherically corrected surface temperatures for the area that stretches from 43°S to 40°S and 200°E to 204°E, which contains Palikir Crater, but is mainly made up of plains surrounding the crater. The local times covered by these data are limited to 1–3 a.m. and 1–3 p.m. The diurnal maximum of the surface temperature is expected within the latter window of local time. Figure 1 shows TES temperatures in comparison to modeled surface temperature (for a flat unobstructed surface) for various thermal inertias. Only the diurnal maxima and minima of the modeled surface temperature are shown. A thermal inertia of 300 tiu provides a reasonable match to the data. (The thermal inertia unit, tiu, is Jm−2 K−1 s−½.) Ojha et al. (2014) Table S2 lists a nighttime thermal inertia of 199 tiu for Palikir Crater, which is also consistent with the data in Figure 1.

Details are in the caption following the image
Thermal Emission Spectrometer (TES)-derived surface temperature over the Martian year compared with the diurnal maxima and minima of model temperatures for several thermal inertias (in units of Jm−2 K−1 s−½). The TES data are from the geographic area 43–40°S, 200–204°E.

The Thermal Emission Imaging System (THEMIS) instrument (Christensen et al., 2004) onboard Mars Odyssey includes a nine-wavelength infrared imaging system with 100 m per pixel resolution. Fergason et al. (2006) produced THEMIS-derived maps of thermal inertia that we accessed through the JMARS software. The area around Palikir is consistent with 300 tiu. The maps show a higher thermal inertia along the crater rim. However, these thermal inertia values are not corrected for slope effects, so they are not quantitatively reliable on the crater walls. Below we will derive a slope-corrected thermal inertia estimate.

The frost point of water in the atmosphere directly above the ground also plays a role. The daytime frost point temperature is about 200 K on Mars. In the southern hemisphere, which is at higher elevation than the northern hemisphere and farther from the north polar water ice cap, the average is about 196 K (Schorghofer & Aharonson, 2005). At sufficiently low mean annual temperature, subsurface ice becomes stable, because it is in equilibrium with the atmospheric vapor content. This threshold temperature is about 7 K below the (daytime) frost point temperature (Schorghofer, 2008), which places the stability threshold at ~189 K. This is the threshold for water ice to be stable at depths beyond the influence of the seasonal thermal wave. For ice to be stable within the seasonal thermal skin depth, even lower annual mean temperatures, or a higher frost point temperature, are required.

2.4 Model Validation

To validate the thermal model, comparisons are made with two types of orbital data: surface temperatures derived from THEMIS IR images and frost patches visible in HiRISE images. A Mars clock has been implemented to compare instantaneous shadows with imagery acquired at a known Earth time (Allison & McEwan, 2000).

Surface temperatures derived from THEMIS IR (100 m per pixel) were obtained through the thmproc.mars.asu.edu website. Target temperatures were selected as output variable and exported as an ISIS cube. Model calculations were carried out with a CTX-based DEM at a downsampled resolution of 32 m per pixel. (The original CTX DEM has a resolution of 18 m per pixel.)

In Figure 2, the measured surface temperatures are compared with model outputs for the same date and time for two thermal inertias. Incidentally, Figure 2 demonstrates the small-scale heterogeneity of temperature, sometimes called “nonisothermality.” THEMIS Image I67450002 (Ls = 324°, sun elevation 13°) was acquired in the late afternoon (local time 17.7 hr). The western rim of the crater is partially in shadow, while the eastern rim faces the late afternoon sun. The temperature of the plains surrounding the crater approximately matches I = 300 tiu, consistent with the thermal inertia derived from the TES data (Figure 1). In contrast, the comparison of model results with the THEMIS measurements suggests that the crater walls as well as the slopes exterior to the uplifted rim have a higher thermal inertia. For steep surfaces, the local phase angle (the angle between the Sun and the spacecraft, as seen from the planet's surface) is high, so the camera likely sees unresolved shadows, and the eastern crater wall may be warmer than it appears on the THEMIS image.

Details are in the caption following the image
Surface temperature derived from Thermal Emission Imaging System IR images (100 m per pixel) in the leftmost column compared with model results (32 m per pixel) for Palikir Crater (42°S, 202°E). The top row shows temperatures for the late afternoon compared with model results for thermal inertias of 300 and 800 tiu. The bottom row shows temperatures for the early morning before sunrise, and model results for thermal inertias of 300 and 400 tiu. Black pixels in the model results are due to missing Digital Elevation Model pixels. North is up.

A second THEMIS image, I60695002, was acquired before sunrise (Ls = 30°, local time 7.0 hr). The Sun is still below the horizon at this time. The temperature of the plains surrounding the crater is roughly consistent with the measured temperature for either thermal inertia, but the crater rim is warmer and requires a higher thermal inertia than 300 tiu, although no single value of thermal inertia matches the THEMIS temperatures.

These two comparisons suggest that for the plains around the crater and the flat interior of the crater, the thermal inertia is about 300 tiu, whereas the slopes interior and exterior to the crater rim, including the areas where the RSL are located, have a higher thermal inertia. HiRISE image data suggest that the crater walls and rim include bedrock outcrops and colluvial rocky material, while the surrounding plains and the flat areas may be covered by dust.

Seasonal frost patches provide another opportunity to test the thermal model. CO2 frost is seen at the study location on pole-facing slopes in HiRISE images. This provides a sensitive test of minimum temperatures in the thermal model, at high spatial resolution. Figure 3 shows a snapshot from the part of the Mars year when CO2 frost is present. The extent of the observed instantaneous shadows provides a validation of the DEM and the geometric horizons calculations. The color portion of the HiRISE image (Figure 3c) reveals bright patches of frost on pole-facing slopes within alcoves, while the frost is not visible in the grayscale portion of the image. For 300 tiu (Figure 3a), the model predicts more expansive CO2 frost than observed. For 600 tiu (Figure 3b), the observed extent of the frost is reproduced in the model calculations.

Details are in the caption following the image
Comparison of observed and modeled frost distribution at Ls = 130.9° (southern winter). (a,b) Model result for thermal inertias of 300 and 600 tiu at 16 m per pixel. Black indicates instantaneous shadows, white indicates CO2 frost, and the background grayscale is proportional to instantaneous insolation. (c) High Resolution Imaging Science Experiment Image ESP_036561_1380_COLOR from the same location and time (Image credit: National Aeronautics and Space Administration/Jet Propulsion Laboratory/University of Arizona). Frost patches are visible in the colored area. The two parallel orange lines in (a) and (b) approximately delineate the colored portion of the High Resolution Imaging Science Experiment image.

CO2 and H2O frost have been identified at Palikir Crater spectroscopically (Vincendon, 2015). The model-predicted extent of H2O frost is dramatically larger than that of CO2 frost (as will be demonstrated below), so the H2O frost is not thick enough to change the visible albedo, whereas the CO2 frost thickness is expected to reach decimeters. Incidentally, present-day gully activity has been observed on the southwest-facing crater wall of Palikir Crater (Dundas et al., 2012; Vincendon, 2015), where seasonal CO2 frost forms.

Results will be presented for a thermal inertia of 600 tiu, which is a reasonable value within the range of inertias present at this site. The dependence of the results on thermal inertia and other input parameters will also be described.

2.5 RSL survey

RSL were manually mapped using the ~25 cm per pixel HiRISE orthoimage produced during ASP DEM processing (from image pair PSP_005943_1380 and ESP_011428_1380), supplemented by HiRISE image data collected at the time of maximum RSL seasonal extent (ESP_022834_1380). RSL top and bottom points were visually identified (Tebolt et al., 2019). Where ASP correlation failed, no data values were reported and were ignored during data analysis. In total 5,505 unique RSL were mapped.

3 Results and Discussion

Results are presented for the east rim of Palikir Crater based on a HiRISE-derived DEM of this area.

Figure 4a shows the maximum duration of continuous shadow over the Mars year. Some of the pole-facing alcoves toward the north end of the crater are shadowed for hundreds of sols. The Mars year has 669 solar days. No area is perennially shadowed. Also plotted in this figure panel are the (top) locations of RSL from the RSL survey (Tebolt et al., 2019). RSL initiation points are not spatially associated with areas showing long-duration seasonal shadow.

Details are in the caption following the image
Model results at a resolution of 8 m per pixel, with about 1.42 million valid pixels. (a) Maximum duration of uninterrupted shadow over a Mars year. Pole-facing slopes of alcoves on the north side of the crater are seasonally shadowed. Red dots mark the start (top) location of recurring slope lineae. The colored rectangle indicates the region shown in Figure 6. (b) Thermal model results for maximum CO2 frost column abundance. (c) Thermal model results for mean annual temperature (white contours represent 190 K, the approximate threshold value for subsurface ice stability). For (b) and (c), the thermal inertia is 600 tiu. Contours are elevation in vertical intervals of 50 m. North is up and toward the equator.

Figure 4b shows the maximum column abundance of CO2 frost over the Mars year, for a thermal inertia of 600 tiu. The area where seasonal CO2 frost accumulates is similar to the area of seasonal shadow. Only about 7% of RSL start at locations that have seasonal CO2 frost at any time of the Mars year. At lower thermal inertia, the CO2 frost becomes thicker and more expansive (Figures 3a and 3b).

At sufficiently low mean annual temperature, subsurface water ice becomes stable. Figure 4c shows the mean annual surface temperature for I = 600 tiu, which reveals that toward the north end of the crater, some pole-facing portions of the alcoves are below 190 K. The presence of subsurface ice delays CO2 frost formation, because it increases the thermal inertia, as is observed elsewhere on the surface of Mars: The timing of low-latitude CO2 frost patches can only be explained with the high thermal inertia typical for subsurface ice (Vincendon, Mustard, et al., 2010). In areas with seasonal CO2 frost, the thermal inertia may be even higher (and layered), but nevertheless, a reasonable match between observations and model output has been accomplished.

At this site, there is a strong relation between seasonal shadowing, seasonal CO2 frost, and subsurface water ice stability (Figures 4a–4c). RSL, on the other hand, are found nearly exclusively outside of these areas, on the warm, poorly shadowed slopes. Figure 5 shows histograms based on the 5,505 RSL top locations. Only about 2% of RSL start at locations with subsurface ice stability, and the median of the mean annual temperature is 215 K, far above the threshold. When the albedo is changed from 0.12 to 0.20, the fraction increases from 2% to almost 4%. When the thermal inertia is reduced from 600 to 400 tiu, the fraction also increases from 2% to almost 4%.

Details are in the caption following the image
Histograms of model results at 5,505 recurring slope lineae (RSL) top locations. (a) Mean annual temperature, (b) annual peak temperature, (c) longest period of continuous H2O frost accumulation (continuous accumulation occurs everywhere for at least about half a sol, at night), and d) areocentric longitude when continuous water frost accumulation stops. (Only RSL locations with at least one sol of continuous accumulation are included, and therefore, the histogram does not sum up to 100%). These results are based on the same area as shown in Figure 4, for 8 m per pixel model resolution and a thermal inertia of 600 tiu. Histograms change with thermal inertia, which may vary locally. DEM = Digital Elevation Model.

Whereas the amount of CO2 frost can be calculated from the surface energy balance, the amount of H2O frost, which forms from a minor constituent of the atmosphere, is difficult to predict. More easily calculated is the duration of uninterrupted H2O frost accumulation, which is the longest period within a Mars year when temperature remains below the frost point. As the air near the surface is saturated over this entire time, water vapor will have to condense (resublimate) onto the surface. The histogram in Figure 5c shows that water frost continuously accumulates for hundreds of sols at many of the RSL start locations. No area remains below the frost point all year.

Results for a subset of the domain are shown in Figure 6, at a resolution of 1 m per pixel. Although temperatures are calculated only for this smaller domain, horizons were determined using the full extent of the HiRISE DEM in Figure 4. The area shown in Figure 6 also includes one of the RSL study areas described in Stillman and Grimm (2018).

Details are in the caption following the image
A section at the southeast side of Palikir Crater, indicated by a colored rectangle in Figure 4a. (a) A portion of High Resolution Imaging Science Experiment Image ESP_022834_1380. (b–d) Thermal model results for a thermal inertia of 600 tiu and a resolution of 1 m per pixel. (b) Mean annual surface temperature. (c) Longest period of uninterrupted H2O accumulation. (d) Latest time of the Martian year (highest areocentric longitude) with water frost accumulation. The dots indicate recurring slope lineae start (top) locations. Contours are elevation spaced by a vertical distance of 5 m. Some of the tightly spaced contours are due to Digital Elevation Model pixel errors. North is up and toward the equator.

Figure 6b shows the annual mean surface temperature. Subsurface water ice is stable below about 190 K. It was already apparent from Figure 4c that subsurface ice is unstable in this region, and it is unstable even at this higher spatial resolution. These mean temperatures are far from the threshold, so even within the uncertainties of frost point temperature, thermal inertia, and albedo, Figures 5a and 6b establish that subsurface ice is unstable at RSL locations, that is, ice, if present, is lost to the atmosphere. The timescale of subsurface ice loss depends on physical parameters, but at this temperature, it is much shorter than the period of orbital variations (Hudson et al., 2007), so there is no known mechanism that could have emplaced ice sufficiently recently.

Figures 5c and 6c show the number of sols with continuous water vapor saturation. Water frost accumulates at large portions of the crater walls, and at some places, it continuously accumulates for hundreds of sols. At lower thermal inertia (not shown), the area with water frost accumulation is smaller. The starting locations of RSL are also plotted in Figures 6b-6d. There is no apparent relation between water frost accumulation and RSL locations. Many RSL starting position do not have even 1 m2 of frost available, at any time of the Mars year.

Figure 6d shows the areocentric longitude of the last day with continuous H2O frost accumulation. At most locations, this period ends in winter, but at steep pole-facing locations within alcoves, it sometimes reaches into early spring. For 373 m2 or 0.1% of the domain shown, frost accumulates beyond Ls = 180°, and accumulation has stopped everywhere by Ls = 195° (early spring). In a subset of the area shown, Stillman and Grimm (2018) identified major RSL lengthening as early as Ls = 271–285° and as late as 285–303° (and small lengthening even before and after these intervals), that is, major RSL activity takes place in early summer. Hence, the end of accumulation for even the last frost patches precedes major RSL activity by a whole season. At lower thermal inertia, water frost accumulation is found to end even earlier.

The Viking 2 Lander (48°N) observed almost continuous early frost 10–20 μm thick and patchy late frost probably 100–200 μm thick, due to recondensation at local cold traps (Svitek & Murray, 1990). All frost had disappeared by the end of northern winter. If this is any guide for Palikir Crater, the water frost layer is first redistributed rather than lost to the atmosphere, but ultimately, frost does not last in the warm season. The sublimation rate depends critically on temperature; near the melting point, loss rates are on the order of millimeters per hour (Chevrier et al., 2007).

These results do not rule out that smaller patches of water frost (i.e., below the DEM resolution) or water frost under bedrock or boulder overhangs could last into late southern spring, but there is no expectation that this is physically plausible at all RSL locations. The HiRISE DEM is a 2.5-D product, meaning orbital viewing geometries cannot resolve putative overhangs or subpixel roughness. The results shown in Figures 4 and 6 exhibit a strong north-south asymmetry for all quantities plotted, whereas RSL activity occurs at Palikir Crater over a large range of slope aspects.

4 Conclusions

A thermal model of the three-dimensional surface energy balance has been implemented for rough topography that can reproduce observed surface temperatures and CO2 frost patches. This model overcomes a computational barrier that existed up to now and has a wide range of potential applications to Mars science. It efficiently incorporates the effect of local slope and horizons in otherwise one-dimensional thermal models. A limitation to such models is that thermal inertia maps (derived from surface temperature measurements) of equally high resolution would be required as input.

Applying the model to Palikir Crater leads to a number of insights. CO2 frost forms in a few alcoves at pole-facing slopes but is absent from nearly all areas with RSL.

Subsurface water ice is unstable at nearly all RSL locations. Annual mean temperatures are too high for water ice to be stable; any perennial water ice is lost to the atmosphere and, therefore, not expected to be available as a source of RSL activity. Seasonal shadowing, seasonal CO2 frost accumulation, and subsurface ice stability are all spatially correlated, but RSL are located almost exclusively outside these areas.

Water frost is deposited continuously for up to hundreds of sols across Palikir, but at 1 m per pixel resolution, some RSL origination points have no water frost available at any time of the Mars year. Others have water frost available, but accumulation ends in early spring at the latest, long before major RSL lengthening typically occurs.

These model calculations of the thermal environment lead to the conclusion that neither CO2 frost, perennial subsurface ice, nor seasonal water frost patches larger than 1 m2 play a major role in RSL activity. The results leave room for the possibility that RSL activity is related to very high surface temperatures or temperature gradients or to unresolved slope or roughness parameters.

Acknowledgments

We thank Sean Corrigan and Michelle Tebolt for RSL mapping and data processing in support of this project and Josh Bandfield, Thuy Vy Luu, and Mathieu Vincendon for insightful comments and contributions. HiRISE and CTX DEMs are publicly available via the Texas Data Repository (Goudge, 2019). Model code and model documentation are publicly available online at GitHub and Zenodo (Schorghofer, 2019). This material is based upon work supported by the National Aeronautics and Space Administration under Grant NNX17AG70G issued through the Habitable Worlds Program. This research has made use of the USGS Integrated Software for Imagers and Spectrometers (ISIS).