Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Planetary Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 05 July 2022

A Retrospective on Mars Polar Ice and Climatefree

A Retrospective on Mars Polar Ice and Climatefree

  • Isaac B. SmithIsaac B. SmithYork University and Planetary Science Institute


The polar regions of Mars contain layered ice deposits that are rich in detail of past periods of accumulation and erosion. These north and south polar layered deposits (NPLD and SPLD, respectively) contain primarily water–ice and ~5% and ~10% dust derived from the atmosphere, respectively. In addition, the SPLD has two known CO2 deposits—one thin unit at the surface and one buried, much thicker unit. Together, they comprise less than 1% of the SPLD volume. Mars also experiences seasonal deposits of CO2 that form in winter and sublimate in spring and early summer. These seasonal caps are visible from Earth and have been studied for centuries.

Zooming in, exposed layers at the PLDs reveal histories of climate change that resulted when orbital parameters such as obliquity, eccentricity, and argument of perihelion changed over tens of thousands to millions of years. Simpler environmental conditions at the NPLD, especially related to seasonal and aeolian processes, make interpreting the history of that polar cap much easier than the SPLD.

The history of Mars polar science is linked by numerous incremental advancements and unexpected discoveries related to the observed geology of both poles, the interpreted and modeled climatic conditions that gave rise to the PLDs, and the atmospheric conditions that modify the surface.


  • Planetary Atmospheres and Oceans
  • Solar System Dynamics and Orbital Structure
  • Planetary Surfaces


The polar regions of Mars are unique and compelling in their science. This is in part because the volatile and lithic materials at the poles are continuously evolving with respect to atmospheric processes in the present climate. Compared to the abundance of rocks on Mars that are billions of years old, having active surface processes and surface evolution means scientists can study what happens on a cold, windy planet in real time and compare that to the geologic and glaciologic record that these processes are creating.

Other than Earth, no planet in the solar system can teach us more about volatile transfer and climate records than Mars, specifically because it has two ice caps that contain thousands of layers in a climate archive. Mars also serves as a very good analogue to the rest of the solar system because it has only dry processes related to H2O and CO2 to create the observable record, similar to the outer moons and dwarf planets. Mars is an excellent place to study polar processes in the solar system because (a) the accessibility and proximity of Mars allow for more data collection than other planets, (b) the wealth and diversity of geologic records provide many clues about the past, and (c) the abundance and intensity of modern activity provide relatively rapid testing of many hypotheses.

Early Investigations of Mars: 1600s–1990s

Ice at the Martian poles was first observed with ground-based telescopes as early at 1666 by Giovanni Domenico Cassini. With improvements of telescope technology, drawings in the 1780s by William Herschel and in the 1860s by William Rutter Dawes contain increasingly high-fidelity depictions of the polar caps and other features on Mars. Improved records start with photographic plates in the late 19th century (Figure 1). Continued observations over the next eight-plus decades demonstrated that Mars has seasonal transport of ice between the poles caused by an orbital tilt of similar magnitude to the Earth (roughly 25°). The seasonal cap moved from north to south, each covering regions as far equator-ward as 50°. Spectral measurements indicated that the seasonal caps were made of CO2, leading researchers to believe that the residual ice—the ice that persisted in polar regions through summer—was also CO2. The composition of the residual ice was debated for decades.

Figure 1. Photographic plates of Mars showing seasonality of albedo that indicates the polar ice grows and retreats due to seasonal frost.

Source: Reproduced with permission from Ted Stryk.

Mariners 6 and 7 were dedicated flyby missions that had superior resolution to all ground-based observations. They also observed the seasonal caps (Figure 2), but it was not until Mariner 9 that new information became available.

Figure 2. Mariner 7 television images showing changing surface ice deposits. The bright white seasonal cap is made of CO2 and comes and goes in each hemisphere similar to the seasons on Earth.

Image credits: NASA/JPL-Caltech.

Some of the most valuable work done before the robotic missions arrived was the seminal paper by Leighton and Murray (1966) that used a simple thermal model to conclude that CO2 was the primary constituent of the Martian atmosphere and that it exchanged between poles throughout a full annual cycle. Water would then be a much smaller component of the atmosphere that exchanged more slowly.

In 1971, Mariner 9 reached Mars, making it the first spacecraft to orbit any planet outside of the Earth–Moon system. Mariner 9 had a high altitude and an orbital inclination of 64.3°, permitting repeated observations of the poles. The images were stunning. For the first time, scientists were able to see the structure of the polar caps (Figure 3), including a giant spiral pattern on the surface, but also “laminated terrain” characterized by “many narrow evenly spaced bands which appear to follow contours” (Murray et al., 1972, p. 335). The laminated terrains are called polar layered deposits (PLDs). Murray et al. (1972) noted that unlike the rest of Mars, the polar terrains had very few craters and were gently sloped, indicative of a young age and recent surface modification. They noted that Mars’ orbit is eccentric, giving rise to colder summers in the north, and suggested that “the laminae may reflect such climatic alternations” (p. 343). This was great insight for being so early in Mars’ exploration.

Figure 3. Mariner 9 television image of the NPLD, the first full image taken in summer at high latitudes in the north. This image showed that the polar terrains were generally smooth, meaning the NPLD had a young surface that had not been bombarded by many craters. In addition, a spiral pattern covered the surface, leading to decades of investigations about their formation.

Image credits: NASA/JPL-Caltech.

The next orbital mission to Mars was Viking 1, but with a lower orbital inclination, Viking 1 did not make observations of the poles. However, with the arrival of the Viking 2 Orbiter mission that had a final orbital inclination of 80.5°, scientists had the opportunity to observe the polar caps on every orbit. This came with improved resolution and coverage over Mariner 9 data, better than 300 m globally and as good as 8 m in targeted observations. One of the more important discoveries from this period was that the PLDs were composed of water ice, not CO2, as was previously considered (Farmer et al., 1976; Kieffer et al., 1976)—meaning that Mars had 1,000 to 100,000 times more water than estimated until then.

Without a subsequent mission, Viking 2 data were poured over for more than two decades between 1976 and 1998, and it was during this period, especially in the mid-1980s, that many critical PLD discoveries and hypotheses were made. Scientists learned that the north and south PLDs (NPLD and SPLD, respectively) were similar in some ways but distinctly different in others. For similarities, the two PLDs both exhibited high albedo polar residual caps (NPRC and SPRC) that persisted throughout their respective summers (Murray et al., 1972). In addition, both PLDs exhibited the laminae or strata that were already being interpreted to contain a climate data archive that corresponded to changes in orbital characteristics (Cutts, 1973a; Cutts & Lewis, 1982; Ward, 1974). Both PLDs even had curvilinear features that were at first interpreted as terraces (Cutts, 1973b; Howard, 1978) but soon recognized (Cutts et al., 1979; Squyres, 1979) and confirmed with photoclinometery measurements as topographic basins, or spiral troughs (Blasius et al., 1982; Howard, Blasius, et al., 1982; Howard, Cutts, et al., 1982).

Differences between the PLDs became more apparent with longer baseline data sets. Radio occultation data from Mariner 9 (Dzurisin & Blasius, 1975; Kliore et al., 1973), which were later confirmed by Viking 2, showed that the NPLD is significantly lower than the SPLD, meaning that they exist in two distinct pressure regimes. At first, this led researchers to believe that the north polar cap would have more stable CO2 (Ward, 1974) because of higher surface pressure, but observations revealed that the NPLD had no permanent or residual CO2 ice (Farmer et al., 1976) and that the SPLD had a residual layer of CO2 that survived the more intense southern summer and lower pressure (Figure 4) (Kieffer et al., 1976). That left a mystery that was mostly unresolved for approximately three decades, and aspects remain unresolved. Another difference, which was not fully known until the early 2000s, is that the SPLD covers a much larger latitudinal band, from the pole down to ~72° S, in contrast to the NPLD’s range of only 90–78° N. Whereas the NPLD’s extent is more obvious from imagery, the expanse of the SPLD was not easily determined because it is covered in dark dust (Figure 4). Based on Viking Infrared Thermal Mapper (IRTM) data, the thermal inertia of the SPLD surface is very low, consistent with a fine-grained surface layer, most likely dust (Paige et al., 1994; Vasavada et al., 2000). This dusty surface blends in with the surrounding terrain, and knowledge of the full SPLD extent had to wait until future missions arrived.

Figure 4. Viking 2 image 407B31 of the bright white SPRC and the dark, surrounding SPLD at LS 341°. The SPRC is composed of CO2 and persists over decades, whereas the NPRC contains no CO2, even with the more favorable lower elevation that should stabilize CO2.

Image credits: NASA/JPL-Caltech.

During this period, atmospheric observations from the Viking 2 Mars Atmospheric Water Detector were used to estimate the water vapor abundance above the north pole (Farmer et al., 1976) and how that affected the planet (Farmer & Doms, 1979). Of note, the 1979 study identified a net transport of water vapor from the southern hemisphere to the north over the course of a year, suggesting that the SPLD was losing ice, while the NPLD was gaining. This was questioned later by Jakosky and Farmer (1982), who observed the vapor gradient strongly favoring transport toward the south and favored a south polar growth at the expense of the north. A larger review of the concurrent knowledge is presented in Jakosky (1985).

Combining acceleration data of the Viking Orbiter spacecraft over the north polar region and estimates of cap volume from areal extent and the poorly resolved thickness allowed teams to calculate the density and therefore ice concentration of the NPLD. A total density of 1 g/cm3 was calculated, implying a purity of 95% and inclusion of ~5% lithic material (Malin, 1986). Because the SPLD extent was still unknown, density measurements had to wait another two decades.

Because the SPLD extent, and thus density, was not fully recognized but the stratified nature was, the polar ice was treated as polar sedimentary material consisting of a mixture of ice and dust and given the name polar layered deposits—a name that was morphology-based rather than material-based. This is the legacy name still in use.

The surfaces of the PLDs have been an important question since their discovery. In the first-ever geologic map of the polar regions, Tanaka and Scott (1987) distinguished between the young, bright residual ice and the surrounding PLDs that contained layers of alternating purity. It was understood that the current surfaces would eventually be included in a buried layer, but active processes kept them fresh and clean from dust in the present epoch. At that point, the bright SPRC (see Figure 4) was established as CO2 ice (Kieffer, 1979), and the bright NPRC (see Figure 3) was known to be water ice (Kieffer et al., 1976). Based on Viking IRTM data, Paige and Ingersoll (1985) concluded that surface reflectivity, also known as albedo, was higher in the south. They then developed a radiative heat balance model and calculated the annual heat balance of the PLDs and found that the SPRC remained much cooler than the NPRC during summer, even with the more intense insolation at perihelion—effectively protecting the SPRC by reflecting away the incoming sunlight. They attributed this to greater atmospheric dust deposition in the north, lowering the albedo and warming the seasonal cap. Although recognized as important decades ago, scientists now know that dust is not the cause of the SPRC/NPRC dichotomy puzzle, but the puzzle is still not fully resolved.

For the portions of the PLDs exposed where there was no residual cap, the story was different. Dust cover prohibited detection of ice visually, and the thickness and origin of the dust (whether a dust lag from sublimation or atmospherically precipitated) were unknown.

Following Paige and Ingersoll (1985) and modeling work by Kieffer (1990) that questioned whether the polar H2O ice was dense and large-grained or dusty and fine-grained, Paige et al. (1994) answered the question by calculating the apparent thermal inertia at the north pole and found the value to be between 600 and 2,000 Jm–2s–1/2K–1, a value consistent with that expected for pure water ice at Martian polar temperatures. This implied that the ice at the surface was dense and large-grained with little admixed dust, a surprising result considering the high albedo and fresh, snow-like appearance of the ices. The north polar cap, then, deposited at full density and had no firn, contrary to processes of terrestrial ice caps that form through densification and metamorphosis of snow into firn and then glacial ice.

The seasonal CO2 caps have been an intriguing feature of Mars for centuries (see Figure 1; see also Slipher [1962] and references as far back as 1781 within), but it was not until better ground-based telescopes were available and favorable viewing geometries in the 1960s that the north polar seasonal cap could be tracked in detail to identify trends in the recession rate (Capen & Capen, 1970). When Mariner 9 arrived, preliminary results confirmed the ground-based observations in the north (Soderblom et al., 1973) but with lower detail because of infrequent observations. South polar seasonal cap retreat was finally studied well with Viking 2 data (James et al., 1979) and followed up with further ground-based observations in the 1986 opposition (James et al., 1990). Observations made by the Hubble Space Telescope in the 1990s were able to further refine the measurements of seasonal cap retreat over multiple Martian years, finding subseason variability (Cantor et al., 1998).

Further investigations of the seasonal cap using Mariner and Viking data employed spectral measurements to constrain the grain size and sintering rates of the CO2 ice (Eluszkiewicz, 1993), dust and H2O impurity contents, and the variability in distribution of the variables (Calvin & Martin, 1994).

The orbital measurements of seasonal cap growth and retreat coincided with the first-ever pressure measurements at the surface made by the Viking landers. Those measurements lasted most of a Mars Year, opening the door for connecting surface measurements to those from orbit (Hess et al., 1979). Hess et al. (1979) confirmed previous estimations from Earth that Mars experienced large swings in surface pressure (Cross, 1971; Woiceshyn, 1974), but a more important result was confirmation that the pressure changes matched perfectly with the growth and retreat of the CO2 seasonal caps, confirming the predictions by Leighton and Murray (1966). This was substantiated by James and North (1982), who were able to apply an energy balance climate model to determine the mass of Mars’ atmosphere and determine the amount of CO2 transported to each pole seasonally. A summary of seasonal cap and CO2 knowledge based on Viking and earlier data can be found in a review by Forget (1998).

Using Viking 2 imagery, the first crater counts of polar ice were done in an attempt to establish quantitative surface ages (Plaut et al., 1988). They found that the PLDs had very few craters (zero craters >300 m in the north) and were much younger, perhaps by billions of years, than the surrounding terrains. Crater counting production functions were established by then (Hartmann, 1981), and they were sufficient to suggest that the SPLD had a surface age of a few 100 million years (Myr). In subsequent work, still using Viking 2 data, Herkenhoff and Plaut (2000) again surveyed craters on both PLDs and revised the ages. The SPLD was closer to 107 years. The NPLD, then, would be two orders of magnitude younger than the SPLD, with a maximum surface age of 105 years, and a site of rapid surface evolution through deposition and erosion. They posited several reasons for this dichotomy but were unable to establish a leading hypothesis. The implications of these results were that both poles had ages consistent with the late Amazonian Period but sat on much older surfaces that were Hesperian in age, leaving a >109 year hiatus between formation of the surrounding terrain and the beginning of deposition of the caps.

Still using invaluable Viking 2 data, some of the activity that kept the north polar surface fresh was determined to be winds (Howard, 2000). Howard identified hundreds of frost streaks on the surface of the PLDs that indicated wind direction and found them to be perpendicular to the orientation of the spiral troughs. Howard, Blasius, et al. (1982) had previously developed a hypothesis for the formation of the spiral troughs that included migration caused by material transport by winds. The winds radiate outwards from the center of the cap, curving because of the Coriolis Force. This new wind orientation mapping bolstered that interpretation (Howard 2000).

The Howard, Blasius, et al. (1982) trough migration hypothesis came on the heels of other hypotheses regarding formation by wind-caused in situ erosion (Cutts, 1973b), construction on terraces (Cutts et al., 1979), and insolation-driven migration (Squyres, 1979) but had an advantage of better explaining the orientations and cross sections of the trough stratigraphy. The 1982 hypothesis was then questioned for decades with alternate hypotheses of fractures from Coriolis-driven glacial surges (Weijermars, 1986; Zeng et al., 2007), “accublation” or ablated ice that was transported to the trough plateaus during glacial flow (Fisher, 1993, 2000), thermal transport within the ice (Pelletier, 2004), and atmospheric oscillations (Ng & Zuber, 2006). The question of spiral trough formation would not be settled until 40 years after their 1973 discovery.

In addition to the spiral troughs, other major features are found at both PLDs: large reentrants called Chasma Australe, Promethei Chasma, and Ultimum Chasma in the south (arrows in Figure 5) and Chasma Boreale in the north (arrow in Figure 6), among other, smaller reentrants. The larger re-entrants have dimensions >500 km long and up to 100 km across and are some of the most striking features when first observing the PLDs (see Figures 5 and 6). Following the work by Cutts (1973b), Howard (1978), and Howard, Blasius, et al. (1982), a consensus emerged that the re-entrants were eroded in place by winds. This was challenged by a hypothesis two decades later that explosive fluvial activity caused by rapid basal melting may have carved the Chasmae (Anguita et al., 2000; Benito et al., 1997), and popular science books suggested that Gemina Lingula, at the southern boundary of Chasma Boreale, was created as an outlet glacier from the Main Lobe, thus forming the Chasma by embayment.

Figure 5. High Resolution Stereo Camera (on Mars Express) image of the south pole and limb. The SPRC is in white, and the SPLD surrounds it, mostly toward the bottom and right. Large re-entrants are visible with arrows: Chasma Australe (red), Promethei Chasma (blue), and Ultimum Chasma (yellow). The margins of the SPLD are difficult to identify because of the reddish dust that covers everything but the very young SPRC.

Image credit: ESA/DLR/FU Berlin.

Figure 6. High Resolution Stereo Camera image of the north pole and limb. The NPRC is in white, and the NPLD is a reddish hue. Black sand comprises the surrounding dune fields. Chasma Boreale (yellow arrow) is the size of the Grand Canyon on Earth. Also visible in this image are water–ice clouds that are driven away from the pole by winds.

Image credit: ESA/DLR/FU Berlin.

In the big picture, three ideas originated or were further developed in the mid-1980s that came to dominate the conversation about the PLDs for decades. The first major idea was that the PLDs were very young compared to the rest of Mars and that they had formed from variations in Mars’ orbit. Those variations would alter insolation, and therefore ice stability, at different latitudes, eventually driving the ice toward high latitudes, where they would be stable in the present orbital configuration. These orbital cycles would create alternating layers of ice and dust at the poles that could be read as a climate archive of recent Mars, namely in the past few hundred million years.

Some of the seminal work was in determining the obliquity (axial tilt with respect to the ecliptic), eccentricity (circularity of Mars’ orbit around the sun), and argument of perihelion (season of Mars at perihelion) and predicting a connection to polar accumulation for the past 10 million years (Cutts & Lewis, 1982). That discussion is ongoing but with more nuance.

The other two major ideas were polar basal melting and glacial flow. Both or either of these ideas, if active on Mars, would have “a profound effect on the evolution of the polar terrains and the long-term climatic cycling of H2O” (Clifford, 1987, p. 9150). Together, these three ideas would play a role in all aspects of PLD evolution, especially the gross and fine-scale morphology, and therefore warranted full consideration with future missions.

Mars Global Surveyor

In 1998, after picking over Viking and Mariner data since the 1970s, scientists would be treated to the first new data from a polar orbit in two decades. This occurred after the 1996 launch and 1998 beginning of the mapping phase of Mars Global Surveyor (MGS). MGS brought with it several instruments that revolutionized the understanding of the planet. It is difficult to overstate the value of the wide- and narrow-angle Mars Orbital Cameras (MOC; Malin et al., 1992), Mars Orbiter Laser Altimeter (MOLA; Smith et al., 2001), and Thermal Emission Spectrometer (TES; Christensen et al., 2001), each of which has contributed to numerous scientific advances with regard to Mars. The latter two data sets have proved useful well beyond their mission duration and have been regularly cited in publications into the 2020s. With a polar orbit, MGS crossed the pole approximately every 2 hours, training these invaluable instruments on the PLDs more than anywhere else.


At the poles, MOLA was used to confirm that the NPLD sat in a basin ~5,000 m below the planetary average, reaching a peak elevation of ~–2,500 m (Zuber et al., 1998). The SPLD, on the contrary, resided on a base more than 1,000 m above the planetary average and reached a peak of ~4,750 m (Smith et al., 2001) (Figure 7). In addition, with the full extent of both PLDs finally known, Fishbaugh and Head (2001) identified an antipodal offset in the center of each cap from the rotational pole and found that the highest elevation of the SPLD was not at the pole. These observations suggested that both caps had experienced asymmetric retreat (Fishbaugh & Head, 2000, 2001).

Figure 7. MOLA colorized topography for the north (left) and south (right) polar regions from Smith et al. (2001). The NPLD resides in a basin that is nearly 5,000 m beneath the planetary average, whereas the SPLD sits in the southern highlands, with a base that is ~1,000 m above the planetary average. This dichotomy in topography influences the stability of volatiles.

Using the topographic data, volumes of the PLDs could finally be measured at 1.14 × 106 km3 for the north and 1.505 × 106 km3 for the south (Smith et al., 2001; Zuber et al., 1998). These volumes, combined with later gravity inversions using the radio science experiment on Mars Reconnaissance Orbiter (MRO), were used to derive the density and dust content of the SPLD at 1220–1271 kg/m3 (Wieczorek, 2008; Zuber et al., 2007), implying upwards of 15% dust content, compared to the lower ~5% dust content estimated for the NPLD (Malin, 1986). Revised estimates for density and dust content in the north would have to wait for radar sounders to arrive on future missions.

MOLA helped determine that the north polar spiral troughs stretched hundreds of kilometers with amplitudes that range from ~400 to 1,000 m, depending on proximity to the cap margins—deeper troughs being farther from the pole. Wavelengths between 20 and 70 km were measured with high precision, and the slopes of the surfaces were finally known to high accuracy (Pathare & Paige, 2005). The northern spiral troughs have asymmetric slopes, with greater values on equator-facing exposures and lower values on pole-facing slopes. Troughs and scarps near the margin of the NPLD have greater slopes, between 15° and 20°, occasionally up to 45°. These slopes were found to be too steep for steady-state glacial flow and supportive of an ablative mechanism that dominates steep exposures. The southern troughs were shorter in length but deeper and broader than their northern counterparts and sometimes had topographic cross sections with raised edges, increasing the already noted asymmetry of the troughs for the south pole (Howard, 2000). The new topographic data, along with MOC narrow-angle imagery, supported a revival of interest in spiral trough formation, and hypotheses related to glacial flow and viscous relaxation related to the troughs (Fisher et al., 2002) and in situ erosion (Kolb & Tanaka, 2001; Rodriguez et al., 2007) were again raised.

The larger re-entrants on both PLDs were again scrutinized, and the new topographic data were used to support the interpretation that massive discharge from subglacial melting first tunneled and then eroded Chasma Boreale (Fishbaugh & Head, 2000, 2002). For both the troughs and the large Chasmae, the creation of those features would remain unsettled until sounding radar arrived some years later.

MOLA Influence on the Topic of Glacial Flow

MOLA also played a major role in the question of glacial flow because for the first time, glacial and ice sheet models could be tested with topographic gradients. The NPLD was generally simpler to interpret, in no small part because the underlying bedrock is not complicated by crater rims and other highland topography, such as in the south (Smith et al., 2001). This meant that the north was the primary subject for glaciological studies. Flow, in various forms, had been invoked to explain the spiral trough morphologies (e.g., Fisher, 1993; Weijermars, 1986). These made testable predictions, especially when flow was balanced by accumulation: a flowing ice cap would require that the layers beneath the spiral troughs bend upwards in response to the lower gravitational stresses at these locations (Fisher, 2000). Following that work, Hvidberg (2003) used topographic data to reinforce the “accublation model” of Fisher but also identified some difficulties with that interpretation, namely that the model broke down at big troughs and that the “troughs divide the cap into dynamically separate units” (p. 367). This again made testable predictions for stratigraphy that waited for testing with radar sounding.

Following those studies, Winebrenner et al. (2008) noted that flow beneath the troughs “required ice to occupy and flow in spaces where troughs currently incise the ice” (morphology that Fisher and Hvidberg predicted) (p. 90). Winebrenner et al. noted that this stratigraphy had not been observed and therefore suggested a separate hypothesis—that the troughs formed very recently, after flow had discontinued for Gemina Lingula. To support this, Winebrenner et al. used MOLA data over the Gemina Lingula region to create “flowlines” that would predict the direction of flow for particles within the ice. Of note, to support the ice sheet flow interpretation, they focused on a portion of the NPLD that is anomalous and ignored the much larger main lobe and even approximately three-fourths of Gemina Lingula. These results, although only applicable in one small region, did make testable predictions of stratigraphy, namely that the dome of ice was incised by spiral troughs—meaning that the eroded layers would line up on either side of a trough, contrary to existing observations made by numerous researchers at many NPLD troughs (Howard, Blasius, et al., 1982 and references within) but consistent with observations of interpretations of troughs at the SPLD (Kolb & Tanaka, 2001).


With the arrival of the MOC narrow-angle camera came high-resolution imagery, as good as 1.5 m per pixel of the surface. This and the new topography data gave many more opportunities to investigate fine-scale features, such as layers and craters. Koutnik et al. (2002), using production functions from Hartmann (1999), updated the estimate of the age of the SPLD surface to be 30–100 million years, an age range that is accepted by the community.

MOC narrow angle was also used to investigate layers on the PLDs at unprecedented resolution, finding layers and sequences that were unresolvable from Viking 2 or Mariner 9. The first MOC observations of polar outcrops provided evidence for a north polar “marker bed” that was thicker and more resistant to erosion than interlayer troughs that eroded more quickly (Figure 8) (Malin & Edgett, 2001).

Figure 8. Images of the spiral troughs in a CTX mosaic and outcrops at the NPLD from MOC Narrow Angle Camera. Layers are contiguous over long distances (Malin & Edgett, 2001). Layers alternate between more and less resistant, possibly indicative of a variable dust to ice ratio. The first “marker bed” was identified (yellow arrows).

Image credit: NASA/MSSS.

Following work with MOC data identified the same marker bed and also supported the earlier conclusion by Howard (2000) that aeolian erosion sculpted portions of both poles (Kolb & Tanaka, 2001). Kolb and Tanaka (2001) found no evidence for glacial flow or trough migration on either PLD, temporarily rebuffing Howard, Blasius, et al. (1982) and Howard (2000). The newly identified marker bed that was laterally extensive over hundreds of kilometers gave rise to the idea that the climate record would be more complete than previously thought. Subsequent work by Fishbaugh and Hvidberg (2006) tracked layers for long distances and identified within the uppermost 500 m upper and lower sequences of layers that were interpreted to store climatic information. Furthermore, they used layer geometry to suggest that “large-scale flow has been so slow, at least during formation of these layers, that mass balance patterns have overprinted its signature in the overall layer structure of at least the upper 500 m of the PLD”—a conclusion that holds without any caveats (pp. 18 of 21).

Soon after MGS arrived, there was a reinvigoration of efforts to demonstrate that the PLDs contained climate records (e.g., Laskar et al., 2002; see the section on “Climate Modeling”). Those efforts made predictions of climatic changes such as ice ages and Milankovitch cycles on Earth. Using the new high-resolution MOC narrow-angle data in combination with MOLA-based global surface roughness maps, Head et al. (2003) identified regions in the mid-latitudes, between 30° and 60° in each hemisphere of Mars, that appear to be losing H2O ice. In response, the PLDs would gain that ice, making the current epoch an interglacial period that followed an “ice age” between 2.1 and 0.4 Ma. They predicted an accumulation of ~1 m global equivalent layer (GEL) (layer thickness if all ice were spread across the planet) at the PLDs.


TES made important discoveries and repeated measurements that answered previous questions and raised new ones.

An important measurement that was made early on by TES was the surface temperature through all seasons and regions. This permitted tracking of the seasonal cap boundary in the south (Kieffer et al., 2000) and the north (Kieffer & Titus, 2001) to greater precision than was observed in previous efforts. These data were also used to track the water ice component of the seasonal cap, a difficult measurement but important for understanding the distribution of water during a Mars year (Bapst et al., 2015; Wagstaff et al., 2008). Wagstaff et al. (2008), using TES data and those from the Thermal Emission Imaging System (THEMIS; Christensen et al., 2004), identified a water ice annulus surrounding the seasonal CO2 in the north; however, Bapst et al. (2015) later found that the water ice was spread in larger concentrations throughout the seasonal cap. The southern seasonal cap is known to have only a water ice annulus.

Kieffer and Titus (2001) also identified a Cold and Bright Anomaly (CABA) at the north pole, near the MGS orbital inclination limit, that appears to remain colder than the surrounding region throughout summer, acting as a cold trap for H2O ice, until it warms in late summer to match the rest of the region. This may be one of the more active portions of the polar plateau. Calvin and Titus (2008) followed this work by measuring albedo across the NPLD throughout a summer. They confirmed the previous CABA results and further identified several regions at high elevations and outlying deposits around the NPLD that defrost in early summer only to re-frost after Ls ~110. Their interpretation included “complex feedback mechanisms” that required cold trapping due to albedo, slope, and elevation effects, in addition to atmospheric activity. The spatial heterogeneity detected by Kieffer and Titus (2001) was followed up by findings of Bapst et al. (2019), who used TES again to show that the NPRC porosity and density were location dependent and that below 0.5 m, the ice was older and denser. Evidence then supported recent accumulation in certain regions and ablation in regions with outcropped layers.

In addition to tracking the southern seasonal cap, Kieffer et al. (2000) made the first spectral measurements of the “Cryptic Terrain,” a portion of the southern seasonal cap that has very low albedo, similar to neighboring regions that have no seasonal ice. They found that the Cryptic Terrain has the same surface temperature as the rest of the seasonal cap and large CO2 crystal sizes (Kieffer et al., 2000; Titus et al., 2001). This observation and interpretation were the first to suggest that the CO2 ice could become fully transparent so that TES would sense the underlying substrate. This led to the “Kieffer model” that explains the Cryptic Terrain “fans.” Fans are granular, geologic material that is deposited onto the icy surface after being transported from below the ice through fractures; it is then released into the air, where it is blown downstream (Kieffer et al., 2006, Figures 9 and 10).

Figure 9. High Resolution Stereo Camera image showing the bright white south polar seasonal cap (yellow arrow) and the darker “Cryptic Terrain” (green arrow) that has albedo and color matching the surrounding terrains. The Cryptic Terrain is likely composed of transparent CO2 ice that permits the dust lag substrate to be visible from orbit. Streaks on the surface (red arrow, Figure 10) are indicative of the Kieffer model, in which the CO2 ice cracks, permitting gas and dust from below to explode as jets (Kieffer et al., 2006).

Image credit: ESA/DLR/FU Berlin.

Figure 10. (Left) THEMIS image showing the bright white south polar seasonal cap streaks from jets that deposit material on the surface. (Right) Enhanced color HiRISE (McEwen et al., 2007) image showing fans at the SPLD during springtime. Fans and streaks are indicative of the Kieffer model, in which the CO2 ice cracks, permitting gas and dust from below to explode as jets (Kieffer et al., 2006).

Image credits: NASA/Arizona State University and NASA/University of Arizona.

Regarding the SPLD, there was still debate about the presence of water ice in the >3–km-thick deposits until TES arrived. Viking 2 had identified stratified material, but it was not confirmed to be composed of H2O. So when TES data, in addition to data from THEMIS, were used to definitively identify H2O ice at the surface of the SPLD for the first time (Titus et al., 2003), it answered long-standing questions. In retrospect, this fits into the larger story about how the PLDs got and kept their names. Without confirmation that the strata were made of water ice, only morphology could accurately characterize them, so a generic name that was created decades prior persists. It also explains why so much contemporaneous effort went into determining whether the SPLD could be composed of CO2 ice instead of H2O (Durham et al., 1999; Nye et al., 2000). Even several years later, other instruments were still being used to identify the constituents of the SPLD (Zuber et al., 2007).

Because TES collected data in so many spectral bands, TES could be made to map out the distribution of dust, water ice, and CO2 ice seasonally, even in polar night, and to estimate the grain size for the ices (Kieffer et al., 2000). TES data were even used to detect CO2 snow for the first time (Titus et al., 2001), raising the question about how much of the surface ice is directly deposited by the atmosphere and how much falls from the sky as ice. This was later constrained by the Mars Climate Sounder (MCS) on MRO at ~15% snowfall annually (Hayne et al., 2014).

It is important to note that TES was used across the entire planet, not just at the poles, and these data were useful in determining numerous aspects of Mars that are relevant to the polar regions. For example, TES seasonal surface temperature data were used to create a global thermal inertia map (Putzig & Mellon, 2007; Putzig et al., 2005). Of course, the polar ice was detected, but there was also evidence of near-surface ice deposits at high and middle latitudes. This jibed well with data from Mars Odyssey’s Neutron Spectrometer that found near-surface in the near subsurface, especially poleward of ±50° (Feldman et al., 2004). This was later supported by MCS in a study that found widespread ground ice at mid- and high latitudes (Piqueux et al., 2019). Furthermore, it was possible to use TES to study water vapor column abundance (Pankine & Tamppari, 2019), the water ice aerosol optical depth, dust distribution (Montabone et al., 2015), and atmospheric temperature (Smith, 2004). Those data sets are still the best available to feed into global circulation models (GCMs) that have been used to consider how Mars’ climate has evolved. Outside of present-day observations of the atmosphere, the polar layered ice deposits are the best test for these models.

A Polar Community

MGS reinvigorated scientists studying Mars polar regions, and in 1999 the first International Conference on Mars Polar Science and Exploration was held (Clifford et al., 2000). That conference had more than 50 attendees and was the first of a series of conferences dedicated to polar science (Becerra et al., 2021; Clifford et al., 2000, 2001, 2005, 2013; Fishbaugh et al., 2008; Smith et al., 2018). Those conferences, reports, and special issues enumerate the advances in science between each conference and the open questions to be answered by subsequent work.

One of the major outcomes of that conference series was to establish a community dedicated to Mars polar science and a forum for participants from throughout the world and of every age to come together to highlight recent discoveries and set priorities for future investigations. These priorities are frequently adopted by the Mars Exploration Program Analysis Group and acknowledged or echoed in the Decadal Survey, a summary of progress and a road map for each decade of exploration. This all began with Clifford et al. (2000), who combined data sets from Mariner 9, Viking, and MGS to put all knowledge of the poles into one document, a status of Mars polar science.

Two Decades of Observations That Converge Toward Consensus

Climate Modeling

Starting with the first identification of layered deposits in the polar regions (Murray et al., 1972), planetary scientists began working on the idea that a record of orbital dynamics forces may be contained in the PLDs (Cutts, 1973a; Cutts & Lewis, 1982; Murray et al., 1973; Sagan et al., 1973; Toon et al., 1980; Ward, 1973, 1974, 1979; Ward et al., 1974). Leading up to that, Brouwer and Van Woerkom (1950) first calculated the accelerations that other planets applied to Mars and how those accelerations periodically affected Mars’ semi-major axis and eccentricity, finding that values of eccentricity ranged from 0.005 to 0.141. Ward (1973) was the first to calculate obliquity going backward through time, eventually finding that it ranged from 14.9° to 35.5° during the past 10 million years (Ward et al., 1974). These obliquity variations did not affect the total sunlight reaching Mars’ surface, but they did determine which pole would favor accumulation or sublimation. Taking into account those variations, Murray et al. (1973) then considered the variation of annual insolation that would reach the surface, eventually forcing transport of ice between the poles and lower latitudes.

In simplest terms, changing insolation from obliquity, eccentricity, and argument of perihelion (which pole is pointing toward the sun at perihelion) would create favorable and unfavorable conditions on Mars to support surface and subsurface H2O ice. When the NPLD received more sunlight, the ice would become unstable and move to other places on the planet, where an equilibrium could be met in that epoch. The same was true for the SPLD and even mid- and low-latitude ice deposits. These periodic changes in insolation could drive accumulation and sublimation cycles that would leave layers to be observed and interpreted.

Similar to the H2O story, these models predicted that dust and CO2 would undergo cycles of similar timescales. For dust, thermal gradients between the poles and the equator would drive wind currents that would bring dust toward the poles. That dust would arrive with ice grains attached (Pollack et al., 1979). For CO2, the poles would be natural locations for deposition when insolation was low. Additional locations for CO2 to be stored include rocks, in which dust would make carbonates, permanently removing the dust from interaction with the atmosphere, or the gas could be absorbed onto basalt or clay that could respond to pressure changes (Davis, 1969; Fanale & Cannon, 1971, 1979). The adsorption of CO2 and that of H2O may compete with each other (Zent & Quinn, 1995).

Unified models that included all of these effects were able to reproduce present-day observations and build up the PLDs (Cutts & Lewis, 1982; Toon et al., 1980). One issue that arose out of the models was an implied young age of the PLDs, fewer than 107 years based on the total thickness of the PLDs. This implied the ice was not present at the poles earlier than that time or the models were wrong and deposition was much slower than modeled.

In the 1990s, new researchers armed with better models and computers were able to pick up the questions from the past and found that Mars experiences chaotic variations in obliquity (Bills, 1990; Laskar & Robutel, 1993; Touma & Wisdom, 1993; Ward & Rudy, 1991). Laskar and Robutel (1993) found that Mars’ obliquity can range from 0° to 60°, even in the past 45 Myr. Recent obliquity values of 60°, which would have obliterated the PLDs, may explain why the PLDs are not as old as Mars itself. Furthermore, Jakosky et al. (1993) tested the differences between the NPLD and the SPLD, finding that the NPLD gains at the expense of the SPLD and vice versa, with the current epoch favoring deposition in the north.

Laskar et al. (2002) used MOC observations to extract a profile of ice-layer radiance in the upper 250 m and compare to modeled north polar insolation dating back more than 1 million years. They found a best-fit correspondence that implied an average accumulation rate of ~0.05 cm/year on the NPLD, a number that has not changed with further investigation except for the error bars. Following that success, Laskar et al. (2004) simulated possible obliquity evolution dating back to 250 Ma. They found that after 20 Ma, the solution diverges and many possible pasts are equally likely; however, during the past 20 Ma, the solutions were self-consistent. This has become the standard reference for insolation through time. Importantly, they found that between 5 Ma and 4 Ma, the average obliquity of Mars decreased by ~10° from ~35° to 25°, the present value (Figure 11). This secular change made conditions for north polar deposition more favorable than in the past. Thus, it is likely that the NPLD are very young, with onset at no more than 5 Ma (Laskar et al., 2004; Levrard et al., 2007).

Figure 11. Model output of the eccentricity and obliquity of Mars dating back to 10 Myr before present (Laskar et al., 2004). The shift from an average obliquity of ~35° to 25° at 5 Ma lowered the insolation at the poles, making them more stable for H2O accumulation.

Following the conclusive work of Laskar et al. (2004) that created insolation curves dating back 20 Myr, several studies were able to develop deposition models for volatiles at the poles. Manning et al. (2006) simulated deposition of CO2 ice at the poles dating back 1 Myr. They found that the 120-kyr periodicity of obliquity minima would yield partial atmospheric collapse, dropping the atmospheric pressure to <1 mbar at times and creating thick CO2 ice deposits at the poles.

In addition, Levrard et al. (2007) used the solutions of Laskar et al. (2004) to insert insolation curves at selected points over the past 10 Myr into a GCM that handled atmospheric dynamics. They were unable to run the GCM for 107 years, but in this way, they were able to simulate accumulation rates at critical points and integrate them through time. They found that the NPLD cannot remain stable with long-term insolation >300 W/m2. This means that prior to the secular drop in average insolation at ~5 Ma, the NPLD would only exist as thin layers that sublimated on the 120-kyr cycle. Then, starting at ~4 Ma, some of the NPLD would survive each high-obliquity excursion, creating the stack of layers observed at the north pole. Much of this ice would have come at the expense of tropical ice, through periods of transient deposition in the mid-latitudes. Levrard et al. (2007) were also able to simulate, for the first time, a synthetic stratigraphic profile that could be compared to observations.

The PLDs were always know to contain layers of variable dust content, even before researchers were sure that ice was present. Some had suggested that lag deposits, after deflation from sublimation of water that was admixed with dust, protected the surface of the PLDs from further vapor diffusion, thermal excursions, and resulting sublimation. After subsequent deposition, these would then explain the variable radiance in layers that Laskar et al. (2002) observed. Now, with results from Levrard et al. (2007) available, lag deposits, with variable ratios of dust to ice, became a major topic of conversation.

All told, between 30 and 50 major layers, namely dust lags that formed at high obliquity excursions, would form. These layers would have wavelengths of ~33 m and 50 m. Deposition rates would range up to 2 mm/year during favorable periods but be negative during unfavorable periods, with an average accumulation rate of ~0.5 mm/year over 4 Myr. Since 2007, the results obtained by Levrard et al. (2007) have been tested extensively, and with few changes, they hold.

It is at this point, around 2007, that data from the Shallow Radar (SHARAD; Seu, Phillips, Alberti, et al., 2007) become available to test these hypotheses. SHARAD quickly becomes an integral part of validation of accumulation models (see the section on “Radar”), and the community begins to coalesce around some solutions.

Greve et al. (2010) used a simpler, one-dimensional stability model that only considered insolation without atmospheric dynamics to generate bulk properties of the NPLD. In general, Greve et al. (2010) found a gross agreement with Levrard et al. (2007) in that the NPLD had experienced four major periods of accumulation with intervening periods of sublimation, but the timing of these periods did not coincide perfectly. These periods of material loss would predict major unconformities that can be tested by radar stratigraphic cross sections and visible observations of outcrops.

Based on the aforementioned radar observations, Hvidberg et al. (2012) created a simple model of accumulation that assumed either constant deposition of water ice or constant deposition of dust following the obliquity history dating back 5 Myr. Results show that in either scenario, a variability in dust to ice concentration would appear in the stratigraphic column. Like Levrard et al. (2007), Hvidberg et al. (2012) found a natural wavelength of ~30 m for dust-rich layers, an average accumulation of 0.55 mm/year, and they were able to produce a dust fraction prediction that matched the distribution of six marker beds.

With the success of Levrard et al. (2007), and considering the size of the undertaking, follow on work was not completed for more than a decade. Emmett et al. (2020) undertook the major task of updating GCM physics to include numerous new routines that had not been developed by 2007, including atmospheric opacity, radiatively active clouds, and dust lag formation with assessment of the volume mixing ratio. The major improvement was in how the model handled dust. Instead of a simple parameterization, Emmett et al. (2020) employed a microphysical coupling of the dust and water cycles and a fully interactive dust lifting and sedimentation scheme. These upgrades allowed for simultaneous quantification of both water ice and dust polar deposition and ablation. In order to save computational time, they focused on the years between 1.7 and 1.0 Ma, when eccentricity was low. This meant they could see how changing obliquity affected the distribution of H2O ice between the poles and other latitudes. In particular, the NPLD grew at the expense of the SPLD and any mid-latitude ice deposits, depending on the recent obliquity history, and rates varied between 0.1 and 1 mm/year, with an average near the 0.5 mm/year that had been reported elsewhere. This result solidifies the idea that obliquity is a major driver of climatic changes and reaffirms the conclusion that the NPLDs are young. Other GCM studies involved tracing of the deuterium to hydrogen ratio (D/H) that would be stored in the poles, making predictions of a stratigraphic sequence that can be examined from the surface (Vos et al., 2019, 2022).

One of the major remaining problems in polar science is the north–south dichotomy. How can the SPLD persist for tens of millions of years during unfavorable conditions? Do dust lags protect the SPLD from sublimation? When did the SPLD form? These questions and others are discussed by Smith et al. (2018), who enumerated 5 major questions and 29 total subquestions that are guiding current studies of polar science.


Radar sounding arrived at Mars via the Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS; Jordan et al., 2009; Picardi et al., 2004) and SHARAD. Immediately, those instruments began providing answers to long-standing questions about the polar regions. It is difficult to overstate the impact of the contributions that radar sounding made to Mars polar science.

Picardi et al. (2005) reported the first subsurface detections of dielectric interfaces at the NPLD and were able to constrain the real (ε‎′) and imaginary parts (ε‎″) of the dielectric permittivity. Based on MOLA topography and an inferred base, they found ε‎′ to be ~3, consistent with that known for pure water ice. In addition, ε‎″, which determines the loss or absorption of radar signals, was found to be quite low, supporting the conclusion that the NPLD is composed of nearly pure water ice. This work was also the first to support an interpretation that the weight of the NPLD did not load the Martian crust sufficiently to push it down, implying either a very thick crust or young age of the NPLD.

Soon after the NPLD was observed for the first time, Plaut et al. (2007) performed detailed mapping of the substrate of the SPLD and estimated ε‎′ to be ~3, similar to what Picardi et al. (2005) found at the NPLD, again implying a bulk volume of nearly pure H2O ice. With many observations, Plaut et al. (2007) were able to create a basal topography from mapping reflections and derive a total volume and thickness of the SPLD at 1.6 ± 0.2 × 106 km3 and 3.7 ± 0.4 km, respectively, revised slightly upward from what Smith et al. (2001) found using only MOLA surface topography. This work also found internal reflections within the SPLD that were assumed to represent boundaries between dusty and dust-free layers. Finally, Plaut et al. (2007) identified bright basal reflections that were brighter than the surface reflection, an uncommon observation (Figure 12).

Figure 12. MARSIS data from observation 2714 across the SPLD. A surface reflection can be traced across the SPLD, and some internal reflections are present. The basal reflection is non-contiguous; however, anomalously bright regions are identified (arrows; e.g., Orosei et al., 2018; Plaut et al., 2007).

Image credit: ASI/ESA/University of Rome.

It is important to note that identifying the bulk properties of the SPLD represented a major step toward understanding the polar environment and climate. It was only 3 years prior that the presence of H2O ice was confirmed in any form at the south pole (Bibring, Langevin, et al., 2004), and MARSIS measurements of the base fed into a determination of the bulk density of the SPLD (Zuber et al., 2007) from radio science data on MRO. Without MARSIS and SHARAD, planetary scientists may still not know the bulk composition of the SPLD—this was a revolution regarding one of the most enigmatic features on the planet.

Work with MARSIS data continued, but another sounding radar arrived in the science orbit in 2006: SHARAD. Similar to MARSIS, the arrival of SHARAD brought entirely new types of investigations to the fore. SHARAD had a much finer vertical resolution. Seu, Phillips, Biccari, et al. (2007) investigated layers as fine as 15 m thick at the SPLD, but this came at the cost of depth of penetration, and SHARAD was unable to detect a basal reflection except in a few locations. With the new reflections came many more detail. Seu, Phillips, Biccari, et al. (2007) interpreted truncated reflections in Promethia Lingula to be indicative of unconformities, or internal boundaries created between episodes of erosion followed by deposition. This result promised to help refine climatic models, but those correlations have proven difficult at the SPLD.

In contrast to the difficulty in forming interpretations at the SPLD, the NPLD was relatively easy to interpret. Phillips et al. (2008) were the first to observe NPLD reflections as fine as 15 m apart, and there were many (Figure 13). SHARAD data were so rich at the NPLD that climatic correlations were made immediately. Phillips et al. (2008) were able to distinguish dozens of reflections that they interpreted to be alternating patterns of dust-rich and dust-poor layers, similar to the climatic scenarios described and predicted in the section on “Climate Modeling.” Furthermore, modeling had predicted four major periods of deposition for the NPLD in the past 4 Myr (Levrard et al., 2007), and there appeared to be four sets of “packets” of layers. It was then easy to suggest ages for these packets, and the search for the “Holy Grail” of Mars climate science—unlocking the climate record stored in the PLDs—began in earnest (Smith et al., 2020).

Figure 13. SHARAD observation 1249501 across the NPLD. Dozens of reflections indicate dielectric contrasts between layers, likely between layers with high and low concentrations of dust.

Phillips et al. (2008) also identified the base of the NPLD in sections, finding at most 100 m of downward deflection caused by the NPLD load, a small number that had to be explained. The outcomes of their geophysical analysis were that the NPLD had to be young and the heat flow of Mars had to be small; otherwise, the downward deflection from the weight of the NPLD would be much greater.

Putzig et al. (2009) followed with a much broader mapping effort using more than 350 SHARAD observations that covered the majority of the NPLD. They found more detail than that found by Phillips et al. (2008), delineating five units of the NPLD, rather than four, including one smaller unit associated with Gemina Lingula. This reconnaissance involved a fair amount of interpolation, but the boundary demarcations permitted volumetric estimates and distributions to be determined for each unit. Of import, this was the first effort to meld the modeling results of Levrard et al. (2007) to radar unit mapping, and Putzig et al. (2009) were the first to suggest specific dates for specific sections observed by SHARAD.

Grima et al. (2009) calculated the SHARAD radar signal loss over the Gemina Lingula portion of the NPLD and found that the NPLD was made of nearly pure (low loss) water ice, at least 95%, leaving 5% for dust and other lithic materials.

Selvans et al. (2010) used the MARSIS techniques developed by Plaut et al. (2007) to separate the NPLD from the lower, and sandier, basal unit (BU) and to detect the base of Planum Boreum (PB) that is composed of both units. PB was found to have a volume of 1.3 ± 0.2 × 106 km3, only slightly larger than the estimate of 1.2 ± 0.2 × 106 km3 by Zuber et al. (1998). Importantly, Selvans et al. (2010) were able to determine the volume of the upper NPLD (7.8 ± 1.2 × 105 km3) and the BU volume at 4.5 ± 1.0 ×105 km3, significantly revised from Byrne and Murray’s (2002) estimate of 4.5 ± 1.0 × 105 km3 for the NPLD.

In 2010, a pair of studies revolutionized how scientists think about features on the NPLD and the sequence of events that created the NPLD. In the first study, Holt et al. (2010) used SHARAD to investigate the stratigraphic sequence of materials surrounding Chasma Boreale, the Grand Canyon–sized re-entrant in the NPLD, and found that after some initial erosive period, much of Chasma Boreale was built up as a constructional feature, meaning that the walls of the canyon got taller as more ice was deposited on the existing NPLD and Gemina Lingula—rather than as an erosional feature, as had been proposed various times previously. They also found a second, but now buried, canyon farther east, demonstrating the complex depositional history of the NPLD.

The second SHARAD study that year found evidence supporting Howard, Blasius, et al.’s (1982) migration hypothesis of the formation of the spiral troughs (Smith & Holt, 2010). The spiral troughs had not formed in place, instead forming at some point in the past and migrating poleward during ~1,000 m of accumulation. This study very quickly eliminated nearly all alternative hypotheses of trough formation related to flowing ice or in situ erosion, leaving only that posited by Howard, Blasius, et al. (1982), which stated the importance of winds in creating and maintaining the spiral troughs of Mars.

In a follow-up study, Smith et al. (2013) demonstrated the complex fluid dynamics that make “katabatic jumps” that repeat in a cyclical pattern to create and then enhance the amplitude of the spiral troughs (Spiga & Smith, 2018). They enumerated 10 trough characteristics that any hypothesis must be able to explain, including the asymmetry of layered terrain on the equator-facing slope and “banded terrain,” or recent deposition on the pole-facing slope. Only wind could create this pattern, and the Coriolis Force drove the winds to create the spiral pattern. Smith and Spiga (2018) found that these winds only occurred during late spring, when the north polar seasonal cap was smallest, implying that trough migration is a periodic process.

Smith and Holt (2015) demonstrated the diversity of the spiral troughs in seven regions and found that the troughs formed in at least two distinct periods. In addition, they tested several scenarios of ice sheet flow proposed earlier and found that none of the predicted stratigraphies relevant to the spiral troughs existing with flow were present at the NPLD or Gemina Lingula. This work followed the detailed analysis and modeling effort of Karlsson et al. (2011), who examined predictions of stratigraphy throughout the column in Gemina Lingula and found no correspondence between predictions and observations. Together, these two studies dealt a major blow to the flow hypotheses of the NPLD.

It is worth noting that the NPLD has been the focus of the various flow hypotheses, even though the SPLD is composed of ~90% ice and has steep slopes that models might predict should flow at detectable rates with current instrumentation. One should also acknowledge that flow predictions are still proposed (e.g., Sori et al. 2016), but each time they are tested, they fail (Fanara et al., 2020).

Smith et al. (2016) identified a cap-wide change in accumulation patterns at ~100 m depth that graded between unconformities and disconformities in low surface slope terrain and a change in inflection of migration trajectory at the spiral troughs and surface undulations. This transition happened simultaneously all over the cap and delineated a change from erosion to rapid accumulation that continues until the current surface. This widespread recent accumulation package had a volume of 80,000 km3 and a maximum thickness of 320 m. A corresponding recent accumulation unit at the SPLD had a volume of 7,000 km3. This thickness and total volume (a GEL of 60 cm) were remarkably consistent with earlier predictions by Head et al. (2003) of 300 m thick and a 100 cm GEL that would transfer to the PLDs at the expense of mid-latitude ice during interglacial periods. Thus, ice ages on Mars were confirmed in the stratigraphic record.

SHARAD mapping was useful in numerous other studies. For the NPLD, one study confirmed the ~30-m spacing of major layers and reflectors in the upper few hundred meters (Christian et al., 2013). Brothers et al. (2015) and then Nerozzi and Holt (2017) updated mapping of the basal contact between the NPLD and the BU, offering major improvements from previous efforts that could only use exposures at the NPLD margin (e.g., Tanaka et al., 2008). Nerozzi and Holt (2017) also mapped out radar reflections in the lowermost section of the NPLD, finding general agreement with the initial emplacement of the NPLD described by Levrard et al. (2007). Nerozzi and Holt (2019) followed up by measuring layer properties in the BU and found evidence for gradations in ice content and alternating units of sand- or water-rich material, possible evidence for climatic transitions prior to 10 Ma.

At the SPLD, Whitten et al. (2017) found evidence for a center of deposition that was geographically limited and accumulated ice more rapidly than other regions. They also found evidence supporting an interpretation by Milkovich and Plaut (2008) that built on work by Byrne and Ivanov (2004) of a deposition sequence of three major units—the bench forming layers, overlying the Promethia Lingula layers, which in turn overlaid an inferred layer sequence that may not have exposures. Whitten and Campbell (2018) followed up this work by developing an “incoherent summing” technique for SHARAD data that enhanced signal to noise. This allowed them to map out numerous other radar facies in the SPLD and identify a “fog” that affected radar transmission and reflection over several regions. This “fog” is not yet understood.

Each of these NPLD and SPLD efforts heavily supplemented mapping of outcrops and exposures, providing significant context to mapping that had been done and inferences about internal structure (see the section on “PLD Layering”).

In order to better understand SHARAD reflections, the reflection coefficient has been used invert for possible layer properties. Nunes and Phillips (2006) were the first to consider how unconformities and layer packets (or finely spaced layers) would appear in a radargram. They created a model that could estimate the strength of a reflection between two contrasting layers and model the response given different radar parameters and processing. Lalich and Holt (2017) continued this work by inverting the measured radar response to determine details about the layer geometry. They found that one large contrasting layer, beneath the resolution of SHARAD, could resemble a set of finely packaged layers when observed; this implied that reflections were not unique identifiers of marker beds (as identified by Malin and Edgett [2001] and the subject of Lalich’s hypothesis), opening the door for more climatic scenarios to create the SHARAD-observed reflections. Lalich et al. (2019) continued this work by studying a reflection identified by Smith et al. (2016) and looked at inverting dust content from the radar reflection strength. Even after these inversions, ambiguity in layer thickness versus dust concentration remains. Courville et al. (2021) developed a forward model to predict radar responses given some input constraints and have largely come to the same conclusions.

New Volatiles: Mid-Latitude Ice and Massive CO2 Ice Deposits

In addition to testing older PLD hypotheses of feature formation and flow, SHARAD has been instrumental in discovering new deposits of volatiles on Mars. Several studies using SHARAD found evidence of widespread, thick H2O ice in the mid-latitudes of Mars at Hellas Basin (Holt et al., 2008), Deuteronilus Mensa (Plaut et al., 2009), Arcadia Planitia (Bramson et al., 2015), and Utopia Planitia (Stuurman et al., 2016). These and numerous follow-on studies that mapped the distribution of mid-latitude ices of both hemispheres were not situated at the poles. Previous work identified ice deposits <0.5 m from the surface in mid-latitudes (Feldman et al., 2004; Piqueux et al., 2019), but these deposits may be transients, with young ages, and could have insufficient quantities of H2O to be useful to human exploration. This made the SHARAD discoveries of thick and widespread H2O ice deposits a major contribution to Mars polar science because the discovery of mid-latitude ice is important for providing context to the climatic history of Mars (e.g., Head et al., 2003; Levrard et al., 2004) and for setting parameters in GCMs that can predict climatic history going backwards (e.g., Emmett et al., 2020; Levrard et al., 2007; Vos et al. (2022).

At the south pole, another volatile was waiting to be discovered by SHARAD, where large vertical sections were found to have extremely low reflectivity (Phillips et al., 2011), contrary to observations elsewhere on the cap. These regions corresponded perfectly with a geologically mapped surface unit that had anomalous sublimation pits not found in the other SPLD units. Using geometry and geophysical arguments, Phillips et al. (2011) calculated the dielectric properties of this unit and found them to be consistent with CO2 ice upwards of 1 km thick. Follow-on work found that the Massive CO2 Ice Deposit (MCID) had three vertically stratified units, separated by bounding layers of H2O ice, indicative of climatic cycles (Bierson et al., 2016) as posited in Manning et al. (2006).

The CO2 ice was studied further by Putzig et al. (2018) following the development of a three-dimensional SHARAD data volume for both NPLD and SPLD (Foss et al., 2016). This new data volume afforded better viewing angles than those provided by MRO and improved clutter mitigation, making interpretation and mapping much easier and faster than mapping in two dimensions. Putzig et al. (2018) found that the volume of CO2 ice, extrapolated to the pole where SHARAD and MOLA data are lacking, exceeded 16,500 km3, consistent with a mass of CO2 greater than that of the current atmosphere (James & North, 1982). Mars, then, had enough CO2 to double the atmospheric pressure, but the ice was trapped in deposits at the south pole.

Alwarda and Smith (2021) followed up by measuring the individual volumes of the units and testing the climatic scenarios that the stratigraphy implied. Instead of supporting the work by Manning et al. (2006, 2019) and partial atmospheric collapse, they found evidence that supported an alternative model that found that the deposition of CO2 was a response to thermal equilibrium in the polar regions and that the bounding layers between CO2 units formed as sublimation lags of H2O ice (Buhler et al., 2020). Buhler et al. (2020), followed by Buhler and Piqueux (2021), used a one-dimensional model and were unable to predict the three-dimensional volumetric distribution of CO2 ice or accurate unit volumes. That explanation came in the form of dry ice glaciers that flowed into the spiral troughs, where the excess thickness was sufficient to withstand long periods of sublimation (Smith et al., 2022). Together, the thermal equilibrium modeling (Buhler & Piqueux, 2021; Buhler et al., 2020) and glacial modeling (Smith et al., 2022) have been able to answer the majority of the questions surrounding the MCID.

Liquid H2O?

Plaut et al. (2007) identified bright reflections at the base of the SPLD using MARSIS data (Figure 12) and ruled out the possibility that these indicated a liquid water layer at the base, but they left the anomalies unexplained. Orosei et al. (2018) and later Lauro et al. (2019) performed inversion calculations that showed that a high dielectric contrast was required between the overlying water ice (ε‎′ ~3.1) and the substrate (ε‎′ >20). They interpreted this to mean that a thin layer of liquid water (ε‎′ ~80) was present. Almost immediately, the community was skeptical. Sori and Bramson (2019) found that the global geothermal flux (<20 mW/m2) was too small to melt ice at the base and that an elevated value of ~70 mW/m2 would be required, even if extremely high concentrations of salts were present. That high flux could only happen with a local heat source. Ojha et al. (2021) followed that, finding that 60 mW/m2 would be possible but that enormous concentrations of salt would be required, on the order of all the salt on Mars. Then, three papers in 2021 made the liquid water interpretations much more difficult to support.

Khuller and Plaut (2021) found that the bright reflection zones were much more abundant than previously reported, with dozens found around the cap. This would require dozens of local heat sources (and much more salt) to maintain liquid water or implied that something else was causing the reflections. In two papers published only weeks apart, Bierson et al. (2021) and Smith et al. (2021) found that Orosei’s interpretation could be remedied if the imaginary part (ε‎″) of the dielectric constant were included in the inversions. The reflection coefficient, which gives rise to the reflection strength, is caused by contrasts in both ε‎′ and ε‎″. Orosei et al. (2018) had effectively locked ε‎″ = 0, meaning that ε‎′ had to make up the difference. If ε‎″ was allowed to vary, then ε‎′ could be smaller. Bierson et al. (2021) and Smith et al. (2021) recognized this missing piece and suggested that materials with high ε‎″ could create the reflections. Smith et al. (2021) took it further by identifying minerals known to be present on Mars, namely smectite clays, and measured their dielectric properties when frozen to cryogenic temperatures. Previously, literature had measured ε‎′ and ε‎″ for smectites at >100 and >3,000, respectively, but the lower temperatures brought those numbers down. Smith et al. (2021) found that when frozen to 230° K, the hydrated smectites had a value of ~15 for both ε‎′ and ε‎″. Finally, Smith et al. (2021) found spectral evidence for smectites at the SPLD using data from MRO’s Compact Reconnaissance Imaging Spectrometer for Mars (CRISM; Murchie et al., 2007) instrument, putting the mineral immediately adjacent to the reflections. The debate continues. Mattei et al. (2022) performed additional measurements of smectites at low temperature and raised questions about the ability of smectites to make the observed signals, undercutting the arguments of Smith et al. (2021). There is no consensus on what creates these reflections; however, it appears that liquid water is no longer the most plausible material to cause the bright reflections.

Residual Cap Monitoring

The residual caps, particularly the SPRC, have been investigated for surface features that belie active processes and stability. Murray et al. (1972) were the first to identify specific features on the SPRC, using Mariner 9 data, and Briggs (1974) suggested that the SPRC would be composed of water ice. Kieffer et al. (1979) used summer temperature data from Viking 2’s Infrared Thermal Mapper (IRTM) to measure a temperature close to the CO2 frost point and was the first to recognize that solid CO2 ice remained on the surface at the south pole throughout a year. This implied that the SPRC must be composed at least in part of CO2. James et al. (1979) agreed that CO2 was more plausible for the composition of the SPRC.

Starting with MOC narrow angle and eventually using data from MRO’s High Resolution Imaging Science Experiment (HiRISE) instrument, the most comprehensive body of work regarding the SPRC has been completed by Thomas et al. (2000, 2005, 2009, 2013, 2016), who teamed up with James et al. (2007). In those articles, Thomas et al. (2000) compared the NPRC with H2O ice to the SPRC that is composed of CO2 and found the textures that are unique to both. Thomas et al. (2005, 2009) developed classification schemes for the numerous units that comprise the SPRC, including thicker, pitted units (“Unit A”) that are older and usually marked by polygonal troughs. Unit A was formerly much more extensive, now persisting mostly as mesas and has moderately steep scarps that border these deposits. The mesas are pockmarked by “Swiss cheese” terrain, or sublimation pits that grow each year (Byrne & Ingersoll, 2003). “Unit B” materials surround the remnants of Unit A and are mostly thinner and smooth-surfaced deposits. Unit B also has layers and sublimation pits, along with “fingerprint terrain” that resembles its namesake. Halos that episodically surround the Swiss cheese pits have been linked to mass balance considerations (Becerra et al., 2015).

The NPRC has been of less interest to the community because it is composed of H2O (Kieffer et al., 1979) and because it changes very little over time. However, spectral measurements from the Observatoire pour la Minéralogie, l’Eau, les Glaces et l’Activité (OMEGA; Bibring, Soufflot, et al., 2004) instrument on Mars Express and from CRISM have been able to watch the NPRC evolve during individual northern summers. Langevin et al. (2005) first identified the changing absorption of H2O bands for the NPRC and interpreted this to mean that the crystal diameter changes from 100 μ‎m to 800 μ‎m over the order of a single month (Ls = 93.3 to Ls = 127.1). This change is expected to be too rapid for ice metamorphism, leaving an interpretation of defrosting of a finer-grained seasonal ice to reveal the larger ice that was below. Appéré et al. (2011) looked earlier in the season with OMEGA to find evidence for a retreat of the seasonal cap that left a thin H2O lag caused in part by aeolian transport and by cold trapping.

Brown et al. (2016) confirmed the net sublimation before Ls = 120 with CRISM but also looked later in the summer to identify a net-positive annual mass balance of the NPRC. Between Ls = 135 and Ls = 164, the grain size of the NPRC decreased, indicative of 70 μ‎m of ice deposition, far more than the 0.6–6 μ‎m they had found in the south (Brown et al., 2014) and enough to compensate for the loss earlier in the season.

The NPRC is highly active with both sublimation and deposition, a cause for the rapid degradation of craters on the NPLD. Work by Herkenhoff and Plaut (2000) examining crater size frequency distributions with Viking 2 data suggested that the NPRC could be as old at 100 ka. That work had two shortcomings. First, it relied on crater production functions established by Hartmann (1981) for the Moon with a multiplier of 4× for Mars—an inaccurate method. Second, the resolution of Viking 2 did not permit identification of all of the NPRC craters. This reduced uncertainty in the resurfacing rate (and age) of the NPRC, but the production function numbers were too approximate to really constrain the young age of the NPRC.

Tanaka (2005) used crater sizes to estimate the age of a lower layered unit and an upper layered unit (LLD and ULD, respectively). He used a smaller multiplier for the crater production function (2 vs. 4) and found that the lower unit could be 3.6 ± 2.5 Myr, consistent with the results of Laskar et al. (2004). The ULD, the bright unit that makes up the NPLD plateaus, had so few craters that the age should be only 8,700 ± 6,200 years. Banks et al. (2010) used data from the Context Camera (Malin et al., 2007) to identify more, smaller craters, finding 130 total, but the age constraints were not improved because the crater production function works best with larger craters and suffers with errors on small craters.

Unexpectedly, nature provided a way to update the problematic small crater production function. Using data from MRO, Daubar et al. (2013) observed new impacts, which led to an update or the modern crater production rate for Mars. This was especially relevant to the NPRC, where all observed craters are small and recent. Landis et al. (2016) were then able to revisit the NPRC craters with a new production function to find the resurfacing rate and surface age. They found that the surface refreshes on a timescale of ~1.5 kyr. A surface refreshes at the rate that erosion removes the craters, so a surface with few and small craters has been recently refreshed. This update made the surface of the NPLD ~70 times younger than Herkenhoff and Plaut (2000) suggested and 4–6 times younger than Tanaka (2005) or Banks et al. (2010) estimated. The NPRC is young and very active.

PLD Layering

The PLD layering that was first discovered with Mariner 9 has been the object of many studies. Starting with the identification of the first marker bed by Malin and Edgett (2001), Fishbaugh and Head (2005) considered layer morphology in the NPLD and the BU. Around this time, detailed mapping with high resolution and large teams became possible. Tanaka et al. (2008) put together the first full stratigraphic sequence of the NPLD. This reference still guides interpretations of the NPLD.

With the arrival of HiRISE, it became possible to see color images at very fine resolution and create high-resolution stratigraphic columns (Fishbaugh, Byrne, et al., 2010). This came with some ambiguity as to what comprises a layer, and the properties of layers were called into question (Fishbaugh, Hvidberg, et al., 2010). Fishbaugh, Hvidberg, et al. (2010) asked if albedo was a good indicator of layering, considering that shading and lag deposits could complicate those measurements. To address that issue, Becerra et al. (2016) developed a protrusion profile method to measure layer topography from HiRISE digital terrain models (DTMs). This had the advantage of coinciding with layer morphology—a better indicator of layer thickness than reflectivity.

Using this new technique, Becerra et al. (2016) were able to tie stratigraphic cross sections from multiple sites of the NPLD, demonstrating that large swaths of the NPLD had the same sequence of deposition, even if the accumulation rates varied by location. Following up on that, Becerra et al. (2017) found patterns of repeating wavelets within sites that linked to astronomically forced deposition as modeled by Hvidberg et al. (2012). Instead of a clear pattern of accumulation following insolation, they found that there must be a nonlinear time–depth relationship for accumulation, with an average value of 0.54 mm/year—very near to previous estimates.

Becerra et al. (2019) then used the Colour and Stereo Surface Imaging System (Thomas et al., 2017) camera on Trace Gas Orbiter to create DTMs of an outcrop of layers at the SPLD margin. Performing the same type of analysis as for the NPLD, they found statistically significant signals of climate forcing in the stratigraphic record and accumulation rates between 0.13 and 0.39 mm/year; however, these rates are related to much older ice than in the NPLD and do not represent values in the present epoch.

Mars Polar Science Going Forward

Of the orbital assets that are collecting data on Mars, none are expected to last into the 2030s. Those missions will continue collecting data as long as the spacecraft and instruments continue functioning properly and as long as funding by their respective agencies continues, but there is a limitation on observations that can be made without new instruments, experiments, and investigations.

Through conference summaries (Becerra et al., 2021; Diniega & Smith, 2020; Smith et al., 2018), reports from analysis groups (Ice and Climate Evolution Science Analysis Group and Next Orbiter Science Analysis Group), reports from workshops (Smith et al. 2020), and white papers to the Decadal Survey (many), the Mars polar community has been able to converge on a number of high-importance measurements that could be made from orbit and from the surface. Many of these measurements, if completed, would advance the state of knowledge greatly. Two classes of measurements are considered—those that improve upon the resolution of current assets and those from instruments that have never observed the poles.

Improving resolution of previous assets to observe the finer scales of the layers at the PLDs would provide much greater detail of the layers themselves and enable testing of many existing, mature hypotheses. This can be done from orbit or from the surface and with cameras or radar sounding. For radar sounding, where SHARAD has the highest vertical resolution of ~10 m, increasing resolution by one order of magnitude to 1 m would improve estimation of the climate record from 20,000 year sampling to 2,000 years. Radar scientists could probe the stratigraphic column in three dimensions and assess the continuity of layers between outcrops at the spiral troughs that are visible at ~1 m resolution in HiRISE. Increasing the vertical radar resolution to 10 cm may require landed or aerial assets with proximity to the surface. With this resolution, scientists may be able to resolve major events in the past ~50–60 kyr of Mars’ recent past. Similar improvements over HiRISE may pay huge dividends.

With those higher resolutions, additional information will be required to interpret the results. New types of measurements, primarily obtained by assets on and near the surface, are required to “ground truth” the active processes present at the PLDs. Active atmospheric processes that need to be measured include drivers of the mass and energy budget of the surface in relationship to sublimation and deposition, isotope fractionation, and horizontal transport by local and global winds. Measuring active processes involved with forming layers is also critical to understand what creates the climate record. Densification, chemical reactions, ice physics, dust exclusion, and vapor diffusion may modify layers deep into the stack. Bettering the knowledge of these present-day processes would enable scientists to understand what goes into forming a layer at the surface and how layers are modified as they become buried—critical knowledge for giving context to the larger climate archive. To obtain these measurements, multiple meteorological stations and cores may be required.

Studying the layers of ice at the Martian poles is very valuable because it would permit scientists to confirm how Mars reached its present state. However, that is not the only benefit to humanity. Mars is one of the terrestrial planets in the solar system and may represent the best example of a class of cold, desert worlds at other stars. The climate archive at the PLDs is much longer than any climate archive on Earth, and therefore it provides an opportunity to improve the collective understanding of climate physics over millions to tens of millions of years. Mars has the additional benefit of not possessing liquid oceans, known biology, or anthropogenic climate changes, making it more pristine and simpler for analysis and interpretation. Through studying the PLDs, scientists can develop knowledge about planetary evolution and share these awe-inspiring discoveries with a public who is eager to understand what makes the universe so amazing.



California Institute of Technology


Deutsches Zentrum für Luft- und Raumfahrt


European Space Agency

FU Berlin

Freie Universität Berlin


Jet Propulsion Laboratory


Malin Space Science Systems


National Aeronautics and Space Administration

Spacecraft (NASA Except Mars Express)

Mariner 6

Flyby July 31, 1969

Mariner 7

Flyby August 5, 1969

Mariner 9

May 30, 1971–October 27, 1972

Viking 1 Orbiter

June 19, 1976–August 7, 1980

Viking 2 Orbiter

August 7, 1976–July 24, 1978

Mars Global Surveyor

March 9, 1999–November 21, 2006

Mars Odyssey

October 24, 2001–present

Mars Express

December 25, 2003–present

Mars Reconnaissance Orbiter

March 10, 2006–present