Geologic characteristics of relatively high thermal inertia intracrater deposits in southwestern Margaritifer Terra, Mars



[1] Twenty-one craters in southwestern Margaritifer Terra exhibit unusually warm interior deposits in nighttime Thermal Emission Imaging System (THEMIS) infrared images. These deposits exhibit nighttime temperatures as high as 223 K and are 5–18° warmer than the surrounding plain. Thermal inertia values, derived from Thermal Emission Spectrometer (TES) data, are greater for the deposits than the plains, with maximum values between ∼455 and 675 J m−2 K−1 s−1/2. Analysis of THEMIS thermal inertia data having nearly an order of magnitude better spatial resolution shows that the deposits can have a thermal inertia as high as ∼1060 J m−2 K−1 s−1/2. Albedo and dust cover index values suggest that both the deposits and the surrounding region are generally dust-free. Compositional analysis with TES and THEMIS data show that though a small number of the deposits may have isolated compositional differences, the majority of the deposits have a composition similar to that of the surrounding plains. The geomorphology of the craters as viewed from Mars Orbiter Camera (MOC) images shows that the deposits are coherent units, rather than sand deposits, an observation that is consistent with the relatively high thermal inertia values. If these deposits are coherent rock units, as the results of our study suggest, possible methods of formation include emplacement of primary igneous material or lithification of sediments from surrounding terrain.

1. Introduction

[2] Our investigation of nighttime thermal infrared images recorded by the Thermal Emission Imaging System (THEMIS) onboard the 2001 Mars Odyssey spacecraft reveals ∼20 craters in the region of southwestern Margaritifer Terra with unusually warm floor deposits relative to the surrounding materials. The appearance of these craters varies from the more typical nighttime appearance of craters in which the rims and possibly the ejecta exhibit warm nighttime temperatures while the crater interior is cooler, more similar to the surrounding terrain. Here we define an intracrater ‘deposit’ as a mass of material emplaced by any process subsequent to crater formation. The deposits evaluated in this study differ from classic intracrater deposits [e.g., Christensen, 1983], in that they do not appear to be eolian deposits of loose sand. To understand their composition and origin, we have analyzed the albedo, spectral, thermophysical, topographic, and geomorphic characteristics of these anomalous deposits using remote sensing data from the Mars Odyssey THEMIS, Mars Global Surveyor Thermal Emission Spectrometer (TES), Mars Orbiter Laser Altimeter (MOLA), and Mars Orbiter Camera (MOC).

[3] Margaritifer Terra is centered at 4.9°S, 335°E; our study area is in the southwestern quadrant, between 11° and 22°S, and 320° and 335°E (Figure 1) and is part of the heavily cratered Noachian plateau sequence [Scott and Tanaka, 1986]. Regional elevation decreases to the northeast, and the Ladon Basin, an ancient multiringed impact basin, is located in the eastern half of the study area. The most recent study of this region was conducted by Grant and Parker [2002]; the remainder of this description of the region is drawn from that work. The topography and geologic evolution of the area is largely dictated by Ladon Basin and other components of the Uzboi-Ladon-Margaritifer mesoscale outflow system (ULM). The ULM is a system of channels and basins that stretch from the Argyre impact basin to the northeast through Uzboi Vallis, Holden Basin, Ladon Valles, Ladon Basin, and Margaritifer Valles, ending in Margaritifer Basin. This large system and its tributaries are estimated to have provided drainage for ∼9% of the Martian surface. Multiple resurfacing events concluding in the Late Noachian overprint the older Holden and Ladon basins. Resurfacing from an additional, more localized event may have continued into the middle Hesperian. Channel and valley formation began in the Late Noachian and extended into the Middle Hesperian, likely occurring in several stages as evidenced by terrace deposits in the channels. The water thought to have pooled in Margaritifer Basin (also an outlet for the Samara and Parana-Loire Valles systems to the east) is suggested to have infiltrated the heavily fractured surface and been released later, beginning in the Middle Hesperian, through the formation of Margaritifer and Iani Chaos, creating the Ares Vallis outflow channel.

Figure 1.

(a) Black box illustrating the location of the study area in southwestern Margaritifer Terra. (b) Study area in more detail. The shaded relief map was created from 128 ppd Mars Orbiter Laser Altimeter topographic data [D. E. Smith et al., 2001].

[4] In a study of intracrater thermal anomalies using Viking IRTM data, Edgett and Christensen [1994] found that deposits in Margaritifer Terra differ from those in the other regions they studied. According to their model, which is based on an inverse relationship between dune cover and thermal inertia, the Margaritifer Terra deposits have a much higher thermal inertia (440 ± 60 J m−2 K−1 s−1/2) and an inferred lack of significant dune cover relative to the other areas studied (where average thermal inertia values range from ∼120 to 300 J m−2 K−1 s−1/2). Limited image resolution available at the time of that study prohibited further investigation as to the cause of these differences from intracrater deposits in other regions.

[5] In this study we investigate the geologic characteristics of these thermally anomalous deposits to determine how their properties differ from those of their surroundings and likely explanations for their occurrence. We analyze and compare temperature, thermal inertia, dust coverage, and morphology to better understand these physical properties. In addition, we examine the spectral properties of these deposits using multispectral and hyperspectral thermal infrared data. Below, we explain the data sets and approach in detail, followed by the results of these analyses and a discussion of their implications to the geologic characteristics of the deposits. We also consider the potential origins of the thermally anomalous deposits suggested by our results.

2. Approach

2.1. Data Sets and Data Selection

[6] The THEMIS instrument on Mars Odyssey consists of two subsystems, the first of which is a thermal infrared imager with 10 bands centered at nine wavelengths in the range of 6.78 to 14.88 μm to produce nighttime and daytime thermal infrared images, each at a resolution of ∼100 m/pixel. THEMIS also carries a visible imaging subsystem that produces images of the surface with five bands centered at wavelengths between 0.425 and 0.860 μm and a resolution of ∼18 m/pixel [Christensen et al., 2004]. The high spatial resolution of THEMIS provides an excellent complement to the high spectral resolution of TES, described below.

[7] The TES instrument onboard Mars Global Surveyor mapped Mars from 1999 to 2006, providing information on the composition and thermophysical properties of the surface. As described in Christensen et al. [2001], TES consists of a Michelson interferometer that measures emitted thermal infrared energy from 5.8 to 50 μm in 143 or 286 channels (10 and 5 cm−1 sampling, respectively), along with thermal (5.1–150 μm) and visible/near-infrared (0.3–2.9 μm) bolometers. Energy is collected by a 3 × 2 array of detectors, with each detector having a spatial resolution of ∼3 × 3 km, though down-track smear increases that to ∼3 × 6 km in the along-track direction [Christensen et al., 2001].

[8] We selected TES emissivity spectra for each crater based on surface temperature (>260 K), total atmospheric dust (<0.22), and ice (<0.07) nadir opacity values. Spectra were also restricted to those between OCK (orbit counter keeper) 1683 and 7000 to avoid the additional noise from the spacecraft introduced in the data after this time in the mission [Bandfield, 2002; Hamilton et al., 2003]. Other constraints, such as HGA motion and solar panel motion, were not used because the number of spectra obtained was small enough to be visibly inspected for quality. To further reduce noise, spectra from individual detectors covering the same region were averaged before being analyzed to produce the results presented here.

[9] We also utilize data from the MOLA and MOC instruments on the Mars Global Surveyor spacecraft. MOLA uses laser altimetry to determine the topography of the Martian surface [D. E. Smith et al., 2001]. The 128 pixel/degree gridded MOLA topography, resampled to THEMIS IR spatial resolution, is used to extract the elevation data for the present work. The topography data is oversampled to allow for direct comparison to the high inertia deposits in the nighttime THEMIS IR images. MOC consists of three visible imaging systems; two cameras provide low-resolution (up to 230 m/pixel), wide-angle context images of the surface and a telescope on the third system creates high resolution (up to 1.5 m/pixel), narrow angle images [Malin et al., 1992; Malin and Edgett, 2001]. The narrow angle images used in this study have a resolution of ∼3 m/pixel.

[10] In our study area, 21 of the 53 craters we examined exhibited high brightness temperatures relative to the surrounding terrain in THEMIS nighttime thermal infrared images, an example of which is shown in Figure 2. These deposits are observed in craters of various diameters (15 to 66 km). For the purpose of this study we label the craters by their south latitude and east longitude (e.g., 14335 indicates the crater at 14°S and 335°E).

Figure 2.

Thermal Emission Imaging System (THEMIS) nighttime thermal infrared images of two craters in southwestern Margaritifer Terra: (a) a crater with a thermally anomalous intracrater deposit and (b) a crater with no anomaly.

2.2. Thermophysical Properties

[11] Thermophysical properties are material properties that relate to heat transfer and storage, such as thermal conductivity, heat capacity, and thermal inertia. Nighttime temperatures primarily represent the thermophysical properties of the top 10s of cm of the surface [e.g., Kieffer et al., 1977; Mellon et al., 2000]. A higher nighttime temperature typically indicates a material that has a higher thermal inertia, absorbing and emitting heat more slowly, than neighboring lower-temperature materials. However, because temperature can be affected by variables such as latitude and season, thermal inertia is used to compare between different areas or separate observations rather than temperature. Thermal inertia variations on Mars commonly can be linked to differences in effective particle size, with larger particles having greater thermal inertia [e.g., Christensen, 1982; Kieffer et al., 1977; Mellon et al., 2000; Presley and Christensen, 1997].

[12] TES bolometric thermal inertia is determined using a lookup table that matches a measured nighttime temperature value with the appropriate season, time of day, latitude, albedo, surface pressure, dust opacity, and thermal inertia [Mellon et al., 2000; Putzig et al., 2005]. An uncertainty of up to 6% is associated with TES bolometric thermal inertia values [Mellon et al., 2000]. The global thermal inertia values determined by Mellon et al. [2000] from TES data fall into three modes centered at 70–80, 180–250, and 240–260 J m−2 K−1 s−1/2 and indicate that the bulk of the planet's surface materials correspond to clay to coarse sand sized particles [Fenton et al., 2003; Presley and Christensen, 1997]. Particularly high values of thermal inertia (>∼1500 J m−2 K−1 s−1/2) could indicate outcrops of bedrock [Fergason et al., 2006; Mellon et al., 2000; Presley and Christensen, 1997], but subpixel mixing would cause the observed values to be lower in remotely sensed measurements of the Martian surface. Additionally, the maximum TES thermal inertia calculated for the lookup table is 800 J m−2 K−1 s−1/2, though some surfaces on Mars may have values far above this.

[13] THEMIS thermal inertia values are determined using a lookup table similar to that for TES data but created from a different thermal model [Fergason et al., 2006]. The high spatial resolution of the THEMIS instrument results in the ability to see smaller-scale anomalies due to less subpixel mixing than with TES. Additionally, the thermal inertia from THEMIS data has been calculated for a possible range from 24 to 3000 J m−2 K−1 s−1/2, higher than that for TES. THEMIS thermal inertia values are calculated directly from the band 9 (12.57 μm) brightness temperatures of a THEMIS nighttime thermal infrared image for each of the craters and have an accuracy of ∼20% [Fergason et al., 2006]. Uncertainty of THEMIS nighttime temperature data is ∼2.8 K (at ∼180 K) [Fergason et al., 2006]. An alternative method of determining thermal inertia from THEMIS data has been developed by Putzig and colleagues [Putzig et al., 2004], but was not used in this study.

[14] In our analysis we found TES thermal inertia values from the solar longitude (Ls) period of ∼150°–270° to be abnormally high and variable relative to the values from the rest of the Mars year. Included in this Ls range is the southern hemisphere summer season, a time of year when high dust opacity is common [M. D. Smith et al., 2001]. Such variations in dust opacity are not taken into account in the TES thermal inertia derivation and can cause artificially high thermal inertia values to be derived [Mellon et al., 2000]. Therefore we exclude values from Ls = 150°–270° in the TES thermal inertia values reported in section 3 below.

2.3. Albedo

[15] Albedo is the ratio of the amount of light reflected by a surface to that of the light incident upon the surface; visible reflectance at Mars, measured by the TES visible/near infrared bolometer, is converted to and presented as Lambert albedo [Christensen et al., 2001]. Albedo can vary due to the effects of particle size on the scattering of light. Typically, for regions away from the poles, higher albedo (brighter) areas are due to dust. We are interested in the distribution of fine particles because their presence can cause scattering of infrared energy and could potentially affect our analysis. Our study has observed that albedo, like TES thermal inertia, can also vary with Ls. The range in average albedo values for the craters from Ls = 150°–270° is higher than the range of values during the rest of the year. These elevated values are likely from the inclusion of data that are affected by variations in dust opacity and possible surface dust removal and deposition during the southern summer season [M. D. Smith et al., 2001]. Because these values are not representative of the albedo during the majority of the Martian year, they are not included.

2.4. Dust Cover Index

[16] Ruff and Christensen [2002] developed a dust cover index (DCI) to complement albedo as a means of evaluating the distribution of fine particulates. On the basis of known spectral trends as a function of particle size, these investigators developed a spectral parameter (the average TES emissivity from 1350 to 1400 cm−1) that can be used as a proxy for dust cover. Values of less than 0.940 indicate a dust-covered area, whereas values greater than 0.962 indicate a dust-free area, and values falling between these numbers suggest partial dust cover.

2.5. Composition

2.5.1. Thermal Emission Imaging System (THEMIS) Compositional Variation

[17] THEMIS daytime thermal infrared calibrated radiance images were used to produce decorrelation stretched (DCS) images of each intracrater deposit identified in the study area. Two deposits have no available daytime IR image coverage, however, and so are not included in the compositional study, though they are part of the thermophysical study. The DCS enhances three user-selected bands by redefining the coordinate system in a way that amplifies variations by removing highly correlated information, and then combines these band images to make a false color image [Gillespie et al., 1986; Kahle et al., 1993]. Multiple DCS images were created for each crater using the 6/4/2 (10.21, 8.56, and 6.78 μm), 8/6/4 (11.79, 10.21, and 8.56 μm), and 9/7/5 (12.57, 11.04, and 9.35 μm) THEMIS spectral band combinations for the red, green, and blue colors in each image. Using multiple band combinations helps to ensure that a spectral difference in any band will be detected. We visually inspected the decorrelation stretched images to determine which crater deposits exhibit spectral variation from their immediate surroundings. For those deposits showing possible spectral variation, we performed further spectral analyses as described below.

2.5.2. Thermal Emission Spectrometer (TES) Compositional Analysis

[18] Each thermal infrared emissivity spectrum is assumed to be a linear mixture of the emissivities of all of the minerals located within that pixel in proportions relative to their abundance [e.g., Ramsey and Christensen, 1998]. At Mars this includes not only the surface, but contributions from the Martian atmosphere as well [Smith et al., 2000a]. By using this linear mixing property, an unknown spectrum can be deconvolved with an iterative least squares method that calculates the best fit of a combination of components and abundances taken from a predetermined spectral library [Smith et al., 2000a]. The user provides the deconvolution algorithm with the measured spectrum, a spectral library of known spectra from which to model the measured spectrum, and constraints on the wavelength range over which to do the deconvolution. The algorithm returns the best-fit model spectrum to the measured spectrum and a list of the library spectra used and their proportions. A blackbody component, used to equalize the differences in spectral contrast, and atmospheric components are subtracted, and the remaining concentrations are then normalized to produce the modeled surface abundances.

[19] The spectral library for our study underwent multiple revisions, with phases not used in the deconvolution removed (e.g., albite, anorthite) and the process repeated until we were confident that the library spectra could model the unknown spectra well. The final set of phases is listed in Table 1 and consists of many typical igneous phases, along with their alteration products. Our spectral library was tailored to the specific area we analyzed and is likely not applicable to other regions or global studies. Individual TES spectra over the specified area were averaged (number of spectra averaged range from 4–25), and the average spectrum was deconvolved from 1300 cm−1 to 235 cm−1 (∼8 to 43 μm) with the exclusion of the region of high atmospheric CO2 opacity, ∼800–500 cm−1 (∼12.5–20 μm).

Table 1. Phases and Atmospheric Components Used in the Spectral Library for This Study
  • a

    Minor transparency features removed.

Dust low CO2 Smith et al. [2000b], Bandfield et al. [2000a]
Dust high CO2 Smith et al. [2000b], Bandfield et al. [2000a]
Water ice cloud (high latitude) Smith et al. [2000b], Bandfield et al. [2000a]
Water ice cloud (low latitude) Smith et al. [2000b], Bandfield et al. [2000a]
CO2 gas Bandfield [2002]
H2O gas Bandfield [2002]
OligoclaseWAR-0234Christensen et al. [2000a], Ruff [1998]
Oligoclase (peristerite)BUR-060Christensen et al. [2000a], Ruff [1998]
Oligoclase (peristerite)BUR-060DChristensen et al. [2000a], Ruff [1998]
LabradoriteWAR-4524Christensen et al. [2000a], Ruff [1998]
LabradoriteBUR-3080AChristensen et al. [2000a], Ruff [1998]
LabradoriteWAR-RGAND01Ruff [1998]
AndesineBUR-240Christensen et al. [2000a], Ruff [1998]
AndesineWAR-0024Christensen et al. [2000a], Ruff [1998]
BytowniteWAR-1384Christensen et al. [2000a], Ruff [1998]
EnstatiteHS-9.4BChristensen et al. [2000a], Hamilton [2000]
BronziteNMNH-93527Christensen et al. [2000a], Hamilton [2000]
HyperstheneALH84001Hamilton et al. [2003]
Pigeonitea Hamilton [2000]
DiopsideBUR-1820Christensen et al. [2000a], Hamilton [2000]
DiopsideNMNH-R15161Christensen et al. [2000a], Hamilton [2000]
DiopsideWAR-6474Christensen et al. [2000a], Hamilton [2000]
AugiteDSM-AUG01Christensen et al. [2000a], Hamilton [2000]
AugiteNMNH-9780Hamilton [2000]
Olivine Fo68 KI3115V. E. Hamilton et al. (manuscript in preparation, 2007)
Olivine Fo60 KI3362V. E. Hamilton et al. (manuscript in preparation, 2007)
Olivine Fo35 KI3373V. E. Hamilton et al. (manuscript in preparation, 2007)
Olivine Fo10 KI3008V. E. Hamilton et al. (manuscript in preparation, 2007)
ForsteriteBUR-3720AChristensen et al. [2000a]
ForsteriteAZ-01Christensen et al. [2000a]
Kaolinite (granular)Kga-1Christensen et al. [2000a], Piatek [1997]
Nontronite (granular)WAR-5108Christensen et al. [2000a], Piatek [1997]
Fe-smectite (granular)SWa-1Christensen et al. [2000a], Piatek [1997]
Illite (granular)Imt-2Christensen et al. [2000a], Piatek [1997]
Halloysite (granular)WAR-5102Christensen et al. [2000a], Piatek [1997]
Na-montmorillonite (granular)SWy-2Christensen et al. [2000a], Piatek [1997]
K-rich glass Wyatt et al. [2001]
Silica glass Wyatt et al. [2001]
HematiteBUR-2600Christensen et al. [2000a]
GypsumS5Christensen et al. [2000a], Lane [2007]
Gypsum (selenite)S8Christensen et al. [2000a], Lane [2007]
AnhydriteS9Christensen et al. [2000a], Lane [2007]

2.5.3. THEMIS Compositional Analysis

[20] We cannot use the linear deconvolution method of atmospheric correction with THEMIS because of its lower spectral resolution. Instead, to derive surface emissivity from THEMIS daytime calibrated radiance images, we removed the effects of atmospheric emission and attenuation through a process outlined by Bandfield et al. [2004]. First, we corrected differences in atmospheric emission due to temperature variation on the surface (i.e., shaded or sunlit slopes) through a constant radiance offset removal. We did this by selecting an area of the THEMIS image that is likely to have no variation in composition but contains slopes exhibiting temperature differences (e.g., a small crater, <∼3 km) as a training region and using an algorithm that isolates the contribution of atmospheric emission, subtracting it from each pixel of the image [Bandfield et al., 2004]. The next step consisted of selecting a training region having uniform emissivity in the THEMIS image, and identifying colocated TES pixels (meeting the quality criteria defined above). We took the average of the TES spectra and linearly deconvolved it to obtain the surface emissivity for the training region. We then convolved the TES-derived surface emissivity spectrum to THEMIS spectral resolution. Assuming that this is the surface emissivity for the training region, we calculated and removed the shape of the atmospheric component in the THEMIS data from each pixel in the image. This process requires the absence of clouds and makes the assumption that the atmosphere is not variable over the image, which is likely valid in regions without significant elevation change (<∼1000 m) [Bandfield et al., 2004]. The topographic deviations of the craters we studied by this method do not appear to affect the analysis.

3. Results

3.1. Thermophysical Properties

[21] The thermally anomalous deposits show average nighttime temperatures in the range of 182 to 213 K (with standard deviations of 2.2 to 5.6), with maximum temperatures as great as 223 K. In contrast, the average and maximum temperatures for the areas outside of the craters are between 171 and 204 K (with standard deviations of 0.9 to 3.0) and 184 and 213 K, respectively. This results in average nighttime temperatures 5 to 18 K greater for the thermally anomalous deposits than the surrounding areas, which is greater than the ∼2.8 K uncertainty for nighttime temperature data [Fergason et al., 2006]. Both the deposits and the adjacent areas exhibit the highest nighttime temperatures during southern summer, when they receive greater insolation during the day.

[22] Figure 3 provides an example of two craters in our study area that have greater values of TES thermal inertia than the surrounding plains. The values here are rounded to the nearest 5 J m−2 K−1 s−1/2. Maximum values for all craters are between 455 and 675 J m−2 K−1 s−1/2 and are plotted with the mean albedo values (described below) in Figure 4. For two craters we are unable to determine their thermal inertia values due to lack of coverage. The average THEMIS-derived thermal inertia also increases for the warm deposits. The maximum values of the deposits measured range between 450 and 1060 J m−2 K−1 s−1/2, with averages from 287 to 616 J m−2 K−1 s−1/2 (standard deviations of 45 to 116).

Figure 3.

Thermal Emission Spectrometer (TES) nighttime bolometric thermal inertia overlaid on a nighttime thermal IR THEMIS mosaic. This figure shows an example of the increase in thermal inertia over the anomalous intracrater deposits. In this image are craters 16322 (left) and 16323 (right).

Figure 4.

Maximum bolometric thermal inertia values plotted with mean Lambert albedo.

3.2. Albedo

[23] The average Lambert albedo values for the craters are shown in Table 2. The values are averages of an area in a box around the crater, which includes the anomalous deposit, the crater interior, and a small amount of the surrounding plains. Individual inspection of the values in each area shows that the albedo does not significantly vary between the intracrater deposit and the surroundings, and therefore the average value reported is a reasonable representation of the albedo over both the deposit and the surrounding area. The average values range between ∼0.120 and 0.142 (standard deviations between 0.011 and 0.025).

Table 2. Average Lambert Albedo
CraterAverageStandard Deviation

3.3. Dust Cover Index

[24] The DCI values are also presented as average values from a small box surrounding the crater and deposit, as with albedo. There is no discernable difference between the deposits and the plains, so an average of values taken from the crater and a small area around it appear to provide a good representation of dust cover. The average DCI values for the craters are between 0.967 and 0.974 (standard deviations of 0.007 to 0.040).

3.4. Composition

3.4.1. THEMIS Compositional Variation

[25] Of the 19 intracrater thermal anomalies analyzed, five show color variations in at least one of the decorrelation stretched images, indicating possible differences in composition between the deposits and their surroundings. The variations are not observed in the same set of bands for every crater; for example, crater 11335 exhibits its most notable variations in the 8/6/4 DCS, whereas crater 16322 has more obvious variation in the 9/7/5 DCS. This suggests that there may be compositional variations among the deposits. Figure 5 gives an example DCS image for each of these five craters. Two of the craters contain DCS color variations within the intracrater deposit and are designated by letter as described in Figure 5. The results of the quantitative compositional analysis of these five craters are presented below.

Figure 5.

Decorrelation stretched (DCS) images of the five craters showing possible compositional variation. The crater's designation for this study is shown above the image. The number of the THEMIS radiance image used in the DCS is given just below each image. The bands used for the red, green, and blue components of the color image are also shown below the image. Crater 16322 exhibits two deposits, one pink and one orange, in this band combination, designated by the letters A and B, respectively. Crater 19331 also has multiple deposits, shown as blue, purple, and orange in this DCS image, designated by A, B, and C, respectively.

3.4.2. TES Compositional Analysis

[26] Figure 6 shows our average surface spectra for each thermally anomalous intracrater deposit and the surrounding plain, along with the Mars surface type 1 and 2 (ST1, ST2) spectra from Bandfield et al. [2000b] for reference. Bandfield et al. [2000b] describe ST1 as a basalt and ST2 as a basaltic andesite or andesite. Wyatt and McSween [2002] demonstrated that ST2 can be modeled as a weathered basalt as well. The Margaritifer spectra resemble the ST1 spectrum more closely than the ST2 spectrum, though none of them appear to be identical to the ST1 shape. In many cases the deepest band minimum (∼1100 cm−1) is shifted to shorter wavelengths than in ST1, and the spectral shape varies at long wavelengths (e.g., lack of a minimum at ∼470 cm−1, location and depth of the emissivity minimum in the 300–400 cm−1 range).

Figure 6.

The average emissivity spectra of each of the units for the five craters. Measured spectra are shown in continuous or dashed lines, while all modeled spectra are shown by dotted lines. The group of spectra for each crater is offset by 0.05 from the previous group. The surface type 1 and type 2 spectra from Bandfield et al. [2000b] are included for reference. The vertical shaded line is present to help in comparison of the spectra. (a) Shows craters 11335, 14332, and 16322; (b) shows craters 16330 and 19331.

[27] The majority of the deposit spectra resemble the spectrum of the plains region surrounding them, though a few are not consistent in shape and spectral contrast, such as the 16330 deposit (e.g., long wavelength variation), the A and B deposits of crater 16322 (e.g., shape in 8–12 μm region), and possibly the A deposit of crater 19331 (e.g., shape in 8–12 μm region). A comparison of the plains spectra adjacent to each crater reveals that all are very similar in shape. Given the general agreement among all of the plains spectra, the deposits deviating from this shape (i.e., 16330, 16322 A and B, 19331 A) are most likely to have a composition differing from that of the plains. The 16330 deposit spectrum contains more noise due to fewer detectors available for averaging, though our confidence in its difference from the plains' spectra and composition is bolstered by the THEMIS data described in the next section.

[28] Deconvolution of the ST1 shape with our library spectra results in slightly different modeled abundances than those of Bandfield et al. [2000b], as shown in Table 3. This is due to small variations in our spectral library, such as the inclusion of more olivine spectra of different compositions and the use of an alternate hematite spectrum. Our modeled phase abundances for each intracrater deposit and surrounding plains material are presented in Table 4, rounded to the nearest 5%. The detection limits and uncertainty of each value is 10–15% [Christensen et al., 2000b]. The values for each mineral group vary slightly between the deposit and surrounding plains and from crater to crater. However, in most cases the variation in abundance between the deposit and surrounding plains and between each of the deposits or each of the plains units is still well within the uncertainty. The deposit and plains associated with crater 16330 provide the only exception, with a difference in abundance between the deposit and plains of 15% for plagioclase feldspar and 25% for pyroxene. The deposit in crater 16330 also varies from the other deposits in plagioclase and pyroxene abundance by up to 20%. Variations of plagioclase and pyroxene between the other four deposits are 5% and 10%, respectively. With the exception of the crater 16330 deposit, the modeled abundances for the thermally anomalous intracrater deposits cannot be said to differ beyond the uncertainty from that of the surrounding plains.

Table 3. Modeled Abundances of the Type 1 Surface Spectruma
 FeldsparPyroxeneOlivineClay/Sheet SilicatesHematiteOtherb
  • a

    Values are reported to the nearest 5%.

  • b

    This includes phases modeled well below the detection limit such as evaporite minerals and glasses.

This study30255201010
Bandfield et al. [2000b]503001505
Table 4. Modeled Abundances for Five Intracrater Deposits and the Surrounding Terrainsa
 11335 Deposit11335 Plains14332 Deposit14332 Plains16322 Deposit A16322 Deposit B16322 Plains16330 Deposit16330 Plains19331 Deposit A19331 Deposit B19331 Deposit C19331 Plains
  • a

    Modeled abundances are derived from the deconvolution of an average spectrum of the area and have been rounded to the nearest 5%. The values have an uncertainty of 10–15%.

Plagioclase feldspar25302525202525402520252525
Sheet silicates00551005105105510
K-rich glass55555105505555
Silica glass0000000000000

3.4.3. THEMIS Compositional Analysis

[29] Our atmospherically corrected surface spectra from THEMIS data are shown in Figure 7. Each spectrum shown is the average of ∼600 or more spectra (∼6 km2). The majority of the deposit and plains spectra are similar in shape, but differences can be seen in the spectra for the 16330 and 16322 A deposits (lower emissivity in THEMIS bands 7 and 8 relative to the other spectra) and the 19331 plains (higher emissivity in band 7 than the other spectra). In the case of the 16330 and 16322 A deposits, this supports the slight differences that we observed in their TES spectra. For the 19331 plains deposit that has no noticeable variation in its TES spectra, the THEMIS spectra may be identifying differences that cannot be resolved at the lower spatial resolution of TES. Any other differences observed between TES spectra may not be significant enough to be apparent in the lower spectral resolution of the THEMIS spectra.

Figure 7.

Atmospherically corrected THEMIS spectra (bands 3–9) for the (a) deposits and (b) surrounding plains of each crater. Each spectrum is an average, and the error bars represent one standard deviation. Spectra are normalized to the same contrast to better evaluate differences in shape.

3.5. Topography

[30] The elevation of the thermally anomalous materials is always lower than the surface adjacent to the crater, as expected for typical Martian craters [Strom et al., 1992]. Mean elevation values for the deposits vary from −491 m to −3103 m and mean values for the surrounding plains vary from ∼1000 m to −2000 m. Figure 8 illustrates this topographic information. Slightly lower deposit elevation in the eastern part of the study region may reflect the overall eastward decrease in elevation of the region. We do not observe any relationship between the deposits and crater diameter, elevation, or location.

Figure 8.

Spatial distribution of anomalous intracrater deposits, along with their average elevation and crater diameter (ranging from 15 to 66 km, indicated by relative size of symbol). No relationship with crater diameter or elevation is observed.

3.6. Geomorphology

[31] Crater 11335 is covered by the MOC narrow angle image S06-01399 (Figure 9a). A dark-toned spot at the eastern edge of the crater interior can be seen in the context image (S06-01400, not shown) and the narrow angle image captures a small part of that area. At ∼3 m/pixel resolution, the dark-toned feature appears to be caused by the filling of small depressions within the lighter-toned, coherent floor with a darker, dune-forming material. The concentration of this darker-toned material is only in this area of the crater, however, and the coherent floor material, which is a characteristic of a large portion of the crater floor, must be the source of the high nighttime temperatures and thermal inertia.

Figure 9.

(a) A portion of the Mars Orbiter Camera (MOC) image S06-01399 showing part of the high thermal inertia intracrater deposit. (b) A section of the MOC image S08-01915 showing the light- and dark-toned high thermal inertia material. (c) A portion of the MOC image S02-01115 covering the junction of fractures within the thermally anomalous deposit. The arrow indicates an example of possible layering. (d) A portion of the MOC image S08-03262 illustrating two of the high thermal inertia deposits in this crater.

[32] MOC narrow angle image S08-01915 (Figure 9b) reveals the detail of an area of the crater 16322 deposits. This narrow angle image shows a high-standing, dark-toned material, which appears to have been eroded in places, exposing a lower-lying light-toned rough terrain. To the best of our ability to determine by comparison with the DCS image, the dark-toned material corresponds to Unit A, whereas the light-toned material corresponds to Unit B. Isolated mesas of the dark material within the light-toned material are also observed. Although there is some infilling material in topographic lows forming dunes, both the light- and dark-toned materials are likely coherent units because of their ability to form steep-sided topography.

[33] Figure 9c, a portion of the MOC narrow angle image S02-01115, shows the intersection of troughs in the middle of crater 16330. From the steep trough walls it is apparent that the crater floor is composed of a coherent unit, homogeneous in tone. The image also shows what might be layering within the deposit, exposed on the sides of the troughs. The 16330 deposit does not appear as rough as some of the other deposits (e.g., the 16322 B deposit) at the ∼3 m/pixel scale, and the only visible dune forms are in the material infilling the troughs and larger craters.

[34] The MOC narrow angle image S08-03262 provides information about the A and B deposits from crater 19331 (the area of the crater containing the C deposit was not imaged). As can be seen in Figure 9d, the A deposit appears to be a high-standing dark-toned material, in places seeming to be stratigraphically above the light-toned B unit. A darker loose material, possibly eroded from the A unit, forms dunes in the pits of the B deposit. Though there is some dune material, the actual unit that makes up the high-temperature B deposit appears to be solid.

[35] From THEMIS VIS image V17875002 (Figure 10), the intracrater deposit in crater 14332 is relatively homogeneous in tone. Higher-resolution imaging of this crater is not available, but the high thermal inertia area seems to be correlated with a coherent unit with fractures creating troughs through it.

Figure 10.

A portion of THEMIS visible image V17875002, showing part of the crater 14332 anomalous deposit.

4. Discussion

[36] Low albedo and high DCI values suggest that dust cover is not significant over the intracrater deposits and their surroundings. The maximum TES thermal inertia values of 455 to 675 J m−2 K−1 s−1/2 for the anomalous deposits correspond to effective particle sizes as large or larger than granules and pebbles [Fenton et al., 2003; Presley and Christensen, 1997]. The higher spatial resolution of the THEMIS thermal inertia calculated for the anomalous deposits returns greater maximum values, as high as 1060 J m−2 K−1 s−1/2.

[37] A plot of the thermal inertia and albedo values of the intracrater deposits (e.g., Figure 4) overlaps the high thermal inertia/low albedo mode B of Mellon et al. [2000], though much of the data extends outside of this mode to higher thermal inertia values. Putzig et al. [2005] divide the albedo/thermal inertia plot further, creating a unit F for thermal inertia values greater than 386 J m−2 K−1 s−1/2, an appropriate category for the deposits of this study. They indicate that unit F is characteristic of rocks, bedrock, duricrust, or polar ice. Thermal inertia unit B (thermal inertia of 160–355 J m−2 K−1 s−1/2) is described by Putzig et al. [2005] as sand, rocks, bedrock, and possibly duricrust. These descriptions suggest that a likely interpretation of the intracrater deposits in the present study, based on their range of thermal inertia and latitude, is that they are composed of rocks, bedrock, or duricrust with the areas of slightly lower thermal inertia containing a higher abundance of smaller, sand sized particles.

[38] An integration of the information gained from thermal inertia values and the deposits' geomorphology helps to determine the most accurate description of the deposits' physical character. The MOC images of the thermally anomalous intracrater deposits show that the high thermal inertia surfaces are predominantly composed of a coherent material with little dune cover, though a more dispersed population of small particles below the image resolution is possible and could explain the spread in thermal inertia values within the deposit, as could isolated accumulations of particles within small depressions, which we have observed.

[39] Intersecting fractures cut half of the craters analyzed in this study, including craters 14332, 16330, and 19331. Floor-fractured craters such as these have been identified on Mars predominantly along the border between the southern highlands and the northern lowlands, and areas surrounding parts of the Valles Marineris and the outflow channels [e.g., Schultz, 1978; Schultz and Glicken, 1979]. The fractures are believed to result from modification of the crater floor due to intrusion of magma through fragmented rock beneath the crater. The extent of this modification may range from concentric grabens and/or polygonal fractures to extrusion of igneous material, as seen in many of the lunar floor-fractured craters [e.g., Schultz, 1976]. In contrast to the Moon, there is also the potential on Mars for the presence of subsurface ice to affect the formation of floor-fractured craters [Schultz, 1978; Schultz and Glicken, 1979]. Schultz [1978] suggests that the release of volatiles from the heat of intrusion could aid in disruption of the crater floor. Differences in regional properties may lead to varying styles of modification in separate areas [Schultz and Glicken, 1979]. The style of floor fractures observed in our study area consists mainly of a polygonal network of shallow fractures across a flat crater floor and/or concentric fractures at the intersection of the crater floor and wall. Though the fractures of most floor-fractured craters on Mars (including our study area) are limited to the crater interior, there is one example within the craters we investigated where the fractures extend beyond the crater rim (crater 16330). We cannot be sure of the process that created this difference in fracture occurrence, but it may be explained by the crater's location within the larger Ladon basin and could be due to an instance of floor fracturing at a more massive scale.

[40] Our compositional analysis of the thermally anomalous intracrater deposits suggests that there is little difference between the deposits and the surrounding region. Our quantitative studies of TES modeled abundances show no distinguishable difference in composition above the uncertainty, even though a few of the spectra vary slightly in shape. Although some degree of compositional deviation exists for deposit of crater 16330, this specific deposit is an exception and not representative of the other 19 deposits in the compositional study.

[41] The region as a whole appears to be broadly similar in composition. The ST1 spectrum was identified by Bandfield et al. [2000b] as one of two spectra that characterize the general composition of the low-albedo Martian surface. The global maps of surface types 1 and 2 concentration given in Bandfield et al. [2000b] indicate that ST1 models the composition in Margaritifer Terra more adequately than ST2. Our results confirm this, but the TES spectral shape and modeled abundances also show that the composition of the region differs noticeably even from that of ST1. For the surface type 1 spectrum, the modeled abundance of plagioclase feldspar is greater than that of pyroxene (Table 3), whereas the modeled abundances from the deposit and plains TES spectra of this study have the opposite trend.

[42] Recent results from Rogers et al. [2007] indicate that the Martian low-albedo regions can be described in further detail by 11 spectral shapes derived from representative regions. The modeled compositions of these representative spectra were used to define four compositional groups [Rogers and Christensen, 2007]. The resulting global maps, Figure 7 of Rogers and Christensen [2007], suggest that Margaritifer Terra is most adequately described by the spectrum and modeled composition of Group 3, consisting of the representative spectra of Tyrrhena Terra, Hesperia Planum, Cimmeria-Iapygia, and Meridiani. The average modeled abundances of phases for Group 3 is fairly consistent with the values we report here for the deposits and plains in southwestern Margaritifer Terra (see Table 5).

Table 5. Comparison of the Range of Modeled Abundances (%) From This Study With the Abundances of Group 3 From Rogers and Christensen [2007]
 Plagioclase FeldsparPyroxeneOlivineHigh-Si PhasesOther
  • a

    Values from this study are reported to the nearest 5%.

This studya20–4025–505–155–155–20
Group 3, Rogers and Christensen [2007]2530111816

[43] With little variation in composition between the deposits and their surroundings, there does not appear to be a correlation between thermal and mineralogical characteristics. Variation in effective particle size is the dominant difference between the anomalous deposits and the plains. If the deposits are solid, coherent units as they appear in visible images and their thermal inertia values suggest, it is likely that they represent outcrops of bedrock. The emplacement of bedrock could have occurred after crater formation by a primary volcanic process (e.g., a lava flow or an upwelling from below the crater) or by lithification of sediments (epiclastic and/or pyroclastic) transported into the crater by eolian or fluvial activity, or mass movements of the crater walls and rim. Such sediments could then be cemented by a material that is present in low abundances not detectable by TES or that lacks significant spectral features in the thermal infrared (e.g., halides).

[44] Impact melts are expected to be produced during impacts on Mars [e.g., Osinski, 2006; Schultz and Mustard, 2004] and may be present within the craters analyzed in our study. However, little to no relief of crater rims and central peaks, along with shallow interior depths, indicate that erosion and infill have caused the degradation of most of these craters. We believe it likely that any impact-generated melts and/or breccias are buried by this later infill. The anomalous deposits are not seen in all the impact craters in this area of Mars, suggesting that they are not linked directly to the impact process (assuming the impact and target conditions are similar throughout this region). Additionally, we do not observe a variation in glass abundance between these deposits and the surrounding plains, and the glass abundances are all below detection limits.

[45] The compositions determined in this study are dominantly basaltic, which might be consistent with either a primary igneous process or lithification of basaltic sediments that have undergone little chemical weathering. However, no lava flow features are observed on the deposits or leading into these craters, and there are no obvious vents nearby to deliver the material. Though floor-fractured craters, like the ones observed in the study area, are thought to be associated with localized magmatic intrusions [Schultz, 1978; Schultz and Glicken, 1979], evidence for surficial igneous activity is lacking (e.g., volcanic dark halo craters such as seen along floor fractures of some craters on the Moon [Head and Wilson, 1979; Schultz, 1976]). Additionally, any relation between the process forming the floor-fractured craters and the high thermal inertia deposits studied here is tenuous because fractures are not observed in all of the craters exhibiting anomalous intracrater deposits, although this logic is complicated by the slight possibility that continued infill of the craters could have obscured fractures or volcanic constructs.

[46] Lithification of basaltic sediments seems to fit more closely with the observations, including possible layering within the 16330 deposit (though layering is not unique to this method of formation). In this hypothesis, water would have interacted with the sediments in the past and resulted in the formation of a coherent unit that retained its bulk composition, adjusted slightly for the small percentage of a cementing phase. The quantity of water likely involved in this type of scenario is unknown and could range from small transitory amounts, including groundwater, to standing bodies of water. The outcome of a similar process has been observed at Meridiani Planum, where sediments are believed to be cemented by sulfates and possibly other evaporites resulting from groundwater fluctuations [e.g., McLennan et al., 2005; Squyres et al., 2004; Squyres and Knoll, 2005].

[47] In light of the identification of sulfates by the MER Opportunity rover in Meridiani Planum, and also by the Observatoire pour la Mineralogie, l'Eau, les Glaces et l'Activité (OMEGA) instrument on the Mars Express spacecraft [e.g., Gendrin et al., 2005], we added the spectra of additional Fe, Mg, and Ca sulfate minerals to our original spectral library. We then repeated the deconvolutions to perform an initial test of the idea of lithification by evaporite cements. With the modified spectral library, the abundance of evaporite minerals was modeled at a consistently higher percentage (up to 35%), though the observed fit of the modeled spectrum was not significantly improved and the RMS error did not show significant improvement above the noise; it therefore cannot be said with certainty which combination of modeled abundances is more accurate. Because deconvolutions using the original spectral library did not result in any obvious misfits in modeled spectra, this observation is not surprising. The information gained from the additional deconvolution does illustrate the possibility that evaporite minerals may be present in significant (>15%) proportions, though there is no clear trend of increased modeled sulfate abundance for intracrater deposits above that of the plains, as would be expected in order to strongly support the idea of sulfate as a cementing agent for the intracrater deposits. It is possible that the differences in sulfate abundance between deposits and plains may be so small that the uncertainty of the data may obscure any potential trend. Analysis of these anomalous deposits by the two visible to near-infrared instruments currently at Mars, OMEGA and the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) on the Mars Reconnaissance Orbiter spacecraft, may be helpful in detecting any cementing material.

[48] Lithification of sediments may be the most straightforward way to explain the presence of these anomalous deposits in low-lying areas, where sediments may become trapped. The region also hosts remnants of the Uzboi-Ladon-Margaritifer mesoscale outflow system [Grant and Parker, 2002], evidence that water has been available there in the past. Even if the deposits formed during a time of no surficial water flow, melting of subsurface ice could be another potential source of fluid, an idea supported by the proximity of chaos regions and outflow channels.

[49] We are unsure as to why these deposits are present in some craters but not in others of similar size in the same region. There does not appear to be a connection with elevation or location. These observations suggest that the occurrence of these high thermal inertia deposits may be the result of a highly localized process. We cannot be certain that a single process or event produced the deposits in this area. The high density of these deposits in southwestern Margaritifer Terra does suggest that the process(es) forming high thermal inertia deposits preferentially occurred, or was preferentially preserved, in this region.

5. Summary and Conclusions

[50] The results and interpretations determined by this work are outlined below.

[51] 1. The thermally anomalous intracrater deposits studied in Margaritifer Terra have higher thermal inertia values than their surroundings. These values can be as large as 1060 J m−2 K−1 s−1/2 and suggest a rocky or bedrock surface.

[52] 2. This proposed rocky to bedrock character is supported by MOC images exhibiting a coherent surface. Smaller, unconsolidated particles likely exist on the surface of these units but are not concentrated in association with the thermal anomalies.

[53] 3. Our compositional analysis of these deposits with TES and THEMIS data show little difference in the composition between the thermally anomalous deposits and that of the surrounding region.

[54] 4. Hypotheses for the origin of these intracrater thermal anomalies include the emplacement of primary igneous material, but lithification of basaltic sediments may be the more likely process.

[55] 5. The process that formed the intracrater deposits seems to have occurred on a local scale.

[56] Thermally anomalous deposits are not restricted to this study area. It is possible that the interpretations derived from this study may be extrapolated to other deposits, implying that they also formed by processes similar to those proposed for the origin of the intracrater thermal anomalies in this area of Margaritifer Terra. Evaluation and comparison to similar deposits in adjacent areas may augment the results of the present study.


[57] The authors would like to thank R. Fergason for the use of the program to derive the THEMIS thermal inertia measurements and M. Lane for additional sulfate spectra. We are also grateful to Malin Space Science Systems for the use of the Mars Orbiter Camera images, available at, specifically images S02-01115, S06-01399, S08-01915, and S08-03262, and their context images obtained through the Public Target Request Site. THEMIS images are provided courtesy of THEMIS Public Data Releases and are available at The helpful discussions and guidance provided by F. S. Anderson and W. Koeppen are greatly appreciated. We are thankful for the thorough reviews by D. Rogers and S. Ruff, which helped to clarify and improve the manuscript. This work was supported by grants from the Mars Data Analysis (NAG5-13421) and Mars Fundamental Research (NAG5-12685) programs. This is HIGP publication 1489 and SOEST publication 7152.