Cornell University, Department of Astronomy, Ithaca, NY 14853.

Space Science Institute, 1540 30th Street, Suite 23, Boulder, CO 80303-1012

Cornell University, Department of Astronomy, Ithaca, NY 14853.

Jet Propulsion Laboratory/Caltech, 4800 Oak Grove Dr., Pasadena CA 91125

Department of Physics and Astronomy, Univ. of Toledo, Toledo, OH 43606

LASP, University of Colorado, Boulder CO

John T. Trauger and Robin W. Evans

Jet Propulsion Laboratory/Caltech, 4800 Oak Grove Dr., Pasadena CA 91125

Submitted to Icarus March 24, 1997

Manuscript #6888

Accepted October 9, 1998

Manuscript Pages: 23

Tables: 2

Figures: 5

Proposed Running Head:

Bell et al.: Near-IR Imaging of Mars from HST

Please address all correspondence to:

Jim Bell

Cornell University

Department of Astronomy

402 Space Sciences Building

Ithaca, NY 14853-6801

phone: (607) 255-5911

fax: (607) 255-9002

email: jimbo@marswatch.tn.cornell.edu

Abstract

We report initial results from Hubble Space Telescope (HST) images of Mars obtained at a near-IR wavelength of 1042 nm between February 1995 and June 1997. The images provide the highest spatial resolution global near-IR measurements of Mars obtained to date. The data have been calibrated to radiance factor (I/F) to an absolute accuracy of 2% to 5% (including new refinements to several HST calibration parameters) and now exist as a set of coregistered, map-projected images spanning a range of phase angles between 10° to 41°. The relatively high spatial resolution (20-30 km/pixel) global maps, when compared with Viking-era global albedo maps, reveal that substantial variations in Mars surface albedo patterns have occurred in the past 20 years. The most obvious changes have involved several classical dark regions becoming substantially brighter, although there are examples in the data of formerly bright regions that have darkened. The normal albedo of typical surface regions in the near-IR is derived from the multiple phase angle observations and is found to be bimodal with typical bright and dark values of 0.36 and 0.18, respectively. The near-IR phase coefficient of surface regions ranges from ~ 0.0 to 0.013 mag/deg, and there is no systematic correlation between phase coefficient and normal albedo. This observation is consistent with the diffusely scattering nature of the Martian atmosphere even during times of low dust opacity. The data presented here may help in the planning and optimization of measurements to be obtained by the Mars Orbiter Laser Altimeter (MOLA) on the Mars Global Surveyor orbiter, scheduled to begin mapping the planet in 1999. These HST data also provide a basis for comparison with future MOLA and Mars Orbiter Camera (MOC) data in searches for seasonal and interannual variations in surface and polar cap albedo variations.

Introduction

Determination of the absolute reflectivity of a planetary surface is critical for deriving photometric properties such as albedo, phase function, macroscopic roughness, and opposition surge magnitude and width (e.g., Hapke, 1993; Helfenstein and Veverka, 1987). In addition, quantitative and realistic determinations of mineral abundances on a planetary surface using spectral mixture modeling require as inputs accurate knowledge of the absolute reflectivity of both the mixed spectrum and of a number of potential endmember materials (e.g., Mustard and Pieters, 1989; Adams et al., 1993). The issue of spectral mixing is relevant to Mars because the detailed surface composition and mineralogy is still largely unknown, and the measurements used to try to redress this gap in knowledge are frequently obtained over a range of spatial scales from centimeters (surface landers) to thousands of kilometers (groundbased telescopes). Measurement of the absolute surface reflectivity in the visible and near-infrared is particularly important for Mars, as there are a number of spectrally-active and geologically important minerals with spectral features in this wavelength range (e.g., Gaffey et al., 1993; Clark et al., 1993), some of which have already been unambiguously detected on the Martian surface from remote sensing measurements (see reviews by Soderblom, 1992; Roush et al., 1993; Bell, 1996).

Recognizing this need for well-calibrated measurements, we embarked upon a series of ultraviolet through near-IR observations of Mars using the Hubble Space Telescope (HST). The observations began in 1990 (James et al., 1994, 1996; Bell et al., 1997), and are designed to take images in many wavelengths diagnostic of atmospheric or mineralogic absorption features. Most importantly for the work presented here, many HST images of Mars have been obtained at 1042 nm, the longest near-IR wavelength possible using the pre-February 1997 HST instrument suite. The major advantages of HST images are that they are free from telluric atmospheric contamination and that they can be accurately calibrated to within a few percent using the results from the rigorous program of frequent HST standard star observations. Another distinct advantage is that imaging with HST in certain modes and at certain times can allow very high (by groundbased standards) spatial resolution measurements to be made of the Martian surface. For example, HST images of Mars obtained during the 1997 opposition achieved a spatial resolution of 22 km/pixel at the sub-Earth point, comparable to the resolution obtained by the Phobos-2 ISM instrument from Martian orbit.

This paper describes: (a) the near-IR observations of Mars that have been obtained between 1995 and 1997 and the methods that have been used to reduce and calibrate these images to radiance factor and to project the data onto a standard map base; (b) the overall nature of the near-IR reflectance of the Martian surface and how it compares to the visible and near-UV reflectances; and (c) some initial determinations of temporal variability in surface albedo marking and near-IR surface photometric properties determined from these measurements.

Observations and Calibration

The raw images were corrected for analog-to-digital conversion errors, bias and dark current, shutter shading effects, and flatfield variations using the standard procedures outlined by Lauer (1989) and Holtzman et al. (1995a,b). The reduced data were then corrected for bad pixels and cosmic ray hits using a fourth-order polynomial least-squares nearest neighbor fitting technique. Finally, in order to take advantage of the full potential spatial resolution of HST, a point spread function deconvolution was performed on the reduced images using 40 iterations of the damped Richardson-Lucy algorithm with a threshold noise parameter of 3 (White, 1994; Wolff et al., 1997).

Observatory VISTA package. We noted that the standard aperture correction factor used in

The I/F images were converted from the normal telescopic perspective (orthographic view) to a coregistered set of map-projected data products. This was achieved through software that automatically finds and fits the elliptical limb of the planet and then uses the associated ephemeris information (sub-Earth latitude and longitude, north pole position angle, geocentric distance) to perform a fully ellipsoidal map projection of the data at a scale of 1°/pixel. Small errors in the ephemerides and in precisely fitting the limb and determining the center pixel in the orthographic images lead to slight latitude and longitude errors for mapped surface features using this approach. In order to achieve greater consistency with existing Mars albedo, topographic, and geologic maps, we developed an iterative scheme to "tweak" the ephemeris information and correlate the resulting map against the Viking Orbiter IRTM albedo map at the same scale (Pleskot and Miner, 1981; Paige et al., 1994). Despite the 20 year time interval and the changes in surface albedo noted below, there is still enough similarity between the gross nature of albedo features in the IRTM and HST data sets to allow matching of the HST mapping to within less than 0.5° of the Viking maps for much of the surface.

Figure 1 Caption

The final data set for use in this paper consists of 28 images calibrated to I/F and projected onto a standard (latitude, longitude) grid. Images from global observing sequences obtained on the same Martian day (Table 1) covered nearly all Martian longitudes and have been merged into global mosaics, constructed using a simple Minnaert photometric correction to reduce seams at longitude overlap regions. The Minnaert correction produces images of the modeled zero-phase albedo, (I/F)_M, in terms of the original I/F such that (I/F)_M = (I/F)[(cos i)^k (cos e)^k-1]^-1, where i and e are the incidence and emission angles for each pixel in the image (Minnaert, 1941). For producing the global mosaics, we used a nominal value of k=0.7, consistent with previous estimates of the average photometric behavior of the Martian surface (e.g., Harris, 1961; Young and Collins, 1971; Binder and Jones, 1972; Erard et al., 1994; de Grenier and Pinet, 1995).

More details and derivations of the reduction, absolute photometric calibration, sources of calibration errors, and map projection of the WFPC2 Mars images can be found in Bell et al. (1997) and Wolff et al. (1997).

Results

Images of the 1042 nm Mars surface reflectance from February 1995 and March 1997 are presented in Figure 1. These global images are mosaics of four images that have been corrected with a simple Minnaert function as described above. Examination of the image histograms reveals the well-known bimodal nature of the surface albedo features. The majority of the surface is composed of bright regions with a relatively narrow distribution of 1042 nm (I/F)_M values from 0.33 to 0.39 (FWHM). The dark regions have a broader distribution of albedos, ranging between 0.15 to 0.25. There are a smaller but still substantial number of regions with albedos between 0.25 and 0.33, and an even smaller subset of the brightest regions, corresponding to the north polar cap, with albedos greater than 0.39. For comparison, a similarly-projected image of the Viking IRTM broadband albedo data (0.3 to 3.0 µm) is shown in Figure 2 (Pleskot and Miner, 1981; Paige et al., 1994; Paige and Keegan, 1994).

Figure 2 Caption

We can assess the short-term and long-term variations in surface albedo features using these new HST data in combination with previous groundbased, HST, and spacecraft images. During the period from February 1995 to June 1997 we saw no evidence for major changes in surface albedo markings, although regional dust storm activity was seen during this time in the northern polar regions (Wolff et al., 1996; James et al., 1996b). The lack of major surface albedo variations is consistent with the observations of James et al. (1994), who saw no variations in the Syrtis Major region in HST imaging during 1990 and 1991 and attributed this to the general absence of major atmospheric dust activity over the past decade. Information on smaller regional surface albedo changes is difficult to obtain by intercomparing this data set because of the large changes in spatial resolution over the course of the observing period. We are, however, able to verify that the only small regional feature known to have changed based on previous HST imaging (Cerberus, near latitude 15°N, longitude 205°W; James et al., 1996a; Lee et al., 1998), has not returned to its dark, pre-HST appearance (cf. Figures 1 and 2).

Figure 3 Caption

Large changes in the relative reflectivity of several areas can be seen in Figure 3 over ~20 years. Many areas have become brighter and/or changed shape considerably. For example, the Icaria/Mare Sirenum region (near 30°S, 120°W) was a typical dark region in 1977, but now has a reflectivity more like a typical bright region. The boundaries of Syrtis Major and Hesperia Planum (15°S to 30°N, 220°W to 320°W) have receded by as much as several hundred km (a change larger than could be accounted for by only variations in effective spatial resolution of the two datasets). Much of northern Mare Cimmerium (near 15°S, 210°W) has changed from an intermediate albedo region to a typical bright region. Cerberus, as discussed above and in Lee et al. (1998), and the elongated Hyblaeus Chaos (30°N to 45°N, 240°W) have both brightened and changed shape considerably. Some regions have become darker over the past 20 years. For example, Mare Erythraeum (20°S, 30°W to 90°W), the northeast margin of the Hellas basin (30°S to 45°S, 270°W to 300°W), and the arrow-shaped southern tip of Utopia Planitia (45°N, 270°W). The changes observed in Figure 3 poleward of 70°N are primarily due to seasonal differences in the extent of the polar cap between the two data sets.

Recently, substantial progress has been made in radiative transfer modeling of the physical and spectroscopic properties of airborne dust (e.g., Clancy and Lee, 1991; Pollack et al., 1995; Clancy et al., 1995; Ockert-Bell et al., 1997; Wolff et al., 1998). All of these models require as inputs accurate estimates of the normal albedo and phase coefficients of the Martian surface and/or airborne dust over a range of reflectivities and viewing geometries. The HST measurements discussed here span a wide range of phase angles and thus represent a new source of information that can be used in such modeling efforts. Determination of the phase function and phase coefficients is not straightforward, however, because of the complicating effects of scattering and absorption by aerosols in the Martian atmosphere (e.g., Clancy and Lee, 1991; Erard et al., 1994), and the variable spatial resolution of the images presented here. However, we present a simplified analysis of the phase coefficient for the surface at 1042 nm as a first step towards more detailed modeling and as a way to make some possibly useful initial results available for ongoing spacecraft investigations and other Mars surface and atmospheric modeling studies (e.g., Wolff et al., 1998).

Figure 4 Caption

The normal albedo at 1042 nm for bright surface regions and the north polar cap ranges from 0.36 to 0.42. The normal albedo for dark regions is typically 0.19 to 0.21. There is a wide range in values for the phase coefficient, although typical values are rather small (0.0035±0.0003 mag/deg). There appears to be little correlation between phase coefficient and normal albedo, reflecting the diffusely-scattering nature of Martian atmospheric dust, even under low opacity circumstances. Both the range of values of the phase coefficient and the lack of a strong correlation between phase coefficient and albedo are consistent with previous groundbased, Mariner, and Viking phase function studies (e.g., Harris, 1961; Binder and Jones, 1972; Thorpe, 1977; Pleskot and Kieffer, 1977; Clancy and Lee, 1991). These new HST analyses, however, extend the photometric study of Mars into a higher spatial resolution near-IR realm than has been possible using previous methods, and they also allow for the long-term comparison of photometric properties among surface units.

Figure 4 Caption

Discussion

In the 20 year hiatus between the Viking missions and the Mars Pathfinder and Mars Global Surveyor missions, the only global-scale data with which to assess seasonal and longer timescale variations in the Martian surface and atmosphere have come from HST and other Earthbased telescopes. Determining the magnitude and extent of such variations, whether they be in atmospheric temperature structure, dust opacity, or in surface reflectivity variations, can help to constrain models of Martian atmospheric dynamics as well as of short- and long-term climate variations. In addition to filling this gap in time coverage, HST data are also providing the first relatively high spatial resolution global measurements of the planet in the near-IR. This paper has presented some of the first results from these near-IR images.

The variations in surface albedo markings observed in Figure 3 between 1977 and 1997 can be compared with observations and results derived from long-timescale groundbased telescopic observations and short-timescale Viking orbiter measurements of surface albedo changes. Slipher (1962) and Martin et al. (1992) provided extensive reviews of historical and more recent observations of surface albedo variations. The data reveal that the boundaries of Martian surface albedo features are continually changing and are usually not exactly reproduceable from year to year. However, there are a number of perennial large-scale albedo features, such as Syrtis Major, that have persisted in general form throughout the historical record despite local changes in their specific outlines. In that regard, the magnitude of the 20-year albedo variations observed in Figure 3 are not inconsistent with similar variations observed telescopically for hundreds of years.

The origin of large-scale albedo variations has been linked to dust deposition and removal processes by a number of groundbased and spacecraft investigations (e.g., Pollack and Sagan, 1967; Sagan et al., 1973; Christensen, 1986, 1988; Martin and Zurek, 1993). In particular, Viking IRTM albedo variations were interpreted by Christensen (1986, 1988) to indicate that low-albedo southern and mid-latitude regions are post-storm sources of dust, and that the dust preferentially accumulated in higher-albedo northern hemisphere regions during the course of the IRTM late 1970s observations. Our analysis of the differences between the 1997 HST and 1977 IRTM albedo data is consistent with continued general north to south accumulation of dust (many formerly intermediate albedo northern hemisphere regions appear to have brightened, and much of the classical low-albedo southern hemisphere has darkened) in the intervening 20 years. However, most of the large-scale regional increases in albedo (Icaria/Mare Sirenum, Mare Cimmerium) occurred in the southern hemisphere and are interpreted as regions of enhanced dust deposition since Viking. A logical prediction based on the IRTM and previous groundbased observations would be that these regions should be active sources of potential dust storms in the near future, eventually returning to their low-albedo appearance. Southern hemisphere dust activity reported in early MGS MOC and TES observations (Malin et al., 1998; Christensen et al., 1998) is consistent with this expectation, although the specific geographic regions where dust activity was noted by MGS (Noachis, Hellas) are not flagged in Figure 3 as regions of enhanced dust deposition in the past 20 years, possibly because such higher southern latitude regions were not well observable from Earthbased observations during the last few apparitions.

These data may prove useful to those working with the Mars Global Surveyor orbiter MOLA instrument, because they provide the most recent measurements of the Martian surface in a near-IR wavelength very close to that being used by MOLA to measure surface elevation, vertical ruggedness, and surface reflectivity variations (Zuber et al., 1992; Smith et al., 1998). Existing telescopic and spacecraft spectra of Mars show little spectral difference between 1042 nm and 1064 nm (e.g., Roush et al., 1993; Mustard and Bell, 1994), reinforcing the assumption that data at these two wavelengths are directly comparable and likely to be influenced by the same surface and/or atmospheric effects. Knowledge of the near-IR reflectivity of the surface is an important factor in planning for the MOLA measurements because the instrument's signal to noise ratio (and thus the vertical precision of the data) is partly a function of within-footprint (140 m diameter) surface reflectivity, although the vertical structure of terrain at footprint scales is the largest factor which influences final vertical precision (Zuber et al., 1992).

Figure 5 Caption

HST measurements like those presented here also extend the time span for surface and polar cap albedo variation studies that will be substantially extended by future MOLA near-IR surface reflectivity measurements and very high resolution images from the MGS/MOC (Malin et al., 1998). And finally, previous studies have shown how these types of near-IR measurements, when combined with images at shorter wavelengths, could be used to search for and map the distribution and abundance of minerals like pyroxene, olivine, or iron oxides on the Martian surface (e.g., James et al., 1996; Bell et al., 1997).

Conclusions

(1) Measurements of the absolute reflectivity of Mars at a near-IR wavelength near 1042 nm have been obtained from the Hubble Space Telescope. Refinements have been made to the previously-published HST calibrations for the F953N and F1042M filters, resulting in data that have an accuracy of 2% to 5% and that represent the highest spatial resolution global near-IR measurements of Mars obtained to date.

(2) The images, obtained over a range of phase angles from 10° to 41°, reveal that the normal albedo of the surface in the near-IR is bimodal with typical bright and dark values of 0.36 to 0.42 and 0.19 to 0.21, respectively.

(3) Substantial variations in Mars surface albedo patterns have occurred in the past 20 years, despite much of this time being regarded by groundbased observers as a fairly "low dust" period. The most obvious changes have involved several classical dark regions (Icaria, Mare Sireumn, Mare Cimmerium, Cerberus) becoming substantially brighter, although there are examples in the data of some formerly bright regions that have darkened.

(4) The average Martian surface phase coefficient near 1042 nm is 0.0035±0.0003 mag/deg. While individual regions show significant variations in phase coefficients, there is no systematic correlation between phase coefficient and normal albedo. These results are consistent with previous groundbased and Viking measurements, and are likely related to the diffusely scattering nature of the Martian atmosphere even during times of low dust opacity.

(5) The data presented here may help in the planning and optimization of Mars Global Surveyor MOLA 1064 nm global topography/reflectivity measurements to begin in 1999, and they also provide a basis for comparison with future MOLA and MOC data in efforts to search for seasonal and interannual variations in surface and polar cap albedo variations.

Acknowledgments. We thank Jim Garvin of the MGS/MOLA instrument team for a formal review of the manuscript and for helping us to determine ways that these HST measurements could most benefit the planned MOLA observations. We also thank Wendy Calvin for detailed and constructive reviews, Dan Schroeder and Holland Ford for allowing us to use their proprietary images for our photometric calibration, David Golimowski for facilitating the data access, and Sylvia Baggett for helpful discussions regarding the SYNPHOT database. Funding for this research was provided by grants from the NASA Planetary Geology and Geophysics Program (NAGW-5062) and the Space Telescope Science Institute. This work is based on observations with the NASA/ESA Hubble Space Telescope obtained at the Space Telescope Science Institute, which is operated by Association of Universities for Research in Astronomy under NASA contract NAS5-26555. All of the data presented here are available in digital form to interested scientists, mission planners, or others; please contact the first author for access information.

References