Chapter 5 - Band Depth Map Results

The first band depth map considered is the 2.010µm feature, using 1.730µm and 2.250µm as the continuum points, which serves primarily as a test because it should show more absorption at the limbs due to the higher atmospheric pathlength as well as more absorption at the polar region due to water and CO₂ ices.  The maps are presented in Figures 5.1 to 5.5 below:

As expected, the value of the band depth does decrease toward the limbs as the amount of CO₂ increases due to atmospheric path length.  The band depth decreases even more at the polar region due to either water ice or dry ice.  In the 14 JAN 95 images one can also see the topographically high region Elysium in the center of the disk and rotating to the evening limb. The feature shows up as less CO₂ absorption due to the much shorter atmospheric pathlength.

The topographically low region of Hellas shows up in these maps quite distinctly.  In the scan 3 map of 14 JAN 95 map it appears in the southwest limb region and in the scan 1 map of 01 FEB 95 it appears in the southeast limb region.  This is due to the much longer pathlength through the CO₂ atmosphere.

There is also darkening along the disk edges in some of these maps.  This is not due to a CO₂ detection but rather an artifact of the registration.  The individual images in a spectral scan have been translated so that ground albedo features align throughout the entire spectral scan.  The difficulty in registering the images is that within the strong absorption features, much of the contrast disappears and the disk changes size and shape due to limb darkening.  These reasons alone make proper registration difficult, but then there is also the problem that Mars is rotating as a spectral cube is being gathered and this can make alignment of images almost impossible, especially when the time between images is on the order of 30 minutes.  Such a difference corresponds to a 7.3° rotation or about 433km at the sub-Earth point which is at the level of the seeing.  Because of these registration problems, edges will tend to become enhanced when the ratio of two images is taken.

The next band depth map considered is really a ratio map.  The ratio of 3.33µm to 2.25µm (or 3.331µm to 2.254µm for scans 2 and 4 of 01 FEB 95 - the Nyquist sampled image cubes) is a general measure of where the volatiles are.  Mars is dark overall in the 3µm range due to surface hydrated mineralogy (Sinton 1967, Houck 1973) but the absorptions due to water and CO₂ will be stronger (Herr and Pimentel 1969).  The maps are presented in Figures 5.6 to 5.12 below:

The features most prominent in the 14 JAN ratio maps are the north polar region, Syrtis Major, and the southern dark albedo regions.  There is also a conspicuous absorption excess near the evening limbs.  The dark region absorption differences are due to hydrated mineral compositional differences between the dark, exposed rock and the bright, dust covered areas (Houck et al. 1973 and Pimentel et al. 1974).  The north polar region absorptions are no doubt due to condensed volatiles.  Since both CO₂ and H₂O are dark at 3.33µm relative to 2.25µm no discrimination can be made in composition.  The excess absorption at the evening limb occurs over high topographic regions.  In scan one the topographic high is Olympus Mons.  Similar cloud structures are seen over the Tharsis volcanoes in Hubble Space Telescope images (Figures 5.34 and 5.35).  In scan two there is no real absorption excess, but in scan three we see absorption over Elysium as evening sets in.  These absorptions could be due to H₂O or CO₂ clouds condensing over the high features due to evening cooling or if prevailing conditions cause air to rise up over the high regions.

Similar effects are seen in the ratio maps of 01 FEB 95.  There is more absorption over the polar regions than anywhere else due to the presence of various ices.  In these images however, there are no excessively high regions so cloud condensation is not as prevalent in the evening. There does appear to be some excess absorption in the morning hemisphere which could be overnight condensation of volatiles either as ground frosts or clouds.  However there is still the question of composition.

To evaluate the composition we can look at a band depth map centered on the CO₂ absorption feature at 3.33µm using 3.270 and 3.400µm as the continuum images (or 3.331µm as the band center and 3.262 and 3.401µm in for the Nyquist sampled images).  By comparison to the previous ratio images we can determine if the above excess absorptions are due to CO₂ or H₂O.  The band depth images are presented in Figures 5.13 to 5.19 below.

Due to the definition of the band depth, low values in these images means more absorption.  In the three images from 14 JAN there is little to no absorption.  The variation from a value of 1.000 is on the order of 3% in scan 1, 5% in scan 2 and at most 8% in scan 3 compared to the 40% to 60% value of a region completely covered by CO₂ frost as shown in the band depth map test in Chapter 3.  Based on this, and the spectral modeling discussed in Chapter 4, we can say that there are no CO₂ clouds with visible optical depths equal to or greater than 1.0.

In the four images from 01 FEB there is a large zero value stripe that is an artifact  due to the fact that in some of the wavelength bands the image of Mars moved off the chip and thus these areas are blank after the images were registered.  Ignoring these stripes we see that low values of the band depth lie in the northern regions and to some smaller degree along the morning and evening limbs.  In scans 1 and 3 the global values of the band depth vary from 1.00 only on the order of ±5% except in the polar regions where there is an absorption on the order of 15%. The appearance of this absorption at 01 FEB where it was not apparent at 14 JAN could indicate that there is water ice cloud polar hood that is beginning to thin as the season progresses, thus allowing the CO₂ seasonal cap below it to be seen.

There is a curious effect in scans 2 and 4 (the Nyquist sampled sets).  The mean value instead of being 1.00 is on the order of 0.875 with a global variation of about 10% with little or no excess absorption in the polar region.  This would tend to indicate that there is CO₂ frost everywhere.  This is a very unlikely situation considering that there is no CO₂ detection in scans 1 and 4.

A possible interpretation is that the polar hood clouds are varying on time scales equivalent to the time it takes to gather a spectral set.  The polar hood is known to be variable during its formation period but is fairly constant once it is fully formed by the northern autumnal equinox (Martin & McKinney 1974, Akabane et al. 1995).  This would indicate that this interpretation is not acceptable.  At this time there is no explanation for the low, but nearly constant, value for scans 2 and 4 and more work on this problem is needed.

Another area where this CO₂ vs. H₂O discrimination can be made is in the 2.2-2.3µm region.  CO₂ ices show only a flat reflectance whereas water has a local maximum as presented in the previous chapter.  The images in Figures 5.20 to 5.26 show a band depth map centered on 2.25µm (2.254µm for the Nyquist sampled sets) with continuum points of 2.131µm and 2.331µm (2.136µm and 2.337µm for the Nyquist sampled sets).

In most of the above band depth maps the north polar region has a value 1.40-1.50 compared to the value for a fully covered region of 1.50-1.60 as shown in the band depth map test in Chapter 3 indicating the local water ice maximum.  This indicates that even in the times where the underlying CO₂ polar cap can be seen, there are still significant amounts of water ice.  No real determinations can be made on areas near the evening and morning limbs, although there does appear to be some degree of correlation between the 2.25µm band depth maps and the 3.33µm/2.25µm ratio map.  It cannot be stated with certainty that there is no CO₂ ice in the evening and morning limbs based on these band depth maps alone; however, taken with all the above maps, it appears that when there are limb volatiles they appear to be only water ice.

Curiously, the scan 1 and 3 band depth maps from 01 FEB 95 appear to have strong correlations with albedo.  The band depth maps from 14 JAN 95 also show this correlation, but to a lesser extent.  This is interesting in that it implies that there may be some unknown surface mineralogical effects.  It is also somewhat troublesome in that such an effect would make this band depth map not fully reliable.

The conclusion from these band depth maps is that the volatile effects in the 3.33µm/2.25µm ratio maps are due mostly to the effect of water ices and surface mineralogy rather than CO₂ ices.  The north polar region is covered by water ices that obscure any CO₂ signatures early in the season, but as time progresses the water has thinned enough to detect the CO₂ ice.  Because of this uncovering effect it is concluded that the CO₂ is a ground ice and is the seasonal polar cap.  The state of the water, whether ground ice or a cloud, is as yet not fully determined, although it can be surmised that in the polar region it is more than likely to be a polar hood cloud.  However at the limbs it could be a condensed ground ice or a water ice cloud formed during the colder nighttime conditions.

To distinguish between clouds and ground ice it is necessary to have some information about the ice grain sizes.  If the spectral signature matches that of a coarse ice, such as the ice cube spectrum of Roush et al. (1990), then it is safe to presume that it is a ground frost.  In the spectral region around 3.69µm it was noted earlier (Figure 3.1) that there is a major difference between coarse and fine ice frosts.  The coarse frost is merely dark across this whole region with little or no spectral variation whereas the fine frosts have a rather large local maximum.  This spectral difference was illustrated in Chapter 3 along with a presentation of the band depth mapping (Figure 3.3) of the laboratory spectra.  With 3.96µm as the band center (3.968µm in the Nyquist set) and the continuua wavelengths of 3.40µm and 3.80µm, the coarse grained water ice frost will have a value less than one whereas the fine grained frost has a value greater than one. The band depth maps are presented in Figures 5.27 to 5.33 below.

]

In all these band depth maps we can see clearly that the north polar region always has a band depth value of 1.2 to 1.4, indicating a fine frost.  In all the images from 14 JAN the evening limb has a value from 1.2 to 1.3, thus indicating that these regions which show up in the 3.33µm/2.25µm ratio map as being volatile rich most likely are composed of a fine water frost.  In the 01 FEB images the limbs are not as similar to one another.  They do not show up in Scan 1 but both show up well in Scan 3 with a value of about 1.275.  The morning limbs in Scans 2 and 4 are extremely high, on the order of 1.4 to 1.5.  This could be due to the previously mentioned registration problem although there is most likely some fine water frost effect as the high values do extend into the disk.

Overall, from the band depth maps it is apparent that much of the volatile effect is due to a fine grained water ice either on the ground or in the form of clouds. Hubble Space Telescope images in Figures 5.34 and 5.35 (James et al. 1996) show many features that appear to be clouds located in a belt around the northern tropics and a north polar hood. Clancy et al. (1996a,b) create models to these images that can be fit using water ice clouds.  Since the detections from this dissertation find fine water frosts in the same regions as the HST detected clouds, we conclude that the band depth map is a good method to detect water ice clouds.

The actual state of the water frosts in the north polar region is unclear, although a polar hood cloud would be consistent with the HST images (James et al. 1996, Clancy et al. 1996a,b). Normally, however, the polar hood is gone by the northern vernal equinox (Martin and McKinney 1974, Martin et al.  1992).  If it is indeed a cloud then this indicates that the polar air is colder than in the past, allowing the hood to remain far longer than before.  The clouds seen in the HST images (James et al. 1996) in Figures 5.34 and 5.35 also indicate that the Martian atmosphere is colder than during the Viking era.  Microwave measurements of water vapor saturation altitudes (Clancy et al. 1996a) also measure a Martian atmosphere that is 15-20K colder than the Viking era with absolute column abundances of water vapor around 8-11pr.µm as compared to the 12-14pr.µm measured at comparable seasons by the Viking MAWD.

So if the atmosphere is so much colder, why is there no evidence of north polar CO₂ ice clouds in the data used in this dissertation?  Kieffer (1970a) showed that even small amounts of H₂O frosts intimately mixed with CO₂ frosts can completely dominate the reflectance spectra beyond 3.0µm.  At particle scales of 5-50µm the mixing ratio that can hide the CO₂ spectral features is only 0.1.  At larger scales of 20-100µm this mixing ratio drops to 0.008 so we could have a case CO₂ being hidden by the H₂O.  There is the question, however, of CO₂ mantles condensed on H₂O seed particles.  Such models or experiments have yet to be performed.

Pollack et al. (1990) showed that there is a correlation between the dust opacity and the amount of CO₂ condensation within the atmosphere.  As the dust opacity increases, so would the CO₂ cloud formation.  If the northern autumn processes are symmetric to those in the spring, then the lack of CO₂ clouds can be explained by the relatively dust-free environment on Mars at the time of these observations (Clancy, et al. 1996b).

If the detected water ice is a ground frost then there is the question of the CO₂ frost feature missing in the 14 JAN images then appearing in the later 01 FEB images.  Since the CO₂ is not in the earlier, seasonally colder, images and then appears in the later, presumably warmer, images then it may be concluded that the fine water frosts are a dissipating polar hood.