Chapter 7 - Summary and Future Work

Regional extracted spectra are useful for comparison of spectral features across the entire spectrum from a small number of locations.  Locations are chosen from large scale differences such as albedo and limb distance.  The spectral features can also be modeled using crude radiative transfer models to set upper limits on various species.  In this work, limits on the sensitivity of CO₂ cloud detection are set at visible optical depths of greater than or equal to 1.0.  These limits are higher the optical depths calculated from Hubble Space Telescope visible images from the same date and so if the tropical clouds seen there were composed of CO₂ ice, the modeling technique would not be able to detect them.  Band depth mapping techniques also indicate that there are no CO₂ clouds with visible optical depth greater than or equal to 1.0 due to the fact that there is little to no spatial variation in the band depth.

Band depth maps are useful for tracing the amount of absorption in a single spectral feature across the entire disk.  The 3.33µm CO₂ ice feature was found to identify the seasonal polar cap after the dissipation of the north polar hood.  A band depth map at 2.25µm was found to be able to locate H₂O frosts due to the local maximum in its spectrum at that point.  The polar hood as well as morning and evening clouds are easily seen.  No local noon, tropical clouds seen in these band depth maps, however they are seen in Hubble Space Telescope images.  This fact sets a limit on the sensitivity of the band depth map to H₂O ice clouds with visible optical depths greater than about 0.3.

The principal components analysis/Linear mixture modeling (PCA/LMM) transform the data cubes into a space spanned by only two principal components.  The first component, accounting for over 97% of the data variance, corresponds to simple albedo variations.  The second component, accounting for about 2% of the data variance, is a measure of ice and temperature.  The extremes from these PCA dimensions allow for the choice of three spectral endmembers for the LMM.  The spectral units are a centrally located, high albedo region, a centrally located, low albedo region and the north polar region.  The evening and morning limb regions show a high value of this final endmember indicating cold, cloudy conditions and an overall wet Martian atmosphere.

The extracted regional spectra are limited to looking at a single pixel or group of pixels at a time, the band depth mapping has some confusion due to possible similar absorptions in a band from substances other than what one is trying to map, and the PCA/LMM is limited to linear processes and by some confusion on whether the polar pixel is a combination of frost types or whether it is a single frost type.

On top of these limitations in the techniques, there were some other problems with the data set that need to be addressed.  One problem was the lack of a true sky endmember.  Since the plate scale of the images caused Mars to fill the detector all of the sky pixels were contaminated by scattered Mars light.

A second problem was the inability to register images perfectly.  The images in a spectral scan were registered using a cross-correlation technique that breaks down in wavelengths where the limbs and pole of Mars disappear.  Spectral scans had to be tweaked by eye after the automated registration using a blink comparison, a ratio image and a difference image in order to reduce edge effects due to misregistration.  Images were only shifted to the nearest whole pixel. Another problem due to edge effects was the inability of the PCA/LMM to properly model limb pixels due to the non-linear processes of Lambert limb darkening and atmospheric absorptions from increasing atmospheric pathlength.

To handle these problems, the Martian disk images need to be limb corrected and projected onto a cylindrical map coordinate system.  The key to doing these two corrections lies in proper limb and center finding of the image disk.  This is no trivial task as the disk size and shape are functions of wavelength.  The size changes are due to Mars atmospheric absorption limb darkening and reflection law limb darkening.  The shape changes are due mainly to polar region disappearance in absorption features.  In these images there is also the problem that Mars slightly overfills the image area and so part of the limb may lie off the image.  Future efforts will attempt to solve these problems.

The data used in this dissertation make up a fraction of a fairly large data set including image sets from nights before and after those presented here that span the entire Martian northern spring season.  Future work will include analyzing the rest of the image sets using these three techniques.  Earlier image sets may show even colder atmospheric conditions with a higher probability of seeing CO₂ clouds.  The thickness of the polar hood may be changing and thinning in later image sets allowing better viewing of the seasonal polar ice cap.

In the extracted regional spectra, more detailed cloud models will need to be made to get better estimates of the amounts of ices in the atmospheric clouds.  These models will require the use of full radiative transfer and ground reflection theories.

Epilogue

I would like to end this chapter with a poem that has held significant meaning for me.  It comes from my favorite literary classic The Lord of the Rings by J. R. R. Tolkien.  The poem was recited by Bilbo Baggins on the occasion of his eleventy-first birthday as he stood on the path leading away from his lifelong home, the Shire, gazing upon it for the last time.  I include it here as I stand on the path leading away from Laramie on the occasion of my thirtieth birthday.

The Road goes ever on and on       Down from the door where it began. Now far ahead the Road has gone,       And I must follow, if I can, Pursuing it with eager feet,       Until it joins some larger way Where many paths and errands meet.       And whither then?  I cannot say.