R. Koenders1, R. C. Lindenbergh1, T. E. Zegers2
1 Delft University of Technology, P.O.Box 5058, 2600 GB Delft, The Netherlands
2 Utrecht University, P.O Box 80125, 3508 TC Utrecht, The Netherlands
The planet Mars has a relatively short human exploration history, while the size of the scientific community studying Mars is also smaller than its Earth equivalent. On the other hand the interest in Mars is large, basically because it is the planet in the solar system most similar to Earth. Several satellites are currently orbiting Mars, and transmit data back in unprecedented detail. In fact, the Martian surface is mapped at up to 5 times higher resolution than the bottom of the ocean here on Earth.
The scientific community studying Mars has already made great discoveries concerning, for example, the variability of the surface (Bibring, 2005), and the presence of water. To learn more about the history of the surface and about the planet as a whole, data generated by different satellite missions will have to be combined. Processing such large, multi-attribute datasets at a global Martian scale requires efficient automated classification methods.
The use of automated classification in combination with geomorphometric data has only recently been possible on Mars with the creation of the global Mars Orbiter Laser Altimeter (MOLA) digital elevation model (DEM) (Smith et al. 2003), as obtained between 1997-2001 by the Mars Global Surveyor. (Bue and Stepinski, 2006) demonstrated the potential of classifying global MOLA DEM data and concluded that similar methodology could be applied on other data sets like the ~60m spatial resolution DEM, as currently under construction from High Resolution Stereo Camera (HRSC) images collected by ESA’s Mars Express (Gwinner, 2007).
On Earth, morphological classification has been used for numerous specific applications (Guzetti and Reichenbach, 1994; Hosokawa and Hoshi, 2001). Also only relatively recent it was demonstrated that attributes like gradient and roughness, as derived from elevation data, can be used to construct a multi-attribute feature vector, that, possibly in combination with other data, like intensity or multi-spectral data, can be consecutively applied in land surface and vegetation classification procedures (e.g. Antonorakis et al., 2008; Bork and Su, 2007; Chust et al., 2008).
Even though the use of automated classification on Martian datasets has great potential, it is not yet being used as intensively by the scientific community studying Mars. The research presented in this abstract therefore formalises the methodology presented by Bue and Stepinski (2006) as the Terrain Fingerprinting Method (TFM) in Section 2. We have applied the TFM to several areas on Mars based on the MOLA DEM, which has a maximum spatial resolution of 400 meters per pixel; HRSC DEM, which has a maximum resolution of 50 meters per pixel; and a combination of the MOLA DEM with data from the Mars Express mineralogical spectrometer (OMEGA). The present abstract focuses on an analysis of the combination of OMEGA and MOLA DEM data as presented in Section 3.