Application of Image Reconstruction Techniques to Mars Neutron Spectroscopy Data

Data sets collected by the Mars Odyssey Neutron Spectrometer contain artifacts from spatial blurring and noise, which are limiting factors in the interpretation of surface physical properties. The spatial response of the spectrometer is about 10 degrees (full width half maximum) of arc length, making comparisons to and synthesis with higher resolution optical data sets and geological maps challenging. In addition, blurring artifacts can cause significant errors in the derived surface properties. Properties constrained by neutron spectroscopy include the abundance of water equivalent hydrogen (WEH), the column abundance of CO2 in the seasonal caps, and the abundance of elements such as Fe and Cl, which have large cross sections for thermal neutron absorption. Computer codes were developed to reconstruct maps of the Martian surface with reduced blurring and noise, which can be subjected to geochemical interpretation with greater reliability. Penalty function and iterative deconvolution techniques (Jansson’s method) were implemented and compared using simulated and measured data sets. These techniques were applied to global reconstruction of the Mars neutron spectroscopy data set. Both techniques were effective in removing noise without introducing unrealistic oscillations in the reconstructed maps. Results of the study are presented along with implications to Mars science.