The threshold values for removing large caters were determined by examining the craters within the study area, referencing previous studies (Molloy and Stepinski, 2007), and some trial and error. After the parameter values are determined, the rest of the process is automated. However, we do anticipate some minimum manual editing may be needed in some complicated terrains when apply it to all of Mars. To minimize the distortion resulted from map projection on global datasets, we will choose an equal area projection by evaluating the options suggested in Steinwand et al. (1995) or conduct geodesic area calculation using software such as “Tools for Graphics and Shapes” (http://www.jennessent.com/arcgis/shapes_graphics.htm)Although post-formational modification to the valleys may be minimum (Williams and Phillips, 2001), there may nonetheless be modifications such as eolian fill and mass wasting (e.g., Grant et al., 2008). Thus the volume estimates derived with PBTH method represents a lower bound. Comparing the estimates from MOLA and HRSC data reveals that MOLA estimate is about 91% of HRSC value. However, MOLA has global coverage whereas HRSC does not. Therefore, for areas where there is only MOLA coverage, the estimate may be scaled upward by 1.1 times. The algorithm has been tested on DEMs with various resolutions (2 m for simulated DEM, 75m for HRSC, and 463m for MOLA). It can certainly be applied to higher resolution DEMs for Mars when they become available, but the threshold values will need to be adjusted.
