The Mars Reconnaissance Orbiter (MRO) arrived at Mars on 10 March 2006 and began its primary science phase in November.  The High Resolution Imaging Science Experiment (HiRISE) on MRO is the largest, most complex camera ever flown to another planet.  Plans call for this scanner to image roughly 1% of Mars by area at a pixel scale of 0.3 m during the next Mars year.  Among the thousands of images will be hundreds of stereopairs that will provide an unprecedented three-dimensional view of the Martian surface at meter scale.  These stereopairs will provide a tremendous amount of information for focused scientific studies, landing site selection and validation, and the operation of landers and rovers.  In this paper, we describe our approach to generating geodetically controlled digital topographic models (DTMs) from such stereopairs, our early results, and plans for future DTM production.

Our approach to the photogrammetric processing of HiRISE images follows that which we have previously described for the MOC and the Mars Express High Resolution Stereo Camera (HRSC).  We use the USGS in-house digital cartographic software ISIS to do initial processing, including ingestion, decompression, and radiometric calibration of the images.  "Three-dimensional" photogrammetric processing steps, including control and DTM creation and editing, are performed on a photogrammetric workstation running the commercial software SOCET SET (® BAE Systems).  Noteworthy departures from past practice are the use of ISIS 3, the object-oriented successor to the older ISIS 2 system, and pre-processing in ISIS to correct geometric complications of the HiRISE images that cannot be modelled in the SOCET sensor model: multiple CCD detectors in the focal plane, optical distortion around an axis far from the detectors, and (ultimately) the small "jitter" motions of spacecraft pointing that distort the images and hence the DTMs.

The first HiRISE stereopair analyzed covered the location of the Opportunity rover near the 750-m crater informally named Victoria in Meridiani Planum.  This scene was extremely unfavorable for automated stereomatching, with extensive areas that are almost featureless, extremely steep, or both, but these problems were offset by the high quality of the HiRISE imagery, permitting us to obtain a 1 m/post DTM that required only limited interactive editing.  Subsequent mapping of the Spirit rover site and a variety of scientifically interesting sites has proven that the greater surface texture found at most places on Mars leads to even better DTMs with even less editing required.  We are currently working to refine and streamline our procedures in order to maximize the number of sites that can be mapped and studied in three dimensions with HiRISE.