Precision and Accuracy of Simultaneously Collected Hirise Digital Terrain Models

Introduction: High precision Digital Terrain Models (DTMs) made from High Resolution Imaging Science Experiment (HiRISE) [1] stereo pairs are being produced for geomorphic and geologic analyses (e.g. [2, 3, 4]) as well as landing site assessment [5, 6, 7]. In many cases, they are being used to orthorectify sequences of images, making change detection of the martian surface possible at unprecedented detail. Stereo pairs separated by one or more Mars years are acquired over selected active sites in order to attempt to quantify volumetric changes in topography such as dune migration [8] and gully activity [9]. Here we present some examples where two HiRISE stereo pairs are controlled simultaneously in order to measure such changes. These special cases provide an opportunity to investigate the internal precision and accuracy of HiRISE DTMs. Quality can be affected by spacecraft jitter, observation angles, image quality, albedo differences and illumination angles. Although the dynamic range and signal-to-noise ratio of HiRISE images are very good, the changes we wish to measure are not large and may be smaller than the differences between the DTMs. This work presents a review of vertical precision, horizontal accuracy and internal precision in HiRISE DTMs, and what effect these values might have on studies of topographic change. Precision and Accuracy in Individual DTMs: Within a single HiRISE DTM, precision and accuracy can be estimated or measured by several methods. Expected vertical precision: This value can be estimated based on the observation geometry and the pixel scale, as well as the quality of the triangulation. The formula used to derive EP is based on [10]: