An assessment of 3D computer models of sand dune surfaces using close range digital photogrammetry

We have been trying to determine the usefulness of close range digital photogrammetry for building three-dimensional models of sand dune surfaces at scales from 5 to 100 meters. Sand dunes are ever-changing environments that show both strengths and weaknesses of digital photogrammetry. In gathering data to create three-dimensional models, we have found that limitations include the fact that sand dunes frequently look homogeneous at our scale of observation, providing few uniquely recognizable points needed to make models. Also, since photogrammetry is a line-of-sight method, foreground vegetation or topographic irregularities can obscure background features. Furthermore, photogrammetry works best with views normal to the surface being modeled, but 3D dune modeling may be limited by relatively low angles, creating less accurate models. Finally, the moving landscape requires extra effort to establish stationary control points for time series comparisons. Strengths include that points are measured remotely, in contrast to GPS and some laser survey methods, therefore leaving the surface undisturbed. Also, data gathering is more rapid than in laser survey methods, and compared to laser scanning, photogrammetry requires less and lighter equipment that is easier to hike through remote dune locations. Of the limitations, the homogeneity of dune surfaces at a distance appears to be the most serious difficulty. A future partial solution to the homogeneity of dunes is to place targets on the dune to improve the density of uniquely recognizable points. Collecting enough data at high enough resolution to compare different time periods and storm events is difficult over large areas but the method can still be useful to monitor changes in dune vegetation or to measure sand movement in smaller views where individual sand grains provide distinct markers for matching points when making models