BUILDING A 3D REPRESENTATION OF OUTCROP HETEROGENEITY FROM LIDAR IMAGERY

Development of models that preserve the influence of heterogeneity at varying scales is vital for modeling solute dispersion in groundwater. Dispersion, the result of varying velocities in a flow field, is, in part, due to material heterogeneity. In order to characterize heterogeneity at the 2-3m scale, a series of high-resolution (millimeter-scale) terrestrial lidar scans were taken of sediments located in braided stream exposures west of Albuquerque, New Mexico. Lidar measures the topography and texture of the outcrop surface by projecting an infrared laser at the outcrop and collecting the intensity of the reflected light from coordinates on the surface. The lidar data are represented as a point cloud of intensity values with an x, y, and z coordinate. We create a 3D characterization of the heterogeneity by successively scraping and scanning the outcrop in approximately 2 cm intervals. Each successive slice is projected into the same coordinate system, thus a 3D point cloud is obtained for the exposed volume. Since most segmentation approaches have been developed for 2D data, we segment sand from gravel from each 2D slice of the outcrop. Combining these segmented slices we develop a 3D grid that delineates the regions of sand versus gravel. At present, our segmentation is binary and uses 2D data. We are currently developing means to interpolate the point clouds into a regular 3D grid and segment facies from this grid. In earlier work, our 2D simulations showed that flow and solute transport are focused by cross-bedding into the coarser-grained units. We expect the 3D simulations to show similar results. Solute transport simulations display a non-Fickian breakthrough character due to the variable velocity field caused by heterogeneity.