Automatic grain sizing of vertical exposures of gravelly deposits

The Grain Size Distribution (GSD) is a key indicator in stratigraphy and morphology; it is a basic tool used by a variety of disciplines: geology, geomorphology, archeology, ecology and engineering. Automatic Grain Sizing (AGS) has been developed to estimate the GSD of surfaces in a rapid and effortless manner. Yet the traditional bulk method, time consuming and laborious, is still in use for the determination of the GSD of coarse-grained sedimentary deposits and of the near-subsurface of landforms, principally river beds. The AGS technique is examined here in its application to the stratigraphic record and to the subsurface. Images processed by AGS and manual samples of coarse-grained deposits in section were compared for evaluation of accuracy and determination of the extent of validation of the AGS technique.The AGS often achieves results very similar to field counting, especially when the largest grains are accounted for in the size of the sampling area and when photographs are taken afar without flash. Best results are provided when two images of the same sediment are taken at close range and also from afar to combine their GSDs, hence overcoming errors generated by resolution, sampling area and grain size. The computed errors are larger than those in prior studies yet remain reasonably small, even though results are not bias corrected, axes exposed in an outcrop very likely differ from those exposed on the surface, and sediment properties and lighting conditions vary. Although results indicate a 25% burial effect, the performance of the AGS for sections is high.Unlike prior application of AGS for riverbed surface openwork gravels, conglomerate sections and the near-subsurface include a matrix, resulting in a somewhat larger mean irreducible error. The technique may be utilized in studies of conglomerate texture, allowing the evaluation of entire GSDs under difficult sampling conditions. An example use of AGS to subsurface GSD evaluation is provided, among others enabling an easy and fast method to calculate extent of armoring.