STOCHASTIC CHANNEL STACKING IN ALLUVIAL ARCHITECTURE: MEASURES OF SPATIAL CLUSTERING APPLIED TO EXPERIMENTAL DATA

The alluvial architecture models introduced by Leeder, Allen, Bridge, and coworkers produce cross-sectional panels in which channel bodies are arranged in space more or less randomly. For instance, as they avulse or migrate laterally, channels may be attracted to previous channels if the latter remain topographically low, or they may be repelled if an abandoned channel belt remains topographically high (as suggested by JRL Allen). The former case would lead to stratigraphic clustering of channel deposits, the latter to avoidance.

Here we examine the degree of randomness of stacking of preserved channels and channel complexes in two-dimensional panels using statistical analyses of spatial clustering. The analyses are based on measures of distance between pairs of channels and points within channels, and on the spatial distribution of channel centers over the stratigraphic panels. We use data from experiments in the Experimental EarthScape subsiding-floor basin facility, St Anthony Falls Laboratory, and associated stratigraphic experiments in which topographic data were collected at high time resolution. The experimental panels show channel densities ranging from <0.1 to nearly 1.0. The combination of video records, topography, and preserved stratigraphy allows us to relate the statistics of channel distribution in the stratigraphic panels to patterns of channel migration, reoccupation, and abandonment. Channels are typically reoccupied several times before they are finally filled in and abandoned. Although reoccupation explains some multi-story sand bodies seen in the field, in the experiments multiple occupation events often do not lead to a distinctive stratigraphic signature.