CSDMS 2014 annual meeting poster Dan Cazanacli
A Simple Rule-Based Model for Distributary Networks
Matthew Wolinsky, Shell Technology Center Houston Texas, United States.
Zoltan Sylvester, Shell Technology Center Houston Texas, United States.
Alessandro Cantelli, Shell Technology Center , United States.
Chris Paola, University of Minnesota Minneapolis Minnesota, United States.
Abstract:
Deltaic networks present a wide variation in almost any geomorphologic aspect (number of channels, channel size distribution, planform sinuosity, shoreline shape). To replicate this complexity we propose a reduced complexity network growth model, based on a set of simple rules some of which are quantitatively anchored in physical processes while others are purely, and connected to the physical process in an implicit way, via observed field correlations among various terms (e.g., Syvitski, 2006). The intent is to keep the number of rules to a minimum necessary to reproduce, in a statistic sense, most but not all of the observed delta styles.
In its most general form, the model generates distributary networks in which planform of individual channels emerge from a correlated random walk algorithm, channel density is a function of a prescribed bifurcation probability and the overall delta shape (i.e., wide vs narrow) is controlled through a dependency on dominant flow direction. Individual channels form through successive addition of short segments (piecewise) each new piece introducing a small direction deflection that is partly correlated to the previous deflection. A preferential flow direction (trend) is factored in as a proxy for the downstream slope. The weight assigned to this direction is small but critical in controlling the overall delta shape. This correlated random walk method is equivalent to the one used and tested by Surkan and van Kan (1969). The key to producing networks is channel bifurcation. Frequent bifurcations result in dense, anabranching channel patterns while more representative deltaic networks are obtained using a small probability bifurcation value (0.01 to 0.05). The proposed network growth model can yield distributary networks of significant morphological variation in terms of shapes, channel planforms, or channel density. The comparison between model outcomes and field analogs will be through a series of metrics such as planform shape of individual channels, delta shape, shoreline shape, or channel density distribution.
We argue that a stochastic approach driven by simple rules is ideal for investigating complex delta distributaries. Though reduced complexity models employs much simpler rules it does demonstrably lead to complex landscape patterns via randomness built in (e.g. Murray & Paola, 1994). Using simple rules to characterize the process, rather than to model it analytically through complex models, also enables scenario testing and makes it easier to understand the important controls and explore the link between process and landscape.
Simple Network Example
* Please acknowledge the original contributors when you are using this material. If there are any copyright issues, please let us know and we will respond as soon as possible.
