Barchan swarm dynamics simulated with a Two-Flank Agent-Based Model

Barchans are often found in swarms spanning hundreds of square kilometres and many tens of thousands of dunes.  The scale of such systems prohibits the use of computationally intensive models of the motion of sediment under fluid flow.  Instead, several agent-based models have been developed to study these vast systems.  Such models consider only idealised symmetric barchans subject to a perfectly unidirectional wind.  In contrast, barchans in nature are often subject to some seasonal variation in the wind direction and rare but strong storms events.  The wind variability as well as interactions between the bedforms means that a typical barchan in a swarm will not be perfectly symmetrical.  To resolve the discrepancy between the modelled and real-world dunes, we have developed a new agent-based model: the Two-Flank Agent-Based Model (TFABM) which can account for both barchan asymmetry and variation in the wind.  The model uses a simple algorithm for determining the outcome of collisions between dunes but which nevertheless is able to reproduce the majority of collisional phase-spaces observed in other studies.  The collision rule is controlled by a single model parameter which also controls the growth of asymmetry and the rate of calving of the simulated barchans.

 Using this model, we have simulated swarms spanning tens of square kilometres and containing thousands of dunes under both unimodal and bimodal winds in both cases generating swarms which have spatially homogeneous size-distributions, something which is observed for real-world swarms but had not been successfully reproduced in earlier agent-based models.  We also find that varying the wind direction changes the dune asymmetry distribution of swarms in a way that cannot be predicted from the asymmetry growth of isolated dunes subject to the same wind regime.  The range of values for the parameter which controls asymmetry and collisions under which we observe spatially homogeneous swarms also coincides with the range of values for which the width of the asymmetry distribution is similar to real-world swarms.  Furthermore, this range of values also matches with the range for which collisional phase-spaces have been successfully reproduced.  The coincidence that several phenomena are reproduced with the same small range of this parameter suggests that the process which it controls may be fundamental to the behaviour of barchans in nature.