Randomness is inherent in nature, and random events play a significant role in shaping the natural world. From genetic mutations to 500-year floods, historical contingency constrains present conditions. Many scientific disciplines incorporate this randomness, known as stochasticity, into numerical modeling to better understand environmental processes. In the Earth sciences, stochastic approaches inform topics ranging from hazard assessment to subsurface hydrology. And now, a new study suggests stochastic models could play a significant role in advancing geomorphology research as well.
Here Davidson and Eaton introduce a new algorithm that models channel geometry under variable streamflow scenarios. The Stochastic Channel Simulator (STOCHASIM) models the interplay between erosion and streamside vegetation under different flow conditions to inform changes in the shape of a stream. The model is intended for use in gravel bed streams.
STOCHASIM differs from other geomorphic models in its treatment of variability. Many geomorphic models rely heavily on so-called regime models, which base stream channel formation on an average flow of water that transports the most sediment over time. In temperate climates, this equates to the bankfull flow that occurs every 1 to 2 years. The underlying equilibrium concept assumes that bankfull conditions most consistently shape the river channel.
Although this approach works well for humid regions—think a coastal, lowland Louisiana—it breaks down in systems that don’t support a well-defined floodplain, like arid ecosystems or small headwater watersheds. In these streams, rare floods can dictate a stream’s shape more than average flow. By drawing from a range of possible flood magnitudes, STOCHASIM more accurately represents conditions found in systems shaped by highly variable flooding.
STOCHASIM generates predictions of channel widths for a focal stream on the basis of different flow regimes, and the authors report that it accurately reproduced erosion rates across several river systems. The new algorithm performed similarly to regime models for rivers with low flow variability but more accurately represented channel widening for streams with a broader range of flow, such as those found in arid environments in Texas, Arizona, and Southern California. The stochastic approach may prove more useful for hazard assessments in regions with high flood variability.
The authors demonstrated the applicability of stochastic models in geomorphology and provided a framework that other researchers can build on. The STOCHASIM model code is available through the authors’ website.