Numerical Model Predictions for Autogenic Fluvial Terrace Formation and Comparison to Natural River Valleys

Abstract:

Terraces cut into bedrock (strath) and sediment (fill-cut) offer key constraints on river evolution over millennial timescales, and are often interpreted to form during phases of increased river vertical incision driven by changes in climate or tectonics. Yet all actively meandering channels evolve their shapes through spatial and temporal changes in lateral erosion rates. Therefore, the sparsest requirement for a meandering river to generate terraces is that the intrinsically unsteady lateral erosion rate be coupled with relief generation by vertical incision, which need not be unsteady. In principle, this basic mechanism for terrace formation by meandering rivers should be possible in all fluvial environments, including for valleys with strath or fill-cut terraces, and may overprint signals from external drivers. We have used a numerical model of a vertically incising, meandering river to identify the age and geometric properties of autogenic terraces. Simulations indicate that autogenic terraces form with a recurrence timescale, set by the rate of relief generation, which may overlap with timescales for climate change. The autogenic terraces also have predictable geometries that can include slope proportional to the ratio of vertical incision rate to lateral erosion rate, pairing, and continuous along-valley extent. We compare these simulation results to data for terrace age and geometry from several well studied natural river valleys that span a wide range in terrace sizes and geometries, rock types, tectonic settings, incision rates, and hypothesized formation mechanisms. In cases, terrace age and geometric properties are consistent with formation by meandering with constant vertical incision rates. These similarities suggest that efforts to distinguish terraces that record signals from climatic and tectonic drivers are best focused on environments where terrace ages and geometries are far different than would be predicted by a constant vertical incision model.