Flood variability in the rock record? Disequilibrium bedform preservation in ancient fluvial stratigraphy

The extent to which the stratigraphic archive preferentially preserves the record of large events such as floods remains contentious. While qualitative approaches exist to address this problem, the way in which disequilibrium morphodynamics is preserved quantitatively in fluvial strata has only recently begun to be investigated. While existing process–product relations for bedform preservation often assume that fluvial cross strata reflect steady-state formative conditions, i.e., bedform evolution equilibrated with the prevailing flow, theory indicates that bedforms may be preferentially preserved in unsteady, or disequilibrium, conditions. Here we test this concept using field data collected from fluvial stratigraphy in the Upper Cretaceous of Utah, USA (Ferron Sandstone and Blackhawk & Castlegate Formations) and the Upper Carboniferous of South Wales, UK (Pennant Formation).

For the US field site, we systematically measured preserved cross-set heights (n = 417) for all three formations, and we observed unanimously low coefficients of variation (CV) across individual co-sets and at a population level (CV = 0.25–0.5). These values are inconsistent with bedform preservation in steady-state conditions (CV = 0.88±0.3), and instead point to bedform preservation in disequilibrium conditions. Similarly in the UK field site, the CV of cross-set height distributions average 0.4, significantly less than the theoretical value for steady-state deposition. In both cases these low values are ubiquitous throughout the stratigraphy studied.

Two independent hypotheses can explain our field observations: (1) short flood recessions, relative to bedform turnover timescale, in flashy flood hydrographs (flood hypothesis); (2) dune evolution in the presence of barforms (hierarchy hypothesis). However, in the Pennant Formation qualitative facies-based evidence such as storm beds containing large woody debris independently demonstrate that flood events clearly did occur. We therefore used our constraints on cross-set size and grain-size to calculate dune height, wavelength and unit bedload flux, in order to quantify bedform turnover timescale. Under the flood hypothesis, our field data are consistent with enhanced bedform preservation driven by flashy flood hydrographs with a duration of a few hours to a few days for both data sets. These durations are consistent with perennial rivers subject to torrential rains and storms. Under the hierarchy hypothesis, our field results would suggest bedform climb angles of 10⁻² to 10⁻¹, and would require rapid bar migration relative to dune migration. We use architectural and palaeohydrological techniques to estimate the size and discharge of the floods that may have formed these deposits and we evaluate the extent to which it is now possible to extract information on flood variability from ancient sedimentary rocks.