Empirical roughness coefficients for moderate floods in an open conduit cave: Fullers stream canyon, Culverson Creek Cave System, West Virginia

Open conduit modeling of cave stream floods can yield useful information about water velocities and shear stresses, which can in turn be used to estimate sediment transport capabilities. All such calculations require roughness coefficients for estimating energy losses and a priori knowledge of either discharge or flow depths to set model boundary conditions. However, the difficulties associated with observing in-cave floods generally preclude measuring discharge; roughness coefficients must be assumed based on channel properties. To overcome these challenges, we monitored stream flow depths in Fullers Cave, Greenbrier County, West Virginia using pressure transducers, and simultaneously measured stage and discharge in a karst window immediately upstream of the cave. Five pressure transducers were deployed opportunistically along a 93-meter-long reach in a 10+ meter high canyon averaging 1.5 to 3 meters wide. Stage-discharge relationships were determined for the karst window using an electromagnetic flow meter for floods with peak discharges of 1.66 m3 s-1 or less. The collected data was used to obtain the empirical Manning’s n roughness values, head losses, and energy gradients. Calculated floodwater velocities are comparable to values obtained from scallops on passage walls. Major energy losses were observed where breakdown partially occludes the passage. At peak flow, Manning n values average 0.053 for reaches typified as cobble-floored canyons, but n was 0.069 in the breakdown reach. Roughness values declined exponentially with increasing discharge, but friction slopes calculated using head losses show more complex relationships with discharge. Notably, n values back calculated using bed gradients differ from those calculated using measured head losses by as little as 12%, so the use of bed gradients in roughness estimations will generally yield reasonable approximations of flow conditions. Fullers Cave experiences significantly larger open conduit floods than we observed, so additional work is needed to estimate roughness coefficients for higher discharges. Our empirical roughness coefficients can be applied to similar cave passages in other caves and contexts, including modeling slot canyon-like channels, and our methods demonstrate a technique for measuring hard to obtain data. The addition of data for open conduit conduits significantly expands the range of environments that can be modeled using empirical data beyond pipe-full caves. Applications include studying flooding, sediment transport, and bedrock erosion process. All of these topics will be addressed in Fullers in the future.