• bedrock channel width;
  • orographic rainfall;
  • Marsyandi River

[1] Pronounced rainfall gradients combined with spatially uniform exhumation of rocks at Quaternary timescales and uniform rock strength make the upper Marsyandi River valley in central Nepal a useful natural laboratory in which to explore variations in bedrock channel width. We focus on small catchments (0.6–12.4 km2) along a more than tenfold gradient in monsoon rainfall. Rainfall data are gathered from a dense weather network and calibrated satellite observations, the pattern of Quaternary exhumation is inferred from apatite fission track cooling ages, and rock compressive strength is measured in the field. Bedrock channel widths, surveyed at high scour indicators, scale as a power law function of discharge (w αQw0.38±0.09) that is estimated by combining rainfall data with 90-m digital topography. The results suggest that power law width scaling models apply (1) to regions with pronounced rainfall gradients, (2) to tributary catchments distributed across a climatically diverse region, and (3) to large, rapidly denuding orogens. An analysis of rainfall data indicates that the regional gradient of rainfall during storms that drive erosive discharge events is about half as large as the gradient of seasonal rainfall across the same area. Finally, numerical models in which the maximum rainfall is displaced significantly downstream from the headwaters predict a midcatchment zone of relatively rapid decreases in channel gradient and increases in channel concavity that are driven by locally enhanced discharge. Because differential rock uplift can produce analogous changes in gradients, the influence of rainfall gradients should be assessed before tectonic inferences are drawn.