Exploring new approaches to understanding channel width and erosion rates in bedrock rivers, Puerto Rico, USA

Abstract

Earth system dynamics produce constant adjustments to sea level, tectonics, and climate. Bedrock rivers communicate these changes throughout mountains by driving landscape and erosional responses that facilitate topographic change. It follows that an improved understanding of bedrock rivers can help us better model and reconstruct the interplay of changes to base level, uplift, and climate from landscapes. Although bedrock channel width plays a first-order role in river stream power and stream power-based landscape evolution models, because of the physical challenges associated with acquiring these data, channel width is often estimated and introduces uncertainty. In addition, the lack of bedrock channel width data has limited our understanding of what factors control channel width. In this dissertation (Chapter 2), I leverage high-resolution topographic data, Mean Annual Precipitation information, and use the HEC-RAS river modeling software to remotely derive bedrock channel width at desired flow scenarios. The accuracy of modeling results is verified for rivers in Puerto Rico using USGS gauging station field measurements, as well as my own channel width field measurements associated with 1-year recurrence interval discharges. As a next step, (Chapter 3) I implement the bedrock width modeling method derived in Chapter 2 to obtain >4,000 channel width measurements from reaches across Puerto Rico. I then compare these bedrock river width values to various factors (e.g. rock type and rock strength, drainage area, Ecozone, and grain size) that have been identified in the literature to scale with or influence channel width. My analyses indicate that, in Puerto Rico, rock type is a dominant control of bedrock channel width in small (≤6-10 km2) drainage areas. Contrary to patterns of rock strength and bedrock width documented in the literature (e.g. Montgomery and Gran, 2001), I find that width doesn't appear to correlate with proxies for bedrock channel strength. Strong granodiorites have the widest low-order channels and the strong volcaniclastics and weak serpentinites have comparably narrow low-order channels. Analysis of limited grain size measurements shows a discernable difference in the coarse grain size distribution between the three rock types, with the volcaniclastic and serpentinite draining rivers having coarser sediment than granodiorite draining streams. These findings suggest that bedrock channel width may be influenced by unmeasured lithological parameters that impact the size of grains delivered to river channels from adjacent hillslopes (i.e. rock fracture density and spacing, as well as weathering). Lastly, (Chapter 4) I spatially analyze in-situ cosmogenic nuclide (10Be in quartz and 36Cl in magnetite) concentrations and find that bedrock erosion rates are higher in the central part of Puerto Rico than toward the east. Analysis of erosion rates compared to other parameters reveals that channel steepness, rather than precipitation or rock type, is positively associated with erosion rates. I further apply these erosion rate data to test the accuracy of four incision models of varying complexity. Model comparisons reveal that drainage area is a better predictor of incision rates in Puerto Rico than a precipitation-weighted drainage area parameter. In addition, whereas an increase in model complexity slightly improves model performance, the model only explains ~35% of the variability in erosion rates. It follows that current incision models are still missing many controlling factors of river incision rates in Puerto Rico.