Underwater sediment density flows, including turbidity currents, are capable of transporting vast amounts of sediment, nutrients and pollutants to the deep-sea. These flows can be powerful, causing damage to seafloor infrastructure. Understanding how the flow velocity and magnitude develops over distance is thus important for risk assessments, as well as determining sediment fluxes. However, as few direct measurements are available, these flows remain rather poorly understood. This thesis aims to use three direct monitoring datasets from three different oceanographic settings worldwide that have captured turbidity currents in unusual detail, allowing for unique analysis of their flow evolution.

Detailed measurements of turbidity currents in Monterey Canyon, offshore California, show that their evolution depends on the initial velocity and the availability of an easily erodible substrate. Turbidity currents exceeding a velocity threshold can plateau with near-uniform velocities, and thus run out over greater distances. A new model is proposed for how these near-uniform velocities are obtained. In the Var Canyon-River system, France, nearshore measurements are used to analyse turbidity current velocity structures, and how these develop over distance. Turbidity currents are shown to self-organise over short distances by amalgamation of velocity peaks, which is partly controlled by erodible substrate availability. This efficient self-organisation occurs within 10 km, after which the original trigger is indiscernible. This has important implications for interpreting turbidity current deposits. Bute Inlet, British Columbia, is one of the most complete studies, where source-to-sink direct measurements are combined with sediment cores. These data allow for a unique analysis of turbidity current activity over space and time. The current-day channelized system is highly active with yearly events, although these events are low magnitude. In contrast, distally the system shows high magnitude events occurring on centennial time scales. These data suggest that infrequent mechanisms control large-scale events currently not observed directly. This thesis provides a detailed analysis of turbidity current development over distance, essential for determination of sediment fluxes and hazard assessment.