Single‐grain post‐infrared infrared stimulated luminescence (SG‐pIRIR) of feldspar has recently been introduced as a method to date Quaternary deposits. The method is particularly appropriate for fluvial deposits that cannot be dated by more conventional quartz optically stimulated luminescence dating and that are heterogeneously bleached (i.e. where only part of the grains is exposed to sufficient light to remove the full luminescence signal). Besides age estimation, single grain equivalent dose (De) distributions also reflect the variable bleaching degree and origins of grains. Thereby, the SG‐pIRIR signal offers a valuable tool to reconstruct sediment pathways. This study builds upon these ideas and develops a dual aspect that combined river terrace dating and SG‐pIRIR sediment pathway reconstruction for fluvial deposits in terraces and modern river floodplain along the Rangitikei River (RR), New Zealand. We found that the RR last aggrading phase (17.4 ± 1.9 ka to 11.6 ± 1.5 ka) was followed by a first phase of fast incision related to knickpoint retreat followed by a steady incision and widening of the RR canyon. The De distribution of fluvial deposits varies accordingly, due in particular to variable input of bedrock particles with saturated pIRIR signal from landsliding of valley flanks. Our study illustrates that the SG‐pIRIR approach is well suited to date terraces and shows how SG‐pIRIR De distributions provide proxies to reconstruct sediment sources and pathways.

Abstract. The upstream propagation of knickpoints in river longitudinal profiles is commonly assumed to be related to discrete changes in tectonics, climate or base level. However, the recognition that some knickpoints may form autogenically, independent of any external perturbation, may challenge these assumptions. We investigate here the genesis and dynamics of such autogenic knickpoints in laboratory experiments at the drainage basin scale, where landscapes evolved in response to constant rates of base level fall and precipitation. Despite these constant forcings, we observe that knickpoints regularly initiate in rivers at the catchments' outlet throughout the duration of experiments. The upstream knickpoint propagation rate does not decrease monotonically in relationship with the decrease in drainage area, as predicted by stream-power-based models, instead the propagation rate first increases until the mid-part of catchments before decreasing. To investigate the dynamics of the knickpoints, we calculated hydraulic information (water depth, river width, discharge and shear stress) using a hydrodynamic model. We show that knickpoint initiation at the outlet coincides with a fairly abrupt river narrowing entailing an increase in their shear stress. Then, once knickpoints have propagated upward, rivers widen causing a decrease in shear stress and incision rate, and making the river incision less than the base level fall rate. This creates an unstable situation which drives the formation of a new knickpoint. The experiments suggest a new autocyclic model of knickpoint generation controlled by river width dynamics independent of variations in climate or tectonics. This questions an interpretation of landscape records focusing only on climate and tectonic changes without considering autogenic processes.