Bedload dynamics and associated snowmelt influence in mountainous and semiarid alluvial rivers

Abstract

The influence of snowmelt in semiarid alluvial rivers is determined by the river bed configuration and erosion processes. Moreover, the large amount of available sediment in the floodplains of these rivers limits traditional bedload sampling methodologies and suggests the use of large control volumes and long-term studies to understand erosive dynamics. In this study, bedload monitoring in a large control volume (0.2 hm3) from2004 to 2010 was used to study erosive processes of the Guadalfeo River, a semiarid alluvial river in southeastern Spain with high mountain influences and a hydrological regime conditioned by snow dynamics. The methodological approach to testing the performance of different transport models included characterization of the forcing agents (precipitation and snowmelt events), hydraulic configuration of the river, determination of the river bed material (surface and substrate), and one-off measurements for particular events. The results indicate a sediment volume–effective runoff relationship that is consistent with the existence of steep channels with a flashy runoff regime, although the existence of an armored layer derived from snowmelt events greatly controls the transport dynamics, with equivalent fractions of cobble–gravel and sand in the measured bedload. This apparent predominance of near-equal mobility was confirmed by the calibrated hiding function exponent (x≈ 0.94–0.97). The thresholds for entrainment and transport efficiency are similar to those observed for mountain rivers that exhibit torrential flow, although notable differences can be observed in the dynamics of the processes. The strong performance of the original Meyer-Peter and Müller model (with c=8) is shown to estimate the total transported volume associated with intense events when the variability of d50 is included in the analysis. However, this model does not sufficiently capture transport because of moderate events (rain or snowmelt) and thus underestimates bedload on medium and long timescales. The calibrated model of Parker–Klingeman adequately represents erosive dynamics, with τr50⁎ values between 0.03 and 0.041 for moderate and intense events, respectively, and a flow threshold of 14.2m3 s−1 that separates these two situations. Nevertheless, these models based on hiding functions present greater dispersion if the variability of diameters related to cobbles and pebbles is considered. For such cases, the Wilcock and Crowe model performed better for both intense and moderate events. The latter is important in the study area, where snowmelt pulses occasionally exceed the critical flow and break the surface layer with significant sediment contributions. These processes differentiate the transport dynamics from other studies and are distinctive characteristics that are more closely related to mountainous influences in semiarid environments where pulses of intense snowmelt are common. Despite its limitations, the proposed methodology, based on a large check-dam, has proven to be adequate and representative for assessment of erosion processes.