Towards prediction of suspended sediment yield from peak discharge in small erodible mountainous catchments (0.45-22 km2) of France, Mexico and Spain

The erosion and transport of fine-grained sediment in small mountainous catchments involve complex processes occurring at different scales. The suspended sediment yields (SSYs) delivered downstream are difficult to accurately measure and estimate because they result from the coupling of all these processes. Using high frequency discharge and suspended sediment data collected in eight small mountainous catchments (0.45-22 km(2)) from four distinct regions, we studied the relationships between event-based SSY and a set of other variables. In almost all the catchments, the event peak discharge (Q(max)) proved to be the best descriptor of SSY, and the relations were approximated by single power laws of the form SSY = alpha Q(max)(beta). The beta exponents ranged between 0.9 and 1.9 across the catchments, while variability in alpha was much higher, with coefficients ranging between 25 and 5039. The broad distribution of alpha was explained by a combination of site-specific physical factors, such as the percentage of degraded areas and hillslope gradient. Further analysis of the factors responsible for data dispersion in each catchment was carried out. Seasonality had a significant influence on variability; but overall, most of the scattering in the SSY-Q(max) regressions was explained by the short-lasting memory effects occurring between successive events (i.e. in-channel temporary storage and remobilization of sediment; antecedent moisture conditions). The predictability of SSY-Q(max) models was also assessed. Simulations of SSY per event and of annual SSY were conducted by using the computed regressions and the measured Q(max). Estimates of SSY per event were very uncertain. In contrast, annual SSY estimates based on the site-specific models were reasonably accurate in all the catchments, with interquartile ranges remaining in the +/- 50% error interval. The prediction quality of SSY-Q(max) relations was partly attributed to the statistical compensation that likely occurred between extreme values over a year; but it also suggests that the complex processes occurring at the event scale were smoothed at the annual scale. This SSY-Q(max) rating appears as a parsimonious predicting tool for roughly estimating SSY in small mountainous catchments. However, in its current form the technique needs further improvement as alpha and beta values need to be better constrained. (C) 2012 Elsevier B.V. All rights reserved.