Modular Modeling of Suspended Sediment Concentrations in a Sandstone Headwater Catchment (luxembourg)

Due to the multitude of adverse effects associated with the transport of suspended sediments in surface waters, it is essential to understand the underlying physical controls that govern their release from catchments and the subsequent transport via watercourses. Previous research has shown that the rate at which suspended sediments are transported in watercourses depends primarily on discharge as the first-order control, but additional factors are though to affect concentrations of suspended sediments as well. Among these, antecedent hydrological and meteorological conditions (e.g. rainfall depth and intensity, discharge prior to an runoff event and the duration of runoff events) may exert a significant transport controlling mechanism. Univariate models using discharge and suspended sediment concentrations (Q-SSC rating curves) often produce large scatter and nonlinearity, because many of hydrological and biotic processes affecting the dynamics of sediments are non-linear and exhibit various dynamic thresholds. The simulation of such highly non-linear processes with dynamic thresholds is therefore an elusive task requiring consideration of several interrelated controlling variables. The aim of this study was to identify major hydrological and meteorological controls determining the dynamics of SSC during storm-runoff events and the magnitude of SSC in a headwater sandstone-dominated catchment in Luxembourg. A parsimonious data-driven model (M5’ modular trees) was used to simulate SSC concentrations in response to the identified controlling variables. Antecedent hydro-meteorological variables (antecedent precipitation depth, and antecedent runoff volume) were determined as the major controls explaining sediment depletion effect and a gradual decline of SSC as a runoff event progresses. We compared the modeling results obtained by M5’ trees to conventional power-law rating curve. The M5’ models outperformed the rating-curve by being successful in describing both the shape and magnitude of sedigraphs. Therefore, we propose that incorporating antecedent hydro-meteorological data into modeling may substantially enhance the accuracy of sediment yield estimates.