Efficiency and representativeness of suspended particulate matter time-integrative sampling in rivers using particle traps

International audience

English. Monitoring hydrophobic organic contaminants and trace metals in surface freshwaters requires measuring contaminant concentrations in the particulate fraction of the water column. Over the past two decades, particle traps (PTs) have been developed to sample suspended particulate matter (SPM) as a cost-efficient, easy-to-operate and time-integrative tool. Whereas the representativeness of SPM samples collected by continuous flow centrifugation was investigated and validated as early as 25 years ago, the representativeness of the particles collected with PTs is still questionable and not fully understood. Since 2009, the Rhône Sediment Observatory (OSR) Program has relied on PTs designed according to the German PT described by Schulze et al. (2007) [1] for the monitoring of particulate contaminants throughout the Rhône River and the outlets of its main tributaries from Lake Geneva to the Mediterranean Sea. The aim of this study was to evaluate the representativeness of SPM characteristics and associated contaminants in samples collected in a large river using this type of PT.To this end, samples collected in the Rhône River using PTs were physically (grain size distribution) and chemically (particulate organic carbon, trace metals, organic compounds) compared with concurrent samples collected using continuous flow centrifugation. Moreover, in order to understand the mechanisms and trapping efficiency of the PTs, experiments were conducted under controlled conditions (direct feeding of the PT and deployment in a laboratory flume). The in situ and laboratory results clearly demonstrate that the grain size distribution was coarser in PTs than in the river. However, PTs captured all classes of particles with a shift, highly dependent on water discharge, from very fine silts towards coarser silts. The particle trapping efficiency could be modeled as a function of particle diameter and approach flow velocity. This relation well predicted the in situ observations and showed that trapping efficiency is independent of SPM concentration. Based on the widely shared assumption that contaminants have more affinity for smaller particles, the influence of particle size bias on the particulate concentrations of different contaminants (polychlorinated biphenyls, mercury and trace metals) was estimated and documented according to hydrological conditions (low water, base flow, flood). Finally, our results suggest that this type of PT is a valuable technique for assessing reliable spatial and temporal trends of particulate contaminants concentrations in large streams and rivers.