Hydrological classification by clustering approach of time-integrated samples at the outlet of the Rhône River: Application to Δ14C-POC.

Within the framework of the Rhône Sediment Observatory, monthly time-integrated samples have been collected by Particle Traps in the last decade to monitor particulate contaminants in the Rhône River and its main tributaries. In this watershed with a contrasted hydrology, a clustering approach is used to classify the samples according to the main hydrological events. This approach has been applied to riverine particulate organic radiocarbon signatures (Δ14C-POC) that are strongly affected by the origin of the material and the occurrence of nuclear power plant releases. Suspended Particulate Matter (SPM) samples were collected near the outlet of the Rhône River and analysed for 14C along with particulate organic carbon (POC), chlorophyll a and tritium contents to confirm Δ14C-POC origins. Cluster Analysis, coupled to Principal Component Analysis, was performed based on monthly average water discharges of the Upper Rhône River and the five main tributaries. The classification obtained by fuzzy C-mean logic of the Rhône River hydrology into 5 clusters is similar to that already observed in the literature with Mediterranean/Cevenol flood, oceanic pluvial flood, nival flood, low-water level and baseflow clusters. The contributions of each cluster among the Δ14C-POC values demonstrate the complexity of hydrological classification of time-integrated samples. First, the samples with a unique and significantly dominant cluster are easily explained with negative Δ14C-POC values observed in the flood clusters due to input of 14C-depleted material from soil or rock weathering, and positive values observed in the low-water level and baseflow clusters due to anthropogenic input by nuclear industry. Second, samples that present a homogeneous mixture between several clusters demonstrate the occurrence of different hydrological events during the sampling periods. This tool appears as a solution to estimate the contribution of each hydrological event in time-integrated samples.