Characterizing intra and inter annual variability of storm events based on very high frequency monitoring of hydrological and chemical variables: what can we learn about hot spots and hot moments from continuous hydro-chemical sensors ?

Abstract:

Storm events are hot moments of emission for several dissolved and particulate chemical species at major stake for water quality (e.g. dissolved organic carbon DOC, suspended sediments, phosphorus, NH4). During such events, the solutes or particles are exported from heterogeneous sources through various pathways to stream or are possibly stored in retention hot spots temporary. This leads to specific integrated signals at the outlet at the scale of storm events. The dynamics of such events are also very short especially in headwater catchments where their total duration ranges over 10h to 3 days, with very quick variations in stream flow and concentrations at the outlet occurring in a few hours. Thus for investigating properly event processes, high frequency monitoring of flow and water quality is required. We analysed 103 storm events in a 5 km² agricultural headwater catchment, part of the AgrHys Observatory, on the basis of a 3-year-long data set which combined meterological (Rainfall), hydrological (flow and piezometry), and water quality (turbidity, conductivity, DOC and NO3 concentrations) data recorded at very high frequencies (from 1 to 20 min) thanks to dedicated sensors. We described the storm events using simple (1 variable) and combined (2 variables) descriptors for characterizing level and dynamics of flow (Q), groundwater levels, and concentrations (C) but also the C-Q relationships. Three intra annual periods have been previously defined for base flow dynamic according to shallow groundwater table variations so that they correspond to different connectivity status in the catchment. The seasonal and inter-annual variability of the storm events have been analysed using the descriptors and based on these predefined periods. Principal component analysis based on storm chemical descriptors led to discriminate these three seasons while storm hydrological descriptors are less variable between them. Finally we used a clustering method to build a typology of storm events based on the descriptors.