Drivers of spatiotemporal variability of river water quality

Abstract We investigated the drivers of spatial and temporal variability of river water quality at different scales using a stochastic modelling approach applied to synthetic river networks comprising hundreds of subcatchments. We simulated daily discharge and solute concentration time series throughout the network, systematically varying hydro-climatic regimes and the spatial configuration of mean solute source concentrations both within and between subcatchments. Locally mobilized discharge and solute loads were hydraulically routed through the network, subject to depth-dependent in-stream processing, to generate water quantity and quality time series for every node in the river network. We show that spatial variability of solute concentrations is predominantly determined by landscape source configuration, which reflect global drivers that remain autocorrelated over long time scales. In contrast, temporal variability was predominantly determined by local drivers that remain correlated to each other only over short (spatial) length scales. We further show that high landscape heterogeneity leads inexorably to persistence of spatial patterns through time, whereas high temporal variability does not necessarily result in synchronized temporal patterns between monitoring locations. Consequently, spatial patterns can be assessed with comparably low effort by conducting temporally sparse, spatially distributed sampling campaigns, especially in landscapes with substantial heterogeneity. These results offer insights on allocating limited resources between spatial and temporal sampling to maximize the information value of water quality monitoring.