A turbidity-based method to continuously monitor sediment, carbon and nitrogen flows in mountainous watersheds

- Turbidity was a good predictor of suspended sediment in a mountainous catchment.
- The same model performed better for particulate organic carbon.
- Serial correlation was the most important factor to obtain good predictions.

SummaryThe aim of this study was to develop a method to continuously monitor sediment, carbon and nitrogen concentrations in streams using turbidity sensors. Field experiments were conducted in an irrigated and intensely cultivated watershed in Northwest Vietnam. Turbidity, discharge and rainfall were monitored during two successive rainy seasons from 2010 to 2011, and manual water samples were collected using a storm-based approach. Samples were analyzed for concentrations of suspended sediment (SSC), particulate organic carbon (POC) and particulate nitrogen (PN). A linear mixed model was developed to account for serial correlation, with turbidity, discharge and rainfall as predictor variables. Turbidity was the most important predictor variable in all models. Fivefold cross-validation showed best model performance for POC with a Pearson's correlation coefficient of 0.91, while predictions for SSC and PN achieved a satisfying correlation of 0.86 and 0.87, respectively. Laboratory testing of the turbidity sensors showed that the turbidity signal is sensitive to differences in organic matter content, and has the smallest variance for fine textures, both of which are correlated to POC and thus supporting the higher predictive accuracy for this variable. The developed methodology is widely applicable and can be used to simultaneously obtain reliable, cost-effective and continuous estimates of SSC, POC and PN with a single sensor.