Potential for using high frequency turbidity as a proxy for total phosphorus in Sävjaån

Abstract: Transport of particles carrying nutrients and contaminants from land to sea is a challenge to monitor due to the high temporal variability in concentrations. Phosphorus is a particle associated nutrient, largely affecting the state of waters due to its effects on eutrophication. When using traditional monitoring methods, there is a prevailing risk of misjudging the state of water and transport of phosphorus due to lack of data. Therefore, this study aimed to evaluate the potential use of continuous high frequency turbidity as a proxy for total phosphorus concentration. An in situ sensor monitoring turbidity every 10th minute was deployed from 2012-2015 in Sävjaån, a river draining a mixed land use catchment in central Sweden. The sub-catchment is an agricultural area dominated by clay soil close to Uppsala in Sweden. The results from measuring high frequency turbidity with the sensor were compared to traditional monthly grab sampling. A significant information loss could be observed when performing grab sampling, the turbidity from the sensor varied more and showed higher maximum values (9-15 times). The correlation between the different parameters was evaluated by linear regression. The results show very high correlation between turbidity and total phosphorus (r2= 0.79) and high correlation between turbidity and total suspended solids (r2= 0.67). The relationships seemed to be affected by calibration of the sensor, spatial variation and the proportion of phosphate and total phosphorus in Sävjaån. The phosphorus load was calculated from the high frequency data and compared to linear interpolation and piecewise constant interpolation of grab samples. Loads calculated from high frequency data was during two years (2012 and 2015) 31% and 17% larger than when using linear interpolation. However, the timing and the instant flow at the time of grab sampling had large impact on the estimated load when using linear interpolation (2013 and 2014). It could be concluded that most P was transported when high discharge and high concentrations coincided. When comparing the calculations of ecological quality ratios from grab sampling and high frequency data the difference was modest.