An evaluation of high frequency turbidity as a proxy for riverine total phosphorus concentrations

Emma E Lannergård; José L J Ledesma; Jens Fölster; Martyn N Futter

doi:10.1016/j.scitotenv.2018.09.127

An evaluation of high frequency turbidity as a proxy for riverine total phosphorus concentrations

Sci Total Environ. 2019 Feb 15;651(Pt 1):103-113. doi: 10.1016/j.scitotenv.2018.09.127. Epub 2018 Sep 11.

Authors

Emma E Lannergård¹, José L J Ledesma², Jens Fölster², Martyn N Futter²

Affiliations

¹ Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, P.O. Box 7050, 75007 Uppsala, Sweden. Electronic address: emma.lannergard@slu.se.
² Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, P.O. Box 7050, 75007 Uppsala, Sweden.

PMID: 30227280
DOI: 10.1016/j.scitotenv.2018.09.127

Abstract

Surface water eutrophication resulting from excessive phosphorus (P) inputs is one of today's most challenging environmental issues. Riverine total phosphorus (TP) concentrations have high temporal variability, which complicates flux estimation. We evaluated the usefulness of high frequency in-situ turbidity measurements as a proxy for TP in Sävjaån, a river draining a mixed land use catchment (722 km²) in central Sweden. Turbidity was monitored every 10th-15th minute during 6 consecutive years (2012-2017). Linear regression showed a good relationship between high frequency turbidity and TP (r² = 0.64) and could hence be used for comparison of flux estimation methods. Predictive power of the turbidity-TP relationship was not improved by adding seasons, hydrograph rising/falling limb or high/low stream discharge to the model which argues for a single transfer function relating turbidity and TP. Both TP and turbidity were log-normally distributed. However, flux estimates were sensitive to data transformation; predicted TP concentrations and fluxes based on log-transformed data were biased towards lower concentrations and fluxes compared to non-transformed data. In five of six years grab sample and high frequency estimated TP fluxes were similar (grab sample estimates -10% to +13% P transport compared to high frequency flux estimates). The exception was in 2013, when a 50-year spring flood occurred, and the grab sample estimated flux was 56% larger than that estimated from high frequency data. Thus, the flux comparisons were mostly affected by stream discharge, which underlines the importance of capturing high discharge episodes with, e.g. in situ sensors. While uncertainties regarding the use of turbidity as a proxy for TP remain, it is clear that credible water chemistry data can be obtained with current high frequency sensors. We conclude that high frequency data can be used to better understand catchment response to external pressures and gain insights into water quality that will be missed with grab sampling.

Keywords: Flux estimations; High frequency monitoring; In situ sensor; Proxy relations; Total phosphorus; Turbidity.