Predicting sediment and nutrient concentrations from high-frequency water-quality data

Abstract

Water-quality monitoring in rivers often focuses on the concentrations of sediments and nutrients, constituents that can smother biota and cause eutrophication. However, the physical and economic constraints of manual sampling prohibit data collection at the frequency required to adequately capture the variation in concentrations through time. Here, we developed models to predict total suspended solids (TSS) and oxidized nitrogen (NOx) concentrations based on high-frequency time series of turbidity, conductivity and river level data from in situ sensors in rivers flowing into the Great Barrier Reef lagoon. We fit generalized-linear mixed-effects models with continuous first-order autoregressive correlation structures to water-quality data collected by manual sampling at two freshwater sites and one estuarine site and used the fitted models to predict TSS and NOx from the in situ sensor data. These models described the temporal autocorrelation in the data and handled observations collected at irregular frequencies, characteristics typical of water-quality monitoring data. Turbidity proved a useful and generalizable surrogate of TSS, with high predictive ability in the estuarine and fresh water sites. Turbidity, conductivity and river level served as combined surrogates of NOx. However, the relationship between NOx and the covariates was more complex than that between TSS and turbidity, and consequently the ability to predict NOx was lower and less generalizable across sites than for TSS. Furthermore, prediction intervals tended to increase during events, for both TSS and NOx models, highlighting the need to include measures of uncertainty routinely in water-quality reporting. Our study also highlights that surrogate-based models used to predict sediments and nutrients need to better incorporate temporal components if variance estimates are to be unbiased and model inference meaningful. The transferability of models across sites, and potentially regions, will become increasingly important as organizations move to automated sensing for water-quality monitoring throughout catchments.

Recent Advances in Information and Communications Technology (ICT) and Sensor Technology for Monitoring Water Quality

“…Spectrometers and water quality probes containing fluorescent detectors can be used to measure downwelling spectral irradiance, surface-water levels of dissolved nutrients, Chl-a, fluorescence, and turbidity at transects, while a towed body probe can be used in underway sampling; surface light sensors can also be used to monitor profile-based ambient fluctuations. For Chl-a measurement, it should be noted that fluorescent quenching to indicate Chl-a is dependent on light intensity [7,21,25]. In Leigh et al's (2018) study of river water quality during high-flow events, in situ automated water quality sensors that contained NTU and CDT sensors were placed at three study sites.…”

“…In Leigh et al's (2018) study of river water quality during high-flow events, in situ automated water quality sensors that contained NTU and CDT sensors were placed at three study sites. Sensors were placed inside flow cells on monitoring stations on the riverbank sides, allowing water to pump through the flow cell and for pressure-induction sensors to record NTU and electrical conductivity [25].…”

“…In Adamo et al's (2014) study, a compliant seawater probe was used to indicate water eutrophication, but it was unable to identify the composition of the vertical water column, providing only information relative to a fixed depth [19]. Pressure-induction automated sensors require linear interpolation of data to provide quality time-matched observations, and the removal of anomalies can result in periods of missing data [25]. Precision calibration equipment (stable reagents and standards), supporting infrastructure, such as flow systems, and interval frequency consistency are among the major issues that affect in situ monitoring.…”

“…It may not always be economically or physically practical to conduct experiments during high water flow events; a solution to these may be to use low-cost in situ automated sensors that can create multi-parameter time series, which can circumvent the need for manual sampling or laboratory analyses if optimized to catch the full range of water-quality conditions happening in transitional waterways [19,23]. However, temporal autocorrection used in most automated systems is not always efficient or wholly indicative of a water body's changes, especially in high-or low-flow events in transitional waterways [25]. Therefore, it is also preferable to design a water quality monitoring system that is tunable, such as enabling researchers to focus on the different spatial distribution of sensor probes, to assess the edges of a possible high-impact pollution area [19].…”

Enhancing the Prediction Accuracy of Data-Driven Models for Monthly Streamflow in Urmia Lake Basin Based upon the Autoregressive Conditionally Heteroskedastic Time-Series Model

“…The prediction of streamflow is essential in many aspects of water resources management, for example, reservoir operation and allocation. Also, streamflow can directly interfere with floods, droughts, agricultural issues, drinkable water supply, and all water-related systems [3]. Predicting streamflow has been conducted by researches on different scales such as hourly scales [4], daily scales [5], 10 days scales [6][7][8], monthly scales [9], and annual scales [10].…”