Abstract
Abstract. Stream water quality is highly variable both across space and time. Water quality monitoring programmes have collected a large amount of data that provide a good basis for investigating the key drivers of spatial and temporal variability. Event-based water quality monitoring data in the Great Barrier Reef catchments in northern Australia provide an opportunity to further our understanding of water quality dynamics in subtropical and tropical regions. This study investigated nine water quality constituents, including sediments, nutrients and salinity, with the aim of (1) identifying the influential environmental drivers of temporal variation in flow event concentrations and (2) developing a modelling framework to predict the temporal variation in water quality at multiple sites simultaneously. This study used a hierarchical Bayesian model averaging framework to explore the relationship between event concentration and catchment-scale environmental variables (e.g. runoff, rainfall and groundcover conditions). Key factors affecting the temporal changes in water quality varied among constituent concentrations and between catchments. Catchment rainfall and runoff affected in-stream particulate constituents, while catchment wetness and vegetation cover had more impact on dissolved nutrient concentration and salinity. In addition, in large dry catchments, antecedent catchment soil moisture and vegetation had a large influence on dissolved nutrients, which highlights the important effect of catchment hydrological connectivity on pollutant mobilisation and delivery.
Highlights
In-stream water quality plays a vital role in influencing the health of freshwater ecosystems (Pérez-Gutiérrez et al, 2017), which in turn underpins environmental, social and economic sustainability (McGrane, 2016; Ustaoglu et al, 2020)
We address the limitations in statistical water quality models by using the following: (1) Bayesian hierarchical modelling was used to investigate water quality temporal variation, which allowed the prediction of water quality in multiple catchments and, simultaneously, quantified parameter uncertainty (Gelman et al, 2013; Rode et al, 2010; Webb and King, 2009); and (2) Bayesian model averaging (BMA) approaches were used to identify the relative importance of the different environmental factors and provide multi-model weighted predictions, which have been shown to better quan
The three key measures that were used to quantify the effect of individual predictors are (1) estimates of posterior inclusion probability (PIP), which quantifies relative importance of individual predictors; (2) posterior model probability (PMP), which estimates differences in plausible model structures; and (3) posterior distributions of coefficients for the key drivers, which measures direction and magnitude of the effect of key predictors on water quality temporal variability
Summary
In-stream water quality plays a vital role in influencing the health of freshwater ecosystems (Pérez-Gutiérrez et al, 2017), which in turn underpins environmental, social and economic sustainability (McGrane, 2016; Ustaoglu et al, 2020). Pollution derived from agricultural land and urban development has led to water quality degradation in streams and lakes in many regions of the world (Ren et al, 2003) Among these water quality issues, coastal regions with high agricultural production have been delivering large amounts of pollutants to the ocean, where marine ecosystems are vulnerable to the evaluated levels of nutrients and sediments (Gorman et al, 2009). It is estimated that 60 % of coastal rivers in the USA have been moderately to severely degraded (Gorman et al, 2009; Howarth et al, 2002) To protect both freshwater and marine ecosystems, better management of catchment-derived pollutants is needed. Spatial differences in water quality concentration can vary markedly due, in part, to heterogeneity of natural landscapes in catchments
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
