Sensor networks for river system monitoring
Catchment management aimed at improving freshwater quality and reducing carbon emission is complicated due to multiple transport pathways that convey water and a wide range of contaminants into rivers (Khamis et al., 2018). These include point sources such as industrial and municipal wastes, and non-point contributions such as agriculture. Identifying hotspot areas (both sources and impacts) is a critical first step in developing adequate intervention measures to improve water quality; however, monitoring is needed to evaluate the effects of these intervention efforts, protect water quality, and meet regulations (Lofton et al., 2023). Sensor networks provide the potential to meet these aims, but operational water quality monitoring programs globally are commonly based on fixed sampling points with periodic manual collection of ”grab” samples and subsequent laboratory analysis for targeted parameters. Due to limitations of personnel, equipment, and access this type of sampling can provide good spatial snapshots of river conditions at the time of sampling (Meyer et al., 2019) but is difficult to implement across entire catchments (Xing et al., 2013). These sampling approaches also largely miss sporadic extreme events, such as contaminant releases or stormflows (Charriau et al., 2016). In response to deficiencies in capturing event-based changes in river ecosystem properties, and with the emergence of more reliable sensor technology, high-frequency monitoring using field deployable sensors and actuators is increasing (Blaen et al., 2016, Bieroza et al., 2023). Autonomous and remotely operated robotic surface vehicles with on-board sensors have increased the achievable spatial resolution of field-deployed water quality sensors(Lee et al., 2023), and show great potential to improve detection of, and response to, short-term changes in river environments (Powers et al., 2018). For example, localisation of a pollution hotspot could trigger reactive behaviours, such as increasing the resolution of data collection or tracking concentration gradients.
In-situ automated systems with multiple sensors that measure at high-frequency (typically 15-60 min resolution but can vary depending on the application) can be used to deliver near real-time data (Meyer et al., 2019, Singh et al., 2022). Various sensors can be deployed to quantify carbon cycling or to supply information on physicochemical drivers (Table 1). However, to advance catchment scale carbon management, networks of these automated systems (i.e. sensor nodes (Figure 3)) are needed to pinpoint areas, such as those with high emissions, and to track event propagation through river basins (Zia et al., 2013). Further potential for enhancing the dimensionality of environmental data is emerging from the development of autonomous robotic platforms to deploy sensors in parts of river systems that are difficult to access. The integration of these approaches and datasets presents a challenge, but these networks offer significant potential for advances in real-time understanding and mitigation of risk for river users, managers, decision-makers, and regulators (Jankowski et al., 2021, O’Grady et al., 2021).