RiverGuard: Proactive Sensing and Rapid Response for Urban Water Quality

A practical early-warning and response system that brings together people, using fixed sensors, open data streams, drones, and aquatic robots to give a live, detailed picture of river health.

Rivers are living systems that can change fast. In cities like London, short, intense pollution events can happen when heavy rain overwhelms sewers, when industrial spills occur, or when dirty runoff washes into the water. These incidents often last only a few hours and can be very local, affecting one stretch of river but not the next. Today’s monitoring methods struggle to keep up. A few fixed stations take readings in set locations, so they can miss a moving “plume” of pollution between them. Manual water sampling takes time and lab work. Standard monitoring on weekly or monthly timescales is likely to ‘miss’ high-impact short-duration pollution events. Community observations can supplement these data, but it is hard to combine with official data. The result is that problems are spotted late, sources are unclear, and decision-makers lack the timely, trusted information they need.

We propose a practical early-warning and response system that brings together people, using fixed sensors, open data streams, drones, and aquatic robots to give a live, detailed picture of river health. Think of it as a layered safety net:
1.    Open meteorological & hydrological data streams can predict when pollution events are more likely to occur.

2.    Fixed sensors provide continuous background monitoring

• Citizen reports widen awareness and help flag issues as they happen. 
• Aerial drones can quickly scan large areas from above to map discoloured water or other surface signs. 
• Underwater and surface robots take measurements beneath the surface, tracing the pollution upstream to find its source.

These tools work as a team. For example, if a sensor detects unusual water quality, it can trigger a drone flight to outline the affected area. A small robot can then be sent into the water to measure at different depths and “follow the trail” to where the problem began. This joined-up approach gives better coverage at the very moment it’s needed and cuts the time between detection and action. It also works in all kinds of weather and on smaller, more complex rivers where satellites or occasional sampling fall short.
Just as important, the system turns complex data into clear, usable guidance. Rather than overwhelming people with raw numbers, it produces simple alerts and severity levels (for instance, “investigate now,” “close intake,” or “issue public advisory”) along with maps that show where the problem is and how it’s moving. Openness and trust are built in: we use standard data formats and transparent checks so that results can be shared, scrutinised, and reused by river managers, regulators, and the public.

This project is inherently collaborative. Our team brings together roboticists, ecologists, hydrologists, and specialists in visualisation and 3D modelling of rivers. We will co-design the system with river managers so it fits real-world workflows from day one. By continuously linking models with measurements, we can focus our efforts where they will reduce uncertainty the most—saving time and money while improving accuracy.

Our long-term vision is an open, affordable platform that anyone—from local community groups to utilities and national agencies—can adopt. Low-cost kits will feed data into an interoperable software layer that detects anomalies in real time and forecasts where pollution will go next. Over time, this will grow into a “digital twin” of river networks: a live model informed by constant measurements, where human teams set goals and rules, robots provide dense evidence, and the system helps everyone act quickly and effectively.

By making the technology accessible and the data transparent, we aim to protect freshwater ecosystems, strengthen public trust, and help communities become more resilient in a changing climate.