Start with the problem
Without continuous, real-time monitoring, threshold breaches and pollution events are detected after the damage, and decisions rely on sparse manual sampling.
A useful first experiment
Start with an available monitoring dataset and assess one forecast or threshold alert with its intended operator.
Evidence to work towards
These are targets described in the brief, not results already achieved or formal government performance indicators. Teams should validate feasibility and agree a baseline with a willing host.
- Continuous monitoring of 8 water-quality parameters per station at 1–5 minute intervals
- 1–72 hour forecasting with threshold-based early-warning alerts
- A 3–6 month pilot across 3–5 stations reporting forecast accuracy
Before a pilot starts
Identify a real user and decision-maker. Agree access, consent, data handling, supervision, costs and the smallest measurable test. Stop or change direction if the evidence does not support the idea.