During this project a complete warning system has been built to warn about bad bathing water quality in Aarhus Harbour using real-time sensor data. The current methods used to measure the bacteria concentration in water is insufficient to warn about bad bathing water quality, due to the long incubation period of the samples of minimum 24 hours. This, combined with the generally bad water quality in the harbour, leads to the need of a warning system.

With a statistical approach the suitable indicators for bathing water quality have been investigated. The conclusion is that the best indicator is the electrical conductivity of the water. This is due to the fact that the main source of pollution to the harbour is the river, Aarhus Å, which has its delta in the harbour. Aarhus Å acts as a recipient of a large amount of CSO's and outlets from wastewater treatment plants. The electrical conductivity therefore acts as a tracer of how much polluted fresh water is present at the place of measurement. The correlation between the bacteria concentration and the electrical conductivity has been modelled with an exponential regression with good results. The performance has been improved by including rain depth and intensity to a new model based on machine learning. Besides the improved performance of the model this results in a significant reduction in the number of closing days.

To complete the warning system a measuring station to monitor the electrical conductivity has been developed. The main task of the new measuring station is to avoid the unwanted growth of algae on the sensor. This problem is solved by placing the sensor in a chamber which can be emptied from water with compressed air.